A Multi-Scale Digital Rock Nuclear Magnetic Resonance Response Spectrum Simulation Method

Abstract

This invention discloses a multi-scale digital rock nuclear magnetic resonance (NMR) response spectrum simulation method, relating to the field of oil and gas resource geological exploration and development evaluation technology. The method includes: large-scale and small-scale multi-resolution CT scanning experiments of rocks; small-scale digital rock NMR response spectrum simulation; construction of a small-scale digital rock NMR response spectrum library; and multi-spectral synthesis of multi-scale digital rock NMR response spectra. This method achieves multi-scale digital rock NMR response spectrum simulation, effectively integrating pore structure characteristics and NMR response spectrum characteristics at different scales. It avoids the problem of unrepresentativeness in single-pore-scale NMR response spectrum simulation, overcomes the limitation of traditional digital core NMR response spectrum simulation in characterizing multi-scale pore structures, improves the accuracy of rock NMR response spectrum simulation results, and easily reveals the rock NMR response mechanism under multi-scale and multi-parameter configuration conditions, laying the foundation for the refined evaluation of the physical properties of oil and gas reservoir rocks.

Description

Technical Field

[0001] This invention relates to the field of oil and gas resource geological exploration and development evaluation technology, and in particular to a multi-scale digital rock nuclear magnetic resonance response spectrum simulation method. Background Technology

[0002] With significant breakthroughs in unconventional oil and gas exploration and development in several basins across my country, unconventional oil and gas has become a replacement resource for increasing domestic oil and gas reserves and production. However, unconventional oil and gas reservoirs generally exhibit strong heterogeneity, showing significant vertical variations in lithology and physical properties across different scales of core samples. In particular, the development of nanopores and micron-sized fractures across different scales makes it difficult to characterize unconventional oil and gas reservoir parameters with high precision at multiple scales, thus limiting the economical and efficient exploration and development of unconventional oil and gas.

[0003] Nuclear magnetic resonance (NMR) is a high-precision technique for evaluating hydrocarbon reservoir parameters. Although NMR experiments can measure the NMR response spectrum of rocks, the interpretation of the internal structural characteristics of rocks from the NMR response spectrum is ambiguous, requiring analysis and understanding of the NMR response mechanism. Because preparing rock samples with different pore structures, mineral compositions, wettability, and other parameters is time-consuming and complex, digital core technology is commonly used. This involves generating different rock sample models using computers, and then employing numerical simulation methods to model the NMR response spectrum, thereby studying the NMR response mechanism. However, traditional digital core technology can only characterize the high-precision, small-scale digital rock NMR response spectrum. Results on multi-scale digital rock NMR response spectrum simulations have not yet been published, resulting in limited effectiveness of NMR response mechanism research results. The main limitation lies in the low resolution and poor pore connectivity of large-scale digital rock models, making effective NMR response spectrum simulation impossible.

[0004] Therefore, there is an urgent need to study a method that fully combines the advantages of reflecting both low-resolution large-scale rocks and high-resolution small-scale rocks, to construct a small-scale digital rock NMR response spectrum library, and to realize multi-scale digital rock NMR response spectrum simulation based on the multi-spectral synthesis method. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention discloses a multi-scale digital rock nuclear magnetic resonance response spectrum simulation method, which can effectively integrate pore structure characteristics and nuclear magnetic resonance response spectrum characteristics at different scales. This avoids the problem that single-pore-scale nuclear magnetic resonance response spectrum simulation is not representative, laying the foundation for the refined evaluation of the physical properties of oil and gas reservoir rocks.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A multi-scale digital rock nuclear magnetic resonance response spectrum simulation method includes the following steps:

[0008] Step S1. Large-scale and small-scale multi-resolution CT scanning experiments on rocks;

[0009] Step S2. Small-scale digital rock nuclear magnetic resonance response spectrum simulation;

[0010] Step S3. Construction of a small-scale digital rock nuclear magnetic resonance response spectrum library;

[0011] Step S4. Multi-spectral synthesis of multi-scale digital rock nuclear magnetic resonance response spectra.

[0012] Optionally, step S1, the steps of the large-scale and small-scale rock multi-resolution CT scanning experiment, includes:

[0013] Step S11. Cut the large-scale rock to be studied into regular cubes or cylinders and wash it with oil.

[0014] Step S12. Perform a low-resolution CT scan on the regular cube or cylinder obtained in step S11;

[0015] Step S13. Based on the CT scan images obtained in step S12, select multiple regions with different structural features for drilling to obtain small-scale sub-rocks;

[0016] Step S14. Perform high-resolution CT scans on the small-scale sub-rocks obtained in step S13.

[0017] Optionally, step S2, the step of small-scale digital rock nuclear magnetic resonance response spectrum simulation, includes:

[0018] Step S21. Based on the CT scan images obtained in step S14, a large number of small-scale digital rocks with the same resolution are generated using a style-based generative adversarial network method.

[0019] Step S22. Combining the random walk method and the finite element method, simulate the nuclear magnetic resonance response spectrum of each small-scale digital rock generated in step S21.

[0020] Optionally, step S3, the step of constructing a small-scale digital rock nuclear magnetic resonance response spectrum library, includes:

[0021] Step S31. Calculate the pore structure of each small-scale digital rock generated in step S21 using the maximum spherical pore network model method;

[0022] Step S32. Establish a porosity-porosity-median pore radius grid with porosity as the abscissa and the median pore radius as the ordinate. Each grid cell represents the nuclear magnetic resonance response spectrum of the corresponding small-scale digital rock generated in step S22 under the conditions of corresponding porosity and median pore radius.

[0023] Step S33. If a certain grid cell in step S32 does not have a corresponding nuclear magnetic resonance response spectrum of small-scale digital rock, then the average value of the nuclear magnetic resonance response spectra of the surrounding 8 grid cells is assigned to that grid cell to realize the construction of the small-scale digital rock nuclear magnetic resonance response spectrum library.

[0024] Optionally, step S4, the step of multi-spectral synthesis of multi-scale digital rock nuclear magnetic resonance response spectra, includes:

[0025] Step S41. Using the low-resolution CT scan image of the large-scale rock in step S12 and the high-resolution CT scan image of the small-scale rock in step S14 as training samples, the resolution of the large-scale image is improved to the resolution of the small-scale rock CT scan in step S14 using the recurrent generative adversarial network method.

[0026] Step S42. Divide the large-scale digital rocks obtained from step S41 with increased resolution to obtain n×n×n small-scale digital rocks;

[0027] Step S43. Calculate the pore structure of each small-scale digital rock in step S42 using the maximum spherical pore network model method;

[0028] Step S44. Using the small-scale digital rock NMR response spectrum library constructed in step S33 and the pore structure results in step S43, the NMR response spectrum of each small-scale digital rock is obtained. Then, the NMR response spectra of all small-scale digital rocks are superimposed in a manner corresponding to the transverse relaxation time values ​​to achieve multi-spectral synthesis of multi-scale digital rock NMR response spectra.

[0029] The beneficial effects of this invention are that the method of this invention realizes multi-scale digital rock nuclear magnetic resonance response spectrum simulation, effectively integrates the pore structure characteristics and nuclear magnetic resonance response spectrum characteristics at different scales, avoids the problem that single-pore-scale nuclear magnetic resonance response spectrum simulation is not representative, overcomes the limitation of traditional digital core nuclear magnetic resonance response spectrum simulation that cannot characterize multi-scale pore structure information, improves the accuracy of rock nuclear magnetic resonance response spectrum simulation results, and easily reveals the rock nuclear magnetic resonance response mechanism under multi-scale and multi-parameter configuration conditions, laying the foundation for the fine evaluation of the physical properties of oil and gas reservoir rocks. Attached Figure Description

[0030] Figure 1 This is a flowchart illustrating a multi-scale digital rock nuclear magnetic resonance response spectrum simulation according to an embodiment of the present invention;

[0031] Figure 2 This is a large-scale digital rock segmentation image with improved resolution, as shown in an embodiment of the present invention.

[0032] Figure 3This is a simulation diagram of the multi-scale digital rock nuclear magnetic resonance response spectrum, which is a multi-spectral synthesis method, as shown in an embodiment of the present invention. 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 some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] A multi-scale digital rock nuclear magnetic resonance response spectrum simulation method, such as Figure 1 As shown, it includes the following steps:

[0035] Step S1. Large-scale and small-scale multi-resolution CT scanning experiments on rocks

[0036] Specifically, it includes:

[0037] Step S11. Cut the large-scale rock to be studied into regular cubes or cylinders and perform oil washing pretreatment;

[0038] Step S12. Perform a low-resolution CT scan on the regular cube or cylinder obtained in step S11;

[0039] Step S13. Based on the CT scan images obtained in step S12, select multiple regions with different structural features for drilling to obtain small-scale sub-rocks;

[0040] Step S14. Perform high-resolution CT scans on the small-scale sub-rocks obtained in step S13.

[0041] Step S2. Small-scale digital rock nuclear magnetic resonance response spectrum simulation

[0042] Specifically, it includes:

[0043] Step S21. Based on the CT scan images obtained in step S14, a large number of small-scale digital rocks with the same resolution are generated using a style-based generative adversarial network method.

[0044] Step S22. Combining the random walk method and the finite element method, simulate the nuclear magnetic resonance response spectrum of each small-scale digital rock generated in step S21.

[0045] Step S3. Construction of a small-scale digital rock nuclear magnetic resonance response spectrum library

[0046] Specifically, it includes:

[0047] Step S31. Calculate the pore structure of each small-scale digital rock generated in step S21 using the maximum spherical pore network model method;

[0048] Step S32. Establish a porosity-porosity-median pore radius grid with porosity as the abscissa and the median pore radius as the ordinate. Each grid cell represents the nuclear magnetic resonance response spectrum of the corresponding small-scale digital rock generated in step S22 under the conditions of corresponding porosity and median pore radius.

[0049] Step S33. If a certain grid cell in step S32 does not have a corresponding nuclear magnetic resonance response spectrum of a small-scale digital rock, then the average value of the nuclear magnetic resonance response spectra of all grid cells that are in direct contact with that grid cell is assigned to that grid cell, thereby realizing the construction of a small-scale digital rock nuclear magnetic resonance response spectrum library.

[0050] Step S4. Multi-spectral synthesis of multi-scale digital rock nuclear magnetic resonance response spectra

[0051] Specifically, it includes:

[0052] Step S41. Using the low-resolution CT scan image of the large-scale rock in step S12 and the high-resolution CT scan image of the small-scale rock in step S14 as training samples, the Recurrent Generative Adversarial Network (RGAN) method is used. That is, the low-resolution CT scan image of the large-scale rock is used as the X domain and the high-resolution CT scan image of the small-scale rock is used as the Y domain. Then, the adversarial loss function of the RGAN algorithm is used to learn the mapping function from the X domain to the Y domain, so as to realize the mapping from the low-resolution CT scan image of the large-scale rock to the high-resolution CT scan image of the small-scale rock, thereby improving the resolution of the large-scale image to the resolution of the small-scale rock CT scan in step S14.

[0053] Step S42. Divide the large-scale digital rocks obtained from step S41 with increased resolution to obtain n×n×n small-scale digital rocks;

[0054] Step S43. Calculate the pore structure of each small-scale digital rock in step S42 using the maximum spherical pore network model method;

[0055] Step S44. Using the small-scale digital rock NMR response spectrum library constructed in step S33 and the pore structure results in step S43, the NMR response spectrum of each small-scale digital rock is obtained. Then, the NMR response spectra of all small-scale digital rocks are superimposed to achieve multi-spectral synthesis of multi-scale digital rock NMR response spectra.

[0056] Application examples

[0057] Taking rock A collected from an oilfield in China as an example, a large-scale digital rock nuclear magnetic resonance response spectrum simulation method is adopted, including the following steps:

[0058] Step S1. Large-scale and small-scale multi-resolution CT scanning experiments on rocks

[0059] Rock A collected from an oilfield in China was cut into regular cubes of 24mm×24mm×24mm and pretreated with oil and salt washing. Then, a CT scan with a resolution of 0.06mm was performed. Three regions with different structural features were selected from the CT scan images, and three small-scale cylindrical sub-rocks with a diameter of 8mm were drilled accordingly. Each sub-rock was then subjected to a CT scan with a resolution of 0.01mm.

[0060] Step S2. Small-scale digital rock nuclear magnetic resonance response spectrum simulation

[0061] Based on 3000 valid CT scan images of small-scale rocks with a resolution of 0.01 mm, this study utilizes a style-based generative adversarial network (GAN) approach. This involves constructing generator and discriminator networks to ensure the network structure adapts to the 0.01 mm image resolution. The dataset is then trained using these networks, enabling the generator to produce realistic small-scale digital rock images and the discriminator to accurately distinguish between real and generated images. After training, the style of the generated images is controlled by adjusting the input noise vector, resulting in 500 small-scale digital rock images, each with a resolution of 0.01 mm. The NMR response spectra of these 500 small-scale digital rocks are then simulated using a combination of random walk and finite element method (FEM) techniques. For the random walk method, the parameters of the simulated NMR system are determined, including surface relaxation rate and the strength and direction of the external magnetic field. The motion of particles in the pores is simulated through three steps: volume relaxation, surface relaxation, and diffusion relaxation, generating NMR decay curves. Finally, singular value decomposition (SVD) is used to obtain the NMR response spectra. By using the finite element method to solve Maxwell's equations, the magnetic field strength of each voxel element in the digital core is obtained, making the diffusion relaxation simulation process more accurate.

[0062] Step S3. Construction of a small-scale digital rock nuclear magnetic resonance response spectrum library

[0063] The pore structure of each small-scale digital rock generated in step S2 is calculated using the maximum spherical pore network model method. This involves dividing the pore space into a series of spherical structural units, selecting the largest sphere radius to match the characteristics of the actual pore structure, and determining the pore radius distribution by calculating the volume of each sphere. A porosity-porosity median pore radius grid is then established with porosity as the abscissa and the median pore radius as the ordinate. Each grid unit represents the nuclear magnetic resonance (NMR) response spectrum of the corresponding small-scale digital rock generated in step S2 under the conditions of corresponding porosity and median pore radius. If a grid unit does not have a corresponding NMR response spectrum of a small-scale digital rock, the average value of the NMR response spectra of the surrounding 8 grid units is assigned to that grid unit. This achieves the construction of a small-scale digital rock NMR response spectrum library based on the NMR response spectra of 500 small-scale digital rocks.

[0064] Step S4. Multi-scale digital rock nuclear magnetic resonance response spectrum simulation using multi-spectral synthesis

[0065] Using the CT scan images of large-scale rocks with a resolution of 0.06 mm and small-scale rocks with a resolution of 0.01 mm from step S1 as training samples, the resolution of the large-scale images is increased by 6 times to 0.01 mm using a recurrent generative adversarial network (GAN) method. Then, the large-scale digital rocks are segmented to obtain 64 small-scale digital rocks in a 4×4×4 pattern, as shown below. Figure 2 As shown. The pore structure of 64 small-scale digital rocks was calculated using the maximum spherical pore network model method. Their parameters were then marked in the small-scale digital rock NMR response spectrum library in step S3, obtaining the NMR response spectrum of each small-scale digital rock. Finally, all small-scale NMR response spectra were superimposed in a manner corresponding to the transverse relaxation time values, achieving multi-spectral synthesis and simulation of multi-scale digital rock NMR response spectra. Figure 3 As shown.

[0066] It should be noted that the terms "large scale" and "small scale," "high resolution" and "low resolution" mentioned in this invention are all relative terms. For the specific range of values, please refer to the industry consensus.

[0067] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A multi-scale digital rock nuclear magnetic resonance response spectrum simulation method, characterized in that, Includes the following steps: Step S1. Large-scale and small-scale multi-resolution CT scanning experiments on rocks; Step S2. Small-scale digital rock nuclear magnetic resonance response spectrum simulation; Step S3. Construction of a small-scale digital rock nuclear magnetic resonance response spectrum library; Step S4. Multi-spectral synthesis of multi-scale digital rock nuclear magnetic resonance response spectra; Step S2, the steps of small-scale digital rock nuclear magnetic resonance response spectrum simulation, include: Step S21. Based on the CT scan images obtained in step S14, a large number of small-scale digital rocks with the same resolution are generated using a style-based generative adversarial network method. Step S22. Combine the random walk method and the finite element method to simulate the nuclear magnetic resonance response spectrum of each small-scale digital rock generated in step S21; Step S3, the steps for constructing the small-scale digital rock nuclear magnetic resonance response spectrum library, include: Step S31. Calculate the pore structure of each small-scale digital rock generated in step S21 using the maximum spherical pore network model method; Step S32. Establish a porosity-porosity-median pore radius grid with porosity as the abscissa and the median pore radius as the ordinate. Each grid cell represents the nuclear magnetic resonance response spectrum of the corresponding small-scale digital rock generated in step S22 under the conditions of corresponding porosity and median pore radius. Step S33. If a certain grid cell in step S32 does not have a corresponding nuclear magnetic resonance response spectrum of small-scale digital rock, then the average value of the nuclear magnetic resonance response spectra of the surrounding 8 grid cells is assigned to the grid cell to realize the construction of the small-scale digital rock nuclear magnetic resonance response spectrum library. Step S4, the step of multi-spectral synthesis of multi-scale digital rock nuclear magnetic resonance response spectra, includes: Step S41. Using the low-resolution CT scan image of the large-scale rock in step S12 and the high-resolution CT scan image of the small-scale rock in step S14 as training samples, the resolution of the large-scale image is improved to the resolution of the small-scale rock CT scan in step S14 using the recurrent generative adversarial network method. Step S42. Divide the large-scale digital rocks obtained from step S41 with increased resolution to obtain n×n×n small-scale digital rocks; Step S43. Calculate the pore structure of each small-scale digital rock in step S42 using the maximum spherical pore network model method; Step S44. Using the small-scale digital rock NMR response spectrum library constructed in step S33 and the pore structure results in step S43, the NMR response spectrum of each small-scale digital rock is obtained. Then, the NMR response spectra of all small-scale digital rocks are superimposed in a manner corresponding to the transverse relaxation time values ​​to achieve multi-spectral synthesis of multi-scale digital rock NMR response spectra.

2. The multi-scale digital rock nuclear magnetic resonance response spectrum simulation method as described in claim 1, characterized in that, Step S1, the steps of the large-scale and small-scale rock multi-resolution CT scanning experiments, include: Step S11. Cut the large-scale rock to be studied into regular cubes or cylinders and perform oil washing pretreatment; Step S12. Perform a low-resolution CT scan on the regular cube or cylinder obtained in step S11; Step S13. Based on the CT scan images obtained in step S12, select multiple regions with different structural features for drilling to obtain small-scale sub-rocks; Step S14. Perform high-resolution CT scans on the small-scale sub-rocks obtained in step S13.

Images