A method and system for optimizing proppant injection strategy under multi-layer fracture distribution

Abstract

The application discloses a kind of support agent injection strategy optimization method and system under multilayer fracture distribution.Method includes: obtaining the static attribute set of multilayer fracture, input pre-trained multi-scale coupling model to calculate dynamic response potential set;Based on the potential set, the maximum problem of effective filling volume and flow capacity balance is solved by multi-objective optimization algorithm, to generate the initial injection parameter combination matching each layer potential;During injection, real-time monitoring data is used to invert and online calibration model parameters;Based on the calibrated model, re-optimization is generated to generate updated injection parameter combination, and real-time adjustment of field equipment, including switching interlayer fluid or adjusting support agent particle size timing.The application realizes closed-loop intelligent optimization from design, perception to dynamic tuning, effectively improves the uniformity of support agent distribution in multilayer fracture and fracturing effect.

Description

Technical Field

[0001] This invention relates to the field of petroleum engineering technology, specifically to a method and system for optimizing proppant injection strategies under multi-layered fracture distribution. Background Technology

[0002] In fracturing operations for oil and gas extraction, multi-layered fractures are becoming increasingly common. However, current proppant injection strategies for multi-layered fractures face numerous unresolved issues.

[0003] Traditional proppant injection strategies often employ a single approach, failing to fully consider the significant differences in geometry, permeability, and rock properties among fracture layers. The geometry of multi-layered fractures is extremely complex, with lengths ranging from tens to hundreds of meters, widths varying in millimeters or even micrometers, and significant height differences. Furthermore, the degree of curvature and branching of fractures vary across layers. For example, in some deep oil and gas reservoirs, upper-layer fractures may be relatively short and straight, while lower-layer fractures may be longer and exhibit a complex network structure.

[0004] Meanwhile, the permeability of fractures in different layers varies greatly due to different reservoir characteristics. Low-permeability layers may have only 1 mD, while high-permeability layers can reach over 100 mD. Rock properties are also complex. Hard rock layers can hinder proppant embedding, while brittle rock layers are prone to generating a large number of microfractures during fracturing, altering the pressure transmission path and the proppant filling space.

[0005] Under such complex multi-layered fracture distribution conditions, traditional injection strategies easily lead to uneven proppant distribution. In low-permeability fracture layers, proppant struggles to penetrate smoothly, resulting in insufficient support; while in high-permeability fracture layers, proppant may excessively accumulate due to high flow rates, wasting resources and reducing fracture conductivity. This series of problems causes some fractures to close prematurely, hindering oil and gas well production and keeping extraction costs high, severely restricting the efficient development of the oil and gas extraction industry. Therefore, developing an optimized proppant injection strategy that can adapt to multi-layered fracture distribution is urgently needed.

[0006] In view of this, the present invention is hereby proposed. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention proposes a method and system for optimizing proppant injection strategies under multi-layered fracture distribution, aiming to improve the rationality of proppant distribution and fracturing effect in hydraulic fracturing operations, thereby increasing the productivity of oil and gas wells.

[0008] Specifically, the following technical solution was adopted:

[0009] An optimization method for proppant injection strategy under multi-layered crack distribution includes:

[0010] Step S1: Obtain the static attribute set and dynamic response potential set of the multi-layer fractures in the target well; the static attribute set includes at least the permeability, porosity, average width, length, tortuosity, and branch density of each fracture layer; the dynamic response potential set is obtained by inputting the static attribute set into a pre-trained multi-scale fracture propagation-proppane migration coupling model, used to characterize the fracture propagation trend and proppane carrying capacity under different injection conditions, the multi-scale fracture propagation-proppane migration coupling model includes at least mutually coupled macroscopic fracture propagation sub-model, proppane migration and settlement sub-model, and fracture... The fracture conductivity prediction sub-model includes a macro-crack propagation sub-model that simulates the dynamic geometric changes of cracks in a stress field, and whose output includes at least the crack length, height, and dynamic width distribution over time. The proppant migration and settling sub-model uses the crack dynamic geometry output by the macro-crack propagation sub-model as the input field to simulate the transport, settling, and sandbank formation processes of proppant of different particle sizes. The fracture conductivity prediction sub-model calculates the equivalent permeability and conductivity of the fracture based on the final proppant concentration and distribution output by the proppant migration and settling sub-model.

[0011] Step S2: Based on the dynamic response potential set, apply an optimization algorithm to solve a function with multiple objectives, namely equalizing the conductivity of each fracture layer and maximizing the effective filling volume of the proppant, and output the initial injection parameter combination for different fracture layers in the target well; the initial injection parameter combination includes the injection flow rate, injection pressure and median proppant particle size that match the dynamic response potential of each fracture layer.

[0012] Step S3: During the injection process, real-time monitoring data of fractures in each layer of the target well is acquired through the downhole monitoring network. The real-time monitoring data includes inlet pressure, proppant concentration distribution, and microseismic event data. The real-time monitoring data is input into the multi-scale fracture propagation-proppant migration coupled model, and the fracture geometric parameters and proppant settlement coefficient of the model are inverted and calibrated online.

[0013] Step S4: Using the calibrated multi-scale fracture propagation-proppane migration coupling model, re-execute the optimization solution process of step S2 to generate an updated combination of injection parameters and adjust the on-site injection equipment in real time. The adjustment includes switching the characteristics of the injection fluid between different layers or realizing the temporal variation of the proppant particle size in the same injection fluid to adapt to the dynamic evolution of the fracture.

[0014] As an optional embodiment of the present invention, a proppant injection strategy optimization method under multi-layered fracture distribution, wherein the macroscopic fracture propagation sub-model, the proppant migration and settlement sub-model, and the fracture conductivity prediction sub-model are coupled together, including:

[0015] At each computation time step, the macroscopic crack propagation sub-model transmits the updated crack dynamic width distribution w(x,y,t) to the proppant migration and settlement sub-model as its migration space field.

[0016] The local proppant concentration c(x, y, t) and the height of the sand embankment formed, calculated by the proppant migration and sedimentation sub-model, are fed back to the macroscopic crack propagation sub-model to correct the effective mechanical width and flow resistance of the cracks in the local area.

[0017] The crack conductivity prediction sub-model is activated after the construction simulation is completed, and its input depends entirely on the final output of the proppant migration and settlement sub-model.

[0018] As an optional embodiment of the present invention, a proppant injection strategy optimization method under multi-layered crack distribution is provided. In the macroscopic crack propagation sub-model, the dynamic width calculation of the crack considers the influence of proppant concentration and adopts a correction formula in the following form:

[0019] w_eff(x, t) = w_h(x, t) - β × c(x, t);

[0020] Where w_eff(x, t) is the effective hydraulic width at position x at time t, w_h(x, t) is the theoretical hydraulic width calculated from rock mechanics and fluid pressure, c(x, t) is the proppant volume concentration at the same position and time, and β is the embedding / clogging correction coefficient related to proppant particle size and rock hardness.

[0021] As an optional embodiment of the present invention, a proppant injection strategy optimization method under multi-layered fracture distribution is provided, wherein the proppant transport and sedimentation sub-model is implemented by solving the following set of coupled equations:

[0022] Continuity equation: ;

[0023] Momentum equation: ;

[0024] Proppant settling velocity: v_s = f(d_p, ρ_p, ρ_f, μ, c);

[0025] in, The symbol is for partial differential, and t is time. Let w be the vector differential operator, c be the proppant concentration, v be the average velocity of the proppant-carrying fluid, μ be the fracturing fluid viscosity, p be the intra-fracture pressure, and v_s be the proppant settling velocity. v_s is a function of the proppant particle size d_p, the proppant and fracturing fluid densities ρ_p and ρ_f, the viscosity μ, and the local concentration c.

[0026] As an optional embodiment of the present invention, in step S2 of the present invention, a method for optimizing proppant injection strategy under multi-layered fracture distribution, the function with the multi-objectives of equalizing the conductivity of each layer of fractures and maximizing the effective filling volume of proppant is as follows:

[0027] Maximize: F = α × F_volume + (1-α) × F_balance;

[0028] Where F_volume = (∑ V_i ) / V_total represents the normalized value of the total effective filling volume, V_i is the predicted effective proppant filling volume of the i-th fracture layer, which is calculated by the multi-scale fracture propagation-proppant transport coupling model based on the combination of injection parameters, ∑ V_i represents the summation of i from 1 to n, and V_total is the sum of the volumes of all fracture layers;

[0029] F_balance = 1 / (1 + σ(C_i) / μ(C_i)), representing the conductivity balance index, where C_i is the predicted dimensionless conductivity of the i-th layer of fracture, and σ(C_i) and μ(C_i) are the standard deviation and mean of the conductivity of each layer, respectively. The closer the conductivity balance index value is to 1, the more balanced it is.

[0030] α is a preset weighting coefficient, ranging from 0.3 to 0.7, used to adjust the relative importance of the two optimization objectives.

[0031] As an optional embodiment of the present invention, a proppant injection strategy optimization method under multi-layered fracture distribution is provided. In step S2, an optimization algorithm is applied to solve a multi-objective function with the goals of equalizing the conductivity of each fracture layer and maximizing the effective filling volume of the proppant, including:

[0032] Step S201: Encode the initial injection parameter combination for different fracture layers into a solution vector. The solution vector includes at least the injection flow rate, injection pressure, and median proppant particle size for each layer. Initialize a population containing multiple solution vectors by random generation or heuristic generation based on historical construction cases.

[0033] Step S202: For each solution vector in the population, input it into the multi-scale crack propagation-proppane transport coupling model to calculate the predicted effective proppant filling volume V_i and the predicted dimensionless conductivity C_i for each layer; then, calculate the fitness value of the solution vector according to the multi-objective function F.

[0034] Step S203: Based on the fitness value of the solution vector, select dominant individuals using tournament selection, and perform simulated binary crossover and polynomial mutation operations on them to generate a new offspring population.

[0035] Step S204: Repeat steps S202 and S203 until the fitness value of the optimal solution vector in the population meets the preset value within a consecutive preset number of generations, or reaches the maximum number of iterations. Then, decode the optimal solution vector at this time and output it as the initial injection parameter combination.

[0036] As an optional embodiment of the present invention, in step S3 of the method for optimizing proppant injection strategy under multi-layered crack distribution, the real-time monitoring data is input into the multi-scale crack propagation-proppant migration coupled model, and the crack geometric parameters and proppant settlement coefficient of the model are inverted and calibrated online, including:

[0037] Step S301: Define a state vector X, which includes at least a subset of crack geometric parameters to be calibrated and the proppant settlement coefficient λ in the multi-scale crack propagation-proppant transport coupling model. The subset of crack geometric parameters includes the half-length of cracks in each layer and the dynamic width distribution coefficient. Define an observation vector Y, whose elements correspond to the acquired real-time monitoring data.

[0038] Step S302: Initialize a set containing N model instances, where N is an integer greater than 1; each model instance is a copy of the multi-scale crack propagation-proppant transport coupled model, but each component of the state vector X of each model instance is assigned a random perturbation value based on a preset probability distribution of prior knowledge, thereby generating N different initial state vectors X_1, X_2, ..., X_N; within the current calibration period, let each model instance use its own state vector as the initial condition, run from the previous calibration time to the current time, and perform numerical simulations respectively, thereby obtaining a set of predicted values ​​of state vectors {X^f_i} and a set of predicted values ​​of observation vectors {Y^f_i}, where i = 1,2, ..., N;

[0039] Step S303: Obtain the actual observation vector Y^obs at the current time; according to the update equation of the ensemble Kalman filter, use the difference between the actual observation vector Y^obs and the observation prediction set {Y^f_i} to correct each member in the state prediction set {X^f_i}, and calculate a set of updated state vectors {X^a_i}.

[0040] Step S304: Take the arithmetic mean of the updated state vector set {X^a_i} as the optimal estimate of the crack geometry parameters and proppant settlement coefficient at the current moment, and use this value to update the corresponding parameters of the multi-scale crack propagation-proppant migration coupling model used for decision-making in step S4.

[0041] As an optional embodiment of the present invention, the proppant injection strategy optimization method under multi-layer fracture distribution of the present invention, wherein the inversion and online calibration in step S3 are triggered periodically at a fixed time interval Δt, or triggered when a set engineering event is detected; the set engineering events include: a sudden change in downhole pressure exceeding a threshold, detection of a proppant concentration peak reaching a certain set sensor position, or a microseismic event occurrence rate exceeding a threshold.

[0042] This invention also provides a proppant injection strategy optimization system for multi-layered crack distribution, comprising:

[0043] The data acquisition and potential calculation module acquires the static attribute set and dynamic response potential set of the multi-layered fractures in the target well. The static attribute set includes at least the permeability, porosity, average width, length, tortuosity, and branch density of each fracture layer. The dynamic response potential set is obtained by inputting the static attribute set into a pre-trained multi-scale fracture propagation-proppane migration coupling model, which is used to characterize the fracture propagation trend and proppane carrying capacity under different injection conditions. The multi-scale fracture propagation-proppane migration coupling model includes at least mutually coupled macroscopic fracture propagation sub-models and proppane migration and settlement sub-models. The system includes a fracture conductivity prediction sub-model, wherein the macroscopic fracture propagation sub-model is used to simulate the dynamic geometric changes of fractures in a stress field, and its output includes at least the fracture length, height, and dynamic width distribution over time; the proppant migration and settling sub-model uses the fracture dynamic geometry output by the macroscopic fracture propagation sub-model as the input field to simulate the transport, settling, and sandbank formation process of proppant of different particle sizes; and the fracture conductivity prediction sub-model calculates the equivalent permeability and conductivity of the fracture based on the final proppant concentration and distribution output by the proppant migration and settling sub-model.

[0044] The optimization decision module is communicatively connected to the data acquisition and potential calculation module. Based on the dynamic response potential set, it applies an optimization algorithm to solve a function with multiple objectives, namely balancing the conductivity of each fracture layer and maximizing the effective filling volume of the proppant. The module outputs a combination of initial injection parameters for different fracture layers in the target well. The initial injection parameter combination includes the injection flow rate, injection pressure, and median proppant particle size that match the dynamic response potential of each fracture layer.

[0045] The real-time monitoring and model calibration module is connected to the downhole monitoring network. During the injection process, it acquires real-time monitoring data of fractures in each layer of the target well through the downhole monitoring network. The real-time monitoring data includes inlet pressure, proppant concentration distribution, and microseismic event data. The real-time monitoring data is input into the multi-scale fracture propagation-proppant migration coupled model to invert and calibrate the fracture geometric parameters and proppant settlement coefficient of the model online.

[0046] The dynamic execution and feedback control module is communicatively connected to the optimization decision module, the real-time monitoring and model calibration module, and the field injection equipment control system, respectively. Upon receiving a signal indicating model calibration completion, it triggers the optimization decision module to re-optimize the solution using the calibrated model, generating updated injection parameter combinations, and accordingly sending control commands to the field injection equipment control system. These control commands drive the execution of adjustments such as switching injection fluid characteristics between different layers or implementing temporal changes in proppant particle size within the same injection fluid, to adapt to the dynamic evolution of the fracture.

[0047] This invention provides an optimization method for proppant injection strategy under multi-layer fracture distribution. Through precise analysis of inter-layer fracture properties, formulation of multi-layer injection optimization strategies, and establishment of a real-time feedback adjustment mechanism, it accurately simulates the proppant migration process based on a multi-scale fracture propagation-proppant migration coupling model. This solves the problem of uneven proppant distribution in multi-layer fractures in existing technologies, and has the following technical effects:

[0048] Improving the uniformity of proppant distribution: By accurately analyzing the properties of interlayer cracks and implementing multi-layer injection optimization strategies, the injection parameters of proppant can be reasonably controlled according to the characteristics of different crack layers, so that the proppant can be more uniformly distributed in each crack layer, avoiding the problems of local accumulation and insufficient support.

[0049] Enhanced fracture conductivity: The uniform distribution of proppant effectively improves the conductivity of fractures, allowing oil and gas to flow more smoothly from the formation to the wellbore, thereby increasing the production and efficiency of oil and gas wells.

[0050] Improving the success rate of fracturing operations: The real-time feedback adjustment mechanism can adjust the injection parameters in a timely manner according to the actual injection effect, ensuring that the proppant exists stably in each fracture layer, reducing the risk of fracturing failure due to unreasonable parameters, and improving the success rate of fracturing operations.

[0051] Achieving efficient oil and gas resource extraction: The application of the method of this invention can more fully develop the potential of multi-layer fractured reservoirs, improve the recovery rate of oil and gas resources, reduce extraction costs, and achieve efficient extraction and sustainable utilization of oil and gas resources. Attached Figure Description

[0052] Figure 1 This invention provides a flowchart of a proppant injection strategy optimization method for multi-layered crack distribution. Figure 1 ;

[0053] Figure 2 This invention provides a flowchart of a proppant injection strategy optimization method for multi-layered crack distribution. Figure 2 . Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0055] Therefore, the following detailed description of embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0056] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the embodiments of the present invention can be combined with each other.

[0057] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0058] In the description of this invention, it should be noted that the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. These terms are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0059] like Figures 1 to 2 As shown, this embodiment of the invention provides a method for optimizing proppant injection strategies under multi-layered crack distribution, including:

[0060] Step S1: Obtain the static attribute set and dynamic response potential set of the multi-layer fractures in the target well; the static attribute set includes at least the permeability, porosity, average width, length, tortuosity and branch density of each fracture layer; the dynamic response potential set is obtained by inputting the static attribute set into a pre-trained multi-scale fracture propagation-proppane transport coupling model, and is used to characterize the fracture propagation trend and proppant carrying capacity under different injection conditions.

[0061] Step S2: Based on the dynamic response potential set, apply an optimization algorithm to solve a function with multiple objectives, namely equalizing the conductivity of each fracture layer and maximizing the effective filling volume of the proppant, and output the initial injection parameter combination for different fracture layers in the target well; the initial injection parameter combination includes the injection flow rate, injection pressure and median proppant particle size that match the dynamic response potential of each fracture layer.

[0062] Step S3: During the injection process, real-time monitoring data of fractures in each layer of the target well is acquired through the downhole monitoring network. The real-time monitoring data includes inlet pressure, proppant concentration distribution, and microseismic event data. The real-time monitoring data is input into the multi-scale fracture propagation-proppant migration coupled model, and the fracture geometric parameters and proppant settlement coefficient of the model are inverted and calibrated online.

[0063] Step S4: Using the calibrated multi-scale fracture propagation-proppane migration coupling model, re-execute the optimization solution process of step S2 to generate an updated combination of injection parameters and adjust the on-site injection equipment in real time. The adjustment includes switching the characteristics of the injection fluid between different layers or realizing the temporal variation of the proppant particle size in the same injection fluid to adapt to the dynamic evolution of the fracture.

[0064] This invention provides an optimization method for proppant injection strategy under multi-layer fracture distribution. Through precise analysis of inter-layer fracture properties, formulation of multi-layer injection optimization strategies, and establishment of a real-time feedback adjustment mechanism, it accurately simulates the proppant migration process based on a multi-scale fracture propagation-proppant migration coupling model. This solves the problem of uneven proppant distribution in multi-layer fractures in existing technologies, and has the following technical effects:

[0065] Improving the uniformity of proppant distribution: By accurately analyzing the properties of interlayer cracks and implementing multi-layer injection optimization strategies, the injection parameters of proppant can be reasonably controlled according to the characteristics of different crack layers, so that the proppant can be more uniformly distributed in each crack layer, avoiding the problems of local accumulation and insufficient support.

[0066] Enhanced fracture conductivity: The uniform distribution of proppant effectively improves the conductivity of fractures, allowing oil and gas to flow more smoothly from the formation to the wellbore, thereby increasing the production and efficiency of oil and gas wells.

[0067] Improving the success rate of fracturing operations: The real-time feedback adjustment mechanism can adjust the injection parameters in a timely manner according to the actual injection effect, ensuring that the proppant exists stably in each fracture layer, reducing the risk of fracturing failure due to unreasonable parameters, and improving the success rate of fracturing operations.

[0068] Achieving efficient oil and gas resource extraction: The application of the method in the embodiments of the present invention can more fully develop the potential of multi-layer fractured reservoirs, improve the recovery rate of oil and gas resources, reduce extraction costs, and achieve efficient extraction and sustainable utilization of oil and gas resources.

[0069] In this embodiment, a proppant injection strategy optimization method under multi-layered fracture distribution is described. In step S1, the multi-scale fracture propagation-proppant migration coupling model includes at least a macroscopic fracture propagation sub-model, a proppant migration and settlement sub-model, and a fracture conductivity prediction sub-model that are coupled with each other.

[0070] The macroscopic crack propagation sub-model is used to simulate the dynamic geometric changes of cracks in a stress field, and its output includes at least the crack length, height, and dynamic width distribution over time. The proppant migration and settling sub-model uses the crack dynamic geometry output by the macroscopic crack propagation sub-model as the input field to simulate the transport, settling, and sand dam formation process of proppant of different particle sizes. The crack conductivity prediction sub-model calculates the equivalent permeability and conductivity of the crack based on the final proppant concentration and distribution output by the proppant migration and settling sub-model.

[0071] This invention constructs and applies a multi-scale model that deeply couples macroscopic crack propagation, proppant migration and settling, and crack conductivity prediction. For the first time, it realizes closed-loop feedback and collaborative calculation between crack dynamic geometry, proppant distribution, and final conductivity at the simulation level. This overcomes the prediction distortion problem caused by the use of isolated or simple serial models in traditional methods, and provides a real and reliable physical basis for subsequent model-based optimization decisions.

[0072] Specifically, in the proppant injection strategy optimization method under multi-layered fracture distribution in this embodiment, the macroscopic fracture propagation sub-model, the proppant migration and settlement sub-model, and the fracture conductivity prediction sub-model are coupled together, including:

[0073] At each computation time step, the macroscopic crack propagation sub-model transmits the updated crack dynamic width distribution w(x,y,t) to the proppant migration and settlement sub-model as its migration space field.

[0074] The local proppant concentration c(x, y, t) and the height of the sand embankment formed, calculated by the proppant migration and sedimentation sub-model, are fed back to the macroscopic crack propagation sub-model to correct the effective mechanical width and flow resistance of the cracks in the local area.

[0075] The crack conductivity prediction sub-model is activated after the construction simulation is completed, and its input depends entirely on the final output of the proppant migration and settlement sub-model.

[0076] In the proppant injection strategy optimization method under multi-layered crack distribution in this embodiment, the dynamic width calculation of the crack in the macroscopic crack propagation sub-model takes into account the influence of proppant concentration, and adopts the following correction formula:

[0077] w_eff(x, t) = w_h(x, t) - β × c(x, t);

[0078] Where w_eff(x, t) is the effective hydraulic width at position x at time t, w_h(x, t) is the theoretical hydraulic width calculated from rock mechanics and fluid pressure, c(x, t) is the proppant volume concentration at the same position and time, and β is the embedding / clogging correction coefficient related to proppant particle size and rock hardness.

[0079] The modified formula in this embodiment achieves reverse coupling from the proppant concentration field to the crack geometry field.

[0080] In the proppant injection strategy optimization method under multi-layered fracture distribution in this embodiment, the proppant migration and sedimentation sub-model is implemented by solving the following set of coupled equations:

[0081] Continuity equation: ;

[0082] Momentum equation: ;

[0083] Proppant settling velocity: v_s = f(d_p, ρ_p, ρ_f, μ, c);

[0084] in, The symbol is for partial differential, and t is time. Let w be the vector differential operator (Nabla operator), w be the fracture width, c be the proppant concentration, v be the average velocity of the proppant-carrying fluid, μ be the fracturing fluid viscosity, p be the intra-fracture pressure, and v_s be the proppant settling velocity. v_s is a function of the proppant particle size d_p, the proppant and fracturing fluid densities ρ_p and ρ_f, the viscosity μ, and the local concentration c.

[0085] In this embodiment, the coupled equations explicitly use the crack width w of the crack dynamic geometry as the key parameter, demonstrating strong coupling with the macroscopic expansion model.

[0086] The multi-scale crack propagation-proppane transport coupling model in this embodiment is trained and calibrated through the following steps:

[0087] a) Historical data collection: Collect complete data packages for multiple fractured wells, including pre-fracturing logging and seismic interpretation attributes (input), pumping procedures and real-time monitoring data during fracturing (intermediate input / verification), and production profiles of each layer obtained from production logging or tracer after fracturing (labels).

[0088] b) Initial model construction: Based on physical principles, the framework of the multi-scale crack propagation-proppane transport coupled model is built, and a set of default physical parameters are initialized;

[0089] c) Backpropagation training: Using pre-construction attributes and pumping procedures as input, drive the multi-scale crack propagation-proppane migration coupled model for calculation; compare the proppane distribution or production capacity profile predicted by the model with the actual production profile label, and calculate the loss function; use gradient descent or evolutionary algorithms to backpropagate and adjust the key uncertain parameters in the model (such as the correction coefficient β, the empirical coefficient in the settlement velocity function f, the crack initiation efficiency, etc.) until the loss function is minimized.

[0090] d) Model solidification and application: The model parameter set obtained from training that best reflects the geological and engineering characteristics of the region is solidified into the pre-trained model and used for predicting the dynamic response potential of new blocks.

[0091] This embodiment provides a proppant injection strategy optimization method under multi-layer fracture distribution. In step S1, the physical properties of the fractures, such as permeability, porosity, average width, length, and orientation, are obtained using geological exploration technology and data analysis methods. The curvature and branch density of the fractures are obtained by performing digital image processing and fracture skeleton extraction algorithms on wellbore imaging logging data.

[0092] In this embodiment, a proppant injection strategy optimization method under multi-layered fracture distribution is proposed. In step S2, the function with multiple objectives—equalizing the conductivity of each fracture layer and maximizing the effective proppant filling volume—is as follows:

[0093] Maximize: F = α × F_volume + (1-α) × F_balance;

[0094] Where F_volume = (∑ V_i ) / V_total represents the normalized value of the total effective filling volume, V_i is the predicted effective proppant filling volume of the i-th fracture layer, which is calculated by the multi-scale fracture propagation-proppant transport coupling model based on the combination of injection parameters, ∑ V_i represents the summation of i from 1 to n, and V_total is the sum of the volumes of all fracture layers;

[0095] F_balance = 1 / (1 + σ(C_i) / μ(C_i)), representing the conductivity balance index, where C_i is the predicted dimensionless conductivity of the i-th layer of fracture, and σ(C_i) and μ(C_i) are the standard deviation and mean of the conductivity of each layer, respectively. The closer the conductivity balance index value is to 1, the more balanced it is.

[0096] α is a preset weighting coefficient, ranging from 0.3 to 0.7, used to adjust the relative importance of the two optimization objectives.

[0097] Furthermore, in the proppant injection strategy optimization method under multi-layered fracture distribution of this embodiment, step S2 involves applying an optimization algorithm to solve a multi-objective function that aims to equalize the conductivity of each fracture layer and maximize the effective filling volume of the proppant, including:

[0098] Step S201: Encode the initial injection parameter combination for different fracture layers into a solution vector. The solution vector includes at least the injection flow rate, injection pressure, and median proppant particle size for each layer. Initialize a population containing multiple solution vectors by random generation or heuristic generation based on historical construction cases.

[0099] Step S202: For each solution vector in the population, input it into the multi-scale crack propagation-proppane transport coupling model to calculate the predicted effective proppant filling volume V_i and the predicted dimensionless conductivity C_i for each layer; then, calculate the fitness value of the solution vector according to the multi-objective function F.

[0100] Step S203: Based on the fitness value of the solution vector, select dominant individuals using tournament selection, and perform simulated binary crossover and polynomial mutation operations on them to generate a new offspring population.

[0101] Step S204: Repeat steps S202 and S203 until the fitness value of the optimal solution vector in the population meets the preset value within a consecutive preset number of generations, or reaches the maximum number of iterations. Then, decode the optimal solution vector at this time and output it as the initial injection parameter combination.

[0102] This embodiment provides a proppant injection strategy optimization method under multi-layered fracture distribution. The differences in parameters of each layer of the initial injection parameter combination generated in step S2 are determined by associating it with the dynamic response potential set obtained in step S1 through the following mapping rule:

[0103] For crack layers with concentrated dynamic response potential classified as "high expansion and low closure stress", the median proppant particle size (d50) should be 0.8-1.2 mm, and the injection pressure should be set at 80%-100% of the design upper limit.

[0104] For fracture layers classified as "low propagation, high filtration loss risk", the median proppant particle size (d50) should be 0.2-0.4 mm, and a fracturing fluid system containing a filtration loss reducer should be used.

[0105] For fracture layers classified as "moderately expansive but with complex curvature and branching", the median particle size (d50) of the matched proppant is distributed in a dual-modal manner, i.e., 0.3-0.5 mm and 0.7-0.9 mm are mixed in a preset ratio to balance the support of the main fracture and the ability to enter the micro fracture.

[0106] Based on the above limitations, the optimization step S2 of this embodiment of the invention achieves the following key technical advancements:

[0107] Objective function quantification: This transforms macroscopic optimization objectives (equilibrium, maximization) into specific, computable mathematical expressions. In particular, the construction of F_balance cleverly utilizes statistical concepts to quantify "equilibrium," making it a target that the algorithm can directly process.

[0108] The algorithm process is made explicit: a complete evolutionary algorithm optimization process is clearly outlined, including key steps such as encoding, evaluation, iteration, and convergence. This not only explains "how to optimize" but also emphasizes its iterative optimization characteristics, distinguishing it from simple rule matching or single-step computation.

[0109] Potential-parameter mapping concretization: This provides exemplary mapping rules from abstract "potential" to specific engineering parameters. While these numerical ranges are examples, the claim structure protects the core method of "differentiated parameter matching based on potential classification." In particular, the introduction of "dual-modal particle size distribution" to address complex seam networks is a very specific and inventive technical approach.

[0110] In this embodiment, a proppant injection strategy optimization method under multi-layered crack distribution is described. In step S3, the real-time monitoring data includes direct monitoring data and indirect inversion data.

[0111] The direct monitoring data includes at least the temperature, acoustic vibration frequency, and strain data at the entrance of each fracture segment obtained through a downhole distributed fiber optic sensing (DTS / DAS) system.

[0112] The indirect inversion data includes at least microseismic event point clouds obtained through ground or downhole microseismic monitoring arrays, used to invert the spatiotemporal profile of crack propagation;

[0113] The inversion calibration of the crack geometry parameters is mainly based on the indirect inversion data, and the inversion calibration of the proppant settling coefficient is mainly based on the time-series signal in the direct monitoring data that reflects the arrival time of the sand-carrying fluid and the proppant accumulation characteristics.

[0114] This embodiment provides a method for optimizing proppant injection strategies under multi-layered fracture distribution. In step S3, the real-time monitoring data is input into the multi-scale fracture propagation-proppant migration coupled model, and the fracture geometric parameters and proppant settlement coefficient of the model are inverted and calibrated online, including:

[0115] Step S301: Define a state vector X, which includes at least a subset of crack geometric parameters to be calibrated and the proppant settlement coefficient λ in the multi-scale crack propagation-proppant transport coupling model. The subset of crack geometric parameters includes the half-length of cracks in each layer and the dynamic width distribution coefficient. Define an observation vector Y, whose elements correspond to the acquired real-time monitoring data.

[0116] Step S302: Initialize a set containing N model instances, where N is an integer greater than 1; each model instance is a copy of the multi-scale crack propagation-proppant transport coupled model, but each component of the state vector X of each model instance is assigned a random perturbation value based on a preset probability distribution of prior knowledge, thereby generating N different initial state vectors X_1, X_2, ..., X_N; within the current calibration period, let each model instance use its own state vector as the initial condition, run from the previous calibration time to the current time, and perform numerical simulations respectively, thereby obtaining a set of predicted values ​​of state vectors {X^f_i} and a set of predicted values ​​of observation vectors {Y^f_i}, where i = 1,2, ..., N;

[0117] Step S303: Obtain the actual observation vector Y^obs at the current time; according to the update equation of the ensemble Kalman filter, use the difference between the actual observation vector Y^obs and the observation prediction set {Y^f_i} to correct each member in the state prediction set {X^f_i}, and calculate a set of updated state vectors {X^a_i}.

[0118] Step S304: Take the arithmetic mean of the updated state vector set {X^a_i} as the optimal estimate of the crack geometry parameters and proppant settlement coefficient at the current moment, and use this value to update the corresponding parameters of the multi-scale crack propagation-proppant migration coupling model used for decision-making in step S4.

[0119] In this embodiment, a proppant injection strategy optimization method under multi-layer fracture distribution is described. In step S3, the inversion and online calibration are periodically triggered at a fixed time interval Δt, or triggered when a set engineering event is detected. The set engineering events include: a sudden change in downhole pressure exceeding a threshold, the detection of a proppant concentration peak reaching a certain set sensor position, or the occurrence rate of microseismic events exceeding a threshold.

[0120] In this embodiment, a proppant injection strategy optimization method under multi-layer fracture distribution is described. In step S3, the updated proppant settling coefficient λ is obtained through inversion calibration and is directly used to correct the objective function calculation on which the optimization algorithm in step S4 depends. Specifically, the updated λ is substituted into the settling velocity function f of the proppant migration and settling sub-model, so that in the re-optimization in step S4, a model that better reflects the current downhole fluid and proppant behavior is used for prediction.

[0121] Based on the above limitations, step S3 of the present invention achieves the following technical effects:

[0122] Data specificity: It clarifies the data sources used for calibration and their correspondence with the calibration target, reflecting the sophisticated design of multi-source data fusion to solve specific inversion problems.

[0123] Algorithm advancement: The algorithm introduces Ensemble Kalman Filter (EnKF), an efficient data assimilation algorithm suitable for nonlinear systems, as the implementation method. This is not a simple parameter adjustment, but an optimal data fusion process based on statistical estimation theory. It is technically sophisticated and fundamentally different from trial-and-error methods or simple regression adjustments.

[0124] Intelligent triggering mechanism: Calibration is not a simple timed task, but is combined with engineering event-driven processes, reflecting the system's intelligent perception and response to the construction status.

[0125] Deepening the closed loop: This emphasizes that calibration results not only update the model, but also provide direct and immediate feedback to the optimization decision loop, truly realizing a deep closed loop of "perception-cognition-decision", thus maximizing the value of calibration.

[0126] This invention also provides a proppant injection strategy optimization system for multi-layered crack distribution, comprising:

[0127] The data acquisition and potential calculation module acquires the static attribute set and dynamic response potential set of the multi-layer fractures in the target well. The static attribute set includes at least the permeability, porosity, average width, length, tortuosity, and branch density of each fracture layer. The dynamic response potential set is obtained by inputting the static attribute set into a pre-trained multi-scale fracture propagation-proppane migration coupling model, and is used to characterize the fracture propagation trend and proppant carrying capacity under different injection conditions.

[0128] The optimization decision module is communicatively connected to the data acquisition and potential calculation module. Based on the dynamic response potential set, it applies an optimization algorithm to solve a function with multiple objectives, namely balancing the conductivity of each fracture layer and maximizing the effective filling volume of the proppant. The module outputs a combination of initial injection parameters for different fracture layers in the target well. The initial injection parameter combination includes the injection flow rate, injection pressure, and median proppant particle size that match the dynamic response potential of each fracture layer.

[0129] The real-time monitoring and model calibration module is connected to the downhole monitoring network. During the injection process, it acquires real-time monitoring data of fractures in each layer of the target well through the downhole monitoring network. The real-time monitoring data includes inlet pressure, proppant concentration distribution, and microseismic event data. The real-time monitoring data is input into the multi-scale fracture propagation-proppant migration coupled model to invert and calibrate the fracture geometric parameters and proppant settlement coefficient of the model online.

[0130] The dynamic execution and feedback control module is communicatively connected to the optimization decision module, the real-time monitoring and model calibration module, and the field injection equipment control system, respectively. Upon receiving a signal indicating model calibration completion, it triggers the optimization decision module to re-optimize the solution using the calibrated model, generating updated injection parameter combinations, and accordingly sending control commands to the field injection equipment control system. These control commands drive the execution of adjustments such as switching injection fluid characteristics between different layers or implementing temporal changes in proppant particle size within the same injection fluid, to adapt to the dynamic evolution of the fracture. Example

[0131] Reservoir conditions: A certain reservoir has three fracture layers. The upper fracture layer has a permeability of 10 mD, a porosity of 15%, an average fracture width of 3 mm, and a length of about 80 m. The middle fracture layer has a permeability of 5 mD, a porosity of 12%, an average fracture width of 2 mm, and a length of about 60 m. The lower fracture layer has a permeability of 15 mD, a porosity of 18%, an average fracture width of 4 mm, and a length of about 100 m.

[0132] Implementation process:

[0133] Interlayer crack attribute analysis and region division: Based on the attribute analysis of the three layers of cracks, the upper layer cracks are defined as the general injection area, the middle layer cracks as the fine injection area, and the lower layer cracks as the key injection area.

[0134] Multi-layer injection optimization strategy formulation and implementation: For the lower key injection area, the injection flow rate is set at 30 m³ / min, the injection pressure at 40 MPa, and a proppant with a particle size of 0.8-1.2 mm is selected; for the upper general injection area, the injection flow rate is set at 20 m³ / min, the injection pressure at 30 MPa, and a proppant with a particle size of 0.6-0.8 mm is selected; for the middle fine injection area, the injection flow rate is set at 15 m³ / min, the injection pressure at 25 MPa, and a proppant with a particle size of 0.4-0.6 mm is selected.

[0135] Real-time feedback and adjustment: During the injection process, if the uneven distribution of proppant in local areas of the mid-layer fracture is detected through real-time monitoring, the injection pressure is immediately reduced to 22MPa and the flow rate is adjusted to 12m³ / min to make the proppant distribution gradually uniform.

[0136] Results: An evaluation after injection revealed that the proppant was evenly distributed in the three fracture layers, the fracture conductivity was increased by about 35% compared to the traditional injection strategy, and the oil and gas well production increased by 30%. Example

[0137] Reservoir conditions: A certain reservoir has four fracture layers. The first fracture layer has a permeability of 8 mD, a porosity of 14%, an average fracture width of 2.5 mm, and a length of approximately 70 m. The second fracture layer has a permeability of 3 mD, a porosity of 10%, an average fracture width of 1.5 mm, and a length of approximately 50 m. The third fracture layer has a permeability of 12 mD, a porosity of 16%, an average fracture width of 3.5 mm, and a length of approximately 90 m. The fourth fracture layer has a permeability of 6 mD, a porosity of 13%, an average fracture width of 2 mm, and a length of approximately 65 m.

[0138] Implementation process:

[0139] Interlayer crack attribute analysis and region division: After analysis, the first layer of cracks is classified as a general injection area, the second layer of cracks as a special fine injection area, the third layer of cracks as a key injection area, and the fourth layer of cracks as a secondary key injection area.

[0140] Multi-layer injection optimization strategy formulation and implementation: The third layer key injection area uses a proppant with an injection flow rate of 35 m³ / min, an injection pressure of 45 MPa, and a particle size of 0.9-1.2 mm; the first layer general injection area uses a proppant with an injection flow rate of 22 m³ / min, an injection pressure of 32 MPa, and a particle size of 0.6-0.8 mm; the fourth layer key injection area uses a proppant with an injection flow rate of 18 m³ / min, an injection pressure of 28 MPa, and a particle size of 0.5-0.7 mm; the second layer special fine injection area uses a proppant with an injection flow rate of 10 m³ / min, an injection pressure of 20 MPa, and a particle size of 0.3-0.5 mm.

[0141] Real-time feedback and adjustment: During injection, it was detected that the proppant advance speed in the cracked area of ​​the fourth layer was slow. The injection pressure was increased to 30MPa and the flow rate was adjusted to 20m³ / min to ensure smooth proppant filling.

[0142] Implementation results: The evaluation showed that the distribution of proppant in each fracture layer was reasonable, the overall fracture conductivity was improved by about 40%, and the oil and gas well production increased by 35%.

[0143] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described herein. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the specific embodiments described above. Therefore, any modifications or equivalent substitutions to the present invention, as well as all technical solutions and improvements that do not depart from the spirit and scope of the invention, are covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing proppant injection strategies under multi-layered fracture distribution, characterized in that, include: Step S1: Obtain the static attribute set and dynamic response potential set of the multi-layer fractures in the target well; the static attribute set includes at least the permeability, porosity, average width, length, tortuosity and branch density of each fracture layer. The dynamic response potential set is obtained by inputting the static attribute set into a pre-trained multi-scale crack propagation-proppane migration coupling model. It is used to characterize the crack propagation trend and proppane bearing capacity under different injection conditions. The multi-scale crack propagation-proppane migration coupling model includes at least a macroscopic crack propagation sub-model, a proppane migration and settlement sub-model, and a crack conductivity prediction sub-model that are coupled with each other. The macroscopic crack propagation sub-model is used to simulate the dynamic geometric changes of cracks in a stress field, and its output includes at least the crack length, height, and dynamic width distribution over time. The proppant migration and settling sub-model uses the crack dynamic geometry output by the macroscopic crack propagation sub-model as the input field to simulate the transport, settling, and sand dam formation process of proppant of different particle sizes. The crack conductivity prediction sub-model calculates the equivalent permeability and conductivity of the crack based on the final proppant concentration and distribution output by the proppant migration and settling sub-model. The macroscopic crack propagation sub-model, proppant migration and settlement sub-model, and crack conductivity prediction sub-model are coupled together as follows: at each calculation time step, the macroscopic crack propagation sub-model transmits the updated crack dynamic width distribution w(x, y, t) to the proppant migration and settlement sub-model as its migration space field; the local proppant concentration c(x, y, t) and the height of the formed sand embankment calculated by the proppant migration and settlement sub-model are fed back to the macroscopic crack propagation sub-model to correct the effective mechanical width and flow resistance of the crack in the local area; the crack conductivity prediction sub-model is started after the construction simulation is completed, and its input depends entirely on the final output of the proppant migration and settlement sub-model. Step S2: Based on the dynamic response potential set, apply an optimization algorithm to solve a function with multiple objectives, namely equalizing the conductivity of each fracture layer and maximizing the effective filling volume of the proppant, and output the initial injection parameter combination for different fracture layers in the target well; the initial injection parameter combination includes the injection flow rate, injection pressure and median proppant particle size that match the dynamic response potential of each fracture layer. Step S3: During the injection process, real-time monitoring data of fractures in each layer of the target well is acquired through the downhole monitoring network. The real-time monitoring data includes inlet pressure, proppant concentration distribution, and microseismic event data. The real-time monitoring data is input into the multi-scale fracture propagation-proppant migration coupled model, and the fracture geometric parameters and proppant settlement coefficient of the model are inverted and calibrated online. Step S4: Using the calibrated multi-scale fracture propagation-proppane migration coupling model, re-execute the optimization solution process of step S2 to generate an updated combination of injection parameters and adjust the on-site injection equipment in real time. The adjustment includes switching the characteristics of the injection fluid between different layers or realizing the temporal variation of the proppant particle size in the same injection fluid to adapt to the dynamic evolution of the fracture.

2. The method for optimizing proppant injection strategy under multi-layered crack distribution according to claim 1, characterized in that, In the macroscopic crack propagation sub-model, the dynamic width calculation of the crack takes into account the influence of proppant concentration, and adopts the following modified formula: w_eff(x, t) = w_h(x, t) - β × c(x, t); Where w_eff(x, t) is the effective hydraulic width at position x at time t, w_h(x, t) is the theoretical hydraulic width calculated from rock mechanics and fluid pressure, c(x, t) is the proppant volume concentration at the same position and time, and β is the embedding / clogging correction coefficient related to proppant particle size and rock hardness.

3. The method for optimizing proppant injection strategy under multi-layered crack distribution according to claim 1 or 2, characterized in that, The proppant transport and sedimentation sub-model is achieved by solving the following set of coupled equations: Continuity equation: ; Momentum equation: ; Proppant settling velocity: v_s = f(d_p, ρ_p, ρ_f, μ, c); in, The symbol is for partial differential, and t is time. Let w be the vector differential operator, c be the proppant concentration, v be the average velocity of the proppant-carrying fluid, μ be the fracturing fluid viscosity, p be the intra-fracture pressure, and v_s be the proppant settling velocity. v_s is a function of the proppant particle size d_p, the proppant and fracturing fluid densities ρ_p and ρ_f, the viscosity μ, and the local concentration c.

4. The method for optimizing proppant injection strategy under multi-layered crack distribution according to claim 1, characterized in that, In step S2, the function with the multi-objective goals of equalizing the conductivity of each fracture layer and maximizing the effective proppant filling volume is: Maximize: F = α × F_volume + (1-α) × F_balance; Where F_volume = (∑ V_i ) / V_total represents the normalized value of the total effective filling volume, V_i is the predicted effective proppant filling volume of the i-th fracture layer, which is calculated by the multi-scale fracture propagation-proppant transport coupling model based on the combination of injection parameters, ∑ V_i represents the summation of i from 1 to n, and V_total is the sum of the volumes of all fracture layers; F_balance = 1 / (1 + σ(C_i) / μ(C_i)), representing the conductivity balance index, where C_i is the predicted dimensionless conductivity of the i-th layer of fracture, and σ(C_i) and μ(C_i) are the standard deviation and mean of the conductivity of each layer, respectively. The closer the conductivity balance index value is to 1, the more balanced it is. α is a preset weighting coefficient, ranging from 0.3 to 0.7, used to adjust the relative importance of the two optimization objectives.

5. The method for optimizing proppant injection strategy under multi-layered crack distribution according to claim 4, characterized in that, In step S2, an optimization algorithm is applied to solve a multi-objective function that aims to equalize the conductivity of fractures in each layer and maximize the effective volume of proppant filling, including: Step S201: Encode the initial injection parameter combination for different fracture layers into a solution vector. The solution vector includes at least the injection flow rate, injection pressure, and median proppant particle size for each layer. Initialize a population containing multiple solution vectors by random generation or heuristic generation based on historical construction cases. Step S202: For each solution vector in the population, input it into the multi-scale crack propagation-proppane transport coupling model to calculate the predicted effective proppant filling volume V_i and the predicted dimensionless conductivity C_i for each layer; then, calculate the fitness value of the solution vector according to the multi-objective function F. Step S203: Based on the fitness value of the solution vector, select dominant individuals using tournament selection, and perform simulated binary crossover and polynomial mutation operations on them to generate a new offspring population. Step S204: Repeat steps S202 and S203 until the fitness value of the optimal solution vector in the population meets the preset value within a consecutive preset number of generations, or reaches the maximum number of iterations. Then, decode the optimal solution vector at this time and output it as the initial injection parameter combination.

6. The method for optimizing proppant injection strategy under multi-layered crack distribution according to claim 5, characterized in that, In step S3, the real-time monitoring data is input into the multi-scale crack propagation-proppane migration coupled model, and the crack geometric parameters and proppane settlement coefficient of the model are inverted and calibrated online, including: Step S301: Define a state vector X, which includes at least a subset of crack geometric parameters to be calibrated and the proppant settlement coefficient λ in the multi-scale crack propagation-proppant transport coupling model. The subset of crack geometric parameters includes the half-length of cracks in each layer and the dynamic width distribution coefficient. Define an observation vector Y, whose elements correspond to the acquired real-time monitoring data. Step S302: Initialize a set containing N model instances, where N is an integer greater than 1; each model instance is a copy of the multi-scale crack propagation-proppant transport coupled model, but each component of the state vector X of each model instance is assigned a random perturbation value based on a preset probability distribution of prior knowledge, thereby generating N different initial state vectors X_1, X_2, ..., X_N; within the current calibration period, let each model instance use its own state vector as the initial condition, run from the previous calibration time to the current time, and perform numerical simulations respectively, thereby obtaining a set of predicted values ​​of state vectors {X^f_i} and a set of predicted values ​​of observation vectors {Y^f_i}, where i = 1, 2, ..., N; Step S303: Obtain the actual observation vector Y^obs at the current time; according to the update equation of the ensemble Kalman filter, use the difference between the actual observation vector Y^obs and the observation prediction set {Y^f_i} to correct each member in the state prediction set {X^f_i}, and calculate a set of updated state vectors {X^a_i}. Step S304: Take the arithmetic mean of the updated state vector set {X^a_i} as the optimal estimate of the crack geometry parameters and proppant settlement coefficient at the current moment, and use this value to update the corresponding parameters of the multi-scale crack propagation-proppant migration coupling model used for decision-making in step S4.

7. The method according to claim 6, characterized in that, The inversion and online calibration in step S3 are triggered periodically at a fixed time interval Δt, or when a set engineering event is detected; the set engineering events include: a sudden change in downhole pressure exceeding a threshold, the detection of a proppant concentration peak reaching a certain set sensor position, or the occurrence rate of microseismic events exceeding a threshold.

8. A proppant injection strategy optimization system for multi-layered crack distribution, characterized in that, include: The data acquisition and potential calculation module acquires the static attribute set and dynamic response potential set of the multi-layer fractures in the target well; the static attribute set includes at least the permeability, porosity, average width, length, tortuosity and branch density of each fracture layer. The dynamic response potential set is obtained by inputting the static attribute set into a pre-trained multi-scale crack propagation-proppane migration coupling model. It is used to characterize the crack propagation trend and proppane bearing capacity under different injection conditions. The multi-scale crack propagation-proppane migration coupling model includes at least a macroscopic crack propagation sub-model, a proppane migration and settlement sub-model, and a crack conductivity prediction sub-model that are coupled with each other. The macroscopic crack propagation sub-model is used to simulate the dynamic geometric changes of cracks in a stress field, and its output includes at least the crack length, height, and dynamic width distribution over time. The proppant migration and settling sub-model uses the crack dynamic geometry output by the macroscopic crack propagation sub-model as the input field to simulate the transport, settling, and sand dam formation process of proppant of different particle sizes. The crack conductivity prediction sub-model calculates the equivalent permeability and conductivity of the crack based on the final proppant concentration and distribution output by the proppant migration and settling sub-model. The macroscopic crack propagation sub-model, proppant migration and settlement sub-model, and crack conductivity prediction sub-model are coupled together as follows: at each calculation time step, the macroscopic crack propagation sub-model transmits the updated crack dynamic width distribution w(x, y, t) to the proppant migration and settlement sub-model as its migration space field; the local proppant concentration c(x, y, t) and the height of the formed sand embankment calculated by the proppant migration and settlement sub-model are fed back to the macroscopic crack propagation sub-model to correct the effective mechanical width and flow resistance of the crack in the local area; the crack conductivity prediction sub-model is started after the construction simulation is completed, and its input depends entirely on the final output of the proppant migration and settlement sub-model. The optimization decision module is communicatively connected to the data acquisition and potential calculation module. Based on the dynamic response potential set, it applies an optimization algorithm to solve a function with multiple objectives, namely balancing the conductivity of each fracture layer and maximizing the effective filling volume of the proppant. The module outputs a combination of initial injection parameters for different fracture layers in the target well. The initial injection parameter combination includes the injection flow rate, injection pressure, and median proppant particle size that match the dynamic response potential of each fracture layer. The real-time monitoring and model calibration module is connected to the downhole monitoring network. During the injection process, it acquires real-time monitoring data of fractures in each layer of the target well through the downhole monitoring network. The real-time monitoring data includes inlet pressure, proppant concentration distribution, and microseismic event data. The real-time monitoring data is input into the multi-scale fracture propagation-proppant migration coupled model to invert and calibrate the fracture geometric parameters and proppant settlement coefficient of the model online. The dynamic execution and feedback control module is communicatively connected to the optimization decision module, the real-time monitoring and model calibration module, and the field injection equipment control system, respectively. Upon receiving a signal indicating model calibration completion, it triggers the optimization decision module to re-optimize the solution using the calibrated model, generating updated injection parameter combinations, and accordingly sending control commands to the field injection equipment control system. These control commands drive the execution of adjustments such as switching injection fluid characteristics between different layers or implementing temporal changes in proppant particle size within the same injection fluid, to adapt to the dynamic evolution of the fracture.

Images