Deep reservoir fracture full-support sand adding strength optimization method and system
The permeability and fracture width variations of deep reservoirs were calculated using the lattice-Boltzmann method and the discrete element method. The proppant addition intensity was optimized by combining K-means clustering and the Transformer-MLP model, which solved the proppant embedding and fragmentation problems in deep reservoir fracturing and improved conductivity and oil and gas well production.
Patent Information
- Application Number
- CN202410898240.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-05
- Publication Date
- 2024-11-29
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Fracturing deep reservoirs is difficult, and proppant is prone to embedding or breaking under high stress conditions, resulting in insufficient fracture conductivity and affecting oil and gas development.
By combining the lattice-Boltzmann method and the discrete element method, the permeability and fracture width variation of proppant in deep reservoirs were calculated. By combining K-means clustering analysis and the Transformer-MLP deep learning model, the proppant addition intensity was optimized to improve the conductivity.
It enables precise proppant filling under different reservoir conditions, improves the conductivity of deep reservoirs and the production efficiency of oil and gas wells, and optimizes fracturing operation design.
Smart Images

Figure CN119047358A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method and a system for optimizing the full-support sand-adding strength of deep reservoir fractures, belonging to the technical field of gas reservoir exploitation. Background Art
[0002] Deep oil and gas reserves in my country are huge and widely distributed. Taking shale gas as an example, the technically recoverable resources of deep shale gas are 19.36×10 12 m 3 , accounting for 56.63% of the total shale gas resources. The efficient development of deep reservoirs will be an important development target for the future increase in oil and gas equivalent. Deep reservoirs have the characteristics of great burial depth. For example, the burial depth of deep shale gas reaches 3500-4500, and the formation temperature is 120-145℃, so the pressure and temperature conditions are more stringent. Therefore, the fracturing development technology for deep reservoirs needs to be broken through urgently.
[0003] Deep reservoirs usually need to be transformed by hydraulic fracturing technology to achieve commercial development conditions. However, due to the characteristics of deep reservoirs, such as deep burial depth, high confining pressure, low permeability, weak rock brittleness and local development of natural fractures, the fracturing transformation of deep reservoirs is extremely difficult. Hydraulic fracturing has become a key core technology in the transformation of low-permeability and unconventional oil and gas reservoirs. Fully breaking up the reservoir through fracturing technology helps to realize the "highway" of oil and gas migration. At present, in view of the geological characteristics of deep reservoirs, the deep fracturing process mode of "promoting fracture network and ensuring filling" is mainly adopted. Therefore, the overall conductivity of the fracture is extremely critical to improving the development output of deep reservoir gas. Compared with traditional fracturing transformation, the transformation process with large displacement and large liquid volume used in the process of volume fracturing transformation is more likely to form a very complex fracture network, that is, to produce main fractures with larger fracture width and branch fractures with relatively smaller fracture width. If the proppant can enter the fracture network to achieve full support, the development effect of the fracturing well will be greatly improved. The conductivity of the fracture network is one of the key factors affecting the production of fracturing wells. However, in many cases, especially under high stress conditions, the probability of proppants entering the fractures of deep formations is low, and under high stress conditions, problems such as proppant breakage or embedding are more likely to occur. Secondly, during the proppant injection process, not all fracture networks can achieve effective filling of proppants. Many secondary fractures can only be self-supported by the roughness generated by wall slip, and their overall contribution to production is relatively small. For deep reservoirs, due to the high confining pressure of the formation, the hydraulic fractures of the formation, especially the branch induced fractures, are more likely to close. In order to achieve long-term maintenance of the conductivity of deep reservoir fractures, it is extremely important to optimize the design of fracture conductivity parameters during volume fracturing.
[0004] The K-means algorithm is a commonly used unsupervised learning method that groups data into multiple categories by measuring the mutual distance between samples. This process mainly includes three steps: first, determine the number of clusters K to be formed and initialize the center of each cluster; then, calculate the distance between each data point and these K center points, and assign the data point to the nearest cluster; then, based on the newly assigned cluster members, update the center position of the cluster. This process will be repeated until a condition is met - either the sum of squared errors (SSE) within the cluster is minimized, or the cluster center no longer changes. Deep reservoir gas wells face greater challenges because they are buried deeper and have higher formation temperatures and pressures. Characterizing the mechanical properties of deep shale reservoirs is the basis for ensuring that deep reservoir gas resources can be developed safely and efficiently. Understanding reservoir characteristics at different levels helps optimize mining strategies and technologies. Summary of the invention
[0005] The purpose of the present invention is to propose a method for optimizing the full support and sanding strength of deep reservoir fractures in view of the phenomenon that proppant embedding and crushing effects occur in deep reservoirs due to the high temperature and high pressure environment. The method first uses the lattice-Boltzmann method, which is very different from the traditional CFD method, to calculate the permeability of the deep reservoir from the perspective of microscopic molecular dynamics, and uses the discrete element method to use software to calculate the change in the fracture width while taking into account the embedding and crushing of the proppant. On this basis, the conductivity calculation equation is calculated. The relationship between the conductivity and the concentration of the sand is established. From a macroscopic perspective, an empirical formula for permeability calculation is obtained, and the cluster analysis method is used to perform stratification for reservoirs with different permeabilities, and the relationship between the conductivity and production at different permeability levels is analyzed, thereby obtaining the optimal production at different levels, combining the microscopic and macroscopic to obtain a complete set of optimal sanding strength solutions, and finally the optimal sanding strength is obtained according to the ant colony algorithm.
[0006] The present invention also provides a deep reservoir fracture full-support sand-adding strength optimization system.
[0007] Terminology explanation:
[0008] 1. LBGK model, LBM is based on the kinetic theory of gases, specifically a discrete version of the BGK-Boltzmann transport equation. It evolves a discrete particle distribution function f according to the Boltzmann equation i (x, t). The governing equation of the LBM discrete form is:
[0009] f i (x+c i δ t ,t+δ t )=f i (x,t)+Ω(f i (x,t)
[0010] Where: f i 、c i are the particle distribution on the grid and the discrete particle velocity, respectively; δ t is the evolution time step; Ω is the operator describing particle collisions.
[0011] In the two-dimensional D2Q9 model, nine discrete speeds c are used. i , and the particle distribution f i (x, t) corresponds one to one. The core operations of LBM include flow step and collision step. The flow step makes f i Along c i Migrate to adjacent nodes to simulate particle flow; the collision step considers the interaction between particles within the node and adjusts f i , to achieve the simulation of macroscopic fluid behavior.
[0012] f i (x,t)=f i (x,t)+Ω(f i (x,t)
[0013] f i (x+c i δ t ,t+δ t )=f i ′(x,t)
[0014] f i (x,t)=f i (x,t)+Ω(f i (x, t)), represents the dynamic changes of particles during the collision process, f i (x+c i δ t ,t+δ t )=f i ′(x,t), describes the migration behavior of particles in the system. Once the basic D2Q9 model is determined, the collision operator Ω becomes the key, which carefully describes the evolution of the particle distribution function, which is also the core of the complexity of the lattice Boltzmann method (LBM) equation. Therefore, in many literatures and studies, the choice of Ω has been carefully considered based on various considerations such as asymptotic behavior analysis, computational accuracy, memory usage efficiency, and numerical stability. Especially in the field of low Reynolds number and nearly incompressible fluid dynamics, the LBGK model has been proven to be a reliable and effective LBM model. The LBGK model is a concise LBM model built based on the BGK collision model operator, and the BGK collision model operator can be expressed as:
[0015]
[0016] Where: f eq is the equilibrium distribution; τ is the relaxation time.
[0017] The equilibrium distribution function is expressed as:
[0018]
[0019] Where: i 、c s are the lattice weight and the speed of sound, respectively.
[0020] 2. Particle flow numerical simulation software. In the discrete element method, some commonly used simulation software include DEM (Discrete Element Method), 3DEC (3-Dimensional Discrete Element Code), PFC (Particle Flow Code) and Yade, etc. Among them, PFC (Particle Flow Code) is a widely used discrete element analysis software, which is built on a general discrete element unit model framework. The software can simulate the displacement and rotation behavior between particles and connect these particles through a suitable contact mechanism. Within a limited spatial range, PFC can simulate the movement of particles and the interaction between them, and can automatically detect newly generated contact points during the calculation process. This function is particularly suitable for the simulation of polyhedral particles with constantly changing shapes. The numerical simulations performed in the present invention are all based on PFC software.
[0021] The technical solution of the present invention is:
[0022] A method for optimizing the full support and sanding strength of deep reservoir fractures, comprising:
[0023] From a microscopic perspective, combined with the in-situ stress conditions of deep reservoirs, a permeability calculation equation is established based on the lattice-Boltzmann method;
[0024] The fracture width variation was calculated by discrete element method taking into account the embedding and breakage of proppant;
[0025] The relationship between the concentration of topdressing sand and the flow conductivity under different closing pressures is calculated, that is, the flow conductivity calculation equation;
[0026] Calculate the reservoir permeability of the research block from a macroscopic perspective and further classify the reservoir types;
[0027] Establish the relationship between conductivity and production in case of fine reservoir division;
[0028] Taking the optimal output as the objective function, the optimal sand adding plan is obtained by combining micro and macro perspectives.
[0029] Preferably, according to the present invention, a permeability calculation equation is established based on the lattice-Boltzmann method, including:
[0030] Based on the lattice-Boltzmann method, the microscopic molecular dynamics characteristics of the fluid are focused on, the fluid is divided into discrete particle elements, and the LBGK model is adopted. The evolution equation of the LBGK model is expressed as:
[0031]
[0032] Where: f i (x, t) refers to the particle distribution function; Δt refers to the time step, s; e i is the discrete velocity, m / s; τ is the dimensionless relaxation time; f i eq (x, t) refers to the particle equilibrium distribution function; i refers to the different discrete velocity directions of each grid point, i = 0, 1, ..., n; x refers to the spatial position of the particle; t refers to time, s;
[0033] For the flow of fluids filling fractures, the permeability of deep reservoirs is determined by Darcy's law, which is:
[0034]
[0035] Where: K refers to absolute permeability, 10 -3 μm 2 ; μ refers to fluid viscosity, mPa·s; v a It refers to the average seepage velocity of the fluid, m / s; dp / dx refers to the average pressure gradient, MPa / m.
[0036] Preferably, according to the present invention, the discrete element method is used to calculate the fracture width change value and the conductivity calculation equation in consideration of proppant embedding and crushing; including:
[0037] From a microscopic perspective, the discrete element method is applied, the fish language program is written using the particle flow numerical simulation software (PFC), a deep reservoir proppant embedding model is established based on the three-axis servo principle, the control variable method is used to conduct a sensitivity analysis of the closure pressure, and the change of the fracture width before and after the proppant is embedded in the deep reservoir is calculated;
[0038] Establish a calculation model for the change of proppant crack width: Assume that proppant is laid in multiple layers and arranged in close rhombuses; according to the theory of elastic mechanics, in the case of multiple layers, the total change of crack width is caused by the cumulative deformation of each layer of proppant and the embedding effect of proppant in direct contact with the crack wall, and only the layer of proppant adjacent to the crack wall is embedded;
[0039] Based on the established deep reservoir proppant embedding model and Hertz's hypothesis, the macroscopic proppant fracture width change calculation model is calculated:
[0040]
[0041] Where: α is the change in fracture width, mm; K is the distance coefficient; p is the closing pressure, MPa; D is the proppant particle size, D 1 is the proppant particle size, mm; D 2 is the effective thickness of the crack wall, mm; E 1 、E 2 is the elastic modulus of the model, MPa; v 1 is the Poisson's ratio of proppant 1; v 2 is the Poisson's ratio of proppant 2; n is the number of paving layers; MPa -1 ;
[0042] The seam width change value is calculated by the support seam width change calculation model.
[0043] According to a preferred embodiment of the present invention, the flow conductivity calculation equation is the product of α and K.
[0044] Preferably, according to the present invention, the reservoir permeability of the research block is calculated from a macroscopic perspective, and the reservoir types are further divided; including:
[0045] The proppant fracture width variation calculation model is used to predict the fracture conductivity under different closure pressures.
[0046] Calculate the reservoir permeability of the study area from a macro perspective:
[0047]
[0048] Where: K is the absolute permeability, mD; S wb is the irreducible water saturation, Φ is the porosity, %, and a, b, c are empirical coefficients related to lithology.
[0049] Preferably, according to the present invention, the reservoir types are further divided; including:
[0050] Initial assessment of the permeability properties of different levels of deep reservoir formations, including:
[0051] 1) Collect basic data: Collect basic geological data about the reservoir, including the reservoir's lithology, thickness, burial depth, and structural characteristics;
[0052] 2) Analyze reservoir structure: By analyzing the structural characteristics of the reservoir's bedding, fractures, and caves, preliminarily determine the reservoir's permeability;
[0053] 3) Evaluate the seepage capacity: Evaluate the seepage capacity by calculating the permeability and flow coefficient of the reservoir;
[0054] 4) Classification of permeability levels: According to the evaluation results of seepage capacity, the reservoir is divided into different permeability levels;
[0055] 5) Comprehensively evaluate and compare the penetration performance of different levels;
[0056] Subsequently, the K-means clustering algorithm is used to further refine the reservoir classification, including:
[0057] Assuming that the sample set is {x(1), x(2), … x(n)}, x(i)∈R(n), the distance is calculated according to the Euclidean distance calculation formula shown in formula (6):
[0058]
[0059] x i is the i-th data point; x j is the jth data center; d is the dimension of the data; x ik is x i The value in the kth dimension; x jk is x j The value in the kth dimension;
[0060] Select k cluster centers and calculate the category c to which each sample i belongs according to the formula shown in formula (7): (i) :
[0061] c (i) =argmin||x (i) -μ j || 2 (7)
[0062] μ j is the center of mass; x (i) It belongs to the center of mass μ that is closest to him j Cluster c (i) ;
[0063] For each type of center point μ j , as shown in formula (8), the final cluster center point is obtained by updating the calculation:
[0064]
[0065] Repeat formula (7) and formula (8) until the sum of the squares of the Euclidean distances from all samples to the cluster center of the cluster to which they belong is minimized and finally converges to obtain the global optimal or local optimal cluster set.
[0066] According to the preferred embodiment of the present invention, based on the two factors of closure pressure and reservoir permeability, each factor selects several different values according to actual production conditions;
[0067] A production prediction method based on Transformer-MLP is further used to study the reasonable conductivity of fractures under different parameter combinations, including:
[0068] According to the selected parameters and their corresponding value ranges, the Transformer-MLP deep learning model for production forecasting is configured accordingly;
[0069] Initially set the fracture conductivity to a low value, run the Transformer-MLP deep learning model, and calculate the reservoir production;
[0070] Gradually and evenly increase the value of fracture conductivity, and rerun the numerical simulation for each new value, recording the corresponding cumulative oil production; continue this process until it is observed that the increase in cumulative oil production becomes small as the fracture conductivity increases, that is, it reaches a certain predetermined threshold; at this time, the current fracture conductivity is considered to be close to the optimal;
[0071] The above steps are repeated while considering different reservoir permeabilities and closure pressures until all possible parameter combinations in the parameter table have been tested.
[0072] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a method for optimizing the full-support sanding strength of deep reservoir fractures when executing the computer program.
[0073] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for optimizing the full-support sanding strength of deep reservoir fractures.
[0074] The deep reservoir fracture full support sand strength optimization system includes:
[0075] The permeability calculation equation establishment module is configured as follows: from a microscopic perspective, combined with the deep reservoir in-situ stress conditions, the permeability calculation equation is established based on the lattice-Boltzmann method;
[0076] The fracture width variation value acquisition module is configured to: calculate the fracture width variation value by using a discrete element method taking into account the embedding and breaking of the proppant;
[0077] The flow conductivity calculation equation acquisition module is configured to: calculate the relationship between the concentration of the topdressing sand and the flow conductivity under different closing pressures, that is, the flow conductivity calculation equation;
[0078] The reservoir type classification module is configured to: calculate the reservoir permeability of the research block from a macroscopic perspective and further classify the reservoir types;
[0079] The module for establishing the relationship between conductivity and production is configured to: establish the relationship between conductivity and production in the case of fine reservoir division;
[0080] The optimal sand adding scheme acquisition module is configured to obtain the optimal sand adding scheme by taking the optimal output as the objective function and combining the microscopic and macroscopic perspectives.
[0081] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0082] 1. Compared with the past, the present invention uses the lattice Boltzmann method when calculating reservoir permeability. Compared with the traditional CFD method, the lattice Boltzmann method is simpler in mathematical processing and relatively easy to handle complex boundary conditions.
[0083] 2. Compared with the past, the present invention uses the K-means algorithm to classify reservoirs with different permeability characteristics in detail. Through the K-means algorithm, the reservoir can be divided into multiple precise groups, and the permeability characteristics in each group are relatively consistent. Then, the present invention combines the permeability data of the reservoir to design an optimal conductivity calculation formula for each layer. Since the comprehensive influence of closure pressure and permeability is taken into account, the required sand concentration under different reservoir conditions can be accurately obtained.
[0084] 3. Compared with the past, the present invention can not only provide customized and differentiated sand adding solutions for each type of reservoir, but also ensure the best propped fracture diversion effect under various reservoir conditions, which helps to enhance the production contribution of branch fractures and improve the overall production efficiency of deep formation fracturing wells. At the same time, it optimizes the design of fracturing operations, making sand adding operations more precise and economical. BRIEF DESCRIPTION OF THE DRAWINGS
[0085] Figure 1 It is a schematic diagram of the overall process of fully supporting sanding strength optimization of deep reservoir fractures according to the present invention;
[0086] Figure 2 Schematic diagram of proppant embedding model established based on discrete element method;
[0087] Figure 3 A diagram of a proppant embedding model for a deep reservoir of proppant;
[0088] Figure 4 Schematic diagram of proppant embedment depth changing with closure pressure;
[0089] Figure 5Schematic diagram of the cumulative oil production curve changing with fracture conductivity. DETAILED DESCRIPTION
[0090] The present invention will be further defined below in conjunction with the accompanying drawings and embodiments, but is not limited thereto.
[0091] Example 1
[0092] A method for optimizing the strength of fully supported sand in deep reservoir fractures, such as Figure 1 As shown, including:
[0093] From a microscopic perspective, combined with the in-situ stress conditions of deep reservoirs, a permeability calculation equation is established based on the lattice-Boltzmann method;
[0094] The fracture width variation was calculated by discrete element method taking into account the embedding and breakage of proppant;
[0095] The relationship between the concentration of topdressing sand and the flow conductivity under different closing pressures is calculated, that is, the flow conductivity calculation equation;
[0096] Calculate the reservoir permeability of the research block from a macroscopic perspective and further classify the reservoir types;
[0097] Establish the relationship between conductivity and production in case of fine reservoir division;
[0098] Taking the optimal output as the objective function, the optimal sand adding plan is obtained by combining micro and macro perspectives.
[0099] Example 2
[0100] The difference between the method for optimizing the full support and sanding strength of deep reservoir fractures described in Example 1 is that:
[0101] The permeability calculation equation is established based on the lattice-Boltzmann method, including:
[0102] Based on the lattice-Boltzmann method, the microscopic molecular dynamics characteristics of the fluid are focused on, the fluid is divided into discrete particle elements, and based on this, a mathematical equation describing its motion law is constructed. Here, the LBGK model is used, and the evolution equation of the LBGK model is expressed as:
[0103]
[0104] Where: f i (x, t) refers to the particle distribution function; Δt refers to the time step, s; e i is the discrete velocity, m / s; τ is the dimensionless relaxation time; f i eq(x, t) refers to the particle equilibrium distribution function; i refers to the different discrete velocity directions of each grid point, i = 0, 1, ..., n; x refers to the spatial position of the particle; t refers to time, s;
[0105] On this basis, for the flow of fluids filling the fractures, the permeability of the deep reservoir is determined by Darcy's law, that is:
[0106]
[0107] Where: K refers to absolute permeability, 10 -3 μm 2 ; μ refers to fluid viscosity, mPa·s; v a It refers to the average seepage velocity of the fluid, m / s; dp / dx refers to the average pressure gradient, MPa / m.
[0108] The discrete element method is used to calculate the fracture width change value and the conductivity calculation equation taking into account the embedding and crushing of the proppant; Figure 2 As shown, including:
[0109] From a microscopic perspective, the discrete element method is applied, the fish language program is written using the particle flow numerical simulation software (PFC), and the deep reservoir proppant embedding model is established based on the three-axis servo principle. Figure 3 The proppant embedding model diagram is presented, and the sensitivity analysis of closure pressure is conducted using the control variable method to calculate the change in fracture width before and after the proppant embedding process in deep reservoirs;
[0110] Establish a calculation model for the change of proppant crack width: Assume that proppant is laid in multiple layers and arranged in close rhombuses; according to the theory of elastic mechanics, in the case of multiple layers, the total change of crack width is caused by the cumulative deformation of each layer of proppant and the embedding effect of proppant in direct contact with the crack wall, and only the layer of proppant adjacent to the crack wall is embedded;
[0111] Considering the influence of closure pressure on proppant embedding, the proppant selected was 20 / 40 mesh ceramsite. The test paving concentration was 5kg / m 2 The fluid medium in the conductivity test is distilled water, and the test pressures are 60, 65, 70, and 75 MPa, respectively. The simulated closing pressure is loaded at a rate of 3.45 MPa / min. When the specified test pressure is reached, it is allowed to stabilize for a period of time. Figure 4 The concentration of sand is 5kg / m 2 The 20 / 40 mesh ceramsite proppant embedment test results.
[0112] When calculating the embedding depth, the displacement gauges at both ends of the embedding model are used for measurement, i.e. h t =(W 0-W 1 )h t is the embedding depth, mm; W 0 is the initial fracture width simulated in deep reservoir; W 1 is the fracture width at any time in deep reservoir simulation.
[0113] Based on the established deep reservoir proppant embedding model and Hertz's hypothesis, the macroscopic proppant fracture width change calculation model is calculated:
[0114]
[0115] Where: α is the change in fracture width, mm; K is the distance coefficient; p is the closing pressure, MPa; D is the proppant particle size, D 1 is the proppant particle size, mm; D 2 is the effective thickness of the crack wall, mm; E 1 、E 2 is the elastic modulus of the model, MPa; v 1 is the Poisson's ratio of proppant 1; v 2 is the Poisson's ratio of proppant 2; n is the number of paving layers; MPa -1 ;
[0116] The seam width change value is calculated by the support seam width change calculation model.
[0117] The conductivity of the propped fractures under different closure pressures is analyzed for the reservoir, and the conductivity mathematical equation is calculated. The conductivity of the fracture, that is, the size of the space it provides for fluid flow, is a key indicator to measure the success of this process. Excellent fracture conductivity and its harmonious match with the formation seepage characteristics are crucial to improving the production gain effect of the operation. Therefore, in the design, ensuring that the fracture has high conductivity to optimize the flow of formation fluids is a key factor in maximizing oil and gas production. In this study, the conductivity calculation equation can be listed as the product of α and K.
[0118] According to different stress conditions, the above method is used to obtain the reservoir permeability, fracture width and conductivity.
[0119] The reservoir permeability of the research block is calculated from a macroscopic perspective, and the reservoir types are further divided; including:
[0120] Based on the established macroscopic proppant fracture width variation calculation model, the conductivity under different closing pressure conditions can be effectively evaluated. Closing pressure, as the pressure applied between fracture surfaces due to formation pressure, has a direct impact on fracture width. As the closing pressure increases, the fracture width will decrease accordingly, which will further affect the sand concentration, that is, the mass of proppant laid on the fracture per unit area. Since the role of the proppant is to keep the fracture open to facilitate the flow of formation fluid, the increase in closing stress leads to a decrease in sand concentration, thereby affecting the conductivity of the fracture. The conductivity is directly related to the production and efficiency of oil and gas wells. Therefore, the proppant fracture width variation calculation model is used to predict the conductivity of the fracture under different closing pressures, providing an important basis for optimizing fracturing design and improving oil and gas well performance.
[0121] Calculate the reservoir permeability of the study area from a macro perspective:
[0122]
[0123] Where: K is the absolute permeability, mD; S wb is the irreducible water saturation, Φ is the porosity, %, and a, b, c are empirical coefficients related to lithology.
[0124] Reservoir types are further divided into:
[0125] Initial assessment of the permeability properties of different levels of deep reservoir formations, including:
[0126] 1) Collect basic data: Collect basic geological data about the reservoir, including the reservoir's lithology, thickness, burial depth, and structural characteristics;
[0127] 2) Analyze reservoir structure: The structure of the reservoir has an important influence on the permeability. By analyzing the structural characteristics of the reservoir's bedding, cracks, and caves, the permeability of the reservoir can be preliminarily determined;
[0128] 3) Evaluate the seepage capacity: The seepage capacity is a key indicator for evaluating the permeability performance. The seepage capacity is evaluated by calculating the permeability and flow coefficient of the reservoir;
[0129] 4) Classification of permeability levels: Based on the evaluation results of seepage capacity, the reservoir is divided into different permeability levels; the classification standard can be determined according to actual conditions and engineering requirements, and can usually be divided according to the size of permeability or the range of flow coefficient.
[0130] 5) Comprehensively evaluate and compare the permeability performance of different levels; this step helps to combine the physical properties of the reservoir and then develop a detailed reservoir classification and rating method.
[0131] Subsequently, the K-means clustering algorithm is used to further refine the reservoir classification, so as to achieve a more accurate classification evaluation. Including:
[0132] Assuming that the sample set is {x(1), x(2), … x(n)}, x(i)∈R(n), the distance is calculated according to the Euclidean distance calculation formula shown in formula (6):
[0133]
[0134] x i is the i-th data point; x j is the jth data center; d is the dimension of the data; x ik is x i The value in the kth dimension; x jk is x j The value in the kth dimension;
[0135] Select k cluster centers and calculate the category c to which each sample i belongs according to the formula shown in formula (7): (i) :
[0136] c (i) =argmin||x (i) -μ j || 2 (7)
[0137] μ j is the center of mass; x (i) It belongs to the center of mass μ that is closest to him j Cluster c (i) ;
[0138] For each type of center point μ j , as shown in formula (8), the final cluster center point is obtained by updating the calculation:
[0139]
[0140] Repeat formula (7) and formula (8) until the sum of the squares of the Euclidean distances from all samples to the cluster center of the cluster to which they belong is minimized and finally converges to obtain the global optimal or local optimal cluster set.
[0141] In reservoirs with different permeabilities, the width, length, and surface roughness of the cracks will be different, and the conductivity will also be different. For reservoirs of different levels, in order to achieve the best conductivity, different proppant types and particle sizes are selected for each layer. This helps to optimize the proppant laying plan, so that the proppant can more effectively support the cracks and improve the conductivity of the cracks.
[0142] According to the two factors of closure pressure and reservoir permeability, 5 different values are selected for each factor according to the actual production conditions. The specific values are shown in Table 1.
[0143] Table 1
[0144] Considerations Parameter selection Reservoir permeability, in md 2、4、6、8、10 Closing stress, in MPa 60、65、70、75、80
[0145] Based on the two factors of closure pressure and reservoir permeability, several different values are selected for each factor according to actual production conditions;
[0146] A production prediction method based on Transformer-MLP is further used to study the reasonable conductivity of fractures under different parameter combinations, including:
[0147] According to the selected parameters and their corresponding value ranges, the Transformer-MLP deep learning model for production forecasting is configured accordingly;
[0148] Initially set the fracture conductivity to a low value, run the Transformer-MLP deep learning model, set the time step to 120 days, and calculate the reservoir production;
[0149] The fracture conductivity value was increased gradually and at equal intervals, and the numerical simulation was rerun for each new value, recording the corresponding 120-day cumulative oil production. This process was continued until the increase in cumulative oil production became negligible with increasing fracture conductivity, i.e., a predetermined threshold (25000m3) was reached. 3 ); At this time, it is considered that the current fracture conductivity is close to the optimal;
[0150] The above steps are repeated while considering different reservoir permeabilities and closure pressures until all possible parameter combinations in the parameter table have been tested.
[0151] According to the above production prediction method, the conductivity and production show a positive correlation. As the conductivity increases, the production will also increase accordingly. This is because the increase in conductivity means that the fluid flows more smoothly in the fracture, so that oil and gas can be extracted from the reservoir more effectively. In addition, it is worth noting that the attenuation coefficient of the fracture conductivity will also affect the production. The smaller the attenuation coefficient, the smaller the decrease in the fracture conductivity during the production process, and the slower the decline in production.
[0152] In order to analyze the influence of fracture conductivity on the cumulative oil production of the model under different combinations of factors, the values of fracture conductivity will be adjusted one by one, and the data obtained by the ant colony algorithm will be used to observe the changing trend of cumulative oil production with the increase of fracture conductivity. Figure 5It shows how the cumulative oil production varies with the change of fracture conductivity under a specific combination of parameters.
[0153] Sand concentration refers to the mass of proppant placed on the fracture per unit area, which determines the width and conductivity of the fracture, which is directly related to the production. There is an optimal sand concentration that can provide the best conductivity under different closure pressures. This optimal value depends on many factors, including formation permeability and closure pressure, so the optimal sand intensity can be obtained by combining the influence of the above two factors.
[0154] Evaluate the impact of changes in fracture conductivity on production dynamics, especially in the early and long-term production process. Adjust fracturing operation parameters such as pump pressure, displacement, fracturing fluid type, etc. to optimize fracture morphology and conductivity. Select the best proppant type, concentration and laying method based on the experimental results of fracture conductivity and the production prediction model.
[0155] Taking the optimal output as the objective function, the optimal sand addition scheme is obtained by combining microscopic and macroscopic perspectives, including:
[0156] The sand adding process involves the selection of proppant and fracturing fluid type and quantity, the planning of parameters such as the length, width and shape of the fracture, and the design of the fracturing pumping program. In order to achieve the expected fracture conductivity, it is necessary to accurately calculate the optimal amount of sand adding, and this result has important reference value for the optimization of the sand adding process in actual construction.
[0157] Based on the previous investigation of deep fully supported fracture reservoirs, the reasonable range of the optimal conductivity of single-stage fracturing in the sand-added numerical model has been set. At the same time, the hydraulic fracturing fracture morphology parameters corresponding to different single-stage fracturing models are set. 40 / 70 mesh and 70 / 140 mesh quartz sands are selected as proppants. Under the condition of a closing pressure of 50MPa, 2, 4, 6 and 12 clusters of single-stage fracturing simulations were carried out using 40 / 70 mesh and 70 / 140 mesh quartz sands through numerical simulation methods, and a total of 8 simulations were carried out.
[0158] Next, the amount of sand added was calculated and the fracturing effect was evaluated. During the sand addition process, a unified pumping procedure was adopted and the pumping speed was set at 14m / s. 3 / min, the initial sand concentration is 150kg / m 3 The maximum sand concentration is 240kg / m 3 After obtaining the optimal sand addition amount of the simulated single stage, it was compared with the actual construction situation to obtain the optimal sand addition scale of the full support process for deep reservoir fractures.
[0159] Example 3
[0160] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method for optimizing the full support and sanding strength of deep reservoir fractures described in Example 1 or 2 are implemented.
[0161] Example 4
[0162] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for optimizing the full-support sanding strength of deep reservoir fractures described in Example 1 or 2.
[0163] Example 5
[0164] The deep reservoir fracture full support sand strength optimization system includes:
[0165] The permeability calculation equation establishment module is configured as follows: from a microscopic perspective, combined with the deep reservoir in-situ stress conditions, the permeability calculation equation is established based on the lattice-Boltzmann method;
[0166] The fracture width variation value acquisition module is configured to: calculate the fracture width variation value by using a discrete element method taking into account the embedding and breaking of the proppant;
[0167] The flow conductivity calculation equation acquisition module is configured to: calculate the relationship between the concentration of the topdressing sand and the flow conductivity under different closing pressures, that is, the flow conductivity calculation equation;
[0168] The reservoir type classification module is configured to: calculate the reservoir permeability of the research block from a macroscopic perspective and further classify the reservoir types;
[0169] The module for establishing the relationship between conductivity and production is configured to: establish the relationship between conductivity and production in the case of fine reservoir division;
[0170] The optimal sand adding scheme acquisition module is configured to obtain the optimal sand adding scheme by taking the optimal output as the objective function and combining the microscopic and macroscopic perspectives.
Claims
1. A method for optimizing the strength of fully supported sand in deep reservoir fractures. It is characterized in that include: From a microscopic perspective, combined with the in-situ stress conditions of deep reservoirs, a permeability calculation equation is established based on the lattice-Boltzmann method; The fracture width variation was calculated by discrete element method taking into account the embedding and breakage of proppant; The relationship between the concentration of topdressing sand and the flow conductivity under different closing pressures is calculated, that is, the flow conductivity calculation equation; Calculate the reservoir permeability of the research block from a macroscopic perspective and further classify the reservoir types; Establish the relationship between conductivity and production in case of fine reservoir division; Taking the optimal output as the objective function, the optimal sand adding plan is obtained by combining micro and macro perspectives.
2. A method for optimizing the full support and sanding strength of deep reservoir fractures according to claim 1, It is characterized in that The permeability calculation equation is established based on the lattice-Boltzmann method, including: Based on the lattice-Boltzmann method, the microscopic molecular dynamics characteristics of the fluid are focused on, the fluid is divided into discrete particle elements, and the LBGK model is adopted. The evolution equation of the LBGK model is expressed as: Where: f i (x, t) refers to the particle distribution function; Δt refers to the time step, s; e i is the discrete velocity, m / s; τ is the dimensionless relaxation time; f i eq (x, t) refers to the particle equilibrium distribution function; i refers to the different discrete velocity directions of each grid point, i = 0, 1, ..., n; x refers to the spatial position of the particle; t refers to time, s; For the flow of fluids filling fractures, the permeability of deep reservoirs is determined by Darcy's law, which is: Where: K refers to absolute permeability, 10 -3 μm 2 ; μ refers to fluid viscosity, mPa·s; v a It refers to the average seepage velocity of the fluid, m / s; dp / dx refers to the average pressure gradient, MPa / m.
3. A method for optimizing the full support and sanding strength of deep reservoir fractures according to claim 1, It is characterized in that The discrete element method is used to calculate the fracture width change value and the conductivity calculation equation taking into account the embedding and crushing of the proppant; including: From a microscopic perspective, the discrete element method is applied, the fish language program is written using the particle flow numerical simulation software, a deep reservoir proppant embedding model is established based on the three-axis servo principle, the control variable method is used to conduct a sensitivity analysis of the closure pressure, and the change in the fracture width before and after the proppant is embedded in the deep reservoir is calculated; Establish a calculation model for the change of proppant crack width: Assume that proppant is laid in multiple layers and arranged in close rhombuses; according to the theory of elastic mechanics, in the case of multiple layers, the total change of crack width is caused by the cumulative deformation of each layer of proppant and the embedding effect of proppant in direct contact with the crack wall, and only the layer of proppant adjacent to the crack wall is embedded; Based on the established deep reservoir proppant embedding model and Hertz's hypothesis, the macroscopic proppant fracture width change calculation model is calculated: Where: α is the change in fracture width, mm; K is the distance coefficient; p is the closing pressure, MPa; D is the proppant particle size, D 1 is the proppant particle size, mm; D 2 is the effective thickness of the crack wall, mm; E 1 、E 2 is the elastic modulus of the model, MPa; v 1 is the Poisson's ratio of proppant 1; v 2 is the Poisson's ratio of proppant 2; n is the number of paving layers; The seam width change value is calculated by the support seam width change calculation model.
4. A method for optimizing the full support and sanding strength of deep reservoir fractures according to claim 1, It is characterized in that The flow conductivity calculation equation is the product of α and K.
5. A method for optimizing the full support and sanding strength of deep reservoir fractures according to claim 1, It is characterized in that The reservoir permeability of the research block is calculated from a macroscopic perspective, and the reservoir types are further divided; including: The proppant fracture width variation calculation model is used to predict the fracture conductivity under different closure pressures. Calculate the reservoir permeability of the study area from a macro perspective: Where: K is the absolute permeability, mD; S wb is the irreducible water saturation, Φ is the porosity, %, and a, b, c are empirical coefficients related to lithology.
6. A method for optimizing the full support and sanding strength of deep reservoir fractures according to claim 1, It is characterized in that Further classification of reservoir types is underway; including: Initial assessment of the permeability properties of different levels of deep reservoir formations, including: 1) Collect basic data: Collect basic geological data about the reservoir, including the reservoir's lithology, thickness, burial depth, and structural characteristics; 2) Analyze reservoir structure: By analyzing the structural characteristics of the reservoir's bedding, fractures, and caves, preliminarily determine the reservoir's permeability; 3) Evaluate the seepage capacity: Evaluate the seepage capacity by calculating the permeability and flow coefficient of the reservoir; 4) Classification of permeability levels: According to the evaluation results of seepage capacity, the reservoir is divided into different permeability levels; 5) Comprehensively evaluate and compare the penetration performance of different levels; Subsequently, the K-means clustering algorithm is used to further refine the reservoir classification, including: Assuming that the sample set is {x(1), x(2), … x(n)}, x(i)∈R(n), the distance is calculated according to the Euclidean distance calculation formula shown in formula (6): x i is the i-th data point; x j is the jth data center; d is the dimension of the data; x ik is x i The value in the kth dimension; x jk is x j The value in the kth dimension; Select k cluster centers and calculate the category c to which each sample i belongs according to the formula shown in formula (7): (i) : c (i) =argmin||x (i) -m j || 2 (7) μ j is the center of mass; x (i) It belongs to the center of mass μ that is closest to him j Cluster c (i) ; For each type of center point μ j , as shown in formula (8), the final cluster center point is obtained by updating the calculation: Repeat formula (7) and formula (8) until the sum of the squares of the Euclidean distances from all samples to the cluster center of the cluster to which they belong is minimized and finally converges to obtain the global optimal or local optimal cluster set.
7. A method for optimizing the full support and sanding strength of deep reservoir fractures according to claim 1, It is characterized in that Based on the two factors of closure pressure and reservoir permeability, several different values are selected for each factor according to actual production conditions; A production prediction method based on Transformer-MLP is further used to study the reasonable conductivity of fractures under different parameter combinations, including: According to the selected parameters and their corresponding value ranges, the Transformer-MLP deep learning model for production forecasting is configured accordingly; Initially set the fracture conductivity to a low value, run the Transformer-MLP deep learning model, and calculate the reservoir production; Gradually and evenly increase the value of fracture conductivity, and rerun the numerical simulation for each new value, recording the corresponding cumulative oil production; continue this process until it is observed that the increase in cumulative oil production becomes small as the fracture conductivity increases, that is, it reaches a certain predetermined threshold; at this time, the current fracture conductivity is considered to be close to the optimal; The above steps are repeated while considering different reservoir permeabilities and closure pressures until all possible parameter combinations in the parameter table have been tested.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program. It is characterized in that When the processor executes the computer program, the steps of the method for optimizing the full-support sanding strength of deep reservoir fractures described in any one of claims 1-7 are implemented.
9. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method for optimizing the full-support sanding strength of deep reservoir fractures according to any one of claims 1 to 7 are implemented. 10.Fully supported sand adding strength optimization system for deep reservoir fractures, include: The permeability calculation equation establishment module is configured as follows: from a microscopic perspective, combined with the deep reservoir in-situ stress conditions, the permeability calculation equation is established based on the lattice-Boltzmann method; The fracture width variation value acquisition module is configured to: calculate the fracture width variation value by using a discrete element method taking into account the embedding and breaking of the proppant; The flow conductivity calculation equation acquisition module is configured to: calculate the relationship between the concentration of the topdressing sand and the flow conductivity under different closing pressures, that is, the flow conductivity calculation equation; The reservoir type classification module is configured to: calculate the reservoir permeability of the research block from a macroscopic perspective and further classify the reservoir types; The module for establishing the relationship between conductivity and production is configured to: establish the relationship between conductivity and production in the case of fine reservoir division; The optimal sand adding scheme acquisition module is configured to obtain the optimal sand adding scheme by taking the optimal output as the objective function and combining the microscopic and macroscopic perspectives.
Citation Information
Patent Citations
Method for determining crack permeability of crack-pore seepage experiment model
CN102619502A
Hydrofracture design parameter optimization method based on fine classification of reservoir
CN105003240A
Optimizing method for parameters of hydraulic fracturing proppant for sandstone reservoir
CN108386179A
Permeability prediction method based on block numerical simulation of core scanning images
CN108763711A
Flow conductivity predicting method considering particle size distribution of propping agent
CN111720105A
Cited By
Experimental device and experimental method for fracture evolution and seepage coupling model on confined water
CN120291927A
