A high-precision inversion method for hot dry rock natural fractures based on geology-engineering integration

Abstract

The present application relates to natural fracture distribution law, geological exploration and development technical field, disclose a kind of high-precision inversion method of hot dry rock natural fracture based on geology-engineering integration, comprising the following steps: (one) field data acquisition and preliminary geological modeling: based on the initial fracture model of the geological data of hot dry rock reservoir is constructed;(two) fracturing experiment and engineering inversion correction: on the basis of preliminary fracture model, using field fracturing test data, microseismic information is carried out engineering inversion, inversion correction is carried out to model;(three) multiple iteration closed-loop feedback optimization: the geological modeling result and engineering inversion data are carried out multiple mutual feedback correction, and the high-precision fracture distribution model of hot dry rock reservoir is obtained.The scheme combines geological modeling and engineering inversion organically, constructs a closed-loop feedback mechanism, and realizes high-precision inversion of fracture distribution by multiple iteration optimization.

Description

Technical Field

[0001] This invention relates to the field of dry hot rock reservoir stimulation technology, specifically to a high-precision inversion method for natural fractures in dry hot rock based on geology-engineering integration. Background Technology

[0002] Hot dry rock (HDR) geothermal resources, possessing immense development potential, have garnered significant attention in recent years due to their cleanliness, widespread distribution, and seasonal independence. HDR reservoirs, with their low porosity and permeability, rely heavily on natural fractures as crucial channels for seepage and heat exchange. However, the accurate characterization of natural fractures in HDR reservoirs, controlled by geological factors, presents a significant technical challenge for achieving efficient development.

[0003] Existing techniques for characterizing natural fractures primarily rely on geological modeling and engineering inversion. Geological modeling simulates the spatial distribution of fractures using geological data such as lithology and tectonic stress fields, but it is limited in accuracy and detail due to its incomplete consideration of reservoir fracture parameters, making it difficult to reflect the actual development of fractures. Engineering inversion, on the other hand, dynamically infers fracture distribution characteristics based on wellbore data such as fracturing experiments, but due to the lack of geological model support, the inversion results often lack geological validity. Relying solely on either geological modeling or engineering inversion techniques is insufficient to comprehensively and accurately describe the spatial distribution of natural fractures in hot dry rock reservoirs, and it is difficult to meet the high-precision fracture model requirements for hot dry rock development.

[0004] Therefore, constructing a high-precision inversion method for natural fractures in hot dry rocks based on geology-engineering integration not only effectively makes up for the shortcomings of existing technologies, but also effectively solves the technical problems of long cycle and high cost when studying natural fractures in fractured reservoirs. This is of great significance for the study and utilization of natural fractures in fractured reservoirs. Summary of the Invention

[0005] This invention relates to a high-precision inversion method for natural fractures in dry hot rock reservoirs based on geology-engineering integration, aiming to solve the technical problem of insufficient accuracy in the characterization of natural fractures in existing technologies.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a high-precision inversion method for natural fractures in hot dry rocks based on geology-engineering integration, comprising the following steps:

[0007] (I) Field data acquisition and preliminary geological modeling: Constructing an initial fracture model based on geological data of hot dry rock reservoirs;

[0008] (II) Fracturing Experiment and Engineering Inversion Correction: Based on the preliminary fracture model, engineering inversion is performed using field fracturing test data and microseismic information to correct the model.

[0009] (III) Iterative closed-loop feedback optimization: The geological modeling results and engineering inversion data are mutually fed back and corrected multiple times to obtain a high-precision fracture distribution model of the dry hot rock reservoir.

[0010] Preferably, as an improvement, in (a) field data acquisition and preliminary geological modeling, the method for acquiring the geological data is as follows: collecting natural fracture parameters of the hot dry rock reservoir through multi-scale data analysis, including UAV scanning, core sampling, and SEM scanning.

[0011] Preferably, as an improvement, the natural crack parameters include the crack dip angle, length, width, density, and directional distribution.

[0012] Preferably, as an improvement, in (II) fracturing experiment and engineering inversion correction, the method for obtaining the field fracturing test data is as follows: conduct fracturing experiments in the target well, monitor and record the pressure curve, stress field changes and microseismic signals of the wellbore in real time, and capture the dynamic evolution characteristics of the fractures during the fracturing process.

[0013] Preferably, as an improvement, in (ii) fracturing experiment and engineering inversion correction, the microseismic information monitoring method is: using microseismic monitoring to record the generation, extension direction and depth changes of cracks.

[0014] Preferably, as an improvement, in (ii) fracturing experiment and engineering inversion correction, the engineering inversion method is as follows: by comparing the pressure data and microseismic information collected in real time with the initial model, and by combining machine learning algorithms to adjust the natural fracture parameters in the fracture model, the spatial location and morphology of the fracture are gradually corrected to obtain the engineering inversion data.

[0015] Preferably, as an improvement, in (iii) multiple iterations of closed-loop feedback optimization, a machine learning-based fracturing curve and microseismic fitting method is also included, including the following: using the density, length, height, dip angle, orientation, and opening of natural fractures in hot dry rock reservoirs as variables, combined with an efficient fracturing simulation model, pressure curves and fracture network distribution are output to form a database.

[0016] Preferably, as an improvement, in (II) fracturing experiment and engineering inversion correction, the displacement discontinuity method is used to develop numerical simulation technology for deep reservoir hydraulic fracturing. This fully considers the influence of stress shadows within, between, and between well sections, as well as the effects of natural fractures. The displacement discontinuity is calculated using the following formula:

[0017] D x =u x (x,y,0 - )-u x (x,y,0 + )

[0018] Dy =u y (x,y,0 - )-u y (x,y,0 + )

[0019] D z =u z (x,y,0 - )-u z (x,y,0 + )

[0020] In the formula, D represents a small change, and u represents displacement; D x Specifically, this is expressed as: on the plane z = 0, at a fixed point (x, y), the difference between the partial derivative of u with respect to x when z approaches 0 from the negative direction and the partial derivative of u with respect to x when z approaches 0 from the positive direction; D y D z Same as above;

[0021] Based on the stress function solution, the displacement and stress induced by the displacement discontinuity of the crack element at any point (xyz) in space are calculated as follows:

[0022] u x =C{[2(1-v)I ,z -zI ,xx ]D x -zI ,xy D y -[(1-2v)I ,x +zI ,xz ]D z}

[0023] u y =C{-zI ,xy D x +[2(1-v)I ,z -zI ,yy ]D y -[(1-2v)I ,y +zI ,yz ]D z}

[0024] u z =C{[(1-2v)I ,x -zI ,xz ]D x +[(1-2v)I ,y -zI ,yz ]D y +[2(1-v)I ,z -zI ,zz ]D z}

[0025] σ xx =2C{[2I ,xz -zI ,xxx ]D x +[2vI ,yz -zI ,xxy ]D y +[I ,zz +(1-2v)I ,yy -zI ,xxz ]D zz}

[0026] σ yy =2C{[2vI ,xz -zI ,xyy ]D x +[2I ,yz -zI ,yyy ]D y +[I ,zz +(1-2v)I ,xx -zI ,yyz ]D z}

[0027] σ zz =2C{-zI ,xzz D x -zI ,yzz D y +[I ,zz -zI ,zzz ]D z}

[0028] τ xy =2C{[(1-v)I ,yz -zI ,xxy ]D x +[(1-v)I ,xz -zI ,xyy ]D y -[(1-2v)I ,xy +zI ,xyz ]D z}

[0029] τ yz =2C{-[vI ,xy +zI ,xyz ]D x +[I ,zz +vI ,xx -zI ,yyz ]D y -zI ,yzz D z}

[0030] τ zx =2C{[I ,zz+vI ,yy -zI ,xxz ]D x -[vI ,xy +zI ,xyz ]D y -zI ,xzz D z}

[0031] In the formula, u is the displacement, σ is the tensile stress, and τ is the shear stress.

[0032] The principle of this scheme is as follows:

[0033] This approach employs a multi-stage feedback loop between geological forward modeling (geological modeling) and engineering inversion (fracture correction) to continuously optimize the fracture model, achieving high-precision characterization of fractures in hot dry rock reservoirs. Specifically, this method first uses geological modeling to comprehensively simulate reservoir lithology, stress field characteristics, and sedimentary environment, generating a preliminary fracture distribution model. During the engineering inversion stage, multiple rounds of fracture model correction are performed using real-time acquired data on fracturing stress and wellbore pressure changes, gradually ensuring the model conforms to actual reservoir conditions. Through iterative feedback adjustments, the accuracy and resolution of the fracture model are continuously improved, making the spatial distribution of fractures in the model more closely resemble real geological conditions.

[0034] Based on specific application examples, such as a field survey of the hot dry rock fracture system in the Gonghe Basin of Qinghai Province, this patented method employs multiple techniques, including UAV scanning, core description, SEM scanning, and fracturing simulation, to acquire and establish fracture distribution data at different scales. Through large-scale fracturing simulation and efficient machine learning algorithms, fracture network parameters are adjusted, and data such as pressure curves and fracture network distribution are iteratively output to generate a high-precision fracture model. In particular, the model is dynamically corrected using microseismic monitoring data and fracturing pump pressure curves, achieving a high-precision natural fracture inversion technology based on geology-engineering integration.

[0035] The advantages of this solution are:

[0036] 1. Compared to existing technologies that rely solely on geological modeling or engineering inversion, which struggle to comprehensively and accurately describe the spatial distribution of natural fractures in hot dry rock reservoirs and fail to meet the high-precision fracture model requirements for hot dry rock development, this approach organically combines geological modeling and engineering inversion, constructing a closed-loop feedback mechanism. Through multiple iterative optimizations, it achieves high-precision fracture distribution inversion. Simultaneously, this method balances computational efficiency and simulation accuracy, making efficient fracture characterization possible in large-scale, deep hot dry rock reservoirs. This high-precision natural fracture inversion technology has broad application potential in the development of hot dry rock resources. Its accurate fracture model not only provides a scientific basis for the development design and thermal energy utilization of hot dry rock reservoirs but also significantly improves the efficiency and effectiveness of geothermal extraction and helps promote fracture characterization and resource utilization in other unconventional reservoirs.

[0037] 2. This scheme is the first to construct a high-precision natural fracture inversion method based on a geological-engineering feedback closed loop, achieving a new level of accuracy in characterizing fracture distribution in hot dry rock reservoirs. Specifically, through multiple rounds of iterative feedback from geological and engineering data, not only is the accuracy of the fracture model enhanced, but its geological rationality and engineering feasibility are also significantly improved. Simultaneously, by introducing machine learning algorithms and using fracturing and microseismic data as key constraints, this invention optimizes the model's computational efficiency, enabling accurate reconstruction of fracture distribution within a complex three-dimensional space. Furthermore, this method is applicable to various hot dry rock reservoirs, providing a generalizable high-precision solution for fracture prediction in unconventional reservoirs.

[0038] 3. Utilizing the high-precision fracture inversion method of this invention, an accurate natural fracture model was successfully established in the hot dry rock reservoir of the Gonghe Basin in Qinghai Province. This model can effectively predict the conductivity and thermal conduction characteristics of fractures, providing a scientific basis for the efficient development and thermal energy utilization of the reservoir. This model provides reliable support for subsequent fracturing design, reservoir assessment, and geothermal development, significantly improving the development efficiency and effectiveness of hot dry rock reservoirs. Attached Figure Description

[0039] Figure 1 This is a process flow diagram of a high-precision inversion method for natural fractures in hot dry rocks based on geology-engineering integration in Embodiment 1 of the present invention.

[0040] Figure 2 This is a typical geological feature of the Gonghe Basin studied in Example 1 of the present invention.

[0041] Figure 3 This is a granite core from the Gonghe Basin studied in Example 1 of the present invention.

[0042] Figure 4 The results of electron microscopy of granite from the Gonghe Basin studied in Example 1 of this invention are shown.

[0043] Figure 5 The outcrop characteristics of the Gonghe Basin studied in Embodiment 1 of the present invention.

[0044] Figure 6 for Figure 4 Stratigraphic details of Zhonghuangquan Institute 1 (outcrop features in Gonghe Basin).

[0045] Figure 7 The length, dip angle, and direction of the fractures in the granite profile of the Gonghe Basin studied in Example 1 of this invention are shown.

[0046] Figure 8 The results of extracting fracture features from each granite profile in the Gonghe Basin in Example 1 of this invention.

[0047] Figure 9 The location of the fractures in the downhole core of the Qinghai Gonghe granite studied in Example 1 of this invention is shown in red.

[0048] Figure 10 The results of extracting the dip angle characteristics of natural fractures in Example 1 of this invention.

[0049] Figure 11 The image shows the SEM scan results of the granite outcrops in the Gonghe Basin studied in Example 1 of this invention.

[0050] Figure 12 The results are energy dispersive spectroscopy analysis of granites from the Gonghe Basin studied in Example 1 of this invention.

[0051] Figure 13 This is a well-connected profile of granite exploration wells in the Gonghe Basin studied in Embodiment 1 of the present invention.

[0052] Figure 14 This is the geological model of the Chabcha block obtained in Embodiment 1 of the present invention.

[0053] Figure 15 This is the three-dimensional natural crack model obtained in Embodiment 1 of the present invention.

[0054] Figure 16 This is the initial crack model obtained in the engineering inversion stage of Embodiment 1 of the present invention.

[0055] Figure 17 The pressure curves obtained from different seam mesh parameters during the engineering inversion stage in Embodiment 1 of the present invention are shown.

[0056] Figure 18 This is a comparison chart of the microseismic test and simulation results in the engineering inversion stage of Embodiment 1 of the present invention.

[0057] Figure 19 This is a schematic diagram illustrating the process of obtaining the pressure curve using machine learning training in the engineering inversion stage of Embodiment 1 of the present invention—weighted correlation.

[0058] Figure 20 This is a comparison chart of the stress test and simulation results in the engineering inversion stage of Embodiment 1 of the present invention.

[0059] Figure 21 This is a flowchart of the machine learning algorithm in the engineering inversion stage of Embodiment 1 of the present invention.

[0060] Figure 22 This is the natural crack feature adjustment scheme based on microseismic events in the engineering inversion stage of Embodiment 1 of the present invention.

[0061] Figure 23 This is a diagram of refined crack network parameters obtained through multiple iterations of closed-loop feedback optimization in Embodiment 1 of the present invention.

[0062] Figure 24 This is a diagram of the microseismic model obtained through multiple iterations of closed-loop feedback optimization in Embodiment 1 of the present invention.

[0063] Figure 25 This is a diagram of the crack propagation model obtained by multiple iterations of closed-loop feedback optimization in Embodiment 1 of the present invention.

[0064] Figure 26 This is a comparison chart of pump pressure curves obtained from multiple iterations of closed-loop feedback optimization in Embodiment 1 of the present invention.

[0065] Figure 27 This is the high-precision natural crack DFN obtained after correction in Embodiment 1 of the present invention. Detailed Implementation

[0066] The present invention will be further described in detail below with reference to embodiments, but the implementation of the present invention is not limited thereto. Unless otherwise specified, the technical means used in the following embodiments and experimental examples are conventional means well known to those skilled in the art, and the materials and reagents used can all be obtained commercially.

[0067] In the development of hot dry rock reservoirs, natural fractures are the main channels for seepage and heat conduction. However, due to the complex distribution of fractures, traditional single geological modeling or engineering inversion methods are insufficient to achieve high-precision characterization. This scheme establishes a precise and efficient hot dry rock fracture inversion technology through multiple iterative feedback between geological forward modeling (geological modeling) and engineering inversion (fracturing correction). This invention utilizes a feedback loop between geological information and fracturing data to form a high-precision fracture distribution model applicable to complex hot dry rock reservoirs, providing a scientific basis for the efficient exploitation of hot dry rock reservoir resources. The specific scheme of this invention is illustrated using the hot dry rock reservoir in the Gonghe Basin of Qinghai Province as an example.

[0068] Example 1

[0069] This solution provides a high-precision inversion method for natural fractures in hot dry rocks based on geology-engineering integration. The process flow diagram is as follows: Figure 1 As shown, it includes the following steps:

[0070] (I) Field data acquisition and preliminary geological modeling: also known as the geological forward modeling stage, key geological information of the dry hot rock reservoir is obtained through multiple field surveys and sampling, and the distribution pattern of natural fractures is obtained based on the analysis of natural fractures by outcrop + well logging interpretation.

[0071] Specifically, the study included the following: based on UAV scanning, core description, full-diameter thin sections, and environmental scanning electron microscopy, the structural characteristics of multi-scale thermal reservoirs were systematically investigated, including features such as the length, dip angle, direction, aperture, spacing, and density of fractures. Based on these features, the development and classification of fractures were confirmed, and a preliminary three-dimensional natural fracture model was formed.

[0072] Specifically as follows:

[0073] 1. On-site data collection:

[0074] Figure 2 Showcasing the typical geological features of the Gonghe Basin, Figure 3 Displaying granite core samples from the Gonghe Basin, Figure 4 The results of electron microscopy scans of the granite are presented. Analysis reveals the outcrop characteristics of the Gonghe Basin as follows: Figures 5-6 As shown, the field survey revealed that the granite outcrops were relatively broken, with numerous joints and fissures that crisscrossed longitudinally and laterally, and were filled with mud (partially mixed with quartz). The natural fissure system was dominated by high-angle fissures (fissure density 5 fissures / m) and joints, which were the main channels for heat extraction from dry hot rock.

[0075] (1-1) UAV scanning: UAV scanning was carried out on the dry hot rock outcrops in the Gonghe Basin of Qinghai Province to generate a three-dimensional geological model and identify and measure the geometric features of the cracks, such as the dip angle, direction and spacing.

[0076] Specifically, based on UAV scanning, visual interpretation of the 3D models of 5 outcrops in the Gonghe Basin was conducted, yielding a total of 217 sets of fracture data. A cross-sectional fracture distribution was plotted, detailing information such as fracture length, width, density, dip angle, and direction. The fracture length, dip angle, and direction are shown in the figure below. Figure 7 As shown. The results of extracting the crack features for each profile are as follows. Figure 8 As shown.

[0077] The average gap size (taking length as an example) is calculated using the following formula:

[0078]

[0079] In the formula, n is Number of gaps x is The size of each gap In this embodiment, the average crack size calculated from 217 cracks is 7.6m.

[0080] The probability density function for the gap size is given by the following formula:

[0081]

[0082] (1-2) Core collection and scanning electron microscopy (SEM) analysis: Through core collection and SEM scanning, the porosity, microfracture structure and filling material composition of the dry hot rock reservoir were analyzed in detail, and the characteristic parameters of the fractures were recorded.

[0083] Specifically, the downhole core samples of the Qinghai Gonghe granite, such as... Figure 9 As shown. The dip angle characteristics of natural fractures are as follows: Figure 10 As shown in Table 1, the statistical results of the fracture parameters are detailed in the table.

[0084] Table 1 Statistical results of fracture parameters

[0085] Crack aperture 0.1-1mm Crack count statistics 9 45 Crack spacing 0.5-50m Crack density 1 / m 1 5

[0086] In addition, the SEM scan results of the granite outcrop are as follows: Figure 11 As shown. Energy dispersive spectroscopy analysis of the granite is as follows. Figure 12 As shown.

[0087] The results show that microscopic SEM scanning reveals that the Gonghe dry hot rock is generally dense with a porosity of 1-3%, and the pores consist of dissolution pores and intergranular micropores. Based on the combined characteristics of the downhole core and outcrops, the Qinghai Gonghe dry hot rock exhibits moderately developed fractures, indicating a fractured geothermal reservoir.

[0088] (1-3) Macro-geological characteristics analysis: Record and statistically analyze the density of natural fractures, distribution of high-angle fractures, and joint fracture characteristics of hot dry rock outcrops.

[0089] After obtaining the above multi-scale and multi-dimensional fracture parameters, a preliminary fracture distribution model of the hot dry rock reservoir is constructed. The model includes information such as fracture dip angle, density, length, and aperture to describe the initial spatial distribution and geometric characteristics of the fractures.

[0090] Specifically, by surveying the locations of four exploration wells in the Gonghe dry hot rock area and correcting the bedding planes based on the well profiles, a geological model of the Chabcha block was established. Simultaneously, based on statistical results of fracture characteristics, a preliminary three-dimensional natural fracture model was formed.

[0091] The analysis results for the natural crack identification parameters include the following:

[0092] (S1) Crack location distribution: The location of the geometric center point of the crack surface is determined;

[0093] (S2) Crack orientation, calculated using the following formula:

[0094]

[0095] (S3) Geometric shape, the calculation formula is as follows:

[0096]

[0097] (S4) The seam width range and its distribution are calculated using the following formula:

[0098]

[0099] Specifically, the well-to-well profile results in this embodiment are as follows: Figure 13 As shown, the geological model of the Chabcha block is as follows: Figure 14 As shown, the three-dimensional natural crack model is as follows: Figure 15 As shown, the initial crack model is as follows: Figure 16 As shown.

[0100] (II) Fracturing Experiment and Engineering Inversion Correction, also known as the engineering inversion stage, or the fracturing correction stage, involves conducting fracturing experiments to obtain engineering inversion data based on the preliminary fracture model. The specific steps are as follows:

[0101] (2-1) Field fracturing test: A fracturing test was conducted in the target well (such as GH01 well in the Gonghe Basin of Qinghai Province) (see Table 2 for fracturing simulation parameters). The pressure curve, stress field change and microseismic signal of the wellbore were monitored and recorded in real time to capture the dynamic evolution characteristics of the fracture during the fracturing process.

[0102] Table 2 Fracturing Simulation Parameters

[0103] Poisson's ratio 0.15 matrix tensile strength 11MPa Penetration 0.5mD Tensile strength of natural cracks 3MPa Natural crack cohesion 5MPa Minimum / Maximum Horizontal Stress 110~130 / 100~120MPa

[0104] (2-2) Microseismic Monitoring: Microseismic monitoring was used to record the generation, propagation direction, and depth changes of fractures, and the results were compared with the initial model. Pressure curves obtained from different fracture network parameters are detailed in [link to relevant documentation]. Figure 17 .

[0105] (2-3) Data Inversion and Correction: Engineering inversion is performed using real-time acquired pressure data and microseismic information. Machine learning algorithms are then used to adjust parameters in the crack model (such as crack length, density, and aperture), gradually correcting the spatial location and morphology of the cracks. In this step, the machine learning model iteratively optimizes the crack model parameters based on the pressure curve and microseismic signals, making the model closer to the actual crack characteristics.

[0106] Specifically, a numerical simulation technique for deep reservoir hydraulic fracturing is developed using the displacement discontinuity method, fully considering the effects of stress shadows within, between, and between fracturing sections, as well as natural fractures. The model limits the computational domain to the fracture surface to reduce dimensionality, thereby significantly improving computational efficiency while ensuring simulation accuracy.

[0107] The displacement discontinuities (in the element's local coordinate system) used in this scheme include the following:

[0108] D x =u x (x,y,0 - )-u x (x,y,0 + )

[0109] D y =u y (x,y,0 - )-u y (x,y,0 + )

[0110] D z =u z (x,y,0 - )-u z (x,y,0 + )

[0111] In the formula, D represents a small change, and u represents displacement; D x Specifically, this can be expressed as: on the plane z = 0, at a fixed point (x, y), the difference between the partial derivative of u with respect to x when z approaches 0 from the negative direction and the partial derivative of u with respect to x when z approaches 0 from the positive direction. D y D z Same as above.

[0112] This scheme uses a semi-infinite space Green's function, as follows:

[0113]

[0114] Based on the stress function solution, the displacement and stress induced by the displacement discontinuity of the crack element at any point (xyz) in space are calculated as follows:

[0115] u x =C{[2(1-v)I ,z -zI ,xx ]D x -zI ,xy D y -[(1-2v)I ,x +zI ,xz ]D z}

[0116] u y =C{-zI ,xy D x +[2(1-v)I ,z -zI ,yy ]D y -[(1-2v)I ,y+zI ,yz ]D z}

[0117] u z =C{[(1-2v)I ,x -zI ,xz ]D x +[(1-2v)I ,y -zI ,yz ]D y +[2(1-v)I ,z -zI ,zz ]D z}

[0118] σ xx =2C{[2I ,xz -zI ,xxx ]D x +[2vI ,yz -zI ,xxy ]D y +[I ,zz +(1-2v)I ,yy -zI ,xxz ]D zz}

[0119] σ yy =2C{[2vI ,xz -zI ,xyy ]D x +[2I ,yz -zI ,yyy ]D y +[I ,zz +(1-2v)I ,xx -zI ,yyz ]D z}

[0120] σ zz =2C{-zI ,xzz D x -zI ,yzz D y +[I ,zz -zI ,zzz ]D z}

[0121] τ xy =2C{[(1-v)I ,yz -zI ,xxy ]D x +[(1-v)I ,xz -zI ,xyy ]D y -[(1-2v)I ,xy +zI ,xyz ]Dz}

[0122] τ yz =2C{-[vI ,xy +zI ,xyz ]D x +[I ,zz +vI ,xx -zI ,yyz ]D y -zI ,yzz D z}

[0123] τ zx =2C{[I ,zz +vI ,yy -zI ,xxz ]D x -[vI ,xy +zI ,xyz ]D y -zI ,xzz D z}

[0124] In the formula, u is the displacement, σ is the tensile stress, and τ is the shear stress.

[0125] Microseismic test and simulation results are, for example Figure 18 As shown. For details on the principle of obtaining the pressure curve, please refer to [link / reference]. Figure 19 The results of stress tests and simulations are, for example Figure 20 As shown.

[0126] (III) Multiple iterations of closed-loop feedback optimization: Based on the above-mentioned fracturing curve and microseismic fitting, the geological modeling results and engineering inversion data are repeatedly fed back and corrected to obtain a high-precision fracture distribution model of the dry hot rock reservoir.

[0127] Specifically, it includes the following:

[0128] (3-1) Data integration and feedback correction: Based on the results of each round of fracturing experiments and inversion, the latest data is fed back into the geological model for multiple rounds of correction to update the spatial distribution and geometric characteristics of the fractures.

[0129] (3-2) Machine Learning-Driven Optimization: Using fracturing pressure curves and microseismic signals as inputs, a machine learning model is established. A database is formed through large-scale fracturing simulations, and the geometric parameters of the fracture model are automatically fitted. This step is continuously iterated and adjusted to gradually make the fracture distribution match the actual measurement results.

[0130] (3-3) Model verification and accuracy improvement: After each round of optimization, the model accuracy is verified by comparing it with field data to ensure that the closed-loop optimized fracture model can truly reflect the geometric and mechanical characteristics of fractures in hot dry rock reservoirs.

[0131] In the process, this solution also provides a machine learning-based method for fitting fracturing curves and microseismic data, including the following: using the density, length, height, dip angle, orientation, and opening of natural fractures in hot dry rock reservoirs as variables, combined with an efficient fracturing simulation model, the solution outputs pressure curves and fracture network distribution to form a database.

[0132] Among them, machine learning algorithm programs such as Figure 21 As shown, the specific steps include the following:

[0133] S1, Begin;

[0134] S2. Divide the training set and the test set;

[0135] S3. Input the training set into the LSSVM model. The LSSVM model is mainly implemented using the JAVA algorithm, which includes the following calculation process: initialize the JAVA algorithm parameters and hyperparameters - perform iterative optimization based on the objective function - reach the progress requirement or the maximum number of iterations - output the optimal result;

[0136] S4. Determine the hyperparameters of the LSSVM model;

[0137] S5, End.

[0138] Schemes for adjusting natural fracture characteristics based on microseismic events, such as... Figure 22 As shown, the relevant functions for natural cracks are adjusted specifically based on the mismatch information.

[0139] This scheme employs a fracturing simulation method that balances computational efficiency and simulation accuracy; and a machine learning algorithm that balances training efficiency and prediction accuracy.

[0140] Specifically, by integrating fracturing simulation and machine learning algorithms, and using microseismic data and pump pressure curves from well GH01 as constraints, the distribution characteristics of natural fractures are iteratively adjusted to form a high-precision natural fracture model based on integrated geology and engineering. This approach utilizes a machine learning model to iteratively optimize fracture model parameters based on pressure curves and microseismic signals, making the model closer to actual fracture characteristics. Furthermore, by optimizing the model's computational efficiency, the fracture distribution within the complex three-dimensional space is accurately reconstructed. 。

[0141] Specifically, the results of comparing the refined fracture network parameters, microseismic data, fracture propagation data, and pump pressure curves obtained through iterative adjustments are as follows: Figures 23-26 As shown, the corrected natural crack model is detailed in [link to model]. Figure 27This model can accurately describe key parameters such as the spatial location, dip angle, aperture, and density of fractures, and reconstruct the true distribution structure of fractures in three-dimensional space, conforming to actual geological and engineering conditions. This scheme employs the above approach, using outcrop + well logging interpretation-based natural fracture analysis in the geological forward modeling stage to obtain the distribution law of natural fractures; then, in the engineering inversion stage, using machine learning-based fracturing microseismic analysis + pump pressure curve correction, it obtains a corrected high-precision natural fracture DFN. This scheme uses a fracturing simulation method that balances computational efficiency and simulation accuracy; and a machine learning algorithm that balances training efficiency and prediction accuracy. By combining the geological forward modeling and engineering inversion stages, this scheme obtains a corrected high-precision natural fracture DFN, effectively realizing integrated geological-engineering analysis of fractured reservoir natural fractures, providing a theoretical basis for the research, analysis, and subsequent development of natural fractures in reservoirs.

[0142] The above descriptions are merely embodiments of the present invention, and common knowledge such as specific technical solutions and / or characteristics are not described in detail here. It should be noted that those skilled in the art can make various modifications and improvements without departing from the technical solutions of the present invention, and these should also be considered within the scope of protection of the present invention. These modifications and improvements will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A high-precision inversion method for natural fractures in hot dry rocks based on geology-engineering integration, characterized in that: The steps include the following: (I) Field Data Acquisition and Preliminary Geological Modeling: An initial fracture model is constructed based on the geological data of the hot dry rock reservoir. In (I) Field Data Acquisition and Preliminary Geological Modeling, the geological data is acquired by collecting natural fracture parameters of the hot dry rock reservoir through multi-scale data analysis, including UAV scanning, core sampling, and SEM scanning. The natural fracture parameters include the dip angle, length, width, density, and orientation distribution of the fractures. (II) Fracturing Experiment and Engineering Inversion Correction: Based on the initial fracture model, engineering inversion is performed using field fracturing test data and microseismic information to correct the model. The method for acquiring the field fracturing test data is as follows: fracturing experiments are conducted in the target well, and the pressure curve, stress field changes, and microseismic signals of the wellbore are monitored and recorded in real time to capture the dynamic evolution characteristics of the fractures during the fracturing process. The engineering inversion method is as follows: the pressure data and microseismic information collected in real time are compared with the initial fracture model, and the natural fracture parameters in the fracture model are adjusted by combining machine learning algorithms to gradually correct the spatial location and morphology of the fractures and obtain the engineering inversion data. (III) Iterative closed-loop feedback optimization: The geological modeling results and engineering inversion data are repeatedly fed back and corrected to obtain a high-precision fracture distribution model of hot dry rock reservoirs; it also includes machine learning-based fracturing curve and microseismic fitting methods, including the following: using the density, length, height, dip angle, strike and opening of natural fractures in hot dry rock reservoirs as variables, combined with an efficient fracturing simulation model, pressure curves and fracture network distribution are output to form a database.

2. The high-precision inversion method for natural fractures in hot dry rocks based on geology-engineering integration as described in claim 1, characterized in that: In (II) Fracturing Experiment and Engineering Inversion Correction, the microseismic information monitoring method is: using microseismic monitoring to record the generation, extension direction and depth changes of cracks.

Images