Shale reservoir carbon dioxide fracturing test evaluation method under multi-field coupling condition

Abstract

The application discloses a shale reservoir carbon dioxide fracturing test evaluation method under a multi-field coupling condition and belongs to the technical field of oil and natural gas engineering. The method comprises multi-field coupling data integration and analysis, crack propagation prediction, multi-scale dynamic effect evaluation and optimized decision output. The method realizes accurate modeling of reservoir states in the fracturing process by fusing multi-field heterogeneous data of pressure, temperature, stress, fluid and chemical reaction and constructing a unified high-dimensional feature description with a self-adaptive weight mechanism. A distributed energy flow tracking algorithm is adopted to predict crack propagation paths and fracture probabilities in real time. Through a hierarchical multi-field feedback algorithm, fracturing effects are quantified in a progressive manner at micro, meso and macro levels to dynamically reflect changes in reservoir structure and productivity. Finally, the optimal fracturing parameter combination and decision suggestions are output in combination with an evolutionary optimization mechanism. The application comprehensively improves the intelligent level of crack growth prediction, reservoir reconstruction effect evaluation and fracturing parameter optimization.

Description

Technical Field

[0001] This invention belongs to the field of oil and gas engineering technology, and more specifically relates to a method for evaluating carbon dioxide fracturing tests in shale reservoirs under multi-field coupling conditions. Background Technology

[0002] With the widespread development of shale oil and gas resources, fracturing technology has become a core means to improve the production capacity and economic benefits of shale reservoirs. Traditional fracturing processes mostly use hydraulics as the main medium, but against the backdrop of complex geological environments and increasingly stringent environmental requirements, carbon dioxide fracturing has gradually shown unique advantages. It can not only enhance crack propagation and reservoir stimulation, but also synergistically achieve green development goals such as carbon sequestration. However, shale reservoirs have complex structures and variable compositions. The fracturing process involves the coupling of multiple field variables such as pressure, temperature, stress, fluid, and chemical reactions. The complex inter-field interactions directly affect crack propagation, reservoir seepage, and ultimately, production capacity enhancement. Most existing evaluation methods focus on single-field or dual-field analysis, lacking systematic integration and dynamic prediction of reservoir behavior under multi-field coupling conditions. This results in limited optimization space for fracturing schemes, making it difficult to fully realize the production enhancement potential and controllable reservoir stimulation efficiency of carbon dioxide fracturing.

[0003] Furthermore, shale reservoirs exhibit multi-scale coupled feedback between microscopic pores and fractures, mesoscopic rock masses, and the macroscopic reservoir as a whole. Single-scale evaluations are insufficient to comprehensively reflect crack growth, energy-driven processes, and production response. With the maturation of multi-field data acquisition and analysis technologies, constructing methods for crack propagation prediction and fracturing effect evaluation based on high-dimensional multi-field data has become a pressing challenge for the industry. Existing methods suffer from insufficient multi-field data fusion, inflexible inter-field weight adjustment, and limited crack prediction accuracy, further impacting reservoir production enhancement and intelligent decision-making optimization of fracturing processes. Moreover, traditional fracturing parameter optimization methods lack multi-field feedback and adaptive evolution mechanisms, offering limited guidance for fracture morphology prediction and production enhancement.

[0004] Therefore, there is an urgent need to develop a novel carbon dioxide fracturing test evaluation method for shale reservoirs that can integrate heterogeneous data from multiple fields, dynamically adjust the weights of inter-field influences, model crack propagation based on energy flow, and possess multi-scale progressive feedback evaluation and intelligent parameter optimization capabilities at the micro-meta-macro levels. This method aims to improve reservoir stimulation efficiency, optimize the fracturing process, achieve green production and carbon emission reduction goals, and promote the high-quality development of shale oil and gas development technology. Summary of the Invention

[0005] This invention aims to address the technical challenges of complex multi-field (pressure, temperature, stress, fluid, and chemical reaction) coupled behavior during carbon dioxide fracturing in shale reservoirs, the inability of existing evaluation methods to systematically integrate heterogeneous data from multiple fields and dynamically predict crack propagation paths, the imperfect multi-scale evaluation of fracturing effects, and the lack of multi-field feedback and intelligent decision-making in fracturing parameter optimization. The goal is to achieve scientific quantification of shale reservoir fracturing behavior and production response, accurate prediction of fracture morphology, and intelligent parameter optimization, thereby improving reservoir stimulation efficiency and production enhancement.

[0006] To achieve the above objectives, the present invention employs the following technical solution: the method comprising: Multi-field coupled data integration and preliminary analysis: Collecting multi-field data involved in the fracturing process of shale reservoirs, including but not limited to pressure field, temperature field, stress field, fluid field, and chemical reaction field. Based on the spatiotemporal correlation, the data of each field are transformed into high-dimensional feature vectors of a unified spatiotemporal grid, and the influence of each field is dynamically adjusted through an adaptive weighting mechanism to obtain a description of the shale reservoir state under multi-field coupling conditions. A multi-field driven crack propagation prediction model was constructed. A distributed energy flow tracking algorithm was used to simulate the dynamic distribution and flow of energy along the crack tip and within the region due to multi-field interaction in the fracturing test. Combined with multi-field time-varying driving analysis, the crack propagation path, velocity and fracture probability were predicted in real time. Multi-scale evaluation of fracturing effect under coupled field: A layered multi-field feedback evaluation algorithm is designed. It calculates the changes of reservoir porosity, permeability and stress redistribution index by carbon dioxide fracturing at three levels: micro (pore-fracture), meso (rock block) and macro (reservoir as a whole). The system outputs decision recommendations, combining experimental parameters (injection rate, carbon dioxide concentration, temperature) and multi-field feedback, and searches for the optimal experimental scheme through an adaptive evolutionary mechanism, effectively improving crack propagation and reservoir stimulation efficiency. It outputs optimized decision-making recommendations for specific shale reservoirs and multiple field conditions, including fracture morphology prediction, fracturing parameter adjustment schemes, and production potential estimation.

[0007] In one scheme, the multi-field coupling data integration and preliminary analysis includes: based on the heterogeneous data of pressure field, temperature field, stress field, fluid field and chemical reaction field involved in the fracturing process of shale reservoir, through a dynamic correlation feature fusion algorithm, the data of various fields are processed by spatiotemporal mapping and corresponding interpolation or dimensionality reduction operators, so that all data are uniformly mapped to a preset spatial grid and time node. The dynamic correlation feature fusion algorithm dynamically adjusts the weight of each field's influence on reservoir characteristics at each spatiotemporal grid node by normalizing variance and combining it with an adaptive adjustment factor generated by the collaborative relationship graph, based on a set of spatiotemporal correlation coefficients. Multi-field data, after standardization and weight adjustment, are fused into a unified high-dimensional feature vector to enhance the ability to describe reservoir conditions.

[0008] In one scheme, the construction of a multi-field driven crack propagation prediction model includes: using a distributed energy flow tracing algorithm to map the multi-field feature information of each grid point into an energy tensor containing different types of energy components, including elastic potential energy, thermal energy, fluid kinetic energy and chemical reaction energy, and dynamically calculating the multi-field energy components through a nonlinear mapping function and a variable weighting factor. The distributed energy flow tracing algorithm uses an energy gradient-driven mechanism to predict the direction and speed of crack propagation in real time. Specifically, for each moment of the crack tip and its neighborhood, it automatically selects the region of maximum energy density as the direction of crack tip growth by comparing the energy flow density field point by point, and considers the regulating effect of dynamic synergy of multiple energy fields on crack propagation speed.

[0009] In one scheme, the multi-scale evaluation of fracturing effect under coupled field includes: using a hierarchical multi-field feedback evaluation algorithm to carry out multi-field coupled quantitative evaluation of carbon dioxide fracturing effect at the micro, meso and macro (storage) levels respectively. At the microscale, the algorithm automatically calculates the increments of parameters such as porosity and permeability based on the crack network structure and local coupling characteristics, and combines multi-field energy tensors and crack growth paths to describe the changes in seepage channels and the reorganization of pore structures. At the mesoscale, rock blocks are considered as complex structural units. The mechanical behavior and stress distribution of each block are adaptively adjusted according to the feedback results of the micro-level, dynamically reflecting the changes in block stress redistribution, crack clustering effect and energy production response. At the macro scale, the overall reservoir behavior is evaluated based on feedback from the mesoscopic level. A multi-field response integral model is used to quantify the reservoir's total porosity, productivity enhancement, stress spatial equilibrium, and fracturing zone expansion indicators. Progressive feedback and automatic linkage are achieved among the three levels, comprehensively and scientifically quantifying the combined impact of multi-field synergy on reservoir fracturing behavior and structural reconstruction.

[0010] In one approach, the output decision recommendations include: based on a coupled-field response intelligent optimization algorithm, integrating multi-field and multi-scale evaluation results, and setting key experimental parameters such as injection rate, carbon dioxide concentration, and temperature as adjustable optimization vectors. Through evolutionary population search and adaptive evolution mechanism, the fracture morphology, fracturing parameters and production capacity indicators are fed back and optimized in real time. The target extreme values ​​of fracture propagation, branching simulation and controllable reservoir stimulation are fitted to form a multi-objective fitness function. The parameter weights are automatically adjusted by integrating fracture morphology fitness, production capacity potential and risk constraints, and the optimal parameter combination is dynamically selected based on multiple feedback factors.

[0011] In one approach, by employing a progressive correlation residual reduction process, components dominated by highly correlated fields are extracted from high-dimensional eigenvectors, eliminating multi-field redundancy, enhancing the prominence of high-order correlation features between fields, effectively alleviating multicollinearity and misalignment of primary and secondary relationships, and providing multi-field coupling feature data with both physical meaning and mathematical basis for subsequent multi-field interactive analysis.

[0012] In one approach, a mechanism based on the spatiotemporal energy accumulation and dissipation competition function is adopted to predict the activation and fracture probability of new crack tips. Based on the energy dominance of the crack tip region and historical energy dissipation, the formation of new crack tips and crack branching evolution are determined, thereby achieving adaptive modeling of the crack tip seepage, dissolution or softening response under the action of special carbon dioxide fluid.

[0013] In one embodiment, the microscopic level refers to pores and fractures, the mesoscopic level to rock masses, and the macroscopic level to the entire reservoir.

[0014] Beneficial effects of this invention: This invention employs multi-field coupled data integration, intelligent crack propagation prediction, and a multi-scale progressive effect evaluation and adaptive parameter optimization mechanism, which can significantly improve the scientific rigor and practicality of carbon dioxide fracturing in shale reservoirs. By integrating heterogeneous data from multiple fields such as pressure, temperature, stress, fluid, and chemical reactions, and dynamically adjusting the weights of inter-field influences, it achieves accurate description and real-time updates of reservoir conditions.

[0015] Distributed energy flow tracing algorithms effectively improve the accuracy of crack propagation path and velocity prediction, helping to optimize reservoir stimulation processes and enhance fracture connectivity and the formation of effective seepage channels. Multi-scale stratified feedback evaluation methods comprehensively quantify the impact of fracturing processes on porosity, permeability, stress distribution, and productivity, achieving synergistic optimization of reservoir structure reconstruction and productivity enhancement.

[0016] The intelligent decision-making algorithm, combining multi-field feedback and adaptive evolution mechanisms, can automatically select the optimal combination of fracturing parameters for different reservoir conditions, improving experimental efficiency and production potential while reducing risks and environmental impacts. This invention not only enriches the theoretical system of shale reservoir fracturing evaluation but also provides technical support for green and efficient development, possessing promising prospects for widespread application. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart for multi-field coupled data integration and preliminary analysis. Detailed Implementation

[0018] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0019] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. To facilitate understanding, the invention will now be described more fully with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of the invention more thorough and complete.

[0020] like Figure 1 As shown, a method for evaluating carbon dioxide fracturing tests in shale reservoirs under multi-field coupling conditions specifically includes: Step 1: Multi-field Coupled Data Integration and Preliminary Analysis First, multi-field data involved in the fracturing process of shale reservoirs are collected, including but not limited to pressure, temperature, stress, fluid, and chemical reaction fields. To integrate these heterogeneous data, a dynamic correlation feature fusion algorithm is designed. This algorithm transforms the data from each field into high-dimensional feature vectors within a unified spatiotemporal grid based on spatiotemporal correlation, and dynamically adjusts the influence of each field through an adaptive weighting mechanism, thereby obtaining a description of the shale reservoir state under multi-field coupling conditions. This step provides fundamental data support for subsequent evaluation methods and preliminarily reveals key feature regions of multi-field interaction.

[0021] In the process of multi-field coupled data integration and preliminary analysis, for heterogeneous data such as pressure field, temperature field, stress field, fluid field and chemical reaction field involved in shale reservoir fracturing tests, the first problem to be solved is the seamless mapping and unified representation of multi-source, multi-scale and multi-resolution data. Different fields have different acquisition times, different physical quantities and different areas of influence, and directly stacking the original data often leads to information loss or misplacement of primary and secondary information.

[0022] Therefore, a Dynamic Relevance Feature Fusion (DRFF) algorithm is proposed in this step. The core idea of ​​the algorithm is to highly compress and organically fuse various field data into a unified feature vector through spatiotemporal mapping and dynamic weight adjustment, providing a data foundation for subsequent multi-field interaction mechanism analysis.

[0023] like Figure 2 As shown in S101, a spatial mesh is defined for all collected field data. Time series Using this as a framework, for each type of physical field (k represents the field type, i represents the spatial index, and j represents the time step), and interpolate or reduce its dimension to map it onto a spatial grid. and time nodes ,Right now: in This refers to interpolation or dimensionality reduction operators customized for different field types.

[0024] S102. Next, a dynamic weighting mechanism is needed to adjust the influence share of each field on the overall reservoir characteristics. Here, a set of spatiotemporal correlation coefficients is introduced. Its initial value is obtained by normalizing the variance of each type of field, and the corresponding grid and time are adaptively adjusted according to real-time data feedback. The formula is as follows: in This represents the variance operator, where K is the total number of all field types. It is an adaptive adjustment factor that is updated online as the experiment progresses, based on a synergistic relationship graph (such as a Bayesian causal network or cluster correlation). This mechanism can respond to phenomena such as synergistic enhancement or competition among multiple fields at a certain spatiotemporal point, adjusting the weight of each field in describing the reservoir state at that point.

[0025] S103. After standardization and weighting of all fields, at each grid time, the multi-field data can be fused into a unified high-dimensional feature vector: After this processing, each point Embedded into the feature space, the ability to describe reservoir states is greatly enhanced. Furthermore, to facilitate feature dimensionality reduction, redundancy removal, and highlight higher-order correlations between fields, a Successive Correlation Residual Elimination (S-CRE) process is used to sequentially remove components dominated by highly correlated fields, as shown in the following formula: in This represents the dynamic correlation coefficient between the current field and the extracted fields. It is a purified feature after eliminating redundancy in multiple fields. It can effectively alleviate multicollinearity and misalignment of primary and secondary relationships caused by the superposition of multiple fields, and provide mathematical support for the characterization of non-linear superposition of multiple fields, or even competition / inhibition phenomena.

[0026] Through the aforementioned dynamic fusion and related residual processing, the distribution, interaction relationships, and key controlling regions of different physical fields in shale reservoir fracturing tests were fully revealed. These results not only demonstrate the dominant field effects at different regional and temporal levels, but also intuitively identify the most active and important segments for subsequent crack propagation and fracturing response. Furthermore, they lay a precise, systematic, and highly compressed data foundation for multi-field coupling mechanism analysis and subsequent quantitative models.

[0027] Step 2: Construct a multi-field driven crack propagation prediction model Based on the coupling characteristics identified in the first step, a multi-field coupled crack propagation prediction model is established, employing a distributed energy flow tracing algorithm. Unlike traditional finite element or boundary element methods, this algorithm simulates the dynamic distribution and flow of energy along the crack tip and within the fracturing region due to multi-field interactions during fracturing experiments. Combined with multi-field time-varying driving analysis, it predicts the crack propagation path, velocity, and fracture probability in real time, providing more direct parameter references for evaluating fracturing effectiveness. This model can fully capture the coupling influence of the unique fluid properties of carbon dioxide on crack evolution.

[0028] In constructing a crack propagation prediction model driven by multiple fields, the dynamic distribution and migration of energy flow are the core elements characterizing crack dynamics. Based on the multi-field coupled high-dimensional eigenvectors generated in the first step, this model completely abandons the static, linear stress criterion-based solution methods of traditional finite element and boundary element methods, and proposes a distributed energy flow tracing (DEFT) algorithm. Its basic idea is to combine the coupled multi-field features of each grid point at a given instant. The mapping is a multi-component energy tensor containing elastic potential energy, thermal energy, fluid kinetic energy, chemical reaction energy, and other components. Then, from the perspective of tensor flow, we simulate in real time how energy migrates within the shale microstructure network and accumulates at the crack tip height.

[0029] Specifically, the energy tensor is constructed as follows: These correspond to elasticity, thermal, fluid, and chemical components, respectively. The energy component of each type of field is determined through... The corresponding physical quantities are realized by weighted integrals. For multiple energy component weighting: in This is a nonlinear mapping function that projects high-dimensional features onto the energy domain, and this function is a priori corrected through physical experiments or numerical simulations.

[0030] For the crack tip tracking section, an energy gradient-driven path growth mechanism is employed. This is used for paths where the main crack has already formed. At each time step, the algorithm searches all topological neighborhoods of the crack endpoint mesh. In the process of calculating the forward energy flux vector... : in For grid space gradient operators, The dominant weight of each energy field at the crack tip is adjusted in real time. Through point-by-point comparison of the energy flux density field, the current crack tip growth direction is selected as the dominant direction within the neighborhood. The largest grid point represents the region with the highest energy flux density. The algorithm differs from traditional maximum stress or strain energy criteria, instead reflecting the physical essence of energy extrema after multi-field mixing in real time, making the crack path correspond to the most easily breached point under multi-field conditions.

[0031] Not only is the crack path direction determined in real time, but the crack propagation rate is also determined by the energy flux. This propagation rate... This can be expressed through the following relationship: in The sensitivity coefficient for the flow response at the crack tip. It represents the amplitude of field variable transitions (such as high-frequency disturbances in multiple fields such as stress, pore pressure, and temperature difference). The damping of various disturbances is controlled. In this way, crack propagation only occurs when there is a significant surge in energy flow and all field disturbances cooperate to a sufficient threshold. This algorithm... The mechanisms of seepage promotion, dissolution reduction, and stress-chemical softening of special fluids have natural adaptability and can be simulated and captured. During injection, the high permeability zone at the fracture tip and even the branching phenomenon are induced.

[0032] The criterion for fracture probability is no longer a single R-curve method, but adopts a spatiotemporal energy accumulation-dissipation competitive function. Weighting the uncertainty of microfractures: The instantaneous energy density at the crack tip, the dominance of each field, and historical energy dissipation are considered. If a certain grid point... If the new crack tip is activated, it is considered a new crack tip activation. This mechanism naturally introduces multi-field perturbations and random fluctuations in energy flow, overcoming the shortcomings of the old model's static evaluation and monotonic fracture criteria, and enabling accurate prediction of actual phenomena such as new crack branching, barrier bypassing, and transient stagnation.

[0033] The entire model is implemented using local distributed computing, iterating repeatedly along time and space, and automatically and self-consistently linking multiple fields of energy flow and crack propagation surface. At each moment, the energy flow trajectory can automatically track, emerge, and calculate in parallel the most physically significant fracture front in the real reservoir. Ultimately, this multi-field coupled crack prediction system can output the fracture network morphology, dominant growth path, theoretical propagation velocity, and branch probability distribution in three-dimensional space at any given time, greatly enriching the physical characterization of the carbon dioxide fracturing mechanism and the dynamic evaluation capability of fracturing response.

[0034] Step 3: Multi-scale evaluation of fracturing effect under coupled field Based on the aforementioned model results, a multi-scale fracturing effect evaluation was conducted. A layered multi-field feedback evaluation algorithm was designed, which calculates the changes in reservoir porosity, permeability, stress redistribution, and other indicators at three levels: microscopic (pore-fracture), mesoscopic (rock mass), and macroscopic (reservoir as a whole). Each level automatically feeds back to the previous level, achieving a progressive evaluation under multi-field coupling. This evaluation system fully reflects the impact of multi-field synergy on fracturing behavior and reservoir reconstruction, ensuring that the evaluation results are scientific and comprehensive.

[0035] Multi-scale evaluation of the effects of multi-field coupled fracturing requires not only understanding the fracturing response at different physical scales, but also achieving a progressive and interconnected integration of data, models, and feedback. The impact of the fracturing process extends from microscopic pores and fractures to mesoscopic rock masses and even the macroscopic reservoir as a whole; each level is subject to the synergistic effects of multiple fields, making quantitative evaluation difficult with a single index. To address this, we constructed a Hierarchical Multi-field Feedback Evaluation (HMFE) algorithm, breaking through the constraints of traditional linear transmission and single-level superposition, striving to obtain reliable quantitative and dynamic feedback of multi-scale fracturing effects under complex coupling backgrounds.

[0036] At the microscale, the model maps the predicted fracture network and regional coupling features to pore and fracture parameters. This is based on fracture volume variation on a fragmented mesh set. and pore volume change The local porosity increment is obtained through weighted aggregation of physical field data streams: in The coefficients representing the influence of multiple fields (such as fluid, stress, and chemistry) on local pore volume changes are adaptively updated based on data. The permeability increment is calculated using the reachable channel coefficient under synergistic field perturbation. The algorithm determines the number and connectivity of new channels based on the combined effects of energy flow and crack growth path. It is a path mapping operator. consider The effects of special fluid properties on the expansion, contraction, dissolution, and blockage of seepage channels. Microscopic calculation results not only reflect local structural reorganization but also provide physical basis feedback for the next level of mesoscopic evaluation.

[0037] At the mesoscale, rock masses are considered as complex structures composed of different micro-units, and their mechanical behavior is influenced by porosity, permeability, and multi-field coupled energy flow. A synergistic field-driven mass redistribution model is used to describe stress redistribution, fracture clustering effects, and energy production response. Stress field variations within the rock mass are also discussed. Feedback coupling from microscopic pores and cracks: in For mesoscopic rock blocks, For the included micro-units, The weights are for mesoscopic feedback. This model allows microscopic damage accumulation to induce stress concentration, softening, or deformation in the mesoscopic region, and adaptively adjusts the feedback coefficients based on the multi-field coupled energy tensor, emphasizing the dynamic interaction between microscopic and mesoscopic feedback.

[0038] The macroscopic scale comprehensively reflects the overall behavior of the reservoir. The evaluation algorithm at this level is based on mesoscopic feedback, constructing a multi-field macroscopic response integral model to primarily quantify total reservoir porosity and permeability, productivity enhancement, stress distribution rebalancing, and fracturing zone expansion effects. Changes in total reservoir porosity are weighted by mesoscopic porosity feedback: here These are the influencing factors of different rock blocks. This is obtained through comprehensive microscopic feedback from the mesoscopic bulk. (Production capacity improvement) Introducing a comprehensive weighted average of changes in fracturing zone volume, channel connectivity, and seepage capacity driven by multiple energy fields: in It is a capacity response mapping. for Field distribution weights, The permeability feedback for the reservoir's main control area is specifically inferred from the linkage between the mesoscopic and microscopic layers.

[0039] The multi-scale evaluation system, through a hierarchical feedback mechanism, allows for automatic linkage and sensitive feedback between upper and lower layers: changes in physical quantities at each level not only propagate upwards but also promptly affect detailed parameters downwards. For example, after the mesoscopic layer identifies rock softening, the microscopic model automatically adjusts the local field weights to simulate pore evolution under the new stress distribution; if abnormal fluctuations occur in the macroscopic productivity enhancement prediction, feedback regulation is used to control the mesoscopic fracture cluster generation mechanism, achieving a scientific closed loop. Multi-field synergy characteristics permeate all levels; each sample is based on energy flow, mutually exclusive field variables, and collaborative driving forces, ensuring highly physical realism in the evaluation results while also considering the comprehensiveness of spatiotemporal dynamic indicators.

[0040] Step 4: Output Decision Recommendations Based on multi-level evaluation results, a coupled-field response intelligent optimization algorithm is designed. Combining experimental parameters (injection rate, carbon dioxide concentration, and temperature) with multi-field feedback, the algorithm searches for the optimal experimental scheme through an adaptive evolutionary mechanism, effectively improving crack propagation and reservoir stimulation efficiency. Finally, it outputs optimization decision recommendations for specific shale reservoirs and multi-field conditions, including fracture morphology prediction, fracturing parameter adjustment schemes, and production potential estimation, providing practical guidance and reference for field fracturing tests.

[0041] In the system that outputs decision recommendations, the focus is no longer limited to a passive understanding of fracture and reservoir characterization, but rather to actively extract the optimal parameter combination through intelligent optimization algorithms, driving the fracturing effect to converge toward the target extreme value. The core of the entire process is the Coupled-Field Response-based Intelligent Optimization (CRIO) algorithm. This algorithm highly integrates the aforementioned multi-field, multi-scale evaluation feedback, defining key experimental parameters such as injection rate, carbon dioxide concentration, and temperature as adjustable vectors. Simultaneously, the data stream of multi-field response, fracture morphology and production capacity results throughout the fracturing process is fed back in real time, and an embedded adaptive evolution mechanism is used to continuously screen and evolve the optimal decision solution.

[0042] The algorithm uses a multi-objective fitness function as its optimization engine to construct a joint objective. : in The fitness of the crack morphology predicted by the model (such as main crack length, number of branches, and orientation). Quantify indicators of production capacity potential (such as penetration improvement and fluid output). Risk constraints (such as over-connection, probability of instability induction). Weights for each item. It can be adaptively adjusted according to the actual engineering focus and reservoir geological characteristics. This evaluation standard is used to simulate the overall response of all parameter combinations to multi-field and multi-level feedback.

[0043] The algorithm employs an evolutionary population search approach. In the initial iteration step, each point in the parameter space... As a volume injection model-driven evaluation, multi-field, multi-scale simulation algorithms are used to provide feedback on the entire process output (such as crack distribution, permeability changes, and stress zoning), and the quality is evaluated based on the fitness function. After each iteration, the intelligent algorithm selects parameter volumes with higher fitness rankings and uses mutation, crossover, and grouping optimization mechanisms from genetic algorithms, or dynamically adjusts the parameter search positions under the particle swarm optimization approach, so that the population gradually approaches the global optimal solution set.

[0044] The core adaptive mechanism is reflected in the continuous feedback. For each generation of solutions, the evaluation results are automatically integrated with multiple feedback factors, adjusting the weights of each parameter's influence on multiple fields or scales in real time. For example, if the current injection rate causes excessive disturbance to macroscopic stress, the risk item... As the value increases, the optimization mechanism rapidly reduces the additional weight of the classification parameter and widens the search range. Time-series data of multi-field responses and fracturing capacity are embedded with a class of memory-based directional mutation operators, enabling the model to identify multi-field nonlinear emergence regions and encouraging the exploration of specific parameter windows that drive fracture branching and enhance reservoir stimulation activity under specific conditions. The specific variation of the time-series data is assisted and controlled by Bayesian optimization or deep reinforcement learning prediction modules, forming an essentially coupled field-parameter-objective introspective loop.

[0045] During the optimization and convergence process, the model continuously outputs multi-level decision-making suggestions under specific reservoir and implementation conditions. In the fracture morphology prediction section, the optimal parameter body from the evolutionary process directly drives fracture propagation and branching simulations, and combined with the criterion function, it provides high-value information such as easily propagating zones, main fracture guidance, and expected branch density. Regarding the fracturing parameter adjustment scheme, the optimal injection rate is used... , concentration ,temperature The proposed solutions are typically expressed in terms of ranges or feasible regions, ensuring that the solutions are compatible with practical controllability and safety limits. Simultaneously, by integrating production capacity and risk assessment results, the estimated reservoir production potential and uncertainty boundaries are output, setting dynamic thresholds for construction decisions.

[0046] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0047] It should be understood that the above detailed description of the technical solutions of the present invention with reference to preferred embodiments is illustrative and not restrictive. Those skilled in the art can modify the technical solutions described in the embodiments or make equivalent substitutions for some of the technical features based on reading this specification; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating carbon dioxide fracturing tests in shale reservoirs under multi-field coupling conditions, characterized in that: The method includes: Multi-field coupled data integration and preliminary analysis: Collecting multi-field data involved in the fracturing process of shale reservoirs, including but not limited to pressure field, temperature field, stress field, fluid field, and chemical reaction field. Based on the spatiotemporal correlation, the data of each field are transformed into high-dimensional feature vectors of a unified spatiotemporal grid, and the influence of each field is dynamically adjusted through an adaptive weighting mechanism to obtain a description of the shale reservoir state under multi-field coupling conditions. A multi-field driven crack propagation prediction model was constructed. A distributed energy flow tracking algorithm was used to simulate the dynamic distribution and flow of energy along the crack tip and within the region due to multi-field interaction in the fracturing test. Combined with multi-field time-varying driving analysis, the crack propagation path, velocity and fracture probability were predicted in real time. Multi-scale evaluation of fracturing effect under coupled field: A layered multi-field feedback evaluation algorithm is designed. It calculates the changes of reservoir porosity, permeability and stress redistribution index by carbon dioxide fracturing at the micro, meso and macro levels, respectively. The system outputs decision recommendations and, by combining experimental parameters and multi-field feedback, searches for the optimal experimental scheme through an adaptive evolutionary mechanism, effectively improving crack propagation and reservoir stimulation efficiency. It outputs optimized decision-making recommendations for specific shale reservoirs and multiple field conditions, including fracture morphology prediction, fracturing parameter adjustment schemes, and production potential estimation.

2. The evaluation method for carbon dioxide fracturing tests of shale reservoirs under multi-field coupling conditions according to claim 1, characterized in that: The aforementioned multi-field coupling data integration and preliminary analysis includes: based on the heterogeneous data of pressure field, temperature field, stress field, fluid field and chemical reaction field involved in the fracturing process of shale reservoir, through dynamic correlation feature fusion algorithm, the data of various fields are processed by spatiotemporal mapping and corresponding interpolation or dimensionality reduction operators, so that all data are uniformly mapped to the preset spatial grid and time node. The dynamic correlation feature fusion algorithm dynamically adjusts the weight of each field's influence on reservoir characteristics at each spatiotemporal grid node by normalizing variance and combining it with an adaptive adjustment factor generated by the collaborative relationship graph, based on a set of spatiotemporal correlation coefficients. Multi-field data, after standardization and weight adjustment, are fused into a unified high-dimensional feature vector to enhance the ability to describe reservoir conditions.

3. The evaluation method for carbon dioxide fracturing tests of shale reservoirs under multi-field coupling conditions according to claim 1, characterized in that: The construction of the multi-field driven crack propagation prediction model includes: using a distributed energy flow tracing algorithm to map the multi-field feature information of each grid point into an energy tensor containing different types of energy components, including elastic potential energy, thermal energy, fluid kinetic energy and chemical reaction energy, and dynamically calculating the multi-field energy components through a nonlinear mapping function and a variable weighting factor. The distributed energy flow tracing algorithm uses an energy gradient-driven mechanism to predict the direction and speed of crack propagation in real time. Specifically, for each moment of the crack tip and its neighborhood, it automatically selects the region of maximum energy density as the direction of crack tip growth by comparing the energy flow density field point by point, and considers the regulating effect of dynamic synergy of multiple energy fields on crack propagation speed.

4. The evaluation method for carbon dioxide fracturing tests of shale reservoirs under multi-field coupling conditions according to claim 1, characterized in that: The multi-scale evaluation of fracturing effect under coupled field includes: using a hierarchical multi-field feedback evaluation algorithm to carry out multi-field coupled quantitative evaluation of carbon dioxide fracturing effect at the micro, meso and macro (storage) levels respectively. At the microscale, the algorithm automatically calculates the increments of porosity and permeability parameters based on the crack network structure and local coupling characteristics, and combines multi-field energy tensors and crack growth paths to describe the changes in seepage channels and the reorganization of pore structures. At the mesoscale, rock blocks are considered as complex structural units. The mechanical behavior and stress distribution of each block are adaptively adjusted according to the feedback results of the micro-level, dynamically reflecting the changes in block stress redistribution, crack clustering effect and energy production response. At the macro scale, the overall reservoir behavior is evaluated based on feedback from the mesoscopic level. A multi-field response integral model is used to quantify the reservoir's total porosity, productivity enhancement, stress spatial equilibrium, and fracturing zone expansion indicators. Progressive feedback and automatic linkage are achieved among the three levels, comprehensively and scientifically quantifying the combined impact of multi-field synergy on reservoir fracturing behavior and structural reconstruction.

5. The evaluation method for carbon dioxide fracturing tests of shale reservoirs under multi-field coupling conditions according to claim 1, characterized in that: The output decision recommendations include: based on a coupled-field response intelligent optimization algorithm, integrating multi-field and multi-scale evaluation results, and setting key experimental parameters such as injection rate, carbon dioxide concentration, and temperature as adjustable optimization vectors. Through evolutionary population search and adaptive evolution mechanism, the fracture morphology, fracturing parameters and production capacity indicators are fed back and optimized in real time. The target extreme values ​​of fracture propagation, branching simulation and controllable reservoir stimulation are fitted to form a multi-objective fitness function. The parameter weights are automatically adjusted by integrating fracture morphology fitness, production capacity potential and risk constraints, and the optimal parameter combination is dynamically selected based on multiple feedback factors.

6. The evaluation method for carbon dioxide fracturing tests of shale reservoirs under multi-field coupling conditions according to claim 2, characterized in that: By employing a stepwise correlation residual reduction process, components dominated by highly correlated fields are extracted from high-dimensional eigenvectors, eliminating multi-field redundancy, enhancing the prominent performance of high-order correlation features between fields, effectively alleviating multicollinearity and misalignment of primary and secondary relationships, and providing multi-field coupling feature data with both physical meaning and mathematical basis for subsequent multi-field interactive analysis.

7. The evaluation method for carbon dioxide fracturing tests of shale reservoirs under multi-field coupling conditions according to claim 3, characterized in that: To predict the activation and fracture probability of new crack tips, a mechanism based on the spatiotemporal energy accumulation and dissipation competition function is adopted. Based on the energy dominance of the crack tip region and the historical energy dissipation, the formation of new crack tips and crack branching evolution are determined, so as to achieve adaptive modeling of crack tip seepage, dissolution or softening response under the action of special carbon dioxide fluid.

8. The evaluation method for carbon dioxide fracturing tests of shale reservoirs under multi-field coupling conditions according to claim 4, characterized in that: The microscopic level refers to pores and fractures, the mesoscopic level refers to rock blocks, and the macroscopic level refers to the entire reservoir.

Images