A profile control and displacement program design and effect prediction method based on a capacity resistance model

Abstract

The application provides a profile control and flooding scheme design and effect prediction method based on a capacity resistance model, and belongs to the field of oil and gas field development, and comprises the following steps: step one: establishing a water flooding capacity resistance model of a target oil reservoir block; step two: establishing an advantage channel identification model to realize well and layer selection decision; step three: establishing a profile control and flooding permeability representation model to construct a profile control and flooding capacity resistance model; and step four: performing profile control and flooding scheme design and effect prediction. The application adopts the capacity resistance model and the ensemble Kalman filter inversion method, does not need the traditional complex geological modeling process based on a grid, can quickly perform dynamic quantitative representation on interwell connectivity coefficients, realizes profile control and flooding well and layer selection decision, simultaneously adopts the ensemble Kalman filter method to perform history matching, can systematically reduce the uncertainty of parameters such as stratified connectivity coefficients and stratified water injection ratios, brings the parameters into the profile control and flooding permeability representation model to construct the flooding capacity resistance model, and can realize profile control and flooding scheme effect prediction.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of oil and gas field development, and particularly relates to a profile control and flooding scheme design and effect prediction method based on a capacity resistance model. BACKGROUND

[0002] Currently, the commonly used methods for identifying the advantage seepage channel in the industry mainly include logging methods, interwell dynamic monitoring methods, well testing methods, coring methods and tracer methods. However, the logging and coring methods can only reflect the situation at the test point and cannot reflect the overall development scale of the advantage seepage channel and the interwell situation, and cannot accurately calculate the development scale and interwell size of the advantage seepage channel. The tracer and well testing methods depend on special testing, and have long cycle and high cost. The existing interwell dynamic monitoring methods mostly focus on making qualitative judgment on whether the advantage seepage channel is developed, and lack quantitative description of the parameters. Moreover, the above existing technologies are often limited by the factors such as less field testing, high cost, long cycle, occupation of offshore platform operation space, and thus the application is limited, and it is difficult to meet the actual needs of offshore oilfield development and production.

[0003] The data required for the research on the interwell advantage channel and the prediction of the effect of profile control and flooding by using numerical simulation software is difficult to obtain, has high cost, relatively complex operation and long calculation cycle, and thus some scholars have proposed to use the capacity resistance model to study the formation permeability advantage channel. However, these research works stop at the profile control and flooding well selection decision after the interwell connection relationship is inverted, and the effect prediction and scheme optimization design of profile control and flooding depend on other professional software or numerical simulation software to carry out related work, without further research on the basis of the same oil reservoir capacity resistance model, and thus the technical advantages of the research on the effect prediction of profile control and flooding on the basis of the analysis of the interwell advantage channel of the same oil reservoir model are lost. At present, no research institution and scholar has coupled the capacity resistance model and the profile control and flooding seepage mechanism model to carry out the research on the effect prediction of profile control and flooding. SUMMARY

[0004] Therefore, the present application aims to provide a profile control and flooding scheme design and effect prediction method based on a capacity resistance model, which does not need the traditional complex geological modeling process based on a grid, can quickly dynamically and quantitatively characterize the interwell connection coefficient, realize the profile control and flooding well selection and layer selection decision, and can systematically reduce the uncertainty of parameters such as the layered connection coefficient and the layered water injection ratio, bring the parameters into a profile control and flooding permeability characterization model to construct a profile control and flooding capacity resistance model, and realize the effect prediction of the profile control and flooding scheme.

[0005] To achieve the above object, the technical scheme of the present application is as follows: a profile control and flooding scheme design and effect prediction method based on a capacity resistance model, comprising the following steps:

[0006] Step 1: establishing a water drive capacity resistance model of a target oil reservoir block;

[0007] Step two: Establishing the dominant channel identification model to realize the well selection and layer selection decision;

[0008] Step three: Establishing the profile control and flooding permeability characterization model, and constructing the profile control and flooding capacity resistance model;

[0009] Step four: Designing the profile control and flooding scheme and predicting the effect.

[0010] Further, the step one uses the relative permeability curve, production dynamic data, well site information and small layer data to establish the water drive capacity resistance model of the target oil reservoir block, comprising:

[0011] (1) Establishing the basic data preparation of the capacity resistance model;

[0012] (2) Establishing the basic theory of the capacity resistance model;

[0013] When the interwell interference is ignored and the bottom hole flowing pressure of the oil well changes little, the estimation formula of the liquid production in the capacity resistance model can be simplified to the following discrete form:

[0014]

[0015]

[0016] In the formula, i=1, 2, …, I, I is the total number of water wells; j=1, 2, …, J, J is the total number of oil wells; n is the discrete time point (n=1, 2, 3, …, N), N is the total number of steps; Δt is the time step, d; is the model calculation value of the liquid production of the jth oil well at the nth time point, m 3 / d; is the initial liquid production of the jth oil well, m 3 / d; τ j is the time constant of the control area of the oil well j, which represents the initial production decline rate and injection-production lag of the oil well j, d; λ I,j is the injection distribution coefficient of the water well i to the oil well j in the i-j injection-production direction; I' i,j (n) is the weighted moving average value of the injection of the water well i in the i-j injection-production direction, m 3 / d;

[0017] The oil production formula is

[0018]

[0019] In the formula, q oj (t n ) is the oil production of the production well j at t n ; q oj (t0) is the oil production of the production well j at the initial time; q oj (t n-1) is the oil production of the production well j at time t n-1 ; q j (t n ) is the liquid production of the production well j at time t n ; and

[0020] (3) Set the constraint condition of the capacity resistance model, as shown in the following formula:

[0021]

[0022] The initial liquid production q 0j and the time constant τ j of each oil well are greater than 0; the water injection distribution coefficient is greater than or equal to 0; the sum of the water injection distribution coefficients of the water injection wells in the direction of each oil well is less than or equal to 1.

[0023] Further, the step two uses the ensemble Kalman filtering method to automatically perform history matching, and the inter-well connectivity coefficient and the time constant of each sublayer are obtained by inversion.

[0024] Further, the step two comprehensively considers the reserves of the target area and the generating dynamic condition, performs multi-dimensional cluster analysis on the liquid production and the connectivity coefficient, and determines the well selection and layer selection decision standard.

[0025] Further, the step two specifically includes:

[0026] (1) Ensemble Kalman filtering inversion;

[0027] The ensemble Kalman filtering method mainly includes three stages of recursive process: data collection stage, data prediction stage and data assimilation stage;

[0028] (2) Determine the well selection and layer selection decision standard by using the K_means cluster analysis method;

[0029] The K_means algorithm is a distance-based clustering algorithm, which uses distance as an evaluation index of similarity, that is, the closer the distance between two objects, the greater the similarity.

[0030] Further, the ensemble Kalman filtering inversion specifically includes:

[0031] First, define the extended state vector y n , as shown in the following formula,

[0032] y n = (m, p, d) T (5)

[0033] In the formula: m is an N m -dimensional fitting parameter vector; p is an N p -dimensional dynamic parameter vector; and d is an N​d The dimension to be fitted is historical production data; y n To expand the state vector, the dimension is N. m +N p +N d ;

[0034] Based on set theory, we define the expression and mean of the set state vector Y. The following formulas are shown respectively:

[0035] Y = (y1, y2, ..., y N (6)

[0036]

[0037] In the formula: N is the number of members in the set, indicating the number of elements within the range of the fitted model;

[0038] Therefore, the recursive process of ensemble Kalman filtering can be obtained as follows:

[0039] (1) Prediction stage: Calculate the predicted state vector y using the capacitive resistance model. f ,

[0040] y f =U[y]+ε w (8)

[0041] In the formula: U represents the calculation process of the capacitive-resistive model; ε w This refers to ambient noise generated during model calculations.

[0042] Therefore, the predicted state vector of the set can be obtained:

[0043]

[0044] (2) Predict the state vector y for the nth iteration time step set. f The elements in the vector are assimilated to obtain the assimilated state vector and its set, as shown in the following equation:

[0045]

[0046]

[0047] In the formula: For set-consistent state vectors;

[0048] To ensure the diversity of observation data among members of a ensemble, it is generally necessary to add random perturbation error to the actual observation data to distinguish ensemble members. The expression is:

[0049] d n =d obs +ε n(12)

[0050] where the Kalman gain matrix is expressed as:

[0051]

[0052] where: is the error covariance matrix of the set of predicted state vectors; C D is the error covariance matrix of the set of observation vectors;

[0053] is obtained as: and

[0054]

[0055]

[0056] where: is the mean of the N set of predicted state members at the nth iteration time step;

[0057] Thus the set of assimilated state vectors is:

[0058]

[0059] where: D is the set of observation vectors, D = (d1, d2, …, d n );

[0060] (3) The interwell connectivity coefficient field map of the small layer is obtained by inversion;

[0061] The connectivity between the injection wells and the effective oil wells is quantified to obtain the specific values of the connectivity coefficients of each small layer, and the connectivity field map and the dominant channel distribution map are drawn.

[0062] Further, the K_means clustering analysis method determines the well selection and layer selection decision criteria, and the specific process is as follows:

[0063] (1) Randomly select K documents as center points from N documents;

[0064] (2) Measure the distance of each remaining document to each center point, and classify it into the nearest centroid class;

[0065] (3) Recalculate the center points of each class obtained;

[0066] (4) Iterate the above steps (2)-(3) until the new centroid is equal to or less than the specified threshold value of the original centroid, and the algorithm ends;

[0067] (5) Two-dimensional clustering analysis of fluid production rate and connectivity coefficient is performed to make well selection and layer selection decisions for profile control and flooding.

[0068] Further, the step three applies the profile control and flooding material experiment model, the numerical simulation profile control and flooding scheme sample, adopts the machine learning algorithm to construct the permeability characterization model after the profile control and flooding, and forms the profile control and flooding capacity resistance model by substituting the profile control and flooding permeability characterization model into the water drive capacity resistance model, and the specific is:

[0069] (1) Establish the profile control and flooding permeability characterization model;

[0070] The change of the permeability is used to represent the mechanism of the profile control and flooding scheme, which is mainly based on the physical model experiment data, and the profile control and flooding parameter and the change rule of the permeability are represented by the model;

[0071] (2) Establish the profile control and flooding capacity resistance model;

[0072] The profile control and flooding permeability characterization model is substituted into the water drive capacity resistance model to form the profile control and flooding capacity resistance model, and the new interwell connectivity coefficient and production index after the profile control and flooding are calculated.

[0073] Further, the step four is based on the profile control and flooding capacity resistance model to predict the effect of the profile control and flooding scheme, and the interwell connectivity coefficient and the oil production of the benefited oil well after the profile control and flooding are predicted.

[0074] Further, the step four is specifically:

[0075] (1) Design the profile control and flooding slug process parameters, and establish the profile control and flooding design scheme set;

[0076] (2) Different profile control and flooding design schemes are brought into the profile control and flooding capacity resistance model, the small layer permeability and the oil and water well production data after the profile control and flooding are predicted, and the profile control and flooding measure effect is compared to finally determine the profile control and flooding slug design parameters.

[0077] Compared with the prior art, the profile control and flooding scheme design and effect prediction method based on the capacity resistance model has the following advantages:

[0078] (1) The capacity resistance model and the ensemble Kalman filter inversion method are used in the application, which does not need the traditional complex geological modeling process based on the grid, can quickly dynamically and quantitatively represent the interwell connectivity coefficient, and realizes the profile control and flooding well selection and layer selection decision;

[0079] (2) The ensemble Kalman filter method is used for history matching, which can systematically reduce the uncertainty of the layered connectivity coefficient and the layered water injection proportion;

[0080] (3) The profile control and flooding permeability characterization model is brought into the capacity resistance model to construct the profile control and flooding capacity resistance model, which can realize the profile control and flooding scheme effect prediction on the basis of the same reservoir model, and the profile control and flooding measure effect is more scientific and reliable. BRIEF DESCRIPTION OF DRAWINGS

[0081] The accompanying drawings, which form a part of this application, are intended to provide further understanding of the application and are incorporated herein in their entirety, and together with the general description of the application and the detailed description of specific embodiments thereof, explain the application. In the drawings:

[0082] Figure 1 A schematic diagram of the workflow of the ensemble Kalman filter of the present application;

[0083] Figure 2 A connectivity field map of the present application;

[0084] Figure 3 A dominant channel distribution map of the present application;

[0085] Figure 4 A schematic diagram of the clustering results of the connectivity coefficient and the fluid production rate by the K_means clustering analysis method. DETAILED DESCRIPTION

[0086] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0087] The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0088] The present application is a profile control and flooding scheme design and effect prediction method based on a capacity resistance model. On the basis of material balance and signal processing, a capacity resistance model is established by using production dynamic data of injection-production units, a dynamic quantitative characterization of interwell connectivity coefficients is quickly realized by using an ensemble Kalman filter inversion method to achieve dominant channel identification, and a clustering analysis method is applied to determine well selection and layer selection decision criteria. A profile control and flooding permeability characterization model is established to form a profile control and flooding capacity resistance model, a plurality of profile control and flooding process parameters are designed to form a scheme set, and profile control and flooding scheme effect prediction is performed. The method comprises the following steps:

[0089] Step 1: Establish a water drive capacity resistance model for the target reservoir block.

[0090] Step 2: Establish a dominant channel identification model to achieve well selection and layer selection decision.

[0091] Step 3: Establish a profile control and flooding permeability characterization model to construct a profile control and flooding capacity resistance model.

[0092] Step 4: Perform profile control and flooding scheme design and effect prediction.

[0093] The step 1 uses relative permeability curves, production dynamic data, well location information, and small layer data to establish a water drive capacity resistance model for the target reservoir block. The specific capacity resistance model establishment method and steps are as follows:

[0094] (1) Prepare the basic data for establishing the capacity resistance model, as shown in Table 1.

[0095] Table 1 data types

[0096]

[0097]

[0098] (2) Establish the basic theory of resistance-capacitance model;

[0099] The resistance-capacitance model can be regarded as an improvement of the multiple linear regression model. It is not from the pure mathematical point of view, but from the physical properties of the reservoir engineering point of view, and the essence of the dynamic connectivity characteristics between injection and production wells is considered. In view of the shortcomings of the multiple linear regression model, such as negative weight coefficient and constant bottom hole pressure of production wells, the resistance-capacitance model creatively introduces a time constant to describe the delay and attenuation of the injection signal, and the effects of compressibility, pore volume and oil recovery index are comprehensively reflected in the nonlinear multiple regression model. Therefore, the connectivity weight coefficient and the time constant can respectively reflect the characteristics of the reservoir and the fluid between the injection well and the production well. In addition, the resistance-capacitance model also takes the bottom hole pressure of the production well into the model, and allows the short-term shut-in situation to occur, which also improves the practicability of the model.

[0100] The resistance-capacitance model does not need the traditional complex geological modeling process based on grid, but only needs to establish an equivalent interwell connectivity network model to describe the reservoir percolation system according to well location and injection-production dynamic data, so as to quantitatively characterize the interwell connectivity, which is faster.

[0101] When the well interference is ignored and the bottom hole pressure of the oil well changes little, the estimation formula of the liquid production rate in the resistance-capacitance model can be simplified as the following discrete form:

[0102]

[0103]

[0104] In the formula, i=1, 2, …, I, I is the total number of water wells; j=1, 2, …, J, J is the total number of oil wells; n is the discrete time point (n=1, 2, 3, …, N), N is the total number of steps; Δt is the time step, d; is the model calculation value of the liquid production rate of the jth oil well at the nth time point, m 3 / d; is the initial liquid production rate of the jth oil well, m 3 / d; τ j is the time constant of the control area of the jth oil well, which represents the initial production decline rate and injection-production lag of the jth oil well, d; λ I,j is the water injection rate distribution coefficient of the ith water well to the jth oil well; I' i,j(n) represents the weighted moving average of the water injection rate of well i in the injection-production direction of well ij, m 3 / d;

[0105] The formula for oil production is:

[0106]

[0107] In the formula, q oj (t n ) for production well j in t n Oil production at any given time; q oj (t0) represents the oil production of production well j at the initial moment; q oj (t n-1 ) for production well j in t n-1 Oil production at any given time; q j (t n ) for production well j in t n The amount of liquid produced at any given time;

[0108] (3) Set the constraints for the capacitive-resistive model as shown in the following equation:

[0109]

[0110] Initial production rate q for each oil well 0j and time constant τ j All are greater than 0; the water injection distribution coefficient is greater than or equal to 0; the sum of the water injection distribution coefficients from the water well to each oil well. Less than or equal to 1.

[0111] Step two uses an ensemble Kalman filter method for automatic history fitting, inverting the well connectivity coefficients and time constants of each sub-layer. Taking into account the target area's reserves and dynamic generation conditions, multi-dimensional cluster analysis is performed on the production rate and connectivity coefficients to clarify the well and layer selection decision criteria. Step two specifically includes:

[0112] (1) Ensemble Kalman Filter (EnKF) Inversion;

[0113] Based on the working principle of ensemble Kalman filtering, we can set as follows: Figure 1 The corresponding iterative flowchart shown illustrates the three recursive processes of ensemble Kalman filtering.

[0114] The ensemble Kalman filter method is mainly divided into three recursive stages: the data set stage, the data prediction stage, and the data assimilation stage.

[0115] First, define the extended state vector y. n As shown in the following formula,

[0116] y n= (m, p, d) T (5)

[0117] where m is N m dimensional vector to be fitted; p is N p dimensional dynamic parameter vector; d is N d dimensional vector to be fitted historical production data; y n is the augmented state vector with dimension N m + N p + N d ;

[0118] According to the set idea, the expression and mean of set state vector Y are defined as follows:

[0119] Y = (y1, y2,..., y N ) (6)

[0120]

[0121] where N is the number of set members, indicating the number within the fitting model range;

[0122] Therefore, the recursive process of set Kalman filter can be obtained as follows:

[0123] ① Prediction stage: the predicted state vector y f is calculated by the capacity resistance model

[0124] y f = U[y] + ε w (8)

[0125] where U is the calculation process of the capacity resistance model; ε w is the surrounding noise generated in the model operation process;

[0126] Therefore, the set predicted state vector can be obtained as follows:

[0127]

[0128] ② The elements in the set predicted state vector y f at the nth iteration time step are assimilated to obtain the assimilated state vector and its set, as shown in the following formula:

[0129]

[0130]

[0131] where: is the set assimilated state vector;

[0132] ​To ensure the difference of each member observation data in the set, generally need to add random disturbance error to the actual observation data to distinguish between set members, the expression is:

[0133] d n = d obs + ε n (12)

[0134] The Kalman gain matrix expression is:

[0135]

[0136] In the formula: is the error covariance matrix of the set prediction state vector; C D is the error covariance matrix of the set observation vector;

[0137]

[0138]

[0139]

[0140] In the formula: is the mean of N set prediction state members at the nth iteration time step;

[0141] Therefore, the assimilation state vector is:

[0142]

[0143] In the formula: D is the set observation vector, D=(d1, d2,…, d n );

[0144] Since the prediction state vector and the assimilation state vector are implemented without linear requirements, they are good at dealing with nonlinear problems. Moreover, the ensemble Kalman filter has strong compatibility and can be run on the basis of the resistance model, while it can operate multiple relatively independent models in parallel, shortening the calculation period and improving the inversion efficiency. However, due to the Gaussian assumption premise, the application of the ensemble Kalman filter has certain limitations, and the fitting effect is not good for complex nonlinear relationships and sample quantity restrictions, so it needs to be modified to some extent to adapt to the actual situation for dealing with specific problems.

[0145] ③ The interwell connectivity coefficient field map of the small layer is obtained by inversion;

[0146] The connectivity between the injection well and the effective oil well is quantified to obtain the specific value of the connectivity coefficient of each small layer, as shown in Table 2, and the connectivity field map (as shown in Fig. Figure 2 ) and the dominant channel distribution map (​​Figure 3 The results are shown in FIG. 1.

[0147] Table 2 Specific values of the connectivity coefficients of each sublayer

[0148]

[0149] (2) Using the K_means clustering analysis method to determine the well selection and layer selection decision criteria;

[0150] The K_means algorithm is a distance-based clustering algorithm, which uses distance as an evaluation index of similarity, that is, the closer the distance between two objects, the greater the similarity. The specific process is as follows:

[0151] ① Randomly select K documents as center points from N documents;

[0152] ② Measure the distance of each remaining document to each center point, and classify it into the nearest centroid class;

[0153] ③ Recalculate the center points of each class obtained;

[0154] ④ Iterate the above steps ②-③ until the new centroid is equal to or less than the specified threshold value, and the algorithm ends;

[0155] ⑤ Two-dimensional clustering analysis of liquid production and connectivity coefficients is performed to make well selection and layer selection decisions for profile control and flooding.

[0156] Figure 4 The clustering of the connectivity coefficient and the liquid production graph is shown in FIG. 2. Different points represent different categories. Among them, ★ represents a well with a large connectivity coefficient and a large liquid production, which is a well type that can be considered for profile control and flooding; ▲ represents a well with a large connectivity coefficient and a large liquid production, which is a well type that must be profiled and flooded; ■ represents a well with a medium connectivity coefficient and a medium liquid production, which does not need profile control and flooding measures; and ● represents a well with a small connectivity coefficient and a small liquid production, which does not need profile control and flooding measures and can consider increasing injection measures.

[0157] Step three applies flooding scheme samples to flooding experiments and numerical simulation, uses machine learning algorithm to construct permeability representation model after profile control and flooding, and substitutes the flooding permeability representation model into the water drive capacity resistance model to form a profile control and flooding capacity resistance model, which is specifically:

[0158] (1) Establish a profile control and flooding permeability representation model;

[0159] The change of permeability is used to represent the mechanism of profile control and flooding scheme, which is mainly based on experimental data to establish a representation model of profile control and flooding parameters and permeability variation;

[0160] (2) Establish a flooding capacity resistance model;

[0161] The permeability characterization model after profile control and flooding is substituted into the water drive capacity resistance model to form a profile control and flooding capacity resistance model, and the new interwell connectivity coefficient and production index after profile control and flooding are calculated.

[0162] The step four is based on the profile control and flooding capacity resistance model to predict the profile control and flooding scheme effect, and the interwell connectivity coefficient and oil production of the benefited oil well after profile control and flooding are predicted.

[0163] The step four is specifically:

[0164] (1) Designing profile control and flooding slug process parameters, and establishing a profile control and flooding design scheme set;

[0165] ① Establishing a profile control and flooding scheme set (designing different plugging agent systems, system concentrations, slug numbers, slug doses, liquid volumes, displacement volumes, etc.), and the adjustable parameter ranges are shown in Table 3;

[0166] Table 3 Adjustable parameter ranges

[0167] Serial number Profile control and flooding design scheme parameter 1 Plug 1 (pre-plug) system type 2 Plug 2 (main plug) system a type 3 Plug 2 (main plug) system b type 4 Plug 3 (protective plug) system type 5 Daily injection volume (m 3 / d) 6 Plug 1 system concentration (mg / L) 7 Plug 2 system a concentration (mg / L) 8 Plug 2 system b concentration (mg / L) 9 Plug 3 system concentration (mg / L) 10 Plug 1 system injection volume (m 3 )]]> 11 Plug 2 system a injection volume (m 3 )]]> 12 Plug 2 system b injection volume (m 3 )]]> 13 Plug 3 system injection volume (m 3 )]]>

[0168] ② Designing multiple profile control and flooding process parameter combinations to form a profile control and flooding scheme set;

[0169] (2) Bringing different profile control and flooding design schemes into the profile control and flooding capacity resistance model to predict the small layer permeability and oil and water well production data after different profile control and flooding schemes, and finally determining the profile control and flooding slug design parameters by comparing the profile control and flooding effect.

[0170] ① The profile control and flooding capacity resistance model can be used to predict the small layer permeability and oil and water well production index after different profile control and flooding schemes, i.e., the interwell connectivity coefficient of each layer after flooding, the daily liquid production of each well, the daily oil production, the water cut, the cumulative water injection volume and the cumulative oil production.

[0171] ② Determining the target function value of the profile control and flooding scheme: the one-year incremental oil of the well group, the square incremental oil, comparing the profile control and flooding effects of different schemes, and finally determining the profile control and flooding scheme process parameters to complete the scheme design.

[0172] The above only describes the preferred embodiments of the present application and should not be used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for designing and predicting the effect of a control drive scheme based on a capacitive-resistive model, characterized in that, Includes the following steps: Step 1: Establish a water drive capacitive resistance model for the target reservoir block; Step 2: Establish a superior channel identification model to achieve well and layer selection decisions; Step 3: Establish a permeability characterization model for profile and displacement adjustment, and construct a capacitive resistance model for profile and displacement adjustment; Step 4: Design and predict the effects of profile control and drive adjustment schemes; Step two uses the ensemble Kalman filter method to perform automatic history fitting and invert the well connectivity coefficients and time constants of each sub-layer. Step two comprehensively considers the target area's reserves and dynamic conditions of formation, and performs multidimensional cluster analysis on the production volume and connectivity coefficient to clarify the decision criteria for well and layer selection. Step two specifically includes: (1) Ensemble Kalman filter inversion; The ensemble Kalman filter method is mainly divided into three recursive stages: the data set stage, the data prediction stage, and the data assimilation stage. (2) Use K-means clustering analysis to determine the decision criteria for well and layer selection; The K-means algorithm is a distance-based clustering algorithm that uses distance as a similarity evaluation index, meaning that the closer two objects are, the greater their similarity. Step three involves applying modified flow model experiments and numerical simulations of modified flow scheme samples, and using machine learning algorithms to construct a permeability characterization model after profile and flow modification. This modified flow permeability characterization model is then substituted into the water drive capacitive resistance model to form a profile and flow modification capacitive resistance model. Specifically: (1) Establish a permeability characterization model for profile control and displacement; The mechanism of profile control and displacement schemes is characterized by changes in permeability. This is mainly based on experimental data from physical models, and a characterization model of the relationship between profile control and displacement parameters and changes in permeability is established. (2) Establish a variable-resistance model; The permeability characterization model after profile and displacement adjustment is substituted into the water drive capacitive resistance model to form the profile and displacement capacitive resistance model, and the new well connectivity coefficient and production indicators after profile and displacement adjustment are calculated.

2. The method for designing and predicting the effect of a control drive scheme based on a capacitive-resistive model according to claim 1, characterized in that: Step one involves establishing a water-drive capacitive resistance model for the target reservoir block using relative permeability curves, production dynamic data, well location information, and layer data, including: (1) Preparation of basic data for establishing the capacitive-resistive model; (2) The basic theory for establishing the capacitive-resistive model; When inter-well interference is ignored and the bottomhole flowing pressure of the oil well does not change significantly, the estimation formula for the fluid production in the capacitive resistance model can be simplified to the following discrete form: (1) (2) In the formula, i = 1, 2, ..., I, where I is the total number of water wells; j = 1, 2, ..., J, where J is the total number of oil wells; n is the discrete time point, n = 1, 2, 3, ..., N, where N is the total number of time steps; Δt is the time step size, d; The production rate of the j-th oil well at time n is the model calculated value. ; Let j be the initial production rate of the j-th oil well. ;τ j d; λ represents the time constant of the control area of ​​well j, characterizing the initial production decline rate and injection-production hysteresis of well j. I,j Let be the water injection distribution coefficient from water well i to oil well j; Let be the weighted moving average of the water injection rate of well i in the injection-production direction of well ij. ; The formula for oil production is: (3) In the formula, For production wells exist Oil production at any given moment; For production wells Oil production at the initial moment; For production wells exist Oil production at any given moment; For production wells exist The amount of liquid produced at any given time; (3) Set the constraints for the capacitive-resistive model as shown in the following equation: (4) Initial production volume of each oil well and time constant All are greater than 0; the water injection distribution coefficient is greater than or equal to 0; the sum of the water injection distribution coefficients from the water well to each oil well. Less than or equal to 1.

3. The method for designing and predicting the effect of a control drive scheme based on a capacitive-resistive model according to claim 1, characterized in that, Ensemble Kalman filter inversion specifically includes: First, define the extended state vector. As shown in the following formula, (5) In the formula: for The vector of parameters to be fitted; for 3D dynamic parameter vector; for The historical production data needs to be fitted. To expand the state vector, the dimension is... ; Based on the concept of sets, define a set state vector. The expression and mean The formulas are as follows: (6) (7) In the formula: The number of set members indicates the number of elements within the fitted model's range; Therefore, the recursive process of ensemble Kalman filtering can be obtained as follows: (1) Prediction stage: Calculate the predicted state vector using the capacitive-resistive model. , (8) In the formula: This describes the calculation process for the capacitive-resistive model. This refers to ambient noise generated during model calculations. Therefore, the predicted state vector of the set can be obtained: (9) (2) For the first Iterative time step set predicts state vector The elements in the vector are assimilated to obtain the assimilated state vector and its set, as shown in the following equation: (10) (11) In the formula: For set-consistent state vectors; To ensure the diversity of observation data among members of the set, random perturbation error needs to be added to the actual observation data to distinguish set members. The expression is: (12) The Kalman gain matrix is ​​expressed as follows: (13) In the formula: Let be the error covariance matrix of the predicted state vectors of the set; Let be the error covariance matrix of the set of observation vectors; Calculate using the following formula and : (14) (15) In the formula: For the first Iteration time step The mean of the predicted state members of each set; Therefore, the set-consistent state vector is: (16) In the formula: For the set of observation vectors, ; (3) The inversion yields the field diagram of the inter-well connectivity coefficient in the small layer; The connectivity between the injection well and the affected oil well is quantified to obtain the specific values ​​of the connectivity coefficient of each sub-layer, and a connectivity field map and a dominant channel distribution map are drawn.

4. The method for designing and predicting the effect of a control drive scheme based on a capacitive-resistive model according to claim 1, characterized in that, The K-means clustering analysis method is used to determine the decision criteria for well and layer selection. The specific process is as follows: (1) Randomly select K documents from N documents as center points; (2) Measure the distance of each remaining document to each centroid and assign it to the class of the nearest centroid; (3) Recalculate the center points of each class that have already been obtained; (4) Iterate through steps (2)-(3) until the new centroid is equal to or less than the original centroid, then the algorithm ends; (5) Two-dimensional cluster analysis of production volume and connectivity coefficient is used to make decisions on profile control, well selection, and layer selection.

5. The method for designing and predicting the effect of a control drive scheme based on a capacitive resistance model according to claim 1, characterized in that: Step four is to predict the effect of the adjustment and drive scheme based on the adjustment and drive capacitive resistance model, and to predict the inter-well connectivity coefficient and the oil production of the benefiting oil wells after the adjustment and drive.

6. The method for designing and predicting the effect of a control drive scheme based on a capacitive-resistive model according to claim 5, characterized in that: Step four specifically involves: (1) Design the process parameters for profile and drive section plugs and establish a set of profile and drive design schemes; (2) Different profile control and drive design schemes are introduced into the profile control and drive capacitive resistance model to predict the permeability of the small layer and the production data of oil and water wells after different profile control and drive measures. The effects of the drive measures are compared and the design parameters of the profile control and drive slug are finally determined.

Images