A method and system for predicting rock mechanical response and crack propagation path
By establishing the correspondence between the digital core structure model and the mechanical response and crack propagation path, the problem of the difficulty in reusing simulation results in the existing technology is solved. This enables rapid prediction of rock mechanical response and crack propagation path, improving prediction efficiency and reliability. It is applicable to rock strength evaluation and underground engineering risk analysis.
Patent Information
- Application Number
- CN202611325729.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-28
- Publication Date
- 2026-11-17
AI Technical Summary
In existing rock mechanics tests and digital core numerical simulations, the analysis is mainly conducted on single core samples. The simulation results are difficult to reuse, and the correspondence between structural features, mechanical response, and crack propagation path is unclear, resulting in low prediction efficiency for new rock samples.
By acquiring digital core structural models of rock samples, we identify and classify mesostructural units, assign material and interface parameters, establish the correspondence between structural features, mechanical responses, and crack propagation paths of multiple digital core models, and use weighted similarity matching and clustering techniques to predict the response and propagation path of the rock sample to be tested.
It enables rapid prediction of rock mechanical response and crack propagation path, improves the efficiency of reusability of simulation results, reduces repetitive modeling work, and improves prediction efficiency and reliability. It is applicable to rock strength evaluation, failure mode identification, and underground engineering risk analysis.
Smart Images

Figure CN123120142A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of rock microstructure characterization and mechanical fracture process prediction technology, specifically to a method and system for predicting rock mechanical response and crack propagation path. Background Technology
[0002] Under compression, tension, shear, or unloading disturbances, rocks undergo mechanical responses and fracture processes, including stress-strain changes, strength decay, damage accumulation, crack initiation, propagation, connection, and through-failure. The microstructure of the rock, including mineral grains, pores, fissures, bedding, and mineral interfaces, significantly controls its mechanical response and crack propagation path. Cracks often preferentially initiate in pores, at crack tips, weak interfaces, bedding planes, or mineral grain contact areas, and propagate and connect along relatively weak regions, ultimately leading to macroscopic failure. Therefore, accurately predicting the mechanical response, crack initiation location, propagation path, and through-failure mode of rocks is crucial for rock strength evaluation, failure risk assessment, and engineering support design.
[0003] Traditional studies on rock mechanical response and crack propagation often rely on uniaxial compression, triaxial compression, Brazilian splitting, direct shear, or unloading tests, combined with techniques such as acoustic emission, digital image correlation, CT scanning, or high-speed photography to obtain information on rock deformation, failure, and crack evolution. While these methods can provide a relatively intuitive understanding of the rock's load-bearing failure process, they are costly, and the results are significantly influenced by sample preparation and loading conditions. Furthermore, it is difficult to repeatedly analyze the mechanical response and crack propagation patterns under different structural conditions and loading scenarios within the same sample. Simultaneously, due to the significant heterogeneity of the rock's internal structure, relying solely on macroscopic experimental results often fails to accurately explain strength differences, the causes of damage concentration zones, and crack initiation and propagation paths.
[0004] With the development of digital core technology and numerical simulation methods, simulating the mechanical response and fracture process based on the actual microstructure of rocks has become an important method in rock mechanics research. Digital core models can characterize structural units within rocks, such as mineral grains, pores, fractures, bedding, and mineral interfaces. The finite element method (FEM) can simulate the stress distribution, strain evolution, damage propagation, strength changes, and crack propagation processes of rocks under different loading conditions. By converting the digital core structural model into a finite element computational model, the influence of different mineral compositions, grain morphologies, pore and fracture distributions, bedding orientations, and interface characteristics on the mechanical response and fracture path of rocks can be analyzed at the microscale.
[0005] However, most existing digital core finite element simulation methods focus on performing one or more numerical calculations on a single core model. The simulation results are usually only used to explain the mechanical response or crack propagation phenomenon of the model under specific working conditions. They have not yet fully and systematically correlated the structural characteristics of multiple digital core models with their finite element response results, making it difficult to form a structure-mechanical response-crack propagation correspondence that can be used for rapid prediction of new rock samples. For practical engineering, if a complete digital core modeling, parameter assignment, and finite element simulation are to be performed again for every new rock sample, not only will the computational workload be enormous, but it will also be difficult to meet the needs of rapid evaluation and engineering applications. Summary of the Invention
[0006] To address the aforementioned shortcomings in existing technologies, this invention provides a method and system for predicting rock mechanical response and crack propagation paths. This addresses the problems in existing rock mechanics experiments and digital core numerical simulations, which primarily analyze single core samples, have difficulty reusing simulation results, lack clear correspondence between structural features, mechanical response, and crack propagation paths, and have low prediction efficiency for new rock samples.
[0007] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: In a first aspect, the present invention proposes a method for predicting rock mechanical response and crack propagation path, comprising the following steps: A digital core structure model of a rock sample is obtained; the digital core structure model includes one or more microstructural units such as mineral grains, pores, fractures, bedding, and mineral interfaces; The microstructural units in the digital core structure model are identified, classified, and labeled, and rock structure characteristic parameters are extracted. Based on the mechanical properties of the microstructural units, at least one of the material parameters, interface parameters, damage parameters, or weakening parameters is assigned to the digital core structure model, and it is then transformed into a mechanical numerical calculation model. Loading or unloading boundary conditions are applied to the mechanical numerical calculation model to carry out numerical simulation and obtain the numerical simulation response results of the rock under load. Extract rock mechanical response characteristics and crack propagation path characteristics from the numerical simulation response results; Based on the rock structure characteristic parameters of multiple different digital core structure models and their corresponding rock mechanical response characteristics and crack propagation path characteristics, the correspondence between rock structure characteristics and rock mechanical response characteristics and crack propagation path characteristics is established. A digital core structure model of the rock sample to be tested is obtained and the rock structure feature parameters are extracted. Based on the weights corresponding to each rock structure feature parameter calibrated with the goal of jointly minimizing the prediction error of mechanical response, crack propagation path, and failure mode, the weighted structural distance between the rock sample to be tested and each existing digital core structure model is calculated to determine the structural similarity between the rock sample to be tested and each existing digital core structure model. For each existing digital core structure model whose structural similarity meets the preset threshold, clustering is performed based on the spatial neighborhood overlap between each crack propagation path to determine the predicted rock mechanical response and crack propagation path.
[0008] Furthermore, the rock structural characteristic parameters include one or more of the following: mineral composition, brittle mineral content, mineral grain area, equivalent diameter of mineral grains, mineral grain shape coefficient, orientation of the major and minor axes of mineral grains, contact relationship of mineral grains, porosity, pore distribution, fracture density, fracture length, fracture dip angle, fracture connectivity, bedding orientation, bedding spacing, mineral interface density, and distribution of weak interfaces.
[0009] Furthermore, based on the mechanical properties of the microstructural units, at least one of the following parameters—material parameters, interface parameters, damage parameters, or weakening parameters—is assigned to the digital core structure model: Mineral material parameters are assigned to the mineral particle region, including one or more of the following: elastic modulus, Poisson's ratio, compressive strength, tensile strength, cohesion, internal friction angle, fracture energy, and damage evolution parameters. Assign void properties or low stiffness weakening parameters to porous regions; Assign crack surface properties, contact properties, or weakening parameters to the fractured region; Assign bedding interface parameters to the bedding region; The mineral interface region is assigned one or more of the following parameters: interface stiffness, interface strength, interface friction parameter, interface fracture parameter, or interface damage parameter.
[0010] Furthermore, based on the mechanical properties of the microstructural units, at least one of the following parameters—material parameters, interface parameters, damage parameters, or weakening parameters—is assigned to the digital core structure model: Mechanical parameters of mineral particles and mineral interfaces are obtained through micro-area mechanical testing. The locations of micro-area mechanical testing are spatially registered with mineral phase distribution maps to determine the mineral type or mineral interface type corresponding to each test point. Using the macroscopic test results of rocks as the calibration target, inversion calibration is performed by adjusting the material parameters and interface parameters to ensure that the numerical simulation results simultaneously meet the requirements for macroscopic mechanical response error and microscopic crack propagation characteristic error.
[0011] Furthermore, the rock mechanical response characteristics include one or more of the following: elastic modulus, peak strength, residual strength, Poisson's ratio, stress-strain curve morphology, and damage evolution characteristics. The crack propagation path characteristics include one or more of the following: crack initiation location coordinates, initial crack propagation direction angle, crack propagation path spatial coordinate sequence, total crack length, propagation ratio along mineral interface, propagation ratio through mineral grains, and failure mode type.
[0012] Furthermore, the weights corresponding to each rock structural characteristic parameter are calibrated with the objective of jointly minimizing the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error. These weights include: One of the existing digital core structure models is used as a verification sample, and the remaining existing digital core structure models are used as reference samples. The weighted structural distance and structural similarity between the verification sample and each reference sample are calculated under given candidate weights. The reference sample with the highest structural similarity is selected as the prediction reference. The verification sample is predicted based on the rock mechanical response characteristics and crack propagation path characteristics corresponding to the prediction reference. The weighted sum of the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the verification sample is used as the comprehensive verification error. Under the constraint that the weights are non-negative and the sum of the weights is 1, the weights corresponding to each rock structural characteristic parameter that minimize the comprehensive verification error are obtained.
[0013] Furthermore, the comprehensive verification error is the weighted sum of the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the verification sample, including: For each round of cross-validation, the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the validation sample are calculated respectively. The mechanical response prediction error is calculated as follows: the norm square of the difference between the actual mechanical response feature vector and the predicted mechanical response feature vector of the verification sample is calculated, and the arithmetic mean of the norm squares obtained from each round of cross-validation is taken. The crack propagation path prediction error is calculated as follows: calculate the difference between the predicted crack propagation path and the actual crack propagation path of the verification sample and subtract the path similarity between the two. Then, take the arithmetic mean of the differences obtained from each round of cross-validation. The calculation method for the damage mode prediction error is as follows: when the predicted damage mode type of the verification sample is inconsistent with the actual damage mode type, the value is a preset first value; when they are consistent, the value is a preset second value; and the arithmetic mean of the values obtained from each round of cross-validation is taken. The comprehensive verification error is obtained by multiplying the mechanical response prediction error, the crack propagation path prediction error, and the failure mode prediction error by their respective weights and then summing them.
[0014] Furthermore, existing digital core structure models whose structural similarity meets a preset threshold are clustered based on the spatial neighborhood overlap between crack propagation paths to determine the predicted rock mechanical response and crack propagation path, including: The crack propagation paths of existing digital core structural models whose structural similarity meets a preset threshold are transformed to a unified normalized coordinate system, and the crack path neighborhood is defined with each crack propagation path as the center and a preset distance threshold. The path similarity is calculated based on the ratio of the intersection to the union of the neighborhoods of each crack path, and paths with a path similarity greater than a preset path threshold are classified as the same crack path type. Calculate the weighted frequency of occurrence for each crack path type, and determine the crack path type with the highest weighted frequency as the predicted master crack propagation path; Calculate the weighted probability of the crack path covering each spatial location in the normalized coordinate system, and determine the region with the weighted probability of occurrence greater than a preset probability threshold as the predicted weak area of damage.
[0015] Furthermore, it also includes: When the structural similarity does not meet the preset threshold, the digital core structure model of the new rock sample is assigned parameters, converted into a mechanical numerical calculation model, boundary conditions are applied, and numerical simulation is carried out to extract its rock mechanical response characteristics and crack propagation path characteristics. The rock structural characteristic parameters, rock mechanical response characteristics, and crack propagation path characteristics of the rock sample to be tested are added to the correspondence, the correspondence is updated and expanded, and the weights corresponding to each rock structural characteristic parameter are recalibrated.
[0016] Secondly, this invention proposes a rock mechanical response and crack propagation path prediction system, which applies the rock mechanical response and crack propagation path prediction method described above, including: The structural model acquisition module is used to acquire a digital core structural model of a rock sample; the digital core structural model includes one or more microstructural units among mineral grains, pores, fractures, bedding, and mineral interfaces; The structural feature extraction module is used to identify, classify, and label the mesoscopic structural units in the digital core structural model and extract rock structural feature parameters. The numerical calculation model construction module is used to assign values to at least one of the material parameters, interface parameters, damage parameters, or weakening parameters of the digital core structure model according to the mechanical properties of the microstructural units, and transform it into a mechanical numerical calculation model. The numerical simulation module is used to apply loading or unloading boundary conditions to the mechanical numerical calculation model to carry out numerical simulation and obtain the numerical simulation response results of the rock under load. The response feature extraction module is used to extract rock mechanical response features and crack propagation path features from the numerical simulation response results. The correspondence establishment module is used to establish the correspondence between rock structure features and rock mechanical response features and crack propagation path features based on rock structure feature parameters and their corresponding rock mechanical response features and crack propagation path features from multiple different digital core structure models. The weighted similarity matching and prediction module is used to acquire digital core structure models of the rock samples to be tested and extract rock structure feature parameters. Based on the weights corresponding to each rock structure feature parameter calibrated with the goal of jointly minimizing the prediction error of mechanical response, crack propagation path, and failure mode, the module calculates the weighted structural distance between the rock samples to be tested and each existing digital core structure model, and determines the degree of structural similarity between the rock samples to be tested and each existing digital core structure model. For each existing digital core structure model whose structural similarity meets the preset threshold, clustering is performed based on the degree of spatial neighborhood overlap between each crack propagation path to determine the predicted rock mechanical response and crack propagation path.
[0017] The present invention has the following beneficial effects: (1) Based on the digital core structure model, this invention introduces the microstructure of the rock, such as mineral particles, pores, cracks, bedding and mineral interfaces, into the mechanical response and crack propagation path prediction process, which can more realistically reflect the control effect of the heterogeneous structure inside the rock on the fracture process.
[0018] (2) This invention not only obtains the simulation results of a single digital core finite element model, but also further extracts the structural features, mechanical response features and crack propagation path features of multiple digital core models, establishes the correspondence between the structural features of digital cores and the finite element response results, so that the simulation results can be reused as a basis for subsequent predictions.
[0019] (3) After obtaining the digital core structure characteristics of a new rock sample, the present invention can quickly predict its stress-strain response, strength characteristics, damage evolution, crack initiation location, crack propagation path, penetration mode and weak failure area based on the existing correspondence, thereby reducing the work of repeated modeling and calculation and improving prediction efficiency.
[0020] (4) This invention can be used for rock strength prediction, crack propagation analysis, failure mode identification, weak area judgment and stability evaluation of surrounding rock in underground engineering, providing technical support for rock fracture risk analysis in deep mines, underground caverns, tunnels and slope engineering. Attached Figure Description
[0021] Figure 1 This is a schematic diagram of the process for predicting rock mechanical response and crack propagation path according to the present invention. Figure 2 A schematic diagram illustrating the correspondence between structural features and finite element response results in this invention; Figure 3 This is a flowchart of the prediction process for the mechanical response and crack propagation path of the rock sample under test in this invention. Figure 4 This is a schematic diagram illustrating the retrieval of structural similarity matching and prediction results for the rock sample to be tested in this invention. Detailed Implementation
[0022] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.
[0023] This invention provides a method and system for predicting rock mechanical response and crack propagation path based on digital core simulation. The method first constructs a digital core structural model of the rock sample, identifies microstructural units such as mineral grains, pores, fractures, bedding, and mineral interfaces within the model, and further establishes a digital core finite element model. Subsequently, finite element simulations are conducted under different loading conditions to extract rock mechanical response and crack propagation results. Based on the simulation results of multiple digital core samples, a correspondence between structural features and mechanical response and crack propagation is established. Finally, for new rock samples, the rock mechanical response and crack propagation path are predicted through structural feature input and similarity matching. This invention transforms digital core finite element simulation results from single-calculation results into reusable prediction data, thereby improving the efficiency and reliability of predicting rock mechanical response, crack propagation path, and failure mode.
[0024] like Figure 1 As shown, the rock mechanical response and crack propagation path prediction method provided by this embodiment of the invention includes the following steps S1 to S7: S1. Obtain a digital core structure model of the rock sample; the digital core structure model includes one or more microstructural units among mineral grains, pores, fractures, bedding, and mineral interfaces; In an optional embodiment of the present invention, step S1 involves constructing a digital core structure model. A digital core structure model of the rock sample is obtained. This digital core structure model can be a two-dimensional or a three-dimensional digital core model. The digital core structure model includes microstructural units such as mineral grains, pores, fractures, bedding, and mineral interfaces. For a two-dimensional digital core model, it can be represented as a structural image composed of multiple pixel units; for a three-dimensional digital core model, it can be represented as a spatial structural model composed of multiple voxel units.
[0025] In one optional implementation, different structural units in the digital core structure model are assigned structural labels. For example, different mineral particles such as quartz, feldspar, carbonate minerals, and clay minerals are assigned mineral labels; porous regions are assigned pore labels; natural fracture regions are assigned fracture labels; bedding planes are assigned bedding labels; and contact areas between different mineral particles are assigned mineral interface labels. Through these methods, each pixel, voxel, or computational unit in the digital core structure model has a clearly defined structural type.
[0026] S2. Identify, classify and label the microstructural units in the digital core structure model, and extract the rock structure characteristic parameters; In an optional embodiment of the present invention, the rock structural feature parameters extracted in step S2 include one or more of the following: mineral composition, brittle mineral content, mineral grain area, equivalent diameter of mineral grains, mineral grain shape coefficient, orientation of the major and minor axes of mineral grains, contact relationship of mineral grains, porosity, pore distribution, fracture density, fracture length, fracture dip angle, fracture connectivity, bedding orientation, bedding spacing, mineral interface density, and distribution of weak interfaces.
[0027] This embodiment performs rock structure feature parameter extraction. Based on the digital core structure model obtained in step S1, rock structure feature parameters are extracted. The rock structure feature parameters include one or more of the following: mineral composition, brittle mineral content, mineral grain area, equivalent diameter of mineral grains, mineral grain shape factor, major and minor axis orientation of mineral grains, mineral grain contact relationship, porosity, pore distribution, fracture density, fracture length, fracture dip angle, fracture connectivity, bedding direction, bedding spacing, mineral interface density, and weak interface distribution.
[0028] In one alternative implementation, the rock structural characteristic parameters can be organized into a structural feature vector X, such as... Figure 2 As shown, its expression form is: X = [x1, x2, x3, ..., x n ]; In the formula, X represents the structural feature vector of a certain digital core model; x1, x2, x3, ..., x n These represent different structural characteristic parameters, such as mineral content, porosity, fracture density, fracture dip angle, bedding direction, and interface density. To eliminate the dimensional differences between these parameters, each structural characteristic parameter can be normalized for use in subsequent structural type classification and similarity matching.
[0029] In one specific implementation, porosity can be determined based on the ratio of the number of pore pixels or voxels to the total number of pixels or voxels; fracture density can be determined based on the ratio of the total fracture length to the model area, or the ratio of the total fracture area to the model area; mineral interface density can be determined based on the ratio of the total mineral interface length to the model area, or the ratio of the total mineral interface area to the model volume; fracture connectivity can be determined based on the number of fracture junctions, the number of fracture connecting branches, or the number of through fractures. Through the above processing, the complex microstructure of digital cores can be transformed into calculable and comparable structural characteristic parameters.
[0030] S3. Based on the mechanical properties of the microstructural units, assign values to at least one of the material parameters, interface parameters, damage parameters, or weakening parameters of the digital core structure model, and transform it into a mechanical numerical calculation model. In an optional embodiment of the present invention, step S3 assigns values to at least one of material parameters, interface parameters, damage parameters, or weakening parameters to the digital core structure model based on the mechanical properties of the microstructural units, including: Mineral material parameters are assigned to the mineral particle region, including one or more of the following: elastic modulus, Poisson's ratio, compressive strength, tensile strength, cohesion, internal friction angle, fracture energy, and damage evolution parameters. Assign void properties or low stiffness weakening parameters to porous regions; Assign crack surface properties, contact properties, or weakening parameters to the fractured region; Assign bedding interface parameters to the bedding region; The mineral interface region is assigned one or more of the following parameters: interface stiffness, interface strength, interface friction parameter, interface fracture parameter, or interface damage parameter.
[0031] The mechanical numerical calculation model is one of the following: finite element model, discrete element model, finite difference model, or phase field fracture model.
[0032] Step S3 assigns values to at least one of the following parameters—material parameters, interface parameters, damage parameters, or weakening parameters—to the digital core structure model based on the mechanical properties of the microstructural units: Mechanical parameters of mineral particles and mineral interfaces are obtained through micro-area mechanical testing. The locations of micro-area mechanical testing are spatially registered with mineral phase distribution maps to determine the mineral type or mineral interface type corresponding to each test point. Using the macroscopic test results of rocks as the calibration target, inversion calibration is performed by adjusting the material parameters and interface parameters to ensure that the numerical simulation results simultaneously meet the requirements for macroscopic mechanical response error and microscopic crack propagation characteristic error.
[0033] This embodiment establishes a digital core finite element model. The digital core structure model in step S1 is converted into a finite element calculation model. During the conversion process, a finite element mesh is generated based on the pixels, voxels, or structural boundaries in the digital core structure model, and a correspondence is established between mesh elements and digital core structure labels. For each finite element, its structure type is determined based on the structure label at its location, and corresponding material parameters, interface parameters, damage parameters, or weakening parameters are assigned.
[0034] In one optional embodiment, mineral grain regions are assigned mineral material parameters, including one or more of the following: elastic modulus, Poisson's ratio, compressive strength, tensile strength, cohesion, internal friction angle, fracture energy, and damage evolution parameters; pore regions are assigned void properties or low stiffness weakening parameters; fracture regions are assigned crack surface properties, contact properties, or weakening parameters; bedding regions are assigned bedding interface parameters; and mineral interface regions are assigned interface stiffness, interface strength, interface friction parameters, interface fracture parameters, or interface damage parameters.
[0035] In one optional implementation, the material and interface parameters in the digital core finite element model can be obtained through one or more of the following methods: micro-area mechanical testing, rock mechanics experiments, existing experimental data, and inversion calibration. First, the test locations for nanoindentation, microindentation, nanoscratching, or microscratching are spatially registered with EDS elemental distribution maps, QEMSCAN mineral phase diagrams, or SEM images to determine the mineral type, bedding location, or mineral interface type corresponding to each test point.
[0036] For mineral grain regions, the elastic modulus and hardness of different minerals can be obtained through targeted nanoindentation or microindentation testing. The nanoindentation hardness H can be calculated based on the maximum indentation load and contact area. ; In the formula, The maximum indentation load; This represents the indenter-sample contact area corresponding to the maximum load.
[0037] mineral elastic modulus The initial stiffness can be determined based on the unloading curve: ; In the formula, S is the contact stiffness in the initial stage of the indentation unloading curve; β is a correction coefficient related to the geometry of the indenter.
[0038] mineral elastic modulus It can be further determined based on the reduced elastic modulus: ; In the formula, and ν m These are the elastic modulus and Poisson's ratio of the tested mineral, respectively. and ν i These are the elastic modulus and Poisson's ratio of the indenter material, respectively. For the same mineral, the test results from multiple valid test points can be statistically analyzed, and the average value, median value, or probability distribution parameter can be used as the material parameters of the mineral.
[0039] The elastic modulus, hardness, microscopic strength, and fracture parameters of mineral particles can be determined directly or indirectly through micro-area mechanical tests such as nanoindentation, microcantilever beam, microcolumn compression, micro-splitting, and micro-scratching. Mineral cohesion, internal friction angle, and damage evolution parameters are difficult to obtain directly through conventional macroscopic tests; however, initial values can be determined based on micro-area mechanical test results, existing mineral parameter data, or test results of similar minerals.
[0040] Macroscopic uniaxial compression, Brazilian splitting, and triaxial compression tests were used to obtain the overall stress-strain curves, elastic modulus, peak strength, tensile strength, failure mode, and crack propagation path of the rock specimens. These macroscopic test results were then used as calibration targets for the digital core finite element model. By adjusting mineral material parameters, interface parameters, and damage evolution parameters, the simulated macroscopic mechanical response and crack propagation characteristics were made consistent with the experimental results, thereby enabling the inversion calibration of the model parameters.
[0041] When using the Mohr-Coulomb criterion, the macroscopic cohesion and internal friction angle of the rock can be obtained by fitting the triaxial compressive peak strength of the rock sample under different confining pressures, and these can be used as constraints for the inversion calibration of digital core model parameters. ; In the formula, and , representing the maximum and minimum principal stresses at rock specimen failure; c and φ, representing the macroscopic cohesion and macroscopic internal friction angle of the rock specimen, respectively. For the porous region, when the pores are empty, they are set as void regions or low-stiffness regions that do not participate in load bearing; when the pores contain filling materials or fluids, appropriate elastic parameters, density parameters, or fluid parameters are assigned according to the actual properties of the filling materials or fluids. When using low-stiffness elements to approximate voids, their elastic modulus should be much lower than the elastic modulus of the weakest mineral in the model, and parameter sensitivity analysis should be used to ensure that the influence of this element on the overall load bearing capacity of the model is negligible.
[0042] For naturally fractured regions, the location, length, width, dip angle, and connectivity of fractures can be identified using digital core images. Open fractures are classified as cavities, contact surfaces, or low-stiffness weakened regions. For closed or filled fractures, normal stiffness, tangential stiffness, cohesion, internal friction angle, and friction coefficient are determined based on fracture normal closure tests, direct shear tests, or fracture infill material tests. (Fraction normal stiffness) and tangential stiffness The slopes of the initial linear phases of the normal stress-closed displacement curve and the shear stress-tangential displacement curve can be determined respectively. ; ; In the formula, and These represent the increments of the normal stress and shear stress during the initial linear stage, respectively. and These are the corresponding normal closed displacement and tangential displacement increments, respectively.
[0043] For bedding regions and mineral interface regions, interfacial mechanical parameters can be obtained through nanoindentation, nanoscratching, micro-area tensile testing, micro-area shearing, or interfacial fracture tests across bedding or mineral interfaces. The interfacial tensile strength can be taken as the peak value of the interfacial normal traction force; the interfacial cohesion and internal friction angle can be determined by fitting the peak interfacial shear stress under different normal stress conditions. ; In the formula, This represents the peak shear stress at the interface. For interfacial cohesion; This refers to the normal compressive stress at the interface. The angle of friction within the interface.
[0044] Interfacial fracture energy It can be determined based on the area under the interface traction force-separation displacement curve: ; In the formula, t is the interface traction force; δ is the interface separation displacement.
[0045] When direct test results are lacking for some minerals, bedding planes, or mineral interfaces, existing test parameters from the same rock type or similar mineral assemblages can be selected as initial values. These parameters are then inverted and calibrated using constraints such as the rock's macroscopic stress-strain curve, elastic modulus, peak strength, crack initiation location, crack propagation path, and final failure mode. The calibrated parameters should simultaneously meet the error requirements of both macroscopic mechanical response and microscopic crack propagation characteristics, and be used for parameter assignment in digital core finite element models of the same rock type.
[0046] In one specific implementation, the element attributes in the finite element model can be assigned values according to the following rules: if the element center is located in a mineral grain region, the material parameters corresponding to the mineral type are called; if the element center is located in a pore or fracture region, void parameters, weakening parameters, or contact parameters are assigned; if the element is located near a bedding plane or mineral interface, it is set as an interface element or a weakened element. Through the above method, spatial correspondence and parameter transfer between the digital core structure model and the finite element calculation model are achieved, enabling the finite element model to reflect the heterogeneity of the microstructure within the rock.
[0047] S4. Apply loading or unloading boundary conditions to the mechanical numerical calculation model to carry out numerical simulation and obtain the numerical simulation response results of the rock under load. In an optional embodiment of the present invention, step S4 performs finite element simulations under different loading conditions. Loading or unloading boundary conditions are applied to the digital core finite element model constructed in step S3 to simulate the rock mechanical response and fracture process. The loading or unloading boundary conditions include one or more of uniaxial compression, triaxial compression, Brazilian splitting, direct tension, shear loading, cyclic loading, and unloading disturbance. The loading method can employ displacement control, stress control, or a combined displacement-stress control, such as... Figure 2 As shown.
[0048] During the simulation, the stress field, strain field, displacement field, damage field, plastic zone distribution, crack propagation field, crack penetration path, and stress-strain curves of the digital core finite element model are recorded. For models with tensile and shear failure criteria, the location, quantity, and evolution process of tensile and shear failure elements are recorded respectively. For finite element calculations using damage models, the growth process and spatial distribution characteristics of damage variables with loading steps are recorded. The above simulation outputs provide basic data for subsequent extraction of mechanical response characteristics and crack propagation path characteristics.
[0049] S5. Extract rock mechanical response characteristics and crack propagation path characteristics from the numerical simulation response results; In an optional embodiment of the present invention, the rock mechanical response features extracted in step S5 include one or more of the following: elastic modulus, peak strength, residual strength, Poisson's ratio, stress-strain curve morphology, and damage evolution features. The crack propagation path features extracted in step S5 include one or more of the following: crack initiation location coordinates, initial crack propagation direction angle, crack propagation path spatial coordinate sequence, total crack length, propagation ratio along mineral interface, propagation ratio through mineral grains, and failure mode type.
[0050] This embodiment first performs rock mechanical response feature extraction. Based on the finite element simulation results obtained in step S4, rock mechanical response features are extracted. The rock mechanical response features include one or more of the following: stress-strain response, elastic modulus, peak strength, residual strength, post-peak softening features, damage evolution features, and plastic zone expansion features.
[0051] In one specific embodiment, the elastic modulus can be determined by the slope of the elastic stage of the stress-strain curve; the peak strength can be determined by the maximum stress value in the stress-strain curve; the residual strength can be determined by the stress value in the post-peak stabilization stage; the post-peak softening characteristics can be determined by the difference between the peak strength and the residual strength or the stress decrease rate; and the damage evolution characteristics can be determined by the change of the number of damage elements, the proportion of damage area, the proportion of damage volume, or the average value of damage variables with the loading step.
[0052] In one alternative implementation, the rock mechanical response characteristics can be organized into a mechanical response characteristic vector Y, such as... Figure 2 As shown, its expression form is: Y = [E, σ] p , σ r ,η,D,P]; In the formula, E represents the elastic modulus; σ p Indicates peak intensity; σ r η represents the residual strength; D represents the post-peak softening coefficient; P represents the damage degree index; and P represents the plastic zone expansion index. By constructing the mechanical response eigenvector Y, the complex response results obtained from finite element simulation can be transformed into a storable, comparable, and retrievable parametric form.
[0053] This embodiment then performs crack propagation path feature extraction. Based on the damage field, crack propagation field, and failure element distribution results obtained in step S4, crack propagation path features are extracted. The crack propagation path features include one or more of the following: crack initiation location, crack propagation direction, crack penetration path, damage concentration area, weak failure area, and failure mode.
[0054] In one specific implementation, the crack initiation location is determined based on the location of the element that first reaches the failure criterion or damage threshold during the finite element simulation; the crack propagation direction is determined based on the spatial connection direction of adjacent failure elements, the extension direction of the damage zone, or the propagation direction of the crack tip; the crack penetration path is determined based on the spatial distribution of continuous failure elements, continuous damage elements, or crack elements; and the damage concentration area is determined based on the area with high damage variables or dense distribution of failure elements.
[0055] To facilitate the subsequent establishment of correspondences, the crack propagation path can be transformed into a crack propagation path feature vector C, such as... Figure 2 As shown, its expression form is: C = [x c y c , z c θ c , φ c L c R c M c ]; In the formula, x c y c z c θ represents the coordinates of the crack initiation location or the center of the main crack; c φ c Main crack propagation direction angle; L c R represents the crack propagation length. c Indicates the degree of crack penetration; M c This indicates the failure mode type. For two-dimensional digital core models, x can be used. c y c and θ c This indicates the location of crack initiation and the direction of crack propagation; for three-dimensional digital core models, x can be used. c y c z c and θ c φ c It indicates the spatial location and direction of crack propagation.
[0056] S6. Based on the rock structure characteristic parameters of multiple different digital core structure models and their corresponding rock mechanical response characteristics and crack propagation path characteristics, establish the correspondence between rock structure characteristics and rock mechanical response characteristics and crack propagation path characteristics; In an optional embodiment of the present invention, step S6 establishes the correspondence between structural features and mechanical response and crack propagation. Steps S1 to S5 are repeated for multiple different digital core structural models to obtain the structural feature vector X, mechanical response feature vector Y, and crack propagation path feature vector C corresponding to each digital core model. Subsequently, X, Y, and C of the same digital core model are associated and stored to form the correspondence between structural features and mechanical response and crack propagation. This step corresponds to... Figure 1 The "establishment of the correspondence between structural characteristics and mechanical response, crack propagation" is as follows: Figure 2 As shown.
[0057] In one specific implementation, each digital core model can generate a corresponding set of data: X → Y, C; Specifically, the structural feature X of a digital core corresponds to the mechanical response feature Y and the crack propagation path feature C. By generating multiple sets of corresponding data from multiple digital core models, a structural feature sample library, a mechanical response result library, and a crack propagation path result library can be established, providing a usable data foundation for predicting new rock samples.
[0058] In one optional implementation, multiple digital core models can be classified into structural types based on structural characteristic parameters such as mineral composition, pore and fracture distribution, bedding orientation, and interface density. Digital core models with similar mineral composition, porosity, fracture density, bedding orientation, or interface density are grouped into the same structural type, and typical stress-strain response, peak intensity range, damage evolution law, crack initiation region, dominant propagation path, and failure mode under this structural type are recorded.
[0059] S7. Obtain a digital core structure model of the rock sample to be tested and extract rock structure feature parameters. Based on the weights corresponding to each rock structure feature parameter calibrated with the objective of jointly minimizing the prediction error of mechanical response, crack propagation path, and failure mode, calculate the weighted structural distance between the rock sample to be tested and each existing digital core structure model to determine the structural similarity between the rock sample to be tested and each existing digital core structure model. For each existing digital core structure model whose structural similarity meets a preset threshold, cluster them based on the spatial neighborhood overlap between each crack propagation path to determine the predicted rock mechanical response and crack propagation path, such as... Figure 3 As shown.
[0060] In an optional embodiment of the present invention, step S7, which aims to jointly minimize the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error, includes calibrating the weights corresponding to each rock structure characteristic parameter, which includes: One of the existing digital core structure models is used as a verification sample, and the remaining existing digital core structure models are used as reference samples. The weighted structural distance and structural similarity between the verification sample and each reference sample are calculated under given candidate weights. The reference sample with the highest structural similarity is selected as the prediction reference. The verification sample is predicted based on the rock mechanical response characteristics and crack propagation path characteristics corresponding to the prediction reference. The weighted sum of the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the verification sample is used as the comprehensive verification error. Under the constraint that the weights are non-negative and the sum of the weights is 1, the weights corresponding to each rock structural characteristic parameter that minimize the comprehensive verification error are obtained.
[0061] Step S7 uses the weighted sum of the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the verification sample as the comprehensive verification error, including: For each round of cross-validation, the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the validation sample are calculated respectively. The mechanical response prediction error is calculated as follows: the norm square of the difference between the actual mechanical response feature vector and the predicted mechanical response feature vector of the verification sample is calculated, and the arithmetic mean of the norm squares obtained from each round of cross-validation is taken. The crack propagation path prediction error is calculated as follows: calculate the difference between the predicted crack propagation path and the actual crack propagation path of the verification sample and subtract the path similarity between the two. Then, take the arithmetic mean of the differences obtained from each round of cross-validation. The calculation method for the damage mode prediction error is as follows: when the predicted damage mode type of the verification sample is inconsistent with the actual damage mode type, the value is a preset first value; when they are consistent, the value is a preset second value; and the arithmetic mean of the values obtained from each round of cross-validation is taken. The comprehensive verification error is obtained by multiplying the mechanical response prediction error, the crack propagation path prediction error, and the failure mode prediction error by their respective weights and then summing them.
[0062] Step S7 clusters existing digital core structure models whose structural similarity meets a preset threshold based on the spatial neighborhood overlap between crack propagation paths to determine the predicted rock mechanical response and crack propagation path, including: The crack propagation paths of existing digital core structural models whose structural similarity meets a preset threshold are transformed to a unified normalized coordinate system, and the crack path neighborhood is defined with each crack propagation path as the center and a preset distance threshold. The path similarity is calculated based on the ratio of the intersection to the union of the neighborhoods of each crack path, and paths with a path similarity greater than a preset path threshold are classified as the same crack path type. Calculate the weighted frequency of occurrence for each crack path type, and determine the crack path type with the highest weighted frequency as the predicted master crack propagation path; Calculate the weighted probability of the crack path covering each spatial location in the normalized coordinate system, and determine the region with the weighted probability of occurrence greater than a preset probability threshold as the predicted weak area of damage.
[0063] This embodiment first performs structural feature input and similarity matching of the rock sample to be tested. For the rock sample to be tested, its digital core structure model is first obtained, and its structural feature parameters are extracted according to step S2 to form a new structural feature vector X. new Then X new By matching the established correspondence between structural features and mechanical response and crack propagation, the structural similarity between the new rock sample and the existing digital core sample is determined.
[0064] In one specific implementation, the structural feature vector of the rock sample to be tested can be represented as: X new = [x1, x2, x3, ..., x n ]; The structural feature vector of the j-th digital core sample can be represented as: X j =[x1',x2',x3',...,x n ']; Where x1, x2, x3, ..., x n The normalized structural characteristic parameters of the new rock sample are represented by x1', x2', x3', ..., x. n 'Indicates the normalized structural characteristic parameters of existing digital core samples.
[0065] In one specific implementation, the structural distance dⱼ between the rock sample to be tested and the j-th existing digital core sample can be calculated using weighted Euclidean distance, and its expression is: ; In the formula, dⱼ represents the structural distance between the rock sample to be tested and the j-th existing digital core sample; This represents the i-th normalized structural feature parameter of the rock sample to be tested. This represents the i-th normalized structural feature parameter of the j-th existing digital core sample; Let represent the weight of the i-th structural feature parameter; n represents the number of structural feature parameters. The weight coefficients satisfy: wᵢ≥0, i=1, 2,...,n; .
[0066] In one specific implementation, the weights of the structural feature parameters are determined by cross-validation prediction errors of existing digital core samples. Assuming the number of existing digital core samples is m, the r-th sample is used as a validation sample, and the remaining m-1 samples are used as reference samples. Given a set of candidate weights, the structural distance and structural similarity between the validation sample and each reference sample are calculated. One or more reference samples with the highest structural similarity are selected to predict the mechanical response and crack propagation path of the r-th validation sample.
[0067] For continuous response parameters such as elastic modulus, peak strength, residual strength, and damage degree, a weighted average can be performed based on the structural similarity of similar samples to obtain the predicted mechanical response feature vector of the r-th validation sample. For crack propagation paths, the crack path similarity can be calculated based on the degree of spatial overlap between the predicted path and the actual crack path. For type-based results such as damage modes, it can be determined whether the predicted damage mode is consistent with the actual damage mode.
[0068] The overall verification error J corresponding to the candidate weights can be expressed as: ; Among them, mechanical response prediction error It can be represented as: ; Crack propagation path prediction error It can be represented as: ; Damage mode prediction error It can be represented as: ; In the formula, and These are the actual mechanical response feature vector and the predicted mechanical response feature vector of the r-th verification sample, respectively; Let be the similarity between the predicted crack path and the actual crack path of the r-th validation sample, and its value ranges from 0 to 1. and These represent the actual failure mode and the predicted failure mode, respectively; I(·) is the indicator function, which takes the value 1 when the condition in parentheses is true, and 0 otherwise; α, β and γ are the weights corresponding to the mechanical response, crack propagation path and failure mode prediction error, respectively, and satisfy α≥0, β≥0, γ≥0 and α+β+γ=1.
[0069] Under the conditions that wᵢ≥0 and the sum of the weights of each structural feature parameter is 1, the structural feature parameter weights that minimize the comprehensive verification error J are obtained by using grid search, constraint optimization, genetic algorithm, or particle swarm optimization. ; In the formula, w* represents the optimal weight set of the final determined structural feature parameters. Using these optimal weights to calculate the structural distance allows the structural similarity ranking to simultaneously consider the prediction accuracy of rock mechanical response, crack propagation path, and failure mode.
[0070] When the existing samples only contain mechanical response results, β=0 and γ=0 can be set, and the weights can be determined only based on the mechanical response prediction error. When there are not enough existing samples for cross-validation, each structural feature parameter can be set to equal weight, or the initial weights can be determined based on the parameter sensitivity analysis results, and the weights can be recalibrated as the digital core sample library expands.
[0071] Furthermore, the structural similarity S is calculated based on the structural distance d, and its expression is as follows: S = 1 / (1 + d); In the formula, S represents the degree of structural similarity. The larger the S value, the higher the structural similarity between the new rock sample and the existing digital core samples. For multiple existing digital core samples, S1, S2, ..., S can be calculated separately. k And select the structural type or sample with the highest similarity as the basis for prediction, such as Figure 4 As shown, its expression form is: S max = max(S1, S2, ..., S...) k ); In the formula, S max The maximum structural similarity is indicated by k; k represents the number of existing digital core samples or structural types.
[0072] This embodiment then performs prediction and correspondence update of the rock sample to be tested. Based on the obtained structural similarity, it determines whether the similarity between the new rock sample and the existing structural types meets preset requirements, such as... Figure 3 As shown.
[0073] When the structural similarity meets the preset requirements, i.e., S is greater than or equal to the preset threshold, the finite element response results of the corresponding structural type are used to predict the mechanical response and crack propagation path of the new rock sample. For continuous mechanical response parameters such as elastic modulus, peak strength, and residual strength, the weighted average of similar samples can be used as the prediction result. For spatial or typological results such as crack initiation region, dominant propagation path, and failure mode, the crack paths of similar samples are first transformed to a unified normalized coordinate system. For two-dimensional digital core models, the crack path coordinates are normalized by model length and width respectively; for three-dimensional digital core models, the crack path coordinates are normalized by model length, width, and height respectively, to eliminate the influence of different model sizes on the spatial comparison of paths.
[0074] Let the main crack path extracted from the j-th similar sample be Cⱼ. Set a distance threshold r centered on the crack path. Define the region whose distance from the crack path Cⱼ is no greater than r as the crack path neighborhood Ωⱼ(r), whose expression is: ; In the formula, Ω represents the normalized digital core model region; x represents the spatial position in the model; d(x, Cⱼ) is the shortest distance from position x to the j-th crack path Cⱼ; and r is the radius of the crack path neighborhood. For a two-dimensional digital core model, Ωⱼ(r) represents the band-shaped region formed along the crack path; for a three-dimensional digital core model, Ωⱼ(r) represents the spatial neighborhood formed along the crack line, crack band, or crack surface. The neighborhood radius r can be determined based on the characteristic size of the finite element unit. In one optional embodiment, it can be taken as 1 to 3 times the characteristic size of the finite element unit.
[0075] Path similarity between crack path j and crack path k It can be determined based on the degree of overlap between the crack path neighborhoods of the two, and its expression is: ; In the formula, Let |Ωⱼ(r)∩Ω be the path similarity between the j-th crack path and the k-th crack path; k (r)| represents the area or volume of the overlapping region of the two crack paths; |Ωⱼ(r)| and |Ω k (r)| represents the area or volume of the neighborhood of the two crack paths, respectively. The value ranges from 0 to 1. The closer the value is to 1, the more similar the two crack paths are in terms of spatial location and extension shape; the closer the value is to 0, the lower the degree of spatial overlap between the two crack paths.
[0076] when Greater than or equal to the preset path similarity threshold At that time, the j-th crack path and the k-th crack path are classified as the same crack path type. The path similarity threshold... The calibration can be performed based on verification samples with known failure modes, so that the similarity between crack paths of the same type is higher than the threshold, while the similarity between crack paths of different types is lower than the threshold.
[0077] Crack path clustering is performed on m existing samples that meet the structural similarity requirements. Let the g-th class of crack paths contain... For each sample, the frequency of occurrence of this type of crack path is... It can be represented as: ; In the formula, Let be the frequency of occurrence of the g-th crack path; is the number of samples belonging to the g-th crack path; m is the total number of similar samples participating in the prediction.
[0078] When considering the differences in structural similarity between different existing samples and the rock sample to be tested, the degree of structural similarity can be used as a basis. Determine the weight of the j-th existing sample. : ; In the formula, Sⱼ represents the structural similarity between the rock sample to be tested and the j-th existing sample; the summation range is all m similar samples; wⱼ represents the normalized weight corresponding to the j-th existing sample, and the sum of the weights of all samples is 1.
[0079] Weighted frequency of occurrence of crack path of type g It can be represented as: ; In the formula, Let g be the set of samples contained in the g-th crack path; represents the weighted frequency of occurrence of the g-th crack path. The crack path type with the highest weighted frequency is selected as the controlling crack propagation path type for the new rock sample.
[0080] To statistically analyze the frequency of crack paths at different spatial locations in the model, for any spatial location x, if location x is within the neighborhood Ωⱼ(r) of the j-th crack path, let the path indicator value Iⱼ(x) = 1; otherwise, let Iⱼ(x) = 0. The probability P(x) of a crack path occurring at spatial location x can be expressed as: ; In the formula, P(x) is the weighted probability of the crack path occurring at position x, and its value ranges from 0 to 1; the larger the P(x) is, the more likely the crack in the similar sample is to pass through that position. The region where P(x) is greater than or equal to the preset probability threshold is determined as the predicted crack propagation advantage region, and the center line or center plane of the continuously distributed region with high P(x) is determined as the predicted main crack propagation path.
[0081] For typological results such as damage modes, the weighted frequency of each damage mode in similar samples is calculated, and the damage mode with the highest weighted frequency is taken as the predicted damage mode for the new rock sample. Figure 3 and Figure 4 As shown.
[0082] In an optional embodiment of the present invention, step S7 further includes: When the structural similarity does not meet the preset threshold, the digital core structure model of the new rock sample is assigned parameters, converted into a mechanical numerical calculation model, boundary conditions are applied, and numerical simulation is carried out to extract its rock mechanical response characteristics and crack propagation path characteristics. The rock structural characteristic parameters, rock mechanical response characteristics, and crack propagation path characteristics of the rock sample to be tested are added to the correspondence, the correspondence is updated and expanded, and the weights corresponding to each rock structural characteristic parameter are recalibrated.
[0083] When the structural similarity does not meet the preset requirements, i.e., S is less than the preset threshold, digital core finite element simulation is performed on the new rock sample to extract its mechanical response characteristics and crack propagation path characteristics, and the structural feature vector X of the sample is then used. new Mechanical response eigenvector Y new and crack propagation path feature vector C new By incorporating the established correspondences, the correspondences between structural features and mechanical responses, as well as crack propagation, can be updated and expanded.
[0084] This embodiment then outputs the prediction results. Based on the matched, recalled, or supplemented simulation results, it outputs the predicted rock mechanical response and crack propagation path of the new rock sample. The predicted mechanical response includes one or more of the following: stress-strain response, elastic modulus, peak strength, residual strength, post-peak softening characteristics, damage evolution characteristics, and plastic zone propagation characteristics; the predicted crack propagation path includes one or more of the following: crack initiation location, crack propagation direction, crack penetration path, weak failure region, and failure mode. Figure 3 and Figure 4 As shown.
[0085] In one specific implementation, the prediction results can be output in the form of curves, contour maps, path diagrams, or parameter tables. Specifically, the stress-strain response is output as a curve; the elastic modulus, peak strength, and residual strength are output as parameter tables; the damage concentration area is output as a contour map; the crack propagation path is output as a path line or a connected region of the failure unit; and the failure mode is output as types such as failure propagation along mineral interfaces, failure propagation through mineral grains, failure propagation along bedding planes, failure propagation along existing fractures, pore-fracture interconnection failure, tensile-dominated failure, shear-dominated failure, or combined tensile-shear failure.
[0086] Through the above implementation methods, this invention can transform digital core finite element simulation results from single calculations into reusable predictive data. For new rock samples, after obtaining their digital core structural characteristics, their mechanical response and crack propagation path can be predicted based on the correspondence between existing structural characteristics and mechanical response and crack propagation. When the similarity between the new rock sample and existing structural types is insufficient, the correspondence can be updated by supplementing finite element simulations, thereby improving the applicability and scalability of the method.
[0087] It should be noted that the finite element simulation described in this invention is not limited to any specific finite element software or constitutive model. Those skilled in the art can select elastic damage models, elastoplastic damage models, cohesive zone models, phase-field fracture models, or other finite element models capable of characterizing the rock damage and fracture process, depending on the rock type, loading method, and research objective. As long as it can obtain the rock mechanical response and crack propagation path based on the digital core structure model, and further establish the correspondence between structural features and finite element response results, it falls within the scope of protection of this invention.
[0088] This invention also provides a rock mechanical response and crack propagation path prediction system, which applies the rock mechanical response and crack propagation path prediction method described above, including: The structural model acquisition module is used to acquire a digital core structural model of a rock sample; the digital core structural model includes one or more microstructural units among mineral grains, pores, fractures, bedding, and mineral interfaces; The structural feature extraction module is used to identify, classify, and label the mesoscopic structural units in the digital core structural model and extract rock structural feature parameters. The numerical calculation model construction module is used to assign values to at least one of the material parameters, interface parameters, damage parameters, or weakening parameters of the digital core structure model according to the mechanical properties of the microstructural units, and transform it into a mechanical numerical calculation model. The numerical simulation module is used to apply loading or unloading boundary conditions to the mechanical numerical calculation model to carry out numerical simulation and obtain the numerical simulation response results of the rock under load. The response feature extraction module is used to extract rock mechanical response features and crack propagation path features from the numerical simulation response results. The correspondence establishment module is used to establish the correspondence between rock structure features and rock mechanical response features and crack propagation path features based on rock structure feature parameters and their corresponding rock mechanical response features and crack propagation path features from multiple different digital core structure models. The weighted similarity matching and prediction module is used to acquire digital core structure models of the rock samples to be tested and extract rock structure feature parameters. Based on the weights corresponding to each rock structure feature parameter calibrated with the goal of jointly minimizing the prediction error of mechanical response, crack propagation path, and failure mode, the module calculates the weighted structural distance between the rock samples to be tested and each existing digital core structure model, and determines the degree of structural similarity between the rock samples to be tested and each existing digital core structure model. For each existing digital core structure model whose structural similarity meets the preset threshold, clustering is performed based on the degree of spatial neighborhood overlap between each crack propagation path to determine the predicted rock mechanical response and crack propagation path.
[0089] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0090] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0091] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0092] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.
[0093] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.
Claims
1. A method for predicting rock mechanical response and crack propagation path, characterized in that, Includes the following steps: A digital core structure model of a rock sample is obtained; the digital core structure model includes one or more microstructural units such as mineral grains, pores, fractures, bedding, and mineral interfaces; The microstructural units in the digital core structure model are identified, classified, and labeled, and rock structure characteristic parameters are extracted. Based on the mechanical properties of the microstructural units, at least one of the material parameters, interface parameters, damage parameters, or weakening parameters is assigned to the digital core structure model, and it is then transformed into a mechanical numerical calculation model. Loading or unloading boundary conditions are applied to the mechanical numerical calculation model to carry out numerical simulation and obtain the numerical simulation response results of the rock under load. Extract rock mechanical response characteristics and crack propagation path characteristics from the numerical simulation response results; Based on the rock structure characteristic parameters of multiple different digital core structure models and their corresponding rock mechanical response characteristics and crack propagation path characteristics, the correspondence between rock structure characteristics and rock mechanical response characteristics and crack propagation path characteristics is established. A digital core structure model of the rock sample to be tested is obtained and the rock structure feature parameters are extracted. Based on the weights corresponding to each rock structure feature parameter calibrated with the goal of jointly minimizing the mechanical response prediction error, crack propagation path prediction error and failure mode prediction error, the weighted structural distance between the rock sample to be tested and each existing digital core structure model is calculated, and the structural similarity between the rock sample to be tested and each existing digital core structure model is determined. For existing digital core structure models whose structural similarity meets a preset threshold, clustering is performed based on the degree of spatial neighborhood overlap between each crack propagation path to determine the predicted rock mechanical response and crack propagation path.
2. The method for predicting rock mechanical response and crack propagation path according to claim 1, characterized in that, The rock structural characteristic parameters include one or more of the following: mineral composition, brittle mineral content, mineral grain area, equivalent diameter of mineral grains, mineral grain shape coefficient, orientation of the major and minor axes of mineral grains, contact relationship of mineral grains, porosity, pore distribution, fracture density, fracture length, fracture dip angle, fracture connectivity, bedding orientation, bedding spacing, mineral interface density, and distribution of weak interfaces.
3. The method for predicting rock mechanical response and crack propagation path according to claim 1, characterized in that, Based on the mechanical properties of the microstructural units, at least one of the following parameters—material parameters, interface parameters, damage parameters, or weakening parameters—is assigned to the digital core structure model: Mineral material parameters are assigned to the mineral particle region, including one or more of the following: elastic modulus, Poisson's ratio, compressive strength, tensile strength, cohesion, internal friction angle, fracture energy, and damage evolution parameters. Assign void properties or low stiffness weakening parameters to porous regions; Assign crack surface properties, contact properties, or weakening parameters to the fractured region; Assign bedding interface parameters to the bedding region; The mineral interface region is assigned one or more of the following parameters: interface stiffness, interface strength, interface friction parameter, interface fracture parameter, or interface damage parameter.
4. The method for predicting rock mechanical response and crack propagation path according to claim 1, characterized in that, Based on the mechanical properties of the microstructural units, at least one of the following parameters—material parameters, interface parameters, damage parameters, or weakening parameters—is assigned to the digital core structure model: Mechanical parameters of mineral particles and mineral interfaces are obtained through micro-area mechanical testing. The locations of micro-area mechanical testing are spatially registered with mineral phase distribution maps to determine the mineral type or mineral interface type corresponding to each test point. Using the macroscopic test results of rocks as the calibration target, inversion calibration is performed by adjusting the material parameters and interface parameters to ensure that the numerical simulation results simultaneously meet the requirements for macroscopic mechanical response error and microscopic crack propagation characteristic error.
5. The method for predicting rock mechanical response and crack propagation path according to claim 1, characterized in that, The rock mechanical response characteristics include one or more of the following: elastic modulus, peak strength, residual strength, Poisson's ratio, stress-strain curve morphology, and damage evolution characteristics. The crack propagation path characteristics include one or more of the following: crack initiation location coordinates, initial crack propagation direction angle, crack propagation path spatial coordinate sequence, total crack length, propagation ratio along mineral interface, propagation ratio through mineral grains, and failure mode type.
6. The method for predicting rock mechanical response and crack propagation path according to claim 1, characterized in that, The weights for each rock structural characteristic parameter are determined with the objective of jointly minimizing the prediction errors of mechanical response, crack propagation path, and failure mode. One of the existing digital core structure models is used as a verification sample, and the remaining existing digital core structure models are used as reference samples. The weighted structural distance and structural similarity between the verification sample and each reference sample are calculated under given candidate weights. The reference sample with the highest structural similarity is selected as the prediction reference. The verification sample is predicted based on the rock mechanical response characteristics and crack propagation path characteristics corresponding to the prediction reference. The weighted sum of the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the verification sample is used as the comprehensive verification error. Under the constraint that the weights are non-negative and the sum of the weights is 1, the weights corresponding to each rock structural characteristic parameter that minimize the comprehensive verification error are obtained.
7. The method for predicting rock mechanical response and crack propagation path according to claim 6, characterized in that, The comprehensive verification error is the weighted sum of the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the verification sample, including: For each round of cross-validation, the mechanical response prediction error, crack propagation path prediction error, and failure mode prediction error of the validation sample are calculated respectively. The mechanical response prediction error is calculated as follows: the norm square of the difference between the actual mechanical response feature vector and the predicted mechanical response feature vector of the verification sample is calculated, and the arithmetic mean of the norm squares obtained from each round of cross-validation is taken. The crack propagation path prediction error is calculated as follows: calculate the difference between the predicted crack propagation path and the actual crack propagation path of the verification sample and subtract the path similarity between the two. Then, take the arithmetic mean of the differences obtained from each round of cross-validation. The calculation method for the damage mode prediction error is as follows: when the predicted damage mode type of the verification sample is inconsistent with the actual damage mode type, the value is a preset first value; when they are consistent, the value is a preset second value; and the arithmetic mean of the values obtained from each round of cross-validation is taken. The comprehensive verification error is obtained by multiplying the mechanical response prediction error, the crack propagation path prediction error, and the failure mode prediction error by their respective weights and then summing them.
8. The method for predicting rock mechanical response and crack propagation path according to claim 1, characterized in that, For existing digital core structure models whose structural similarity meets a preset threshold, clustering is performed based on the spatial neighborhood overlap between crack propagation paths to determine the predicted rock mechanical response and crack propagation path, including: The crack propagation paths of existing digital core structural models whose structural similarity meets a preset threshold are transformed to a unified normalized coordinate system, and the crack path neighborhood is defined with each crack propagation path as the center and a preset distance threshold. The path similarity is calculated based on the ratio of the intersection to the union of the neighborhoods of each crack path, and paths with a path similarity greater than a preset path threshold are classified as the same crack path type. Calculate the weighted frequency of occurrence for each crack path type, and determine the crack path type with the highest weighted frequency as the predicted master crack propagation path; Calculate the weighted probability of the crack path covering each spatial location in the normalized coordinate system, and determine the region with the weighted probability of occurrence greater than a preset probability threshold as the predicted weak area of damage.
9. The method for predicting rock mechanical response and crack propagation path according to claim 1, characterized in that, Also includes: When the structural similarity does not meet the preset threshold, the digital core structure model of the new rock sample is assigned parameters, converted into a mechanical numerical calculation model, boundary conditions are applied, and numerical simulation is carried out to extract its rock mechanical response characteristics and crack propagation path characteristics. The rock structural characteristic parameters, rock mechanical response characteristics, and crack propagation path characteristics of the rock sample to be tested are added to the correspondence, the correspondence is updated and expanded, and the weights corresponding to each rock structural characteristic parameter are recalibrated.
10. A rock mechanical response and crack propagation path prediction system, employing the rock mechanical response and crack propagation path prediction method as described in any one of claims 1 to 9, characterized in that, include: The structural model acquisition module is used to acquire a digital core structural model of a rock sample; the digital core structural model includes one or more microstructural units among mineral grains, pores, fractures, bedding, and mineral interfaces; The structural feature extraction module is used to identify, classify, and label the mesoscopic structural units in the digital core structural model and extract rock structural feature parameters. The numerical calculation model construction module is used to assign values to at least one of the material parameters, interface parameters, damage parameters, or weakening parameters of the digital core structure model according to the mechanical properties of the microstructural units, and transform it into a mechanical numerical calculation model. The numerical simulation module is used to apply loading or unloading boundary conditions to the mechanical numerical calculation model to carry out numerical simulation and obtain the numerical simulation response results of the rock under load. The response feature extraction module is used to extract rock mechanical response features and crack propagation path features from the numerical simulation response results. The correspondence establishment module is used to establish the correspondence between rock structure features and rock mechanical response features and crack propagation path features based on rock structure feature parameters and their corresponding rock mechanical response features and crack propagation path features from multiple different digital core structure models. The weighted similarity matching and prediction module is used to acquire the digital core structure model of the rock sample to be tested and extract the rock structure feature parameters. Based on the weights corresponding to each rock structure feature parameter calibrated with the goal of jointly minimizing the prediction error of mechanical response, the prediction error of crack propagation path and the prediction error of failure mode, the weighted structural distance between the rock sample to be tested and each existing digital core structure model is calculated, and the degree of structural similarity between the rock sample to be tested and each existing digital core structure model is determined. For existing digital core structure models whose structural similarity meets a preset threshold, clustering is performed based on the degree of spatial neighborhood overlap between each crack propagation path to determine the predicted rock mechanical response and crack propagation path.
