Method for characterizing dynamic evolution of multi-scale mechanical properties of bedded shale under water-rock interaction

Abstract

This application relates to the interdisciplinary fields of electro-digital data processing and rock mechanics, specifically a method for characterizing the dynamic evolution of multi-scale mechanical properties of layered shale under water-rock interaction. This method collects multi-source digital experimental data under different flowback fluid immersion durations and performs standardized preprocessing. Multi-dimensional correlation analysis is conducted on the data to divide the water-rock interaction evolution stages. The fractal characteristics of fractures and fracture surfaces are calculated and their quantitative mapping relationship with mechanical parameters is established. An FDEM numerical simulation model considering the spatial distribution of multiple minerals and the failure of interface units is constructed and calibrated. Data are fused to construct a water-rock interaction physicochemical coupling damage model and a dynamic evolution prediction model, outputting the microscopic and macroscopic mechanical parameters of the target reservoir at any given time. This method can accurately characterize the dynamic evolution process of multi-scale mechanical properties of layered shale, providing a reliable reference for shale reservoir fracturing operations.

Description

Technical Field

[0001] This application relates to the interdisciplinary fields of electrical digital data processing and rock mechanics, specifically to a method for characterizing the dynamic evolution of multi-scale mechanical properties of bedding shale under water-rock interaction. Background Technology

[0002] Current methods for characterizing the mechanical properties of layered shale under water-rock interaction can mostly only obtain macroscopic mechanical parameters at discrete time points. They cannot achieve continuous dynamic characterization of mechanical properties at multiple scales from micro to macro, nor do they fully consider the influence of multi-mineral heterogeneity and physicochemical coupling effects on mechanical properties. They are difficult to accurately reflect the true evolution of shale mechanical properties during water-rock interaction, and cannot provide accurate dynamic mechanical parameter support for hydraulic fracturing design of shale gas reservoirs, thus affecting the effectiveness and safety of reservoir stimulation. Summary of the Invention

[0003] To address, or at least partially address, the aforementioned technical problems, this application provides a method for characterizing the dynamic evolution of multi-scale mechanical properties of layered shale under water-rock interaction.

[0004] This application provides a method for characterizing the dynamic evolution of multi-scale mechanical properties of layered shale under water-rock interaction, including the following steps:

[0005] S1. Collect multi-source digital test data of layered shale under different flowback fluid soaking times, and perform standardized preprocessing on the multi-source digital test data; the multi-source digital test data includes micro-mineral composition data, micro-structure data, micro-mechanical data, macro-tensile mechanical data, macro-shear mechanical data, and macro-compression mechanical data.

[0006] S2. Perform multi-dimensional correlation analysis on the preprocessed multi-source digital test data to divide the evolution stages of water-rock interaction, and calculate the fractal characteristics of cracks and fracture surfaces based on fractal geometry algorithms, and establish a quantitative mapping relationship between the fractal characteristics and the mechanical parameters corresponding to the micromechanical data, macroscopic tensile mechanical data, macroscopic shear mechanical data, and macroscopic compressive mechanical data.

[0007] S3. Based on the microscopic mineral composition data and the grain boundary mechanical parameters extracted from the microscopic mechanical data, construct an FDEM numerical simulation model that considers the spatial distribution of multiple minerals and the destruction of interface units, and calibrate the FDEM numerical simulation model using the multi-source digital experimental data.

[0008] S4. By integrating the multi-source digital test data with the output of the calibrated FDEM numerical simulation model, a water-rock interaction physicochemical coupling damage model is constructed, and a multi-scale dynamic evolution prediction model of mechanical properties with immersion time is generated.

[0009] S5. Using the dynamic evolution prediction model, output the microscopic and macroscopic mechanical parameters of the target reservoir at any time.

[0010] Optionally, step S1 specifically includes the following steps:

[0011] The microscopic mineral composition data were collected using an X-ray diffractometer, the microscopic structure data were collected using a scanning electron microscope, the microscopic mechanical data were collected using a nanoindenter, the macroscopic tensile mechanical data were collected using a Brazilian splitting test, the macroscopic shear mechanical data were collected using a shear test, and the macroscopic compressive mechanical data were collected using a triaxial compression test.

[0012] All collected data are subjected to outlier removal, data normalization, and format unification to obtain standardized preprocessed multi-source digital experimental data.

[0013] Optionally, the division of the evolutionary stages of water-rock interaction specifically includes:

[0014] Based on the time-series data of elastic modulus, hardness, tensile strength, and compressive strength in the preprocessed multi-source digital test data, an unsupervised clustering algorithm is used for cluster analysis to divide the water-rock interaction into a rapid hardening stage, a gradual softening stage, and a late recovery stage, and output the time nodes and feature thresholds for each stage.

[0015] Optionally, the calculation of the fractal features of the crack and fracture surface based on the fractal geometry algorithm specifically includes:

[0016] The fractal dimension of the plane of the Brazilian splitting crack was calculated using the wire box method, the fractal dimension of the triaxial compression crack body was calculated using the voxel method, and the fractal dimension of the shear fracture surface elevation was calculated using the improved cubic covering method. The fractal characteristics of the crack and the fracture surface were then integrated to obtain the fractal features.

[0017] Optionally, the construction of the FDEM numerical simulation model considering the spatial distribution of multiple minerals and the destruction of interface units specifically includes:

[0018] Based on the aforementioned microscopic mineral composition data, a multi-mineral grid is generated using a random distribution algorithm;

[0019] Zero-thickness cohesive elements are embedded between adjacent mesh elements to characterize interface element failure;

[0020] The elastic modulus and Poisson's ratio parameters of different minerals are assigned in batches, as well as the grain boundary toughness and grain boundary stiffness parameters in the grain boundary mechanical parameters, to obtain the FDEM numerical simulation model.

[0021] Optionally, calibrating the FDEM numerical simulation model using the multi-source digital experimental data specifically includes:

[0022] Extract the digital features of the stress-strain curves and crack propagation morphology output from the FDEM numerical simulation model;

[0023] The similarity between the digital features and the corresponding experimental features in the multi-source digital experimental data is calculated.

[0024] The parameters of the FDEM numerical simulation model are iteratively optimized using the gradient descent method until the similarity error is less than a preset error threshold, thus completing the model calibration.

[0025] Optionally, after calibrating the FDEM numerical simulation model, the following steps are also included:

[0026] The controlled variable method was used to simulate the rock mechanical response under different mineral grain sizes, different grain boundary toughnesses, and different mineral component contents.

[0027] The influence factors of each factor on rock strength and failure mode are calculated, an influence factor weight matrix is ​​generated, and the main controlling mineral factors affecting the mechanical properties of bedding shale are identified.

[0028] Optionally, the construction of the water-rock interaction physicochemical coupling damage model specifically includes:

[0029] Based on the mineral dissolution and precipitation kinetics corresponding to the microscopic mineral composition data, and combined with the Rebindelle effect and Griffith crack propagation theory, a digital damage factor coupling surface energy reduction and structural damage is constructed. The degree of rock damage at different stages of water-rock interaction evolution is calculated, and a water-rock interaction physicochemical coupling damage model is obtained.

[0030] Optionally, the generation of a dynamic evolution prediction model for multi-scale mechanical properties over immersion time specifically includes:

[0031] Using immersion time, initial micro-mineral composition, and initial geostress as input features, and the multi-scale mechanical parameters output by the multi-source digital experimental data and the calibrated FDEM numerical simulation model as labels, a gradient boosting tree prediction model is trained to obtain the dynamic evolution prediction model of multi-scale mechanical properties with immersion time.

[0032] Optionally, the microscopic mechanical parameters include microscopic elastic modulus and microscopic hardness, and the macroscopic mechanical parameters include macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength; S5 specifically includes:

[0033] The micro-elastic modulus, micro-hardness, macro-tensile strength, macro-shear strength, and macro-compressive strength of the target reservoir at any given time are calculated using the dynamic evolution prediction model.

[0034] Generate multi-scale mechanical property evolution curves, rock damage distribution cloud maps, and parameter prediction reports that include the microscopic and macroscopic mechanical parameters;

[0035] The output microscopic and macroscopic mechanical parameters are imported into the reservoir fracturing design numerical simulation system.

[0036] The method provided in this application has the following beneficial effects:

[0037] This application comprehensively covers the microscopic mineral composition, microstructure, micromechanical properties, and macroscopic tensile, shear, and compressive mechanical properties of layered shale by collecting multi-source digital experimental data under different flowback fluid immersion durations and performing standardized preprocessing. Multi-dimensional correlation analysis of the preprocessed multi-source data allows for the classification of the evolution stages of water-rock interaction, accurately reflecting the non-monotonic variation of shale mechanical properties with immersion time. Furthermore, based on fractal geometry algorithms, the fractal characteristics of cracks and fracture surfaces are calculated, and their quantitative mapping relationship with mechanical parameters is established, enabling quantitative characterization of the correlation between rock damage degree and mechanical properties, thus overcoming the shortcomings of traditional qualitative descriptions. Based on microscopic mineral composition data and grain boundary mechanical parameters, an FDEM numerical simulation model considering the spatial distribution of multiple minerals and the failure of interface units is constructed and calibrated using multi-source experimental data. This model can realistically reproduce the heterogeneous structure and crack propagation process within shale, improving the accuracy of the numerical simulation results. By integrating multi-source experimental data with calibrated numerical simulation results, a water-rock interaction physicochemical coupling damage model and a multi-scale dynamic evolution prediction model for mechanical properties are constructed. This model enables continuous prediction of microscopic and macroscopic mechanical parameters at any time, eliminating the need for repeated, time-consuming and laborious laboratory tests, significantly shortening the parameter acquisition cycle, and providing accurate dynamic mechanical parameter references for shale reservoir fracturing operations. Attached Figure Description

[0038] Figure 1 A schematic diagram of the process for characterizing the dynamic evolution of multi-scale mechanical properties of layered shale under water-rock interaction, as provided in an embodiment of this application;

[0039] Figure 2 This is a schematic diagram illustrating the evolution of mineral components in layered shale through water-rock interaction, provided in an embodiment of this application.

[0040] Figure 3 This is a schematic diagram illustrating the microstructural evolution of layered shale through water-rock interaction, provided in an embodiment of this application.

[0041] Figure 4 This is a schematic diagram of the displacement-load characteristics of nanoindentation in layered shale provided in an embodiment of this application; wherein, Figure 4 (a) is a schematic diagram of the displacement-load characteristics of nanoindentation in layered shale with an immersion time of 0 days. Figure 4(b) is a schematic diagram of the displacement-load characteristics of nanoindentation in layered shale after immersion for 1 day. Figure 4 (c) is a schematic diagram of the displacement-load characteristics of nanoindentation in layered shale after immersion for 3 days. Figure 4 (d) is a schematic diagram of the displacement-load characteristics of nanoindentation in layered shale after immersion for 7 days. Figure 4 (e) is a schematic diagram of the displacement-load characteristics of nanoindentation in layered shale after 15 days of immersion. Figure 4 (f) is a schematic diagram of the displacement-load characteristics of nanoindentation in layered shale after immersion for 30 days. Figure 4 (g) is a schematic diagram of the nanoindentation displacement-load characteristics of layered shale after immersion for 45 days;

[0042] Figure 5 A schematic diagram illustrating the evolution of the mechanical properties of Brazilian splitting shale, provided for an embodiment of this application; wherein... Figure 5 (a) is a schematic diagram of strain-stress evolution. Figure 5 (b) is a schematic diagram of the evolution of immersion time-tensile strength;

[0043] Figure 6 This application provides a schematic diagram illustrating the evolution of triaxial compressive mechanical properties of layered shale in an embodiment of the present application; wherein, Figure 6 (a) is a schematic diagram of strain-stress evolution. Figure 6 (b) is a schematic diagram of the evolution of immersion time and compressive strength. Figure 6 (c) is a schematic diagram of the evolution of immersion time and elastic modulus. Figure 6 (d) is a schematic diagram of the evolution of soaking time-Poisson's ratio;

[0044] Figure 7 A schematic diagram of the fractal characteristics of Brazilian splitting fractures in layered shale provided in an embodiment of this application;

[0045] Figure 8 This is a schematic diagram illustrating the verification of the FDEM numerical simulation model of layered shale provided in the embodiments of this application;

[0046] Figure 9 This is a schematic diagram illustrating the influence of mineral grain size on the mechanical properties of bedding shale, provided in an embodiment of this application; wherein, Figure 9 (a) is a strain-stress diagram. Figure 9 (b) is a schematic diagram of mineral size versus compressive strength. Figure 9 (c) is a schematic diagram of time-total crack area. Figure 9 (d) is a schematic diagram of the time-tensile failure ratio;

[0047] Figure 10 This is a schematic diagram illustrating the influence of grain boundary toughness on the mechanical properties of bedding shale, provided in an embodiment of this application; wherein, Figure 10 (a) is a strain-stress diagram. Figure 10 (b) is a schematic diagram of mineral size versus compressive strength. Figure 10 (c) is a schematic diagram of time-total crack area. Figure 10 (d) is a schematic diagram of the time-tensile failure ratio. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, specific embodiments of this application will be described in further detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely for explaining this application and not for limiting it. It should also be noted that, for ease of description, only the parts relevant to this application are shown in the drawings, not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe operations (or steps) as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0049] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0050] See Figures 1 to 10 This application provides a method for characterizing the dynamic evolution of multi-scale mechanical properties of layered shale under water-rock interaction, including the following steps:

[0051] S1. Collect multi-source digital test data of bedding shale under different flowback fluid soaking times, and perform standardized preprocessing on the multi-source digital test data; the multi-source digital test data includes micro-mineral composition data, micro-structure data, micro-mechanical data, macro-tensile mechanical data, macro-shear mechanical data, and macro-compression mechanical data.

[0052] In some implementations, S1 specifically includes the following steps:

[0053] Microscopic mineral composition data were collected using X-ray diffraction, microscopic structure data were collected using scanning electron microscopy, microscopic mechanical data were collected using nanoindentation, macroscopic tensile mechanical data were collected using Brazilian splitting test, macroscopic shear mechanical data were collected using shear test, and macroscopic compressive mechanical data were collected using triaxial compression test.

[0054] All collected data were processed by outlier removal, data normalization, and format unification to obtain standardized preprocessed multi-source digital experimental data.

[0055] This method acquires multi-dimensional experimental data of layered shale under different flowback fluid durations, and combines fractal geometric analysis and numerical simulation to achieve dynamic characterization and prediction of multi-scale mechanical properties during water-rock interactions. The method sequentially completes multi-source digital experimental data acquisition and preprocessing, multi-dimensional data correlation analysis and fractal feature extraction, FDEM (finite element-discrete element method) numerical simulation model construction and calibration considering multi-mineral heterogeneity, physicochemical coupling damage model construction and dynamic evolution prediction model generation, and multi-scale mechanical parameter output.

[0056] Specifically, the first step was to acquire and standardize the experimental data from multiple sources using digital methods. A specific layered shale formation from a certain block was selected as the research object. Cylindrical and square standard specimens were prepared according to rock mechanics testing standards. The Brazilian splitting specimens were φ25mm×12.5mm, the triaxial compression specimens were φ25mm×50mm, and the shear specimens were 50mm×50mm×50mm. All specimens were taken from intact cores at the same depth, and specimens with obvious natural cracks and defects were discarded. The prepared specimens were then soaked in flowback fluid collected in the field. The flowback fluid was prepared by mixing fracturing fluid stock solution and water at a volume ratio of 3:1000, and then stirred at high speed until homogeneous. Multiple soaking time gradients were set: 0 days, 1 day, 3 days, 7 days, 15 days, 30 days, and 45 days. Three sets of parallel specimens were set for each time gradient to reduce experimental error. Microscopic mineral composition data, microstructural data, micromechanical data, macroscopic tensile mechanical data, macroscopic shear mechanical data, and macroscopic compressive mechanical data were collected from samples under different immersion times. Microscopic mineral composition data were acquired using X-ray diffraction. Samples were dried at 105℃ for 24 hours, ground through a 200-mesh sieve, and then pressed into pellets to obtain the relative contents of minerals such as quartz, calcite, illite, and kaolinite in the samples under different immersion times. Figure 2 As shown, the original shale was mainly composed of calcite (50.37%) and quartz (33.91%), with carbonate minerals accounting for 56.86% and clay minerals accounting for only 7.56%. During the soaking process, the calcite content increased slightly in the initial stage and then stabilized, the dolomite content decreased briefly and then recovered, and the illite content showed a dynamic equilibrium of first decreasing (to 4.68% after 7 days), then increasing and then stabilizing, which intuitively reflects the evolution law of mineral dissolution and precipitation under water-rock interaction. Microstructural data were collected using a scanning electron microscope. The samples were cut into φ10mm×3mm thin sections and polished to a roughness ≤100nm before surface morphology observation, observing the development and evolution process of surface pores and cracks. Figure 3As shown, the original shale (original sample) exhibits a dense and homogeneous structure with only a small amount of primary pores. After soaking for 1 day, initial microcracks and cementation spots appear. From 3 to 7 days, cementation and erosion compete. From 15 to 45 days, erosion dominates pore expansion and fracture connectivity, ultimately forming a porous and easily cracked structure, clearly demonstrating the evolution of the microstructure with water-rock interaction. Micromechanical data were collected using a nanoindenter with a Belkovich indenter, a maximum loading force of 8000 μN, a loading time of 10 s, a holding time of 5 s, and an indentation spacing of 25 μm. Twenty test points were randomly selected on the surface of each sample, and the microelastic modulus and microhardness of each test point were calculated using the Oliver-Fal method. Figure 4 As shown, Figure 4 (a) to Figure 4 (g) correspond to different soaking times, from Figure 4 (a) to Figure 4 (g) The soaking times are 0, 1, 3, 7, 15, 30, and 45 days, respectively. The horizontal axis in each graph represents the indentation depth (displacement), and the vertical axis represents the indentation load. With prolonged soaking time, the displacement-load curve shifts towards greater displacement during the loading stage, the slope of the unloading stage decreases, and the residual indentation depth increases, indicating that the shale's resistance to indentation and its elastic recovery gradually weaken. The short-term soaking curve (1-7 days) shows identifiable softening, while the dispersion further increases in the medium- to long-term (15-45 days) stages, reflecting intensified microscopic heterogeneity. Macroscopic tensile mechanical data were collected through Brazilian splitting tests at a loading rate of 0.002 mm / s. The displacement-load curves during the sample failure process were recorded, and the macroscopic tensile strength was calculated. Macroscopic shear mechanical data were collected through shear tests at a loading rate of 0.002 mm / s. The displacement-load curves during the shear process were recorded, and the macroscopic shear strength, cohesion, and internal friction angle were calculated. Macroscopic compressive mechanical data were collected through triaxial compression tests. The confining pressure was 60 MPa, the confining pressure loading rate was 0.05 kN / s, and the axial loading rate was 0.05 kN / s. The stress-strain curves during the compression process were recorded, and the macroscopic compressive strength, macroscopic elastic modulus, and Poisson's ratio were calculated.

[0057] Next, all collected raw data underwent standardized preprocessing. First, outlier removal was performed using the Grubbs' test to identify and remove abnormal data from each group of parallel samples, retaining only valid data. Then, data normalization was performed, mapping all parameters to the 0-1 range to eliminate differences in the dimensions and orders of magnitude of different parameters. Finally, format unification was performed, converting all data into a unified digital format to generate a structured multi-source digital experimental dataset. This step ultimately outputs the standardized multi-source digital experimental dataset, containing a matrix of microscopic mineral components, a set of microscopic structure images, sequences of microscopic mechanical parameters, sequences of macroscopic tensile mechanical parameters, sequences of macroscopic shear mechanical parameters, and sequences of macroscopic compressive mechanical parameters under different immersion times, for subsequent data correlation analysis and numerical simulation calculations.

[0058] S2. Perform multi-dimensional correlation analysis on the preprocessed multi-source digital test data to divide the evolution stages of water-rock interaction. Based on fractal geometry algorithm, calculate the fractal characteristics of cracks and fracture surfaces, and establish a quantitative mapping relationship between fractal characteristics and corresponding mechanical parameters of micromechanical data, macroscopic tensile mechanical data, macroscopic shear mechanical data, and macroscopic compressive mechanical data.

[0059] In some implementations, the evolutionary stages of water-rock interaction are divided, specifically including:

[0060] Based on the time series data of elastic modulus, hardness, tensile strength and compressive strength in the preprocessed multi-source digital test data, an unsupervised clustering algorithm is used for cluster analysis to divide the water-rock interaction into a rapid hardening stage, a gradual softening stage and a late recovery stage, and output the time nodes and feature thresholds of each stage.

[0061] In some implementations, the fractal characteristics of cracks and fracture surfaces are calculated based on fractal geometry algorithms, specifically including:

[0062] The fractal dimension of the plane of the Brazilian splitting fracture was calculated using the wire box method, the fractal dimension of the triaxial compression fracture body was calculated using the voxel method, and the fractal dimension of the shear fracture surface elevation was calculated using the improved cubic covering method. The fractal characteristics of the fracture and fracture surface were then integrated to obtain the results.

[0063] After obtaining the standardized multi-source digital experimental dataset, multi-dimensional data correlation analysis was conducted based on the standardized multi-source digital experimental dataset. This integrated the dynamic changes of micro-mineral components, the evolution characteristics of microstructure, and the time-series changes of micro-mechanical parameters and macro-mechanical parameters. Continuous change information of key parameters such as elastic modulus, hardness, tensile strength, shear strength, and compressive strength under different immersion times was extracted. Combined with the physicochemical processes of mineral dissolution, secondary precipitation, and structural reconstruction within shale, the synergistic changes and coupling response laws among various parameters were analyzed. The time-series variation data of elastic modulus, hardness, tensile strength, and compressive strength in the dataset are extracted. The K-means unsupervised clustering algorithm is used to perform cluster analysis on the above multi-dimensional time-series data. The input feature vector is [elastic modulus, hardness, tensile strength, compressive strength]. The algorithm identifies the natural cluster boundaries of the data distribution by calculating the Euclidean distance of the data points. Without requiring manual pre-defined partitioning criteria, it automatically distinguishes the stages of water-rock interaction evolution, obtaining complete partitioning results for the rapid hardening stage, gradual softening stage, and late-stage recovery stage. Simultaneously, it outputs the corresponding time nodes and feature thresholds for each stage. Figure 5 As shown ( Figure 5 (a) The horizontal axis ε represents strain, and the vertical axis σ1 represents stress. Different colored curves represent different soaking times. Figure 5 (b) The horizontal axis T represents the soaking time in days, and the vertical axis σ2 represents the tensile strength. All samples exhibited brittle fracture characteristics, with the tensile strength showing a non-monotonic evolution of short-term increase, mid-term weakening, and late-term recovery: it rose to 12.92 MPa in 1 day, dropped to a minimum of 7.50 MPa in 15 days, and recovered to 10.74 MPa in 45 days, consistent with the three-stage evolution of water-rock interaction. The rapid hardening stage occurred in the early stage of soaking. After the backflow fluid invaded the pores inside the shale, it triggered the instantaneous adsorption and volume expansion of mineral particles, resulting in a brief stress concentration inside the pores. The elastic modulus and hardness in the micromechanical parameters showed a short-term increase, and the macroscopic tensile strength and macroscopic compressive strength also showed a slight upward trend. During this stage, the microstructure of the shale remained dense, with no obvious pore expansion or fracture development. Calcite and dolomite in the mineral composition only underwent slight short-term dissolution, and the dissolution rate of illite was at a low level. The overall mechanical properties of the rock showed short-term strengthening characteristics. The gradual softening stage corresponds to the middle stage of immersion. During this stage, the chemical dissolution effect of the flowback fluid intensifies, illite undergoes a significant dissolution process, the cementation structure between mineral particles is gradually destroyed, pores in the shale microstructure continue to expand, fracture connectivity increases, and structural damage within the rock continues to accumulate. Both microscopic and macroscopic mechanical parameters show a continuous downward trend, and the overall load-bearing capacity and deformation resistance of the rock gradually decrease. At this stage, the Repintier effect plays a dominant role; water molecules are adsorbed at mineral interfaces and bedding planes, reducing the rock's surface energy, promoting the initiation and propagation of microcracks, and further exacerbating the weakening of mechanical properties. Figure 6 As shown ( Figure 6 (a) The horizontal axis ε represents strain, and the vertical axis σ1 represents stress; Figure 6 (b) The horizontal axis T represents the soaking time in days, and the vertical axis σ3 represents the compressive strength. Figure 6 (c) The horizontal axis T represents the soaking time in days, and the vertical axis E represents the elastic modulus. Figure 6 (d) The horizontal axis T represents the soaking time in days, and the vertical axis v represents Poisson's ratio. The triaxial compressive strength also exhibits a three-stage evolution: from 0 to 7 days, it drops sharply due to hydration and dissolution, reaching 256.92 MPa at 7 days; from 15 to 45 days, it gradually recovers with the precipitation of secondary minerals, reaching 326.40 MPa at 45 days. The elastic modulus and Poisson's ratio also show synchronous dynamic changes. The later recovery stage corresponds to the long-term soaking stage. The fluid chemical environment inside the pores gradually tends to equilibrium, and the dissolution and formation process of illite reaches a dynamic equilibrium state. Secondary minerals precipitate and fill part of the pores and fractures, and the compactness of the shale microstructure shows a local recovery. Both microscopic and macroscopic mechanical parameters show a slight upward trend, and the mechanical properties of the rock are locally improved. The secondary precipitated minerals reconstruct the stress transmission path between particles, partially compensating for the mechanical deterioration caused by structural damage.

[0064] Based on fractal geometry algorithms, fractal features of cracks and fracture surfaces formed under different experimental failure modes are calculated. For planar cracks formed by the Brazilian splitting test, the fractal dimension is calculated using the box method. The crack plane is divided into multiple square boxes of different sizes, and the number of boxes required to cover the crack outline is counted one by one. The fractal dimension of the Brazilian splitting crack plane is obtained through double log-linear regression of box size and corresponding box number. For three-dimensional crack bodies formed by triaxial compression tests, the voxel method is used to calculate the fractal dimension. The crack body is discretized into cubic voxel units of different sizes, and the number of voxels containing crack structures is counted. Combining the variation law of voxel size and voxel number, the fractal dimension of the triaxial compression crack body is obtained through double log-linear regression. For fracture surfaces formed by shear tests, an improved cubic covering method is used to calculate the fractal dimension. After obtaining the three-dimensional elevation data of the fracture surface, cubes of different sizes are used to cover the fracture surface area, and the number of cubes containing fracture surface structures is counted. The fractal dimension of the shear fracture surface elevation is obtained through double log-linear regression. The fractal dimension calculation results obtained by integrating the wirebox method, voxel method, and improved cubic covering method are used to form the complete fractal features of cracks and fracture surfaces, such as... Figure 7As shown (the horizontal axis T represents immersion time in days, and the vertical axis H represents fractal dimension), the fractal dimension of the Brazilian splitting fracture first decreases, then increases, and then decreases again with immersion time: it drops to a minimum of 1.02 after 3 days, gradually increases from 7 to 15 days, reaches a peak of 1.10 after 30 days, and slightly decreases to 1.05 after 45 days. A higher fractal dimension indicates a more complex fracture and is significantly negatively correlated with tensile strength. The fractal dimension can quantitatively characterize the complexity of fractures and fracture surfaces. A higher fractal dimension value represents a higher number of fracture branches, greater tortuosity, and higher roughness of the fracture surface. This characteristic reflects the structural damage and crack propagation degree of shale under water-rock interaction. In the early stage of immersion, the fracture structure is simple, and the fractal dimension is at a low level; in the middle stage of immersion, structural damage intensifies, crack branches increase, and the fractal dimension gradually increases; in the long-term immersion stage, secondary minerals fill part of the cracks, and the fractal dimension shows a slight decrease, consistent with the evolution law of mechanical properties.

[0065] A systematic regression analysis was conducted on the fractal characteristics of cracks and fracture surfaces and the corresponding microscopic, macroscopic tensile, macroscopic shear, and macroscopic compressive mechanical data to establish a quantitative mapping relationship between fractal characteristics and various mechanical parameters. Using fractal dimension as the independent variable and microscopic elastic modulus, microscopic hardness, macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength as dependent variables, univariate linear regression models and multivariate nonlinear regression models were constructed. The goodness of fit and residual distribution of different models were compared, and the model with the best fit was selected as the final quantitative mapping relationship. The analysis results show that fractal dimension exhibits a significant negative correlation with both microscopic and macroscopic mechanical parameters. As the fractal dimension increases, the microscopic elastic modulus and microscopic hardness continuously decrease, while the macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength also decrease synchronously. This pattern perfectly matches the structural damage and crack propagation process of shale under water-rock interaction. During the rapid hardening stage, the shale has few internal cracks and a complete structure, with a low fractal dimension and correspondingly high mechanical parameters. In the gradual softening stage, mineral dissolution and interface weakening lead to the initiation, propagation, and connection of numerous cracks, increasing the complexity of fractures and fracture surfaces. The fractal dimension continues to rise, while mechanical parameters decrease synchronously. This process aligns with Griffith's crack propagation theory, where the elastic strain energy released by crack propagation exceeds the energy required for the formation of new fracture surfaces, resulting in a rapid decrease in rock mechanical strength. In the later recovery stage, secondary mineral precipitation fills pores and cracks, reducing the complexity of fractures and fracture surfaces. The fractal dimension slightly decreases, and mechanical parameters show a local recovery. The quantitative mapping relationship accurately reflects the dynamic correlation between fractal characteristics and mechanical properties under water-rock interaction, providing stable data support for subsequent numerical simulation model construction and mechanical parameter prediction.

[0066] S3. Based on microscopic mineral composition data and grain boundary mechanical parameters extracted from microscopic mechanical data, an FDEM numerical simulation model considering the spatial distribution of multiple minerals and the destruction of interface units is constructed, and the FDEM numerical simulation model is calibrated using multi-source digital experimental data.

[0067] In some implementations, an FDEM numerical simulation model is constructed that considers the spatial distribution of multiple minerals and the destruction of interface elements, specifically including:

[0068] Based on microscopic mineral composition data, a random distribution algorithm is used to generate multi-mineral grids;

[0069] Zero-thickness cohesive elements are embedded between adjacent mesh elements to characterize interface element failure;

[0070] By batch assigning values ​​to the elastic modulus, Poisson's ratio, and grain boundary toughness and stiffness parameters of different minerals, the FDEM numerical simulation model is obtained.

[0071] In some implementations, the FDEM numerical simulation model is calibrated using multi-source digital experimental data, specifically including:

[0072] Digital features of stress-strain curves and crack propagation morphology output from FDEM numerical simulation models are extracted.

[0073] The similarity between the digital features and the corresponding experimental features in the multi-source digital experimental data is calculated.

[0074] The parameters of the FDEM numerical simulation model are iteratively optimized using the gradient descent method until the similarity error is less than a preset error threshold, thus completing the model calibration.

[0075] In some implementations, after the FDEM numerical simulation model has been calibrated, the following steps are also included:

[0076] The controlled variable method was used to simulate the rock mechanical response under different mineral grain sizes, different grain boundary toughnesses, and different mineral component contents.

[0077] The influence factors of each factor on rock strength and failure mode are calculated, an influence factor weight matrix is ​​generated, and the main controlling mineral factors affecting the mechanical properties of bedding shale are identified.

[0078] Based on previously acquired microscopic mineral composition data and grain boundary mechanical parameters extracted from micromechanical data, an FDEM numerical simulation model considering the spatial distribution of multiple minerals and the failure of interface units was constructed. This model, with the finite element method (FEM) as its core, integrates the heterogeneous distribution characteristics of multiple minerals within shale with the failure patterns of interface units, enabling the reconstruction of the stress-deformation and crack propagation processes of bedding shale under water-rock interaction. The microscopic mineral composition data includes the proportions of minerals such as quartz, calcite, illite, kaolinite, and dolomite under different immersion times. Grain boundary mechanical parameters, extracted from nanoindentation results in the micromechanical data, contain key parameters for grain boundary resistance to deformation and failure, reflecting the mechanical properties of the interfaces between mineral particles and providing reliable data support for parameter assignment in the numerical simulation model. During model construction, a multi-mineral mesh consistent with the actual mineral distribution in shale was generated using a random distribution algorithm based on the proportions of various minerals in the microscopic mineral composition data. The mesh division follows an unstructured triangular mesh format, with unit sizes matching the actual mineral grain sizes, adapting to the heterogeneous structure of shale and avoiding simulation biases caused by regular meshes. Zero-thickness cohesive elements are embedded between adjacent mesh elements. These elements characterize the interface failure behavior between mineral grains within shale. The constitutive behavior of the cohesive elements is defined using a bilinear traction separation model. Before damage evolution, the elements are in a linear elastic stress state, and the stress-strain relationship in the normal and shear directions follows fixed physical laws. The failure of the elements includes both tensile and shear modes. The bilinear traction separation criterion is used to describe the damage evolution process. The damage factor is calculated based on the ratio of the actual displacement of the element to the critical failure displacement. The calculation formula is as follows: ,in As a damage factor, This represents the maximum displacement value reached by the element during the loading process. This represents the displacement value corresponding to the initial damage of the element. The formula represents the displacement value corresponding to the complete failure of a unit. This formula can be used to determine whether new cracks are generated in the interface unit, thus reconstructing the entire process of crack initiation and propagation within the shale. After completing mesh construction and embedding cohesive units, batch parameter assignments are performed on the model. Based on the physical and mechanical properties of different minerals, the elastic modulus and Poisson's ratio parameters are assigned to minerals such as quartz, calcite, and illite. Simultaneously, the extracted grain boundary mechanical parameters, including grain boundary toughness and grain boundary stiffness, are assigned to the cohesive units. Grain boundary toughness determines the maximum allowable deformation of the interface unit before failure, while grain boundary stiffness reflects the deformation resistance of the interface unit. These two types of parameters jointly regulate the crack propagation behavior of the mineral interface. After parameter assignment, a complete FDEM numerical simulation model is obtained, which can simultaneously simulate both continuous deformation and discontinuous failure processes of shale.

[0079] Subsequently, multi-source digital experimental data was used to calibrate the constructed FDEM numerical simulation model, eliminating simulation errors caused by model parameter deviations. During calibration, the FDEM numerical simulation model was first run to perform Brazilian splitting and triaxial compression simulations. Digital features of the stress-strain curve and crack propagation morphology were extracted from the model output. The digital features of the stress-strain curve included key parameters such as peak stress, elastic modulus, and Poisson's ratio. The digital features of crack propagation morphology included crack propagation path, crack distribution range, and the proportion of tensile and shear failure. The proportion of tensile and shear failure was determined using a tensile / shear failure ratio factor. judge, Shear failure, For tensile failure, , , These represent the crack energies in the normal and two shear directions, respectively. The simulated digital features are compared one by one with the corresponding experimental features in the multi-source digital experimental data. A similarity calculation method is used to quantify the deviation between the simulation and experimental results. The similarity calculation focuses on the difference in parameter values, and a comprehensive error index is constructed by considering the deviations of key parameters such as peak stress and elastic modulus. Using the comprehensive error index as the optimization target, the gradient descent method is used to iteratively optimize the parameters of the FDEM numerical simulation model. The optimization objects include the critical failure displacement of cohesive units, grain boundary toughness, grain boundary stiffness, and the mechanical parameters of mineral phases. After each iteration, the model is rerun and the comprehensive error index is calculated, gradually reducing the deviation between the simulation and experimental results until the comprehensive error index is less than a preset error threshold, at which point the model calibration is considered complete. Figure 8 As shown (where ε is strain on the horizontal axis and σ1 is stress on the vertical axis), the stress-strain curve output by the FDEM numerical simulation is highly consistent with the indoor Brazilian splitting test curve in terms of the rising trend, peak stress, and post-peak softening characteristics. The similarity error is less than the preset threshold, verifying the effectiveness of the model calibration. The calibrated model maintains a high degree of consistency with the indoor test results in terms of the evolution trend of the stress-strain curve, the peak parameter value, and the crack propagation mode. It can realistically reproduce the mechanical response and crack propagation law of bedding shale under water-rock interaction. Whether it is the Brazilian splitting process dominated by tensile failure or the triaxial compression process dominated by shear failure, the model can accurately simulate both.

[0080] After calibrating the FDEM numerical simulation model, multiple simulations were performed using the controlled variable method to analyze the influence of different mineral parameters on the mechanical properties of bedding shale and identify the dominant mineral factors affecting these properties. During the implementation of the controlled variable method, external conditions such as boundary conditions, loading rate, and geostress parameters were kept constant, and only parameters related to a single mineral were adjusted to simulate the rock mechanical response under different mineral grain sizes, grain boundary toughnesses, and mineral component contents. When simulating the influence of different mineral grain sizes, multiple sets of gradient mesh element sizes were set to correspond to different mineral grain sizes. The influence of mineral grain size variations on stress transmission and crack propagation modes was analyzed. The mineral grain sizes were set to 0.0003 mm (Model 1), 0.0005 mm (Model 2), 0.0007 mm (Model 3), 0.0009 mm (Model 4), and 0.001 mm (Model 5), respectively. Figure 9 As shown ( Figure 9 (a) The horizontal axis ε represents strain, and the vertical axis σ1 represents stress; Figure 9 (b) The horizontal axis represents minerals of different sizes, and the vertical axis σ3 represents compressive strength; Figure 9 (c) The horizontal axis represents time in seconds, and the vertical axis represents the total area of ​​the cracks; Figure 9 (d) The horizontal axis represents time in seconds, and the vertical axis represents the tensile failure ratio. Smaller mineral grain sizes result in higher peak rock strength (up to 18.62 MPa in Model 1), more delayed crack propagation, and a higher tensile failure ratio (0.8~0.9). Larger grain sizes lead to decreased strength, increased brittleness, and more pronounced tensile-shear mixed failure characteristics. In fine-grained structures, the stress distribution is more uniform, resulting in higher overall rock bearing capacity and crack propagation primarily in a tensile mode. In coarse-grained structures, local stress concentration is more likely, and the crack propagation mode shifts from a single tensile stress to a combined tensile-shear mode. When simulating the effects of different grain boundary toughness, adjusting the critical failure displacement parameter of the cohesive unit can improve grain boundary toughness, inhibiting early microcrack propagation, delaying crack initiation time, and enhancing the peak strength and ductility of the rock. Figure 10 As shown ( Figure 10 (a) The horizontal axis ε represents strain, and the vertical axis σ1 represents stress; Figure 10 (b) The horizontal axis represents minerals of different sizes, and the vertical axis σ3 represents compressive strength; Figure 10 (c) The horizontal axis represents time in seconds, and the vertical axis represents the total area of ​​the cracks; Figure 10(d) The horizontal axis represents time in seconds, and the vertical axis represents the tensile failure ratio. The larger the critical failure displacement of the grain boundary, the higher the peak rock strength (15.98 MPa in Model 5). The later the crack initiation time, the closer the tensile failure ratio is to 0.9, indicating that improved grain boundary toughness can effectively inhibit the early propagation of microcracks, while decreased grain boundary toughness leads to premature crack initiation and enhanced rock brittleness. When simulating the influence of different mineral composition contents, the proportion of various minerals is adjusted according to the dynamic change law of mineral composition under water-rock interaction. When the proportion of high-stiffness brittle minerals is increased, the stress concentration in the model is more concentrated, and crack propagation is mainly a straight main crack. When the proportion of clay minerals is increased, the stress distribution is more dispersed, and cracks are more prone to deflection and branching. After completing multiple sets of simulations, key indicators such as rock strength, failure mode, and crack propagation parameters were extracted from the simulation results. The influence factors of each mineral parameter on rock strength and failure mode were calculated. The values ​​of these influence factors reflect the degree of influence of single parameter changes on mechanical properties. All influence factors were integrated to generate an influence factor weight matrix. By analyzing the values ​​of each factor in the weight matrix, the dominant factors influencing the mechanical properties of layered shale—mineral grain size, grain boundary toughness, and mineral composition content—were intuitively identified. Mineral composition content had the highest weight and was the core factor influencing mechanical properties, followed by grain boundary toughness and mineral grain size, which together constituted the main regulating factors of the mechanical properties of layered shale. This analysis clarifies the effect of mineral heterogeneity on the mechanical properties of shale under water-rock interaction, providing theoretical support for subsequent damage model construction and dynamic evolution prediction.

[0081] S4. By integrating the output results of multi-source digital test data and the calibrated FDEM numerical simulation model, a water-rock interaction physicochemical coupling damage model is constructed, generating a dynamic evolution prediction model of multi-scale mechanical properties with immersion time.

[0082] In some implementations, a water-rock interaction physicochemical coupling damage model is constructed, specifically including:

[0083] Based on the mineral dissolution and precipitation kinetics corresponding to microscopic mineral composition data, and combined with the Rebindelle effect and Griffith crack propagation theory, a digital damage factor coupling surface energy reduction and structural damage is constructed. The degree of rock damage at different stages of water-rock interaction evolution is calculated, and a water-rock interaction physicochemical coupling damage model is obtained.

[0084] In some implementations, a dynamic evolution prediction model for multi-scale mechanical properties over immersion time is generated, specifically including:

[0085] Using immersion time, initial micro-mineral composition, and initial geostress as input features, and multi-scale mechanical parameters output from multi-source digital experimental data and calibrated FDEM numerical simulation models as labels, a gradient boosting tree prediction model is trained to obtain a dynamic evolution prediction model of multi-scale mechanical properties with immersion time.

[0086] By integrating multi-source digital experimental data with the output of a calibrated FDEM numerical simulation model, a water-rock interaction physicochemical coupling damage model is constructed. This model integrates chemical and mechanical damage effects during water-rock interactions, enabling quantitative characterization of the degree of damage accumulation within shale at different evolution stages. The model is built upon the kinetics of mineral dissolution and precipitation, incorporating dynamic changes in mineral composition over different immersion times obtained from previous experiments. This establishes a mathematical relationship between mineral dissolution and precipitation rates over time, quantifying the degree of chemical modification of the shale's internal structure. Simultaneously, the Repintier effect and Griffith crack propagation theory are introduced to analyze the influence of water molecule adsorption on rock surface energy and the energy conditions for crack propagation, establishing a coupling relationship between chemical interaction and mechanical damage. Based on these theoretical foundations, a digital damage factor coupling surface energy reduction and structural damage is constructed, calculated using the following formula: ,in The total damage factor, The chemical damage factor, calculated from the mineral dissolution and precipitation kinetics, reflects the structural damage caused by chemical processes. The mechanical damage factor is calculated from the crack propagation area and the number of element failures obtained from FDEM numerical simulation, and reflects the structural damage caused by mechanical action. and The weighting coefficients for chemical and mechanical damage were determined by fitting multi-source experimental data with simulation results. Using this digital damage factor, the degree of rock damage at different stages of water-rock interaction was calculated. In the rapid hardening stage, the damage factor increased slowly, with only minor chemical damage and a few microcracks existing within the rock. In the gradual softening stage, the damage factor increased rapidly, with chemical and mechanical damage accumulating simultaneously, resulting in severe damage to the internal rock structure. In the later recovery stage, the increase in the damage factor tended to level off, with some areas showing damage repair due to secondary mineral precipitation, and the total damage factor showing a slight decreasing trend. This model enables continuous quantitative characterization of the damage degree throughout the entire water-rock interaction process, accurately reflecting the dynamic evolution of damage to the internal structure of shale.

[0087] Based on the established water-rock interaction physicochemical coupling damage model, and combined with all previous multi-source digital experimental data and FDEM numerical simulation results, a dynamic evolution prediction model of multi-scale mechanical properties with immersion time was generated. This model is trained using a gradient boosting tree algorithm, which can effectively handle multi-dimensional and nonlinear data relationships and capture the complex evolution of multi-scale mechanical parameters under water-rock interaction. During model training, input features and output labels are first determined. Input features include immersion time, initial micro-mineral composition, and initial geostress. The initial micro-mineral composition includes the initial proportions of minerals such as quartz, calcite, illite, kaolinite, and dolomite. The initial geostress includes the maximum horizontal principal stress, minimum horizontal principal stress, and vertical principal stress. Output labels are the multi-scale mechanical parameters output from the multi-source digital experimental data and the calibrated FDEM numerical simulation model, including micro-elastic modulus, micro-hardness, macro-tensile strength, macro-shear strength, macro-compressive strength, and total damage factor. All data are divided into training and testing sets in a 7:3 ratio. The training set is used for training and optimizing model parameters, while the testing set is used to verify the model's prediction accuracy. During training, a decision tree is used as the base learner. The model's fitting ability is continuously improved through iterative training. Each iteration builds a new decision tree based on the prediction residuals of the previous iteration, gradually reducing the model's prediction error. At the same time, a 5-fold cross-validation method is used to prevent the model from overfitting. The model's generalization ability is optimized by adjusting hyperparameters such as the depth of the decision tree, the learning rate, and the number of iterations.

[0088] After training, the accuracy of the dynamic evolution prediction model was verified using test set data. The multi-scale mechanical parameters predicted by the model were compared with the actual experimental data in the test set, and the mean absolute error (MAE) and root mean square error (RMSE) were used as the model accuracy evaluation indicators. The verification results show that the prediction errors of the model for microscopic elastic modulus, microscopic hardness, macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength are all within the allowable range. It can accurately reflect the three-stage evolution law of multi-scale mechanical properties under water-rock interaction. The parameter increase trend in the rapid hardening stage, the parameter decrease trend in the gradual softening stage, and the parameter recovery trend in the later recovery stage are all highly consistent with the actual experimental results. This model can integrate the influence of multiple factors such as mineral heterogeneity, water-rock interaction chemical effects, and mechanical damage effects, breaking through the limitation of traditional methods that can only obtain mechanical parameters at discrete time points, and realizing continuous prediction of multi-scale mechanical parameters under arbitrary immersion time.

[0089] S5. Through the dynamic evolution prediction model, output the microscopic and macroscopic mechanical parameters of the target reservoir at any time.

[0090] In some embodiments, the microscopic mechanical parameters include microscopic elastic modulus and microscopic hardness, and the macroscopic mechanical parameters include macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength; S5 specifically includes:

[0091] The micro-elastic modulus, micro-hardness, macro-tensile strength, macro-shear strength, and macro-compressive strength of the target reservoir at any time are calculated using a dynamic evolution prediction model.

[0092] Generates multi-scale mechanical property evolution curves, rock damage distribution cloud maps, and parameter prediction reports, including microscopic and macroscopic mechanical parameters;

[0093] The output microscopic and macroscopic mechanical parameters are imported into the reservoir fracturing design numerical simulation system.

[0094] By constructing a multi-scale dynamic evolution prediction model for mechanical properties, the model outputs the microscopic and macroscopic mechanical parameters of the target reservoir at any given time. During the output process, the initial microscopic mineral composition, initial geostress, and the soaking time to be predicted for the target reservoir are first input. The model then calculates and outputs the complete set of mechanical parameters for the corresponding time point. The microscopic mechanical parameters include microscopic elastic modulus and microscopic hardness, while the macroscopic mechanical parameters include macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength.

[0095] Based on the output continuous time-series mechanical parameters, multi-scale mechanical property evolution curves, rock damage distribution cloud maps, and parameter prediction reports are generated. The multi-scale mechanical property evolution curves show the continuous variation trends of microscopic elastic modulus, microscopic hardness, macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength with immersion time, presenting the parameter variation characteristics of the three evolution stages of water-rock interaction, and marking the key time nodes and characteristic parameter values ​​for each stage. The rock damage distribution cloud map shows the distribution of total damage factors in different regions of the target reservoir at corresponding immersion times, reflecting the spatial differences in the degree of damage within the reservoir. The parameter prediction report summarizes all calculated mechanical parameters, evolution curves, and damage distribution information, while also indicating the prediction accuracy and applicable scope of each parameter, providing a complete reference for engineering applications.

[0096] The output microscopic and macroscopic mechanical parameters are imported into a reservoir fracturing design numerical simulation system, providing dynamic mechanical parameter support for determining fracturing operation parameters. Traditional fracturing design typically uses static mechanical parameters that do not consider water-rock interactions, failing to reflect the dynamic changes in reservoir mechanical properties during flowback, easily leading to discrepancies between the fracturing design and actual reservoir conditions. The dynamic mechanical parameters output by this method accurately reflect the actual mechanical state of the reservoir under different flowback durations, and can be used in fracturing design numerical simulation systems for fracture propagation simulation, geostress calculation, and construction parameter optimization, helping to determine reasonable injection rate, pump pressure, perforation length, and other key construction parameters. Simultaneously, by combining rock damage distribution cloud maps, areas with high levels of damage within the reservoir can be identified, allowing for targeted adjustments to the fracturing operation plan, avoiding high-intensity fracturing operations in severely damaged areas, and reducing construction risks.

[0097] This method integrates multi-scale experimental data, numerical simulation results, and machine learning algorithms to achieve full-chain dynamic characterization and prediction of the multi-scale mechanical properties of layered shale under water-rock interaction. It solves the problems of traditional methods, such as the inability to obtain mechanical parameters at continuous time points and the failure to consider the heterogeneity and physicochemical coupling effects of multiple minerals. The output mechanical parameters are characterized by high accuracy, fast acquisition speed, and wide applicability, providing reliable technical support for hydraulic fracturing operations in shale gas reservoirs and improving the effectiveness and economic benefits of reservoir stimulation.

[0098] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application.

Claims

1. A method for characterizing the dynamic evolution of multi-scale mechanical properties of layered shale under water-rock interaction, characterized in that, Includes the following steps: S1. Collect multi-source digital test data of layered shale under different flowback fluid soaking times, and perform standardized preprocessing on the multi-source digital test data; the multi-source digital test data includes micro-mineral composition data, micro-structure data, micro-mechanical data, macro-tensile mechanical data, macro-shear mechanical data, and macro-compression mechanical data. S2. Perform multi-dimensional correlation analysis on the preprocessed multi-source digital test data to divide the evolution stages of water-rock interaction, and calculate the fractal characteristics of cracks and fracture surfaces based on fractal geometry algorithms, and establish a quantitative mapping relationship between the fractal characteristics and the mechanical parameters corresponding to the micromechanical data, macroscopic tensile mechanical data, macroscopic shear mechanical data, and macroscopic compressive mechanical data. S3. Based on the microscopic mineral composition data and the grain boundary mechanical parameters extracted from the microscopic mechanical data, construct an FDEM numerical simulation model that considers the spatial distribution of multiple minerals and the destruction of interface units, and calibrate the FDEM numerical simulation model using the multi-source digital experimental data. S4. By integrating the multi-source digital test data with the output of the calibrated FDEM numerical simulation model, a water-rock interaction physicochemical coupling damage model is constructed, and a multi-scale dynamic evolution prediction model of mechanical properties with immersion time is generated. S5. Using the dynamic evolution prediction model, output the microscopic and macroscopic mechanical parameters of the target reservoir at any time.

2. The method according to claim 1, characterized in that, S1 specifically includes the following steps: The microscopic mineral composition data were collected using an X-ray diffractometer, the microscopic structure data were collected using a scanning electron microscope, the microscopic mechanical data were collected using a nanoindenter, the macroscopic tensile mechanical data were collected using a Brazilian splitting test, the macroscopic shear mechanical data were collected using a shear test, and the macroscopic compressive mechanical data were collected using a triaxial compression test. All collected data are subjected to outlier removal, data normalization, and format unification to obtain standardized preprocessed multi-source digital experimental data.

3. The method according to claim 1, characterized in that, The stages of water-rock interaction evolution are specifically divided as follows: Based on the time-series data of elastic modulus, hardness, tensile strength, and compressive strength in the preprocessed multi-source digital test data, an unsupervised clustering algorithm is used for cluster analysis to divide the water-rock interaction into a rapid hardening stage, a gradual softening stage, and a late recovery stage, and output the time nodes and feature thresholds for each stage.

4. The method according to claim 1, characterized in that, The calculation of the fractal characteristics of cracks and fracture surfaces based on fractal geometry algorithms specifically includes: The fractal dimension of the plane of the Brazilian splitting crack was calculated using the wire box method, the fractal dimension of the triaxial compression crack body was calculated using the voxel method, and the fractal dimension of the shear fracture surface elevation was calculated using the improved cubic covering method. The fractal characteristics of the crack and the fracture surface were then integrated to obtain the fractal features.

5. The method according to claim 1, characterized in that, The construction of the FDEM numerical simulation model considering the spatial distribution of multiple minerals and the destruction of interface units specifically includes: Based on the aforementioned microscopic mineral composition data, a multi-mineral grid is generated using a random distribution algorithm; Zero-thickness cohesive elements are embedded between adjacent mesh elements to characterize interface element failure; The elastic modulus and Poisson's ratio parameters of different minerals are assigned in batches, as well as the grain boundary toughness and grain boundary stiffness parameters in the grain boundary mechanical parameters, to obtain the FDEM numerical simulation model.

6. The method according to claim 1, characterized in that, The calibration of the FDEM numerical simulation model using the multi-source digital experimental data specifically includes: Extract the digital features of the stress-strain curves and crack propagation morphology output from the FDEM numerical simulation model; The similarity between the digital features and the corresponding experimental features in the multi-source digital experimental data is calculated. The parameters of the FDEM numerical simulation model are iteratively optimized using the gradient descent method until the similarity error is less than a preset error threshold, thus completing the model calibration.

7. The method according to claim 1, characterized in that, After calibrating the FDEM numerical simulation model, the following steps are also included: The controlled variable method was used to simulate the rock mechanical response under different mineral grain sizes, different grain boundary toughnesses, and different mineral component contents. The influence factors of each factor on rock strength and failure mode are calculated, an influence factor weight matrix is ​​generated, and the main controlling mineral factors affecting the mechanical properties of bedding shale are identified.

8. The method according to claim 1, characterized in that, The construction of the water-rock interaction physicochemical coupling damage model specifically includes: Based on the mineral dissolution and precipitation kinetics corresponding to the microscopic mineral composition data, and combined with the Rebindelle effect and Griffith crack propagation theory, a digital damage factor coupling surface energy reduction and structural damage is constructed. The degree of rock damage at different stages of water-rock interaction evolution is calculated, and a water-rock interaction physicochemical coupling damage model is obtained.

9. The method according to claim 1, characterized in that, The dynamic evolution prediction model for generating multi-scale mechanical properties over immersion time specifically includes: Using immersion time, initial micro-mineral composition, and initial geostress as input features, and the multi-scale mechanical parameters output by the multi-source digital experimental data and the calibrated FDEM numerical simulation model as labels, a gradient boosting tree prediction model is trained to obtain the dynamic evolution prediction model of multi-scale mechanical properties with immersion time.

10. The method according to claim 1, characterized in that, The microscopic mechanical parameters include microscopic elastic modulus and microscopic hardness, and the macroscopic mechanical parameters include macroscopic tensile strength, macroscopic shear strength, and macroscopic compressive strength; S5 specifically includes: The micro-elastic modulus, micro-hardness, macro-tensile strength, macro-shear strength, and macro-compressive strength of the target reservoir at any given time are calculated using the dynamic evolution prediction model. Generate multi-scale mechanical property evolution curves, rock damage distribution cloud maps, and parameter prediction reports that include the microscopic and macroscopic mechanical parameters; The output microscopic and macroscopic mechanical parameters are imported into the reservoir fracturing design numerical simulation system.

Images