Oil reservoir injection-production optimization method based on ensemble Kalman filtering

Abstract

The invention relates to the technical field of oil and gas field development, and provides an oil reservoir injection-production optimization method based on ensemble Kalman filtering. Comprising the following steps: constructing a state variable by adopting a well control degree and a maximum NPV; operating the reservoir numerical simulator to obtain the NPV under the current well control degree; calculating a Kalman gain matrix; performing Kalman filtering updating; when the number of updating times reaches the maximum set number, the optimal variable and the NPV are output; according to the method, an ensemble Kalman filtering state variable is established, an economic net present value is taken as a parameter, the technology is based on a fitted oil reservoir geologic model, an injection and production scheme is designed by taking a single well as an object, an internal file of a simulator is automatically modified through a program, and automatic injection and production parameter optimization design is performed by adopting ensemble Kalman filtering. And an optimal single-well injection-production scheme is obtained.

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas field development and proposes an oil reservoir injection and production optimization method based on ensemble Kalman filtering. Background Art

[0002] Reservoir injection and production optimization plays a vital role in the oil and gas industry. It is not only a science, but also an art, allowing geologists and engineers to explore and understand the mysteries of the underground world. By flexibly applying a series of conventional technical means, we can continuously optimize reservoir development strategies to achieve efficient, economical and environmentally friendly oil production.

[0003] First, it is crucial to have a deep understanding of reservoir characteristics. This includes studying geological structures, rock layer distribution, fluid properties, and key parameters such as formation pressure and temperature. By combining geological exploration, core analysis, well testing, and numerical simulation, we can obtain comprehensive and accurate reservoir information. This information is the basis for formulating reasonable injection and production plans and provides us with a basis for scientific decision-making.

[0004] In the process of formulating the injection and production plan, we need to fully consider the actual situation and development goals of the reservoir. This involves key links such as the optimized layout of water injection wells and oil production wells, the reasonable allocation of injection and production volume, and the careful setting of the injection and production pressure difference. Through careful analysis, we can formulate a personalized development plan suitable for the characteristics of the reservoir and provide guidance for subsequent implementation.

[0005] In the process of implementing injection-production adjustments, we need to make timely adjustments and optimizations to the injection-production parameters according to the actual response of the reservoir and changes in engineering requirements. This includes changing the water injection volume, adjusting the injection-production pressure difference, and increasing or decreasing the number of oil production wells. These adjustments can not only improve the development effect of the reservoir, but also reduce engineering costs and maximize economic benefits.

[0006] In addition, real-time monitoring of reservoir pressure, temperature, flow and other key parameters, as well as regular collection of oil samples for testing and analysis, are key links in optimizing reservoir injection and production. By analyzing these data, we can gain a deep understanding of the actual response and changing trends of the reservoir, and thus timely evaluate the implementation effect of the injection and production plan. This helps us identify problems and adjust strategies in a timely manner to ensure the effective implementation of the optimization plan.

[0007] Finally, we need to make necessary adjustments and optimizations to the injection-production scheme based on the results of monitoring and analysis. This may involve measures such as rearranging injection and production wells, adjusting water injection and oil production, and changing the injection-production pressure difference. At the same time, in order to improve optimization efficiency and accuracy, we can also introduce advanced optimization algorithms and technologies, such as numerical simulation and artificial intelligence. The application of these new technologies will bring us more opportunities and challenges, allowing us to better cope with the complexity and uncertainty of reservoir development.

[0008] In short, reservoir injection and production optimization is a complex and changeable process, which requires us to comprehensively apply knowledge from many aspects such as geology, engineering, and economics. In actual operation, we need to flexibly use various conventional and new technical means, and make appropriate adjustments and improvements according to actual conditions. By constantly learning and exploring new methods and technologies, we can achieve efficient development of reservoirs and optimal economic benefits.

[0009] The Chinese invention patent with the patent number CN201911085801.1 discloses a method for optimizing the energy consumption of the injection and production system of a water-flooding oilfield. The method comprises the following steps: step 1, determining the decision variables of the optimization model; step 2, determining the objective function of the optimization model; step 3, determining the constraints of the optimization model; and step 4, solving the optimization model by using a particle swarm algorithm combined with numerical simulation. The method for optimizing the energy consumption of the injection and production system of a water-flooding oilfield takes the reservoir system as the hub, comprehensively considers the energy consumption of the water-flooding, reservoir, and lifting system, obtains relevant parameters by means of numerical simulation, establishes an overall optimization model of the injection and production system, optimizes the injection and production plan of the reservoir, and optimizes the total energy consumption of the injection and production system as a whole under the conditions of the reservoir plan, thereby achieving further energy saving and consumption reduction under the conditions of the reservoir plan, and providing a new method for energy saving and consumption reduction in oilfields.

[0010] The Chinese invention patent with patent number CN201910156773.1 discloses a method and device for optimizing water injection strategy in oil field production. The method includes: collecting the first characteristic parameter in the crude oil recovery process in the real environment, wherein the first characteristic parameter includes: characteristic data in the water injection strategy, the recovery rate of crude oil obtained based on the water injection strategy, and the value assessment value of the water injection strategy; updating the first characteristic parameter to the virtual environment, wherein the virtual environment is used for learning the planning algorithm for optimizing the water injection strategy; training the first characteristic parameter for a first predetermined number of times in the virtual environment to obtain a second characteristic parameter; updating the second characteristic parameter to the real environment, and optimizing the water injection strategy in the first characteristic parameter using the second characteristic parameter to obtain an optimized water injection strategy. The present invention solves the technical problem that the water injection method used in the related art during crude oil recovery is unreliable, easily causes damage to crude oil, and affects the recovery rate of crude oil.

[0011] The Chinese invention patent with the patent number CN202010647863.3 discloses a smart oilfield injection and production real-time optimization and control system, including: intelligent control wellbore for production wells, intelligent control wellbore for injection wells, surface control device for production wells, surface control device for injection wells, and surface data acquisition-analysis-decision-making computer processing system; the intelligent control wellbore for production wells, surface control device for production wells, intelligent control wellbore for injection wells, and surface control device for injection wells are respectively connected to the surface data acquisition-analysis-decision-making computer processing system to realize data acquisition and control instruction transmission, and the surface control device for production wells is connected to the intelligent control wellbore for production wells, and the surface control device for injection wells is connected to the intelligent control wellbore for injection wells to realize flow control. The present invention realizes the seamless integration of injection and production optimization software and injection and production control hardware, realizes the historical matching and production plan optimization of large-scale oil reservoir production dynamics within a few hours, and performs real-time control of injection and production well flow.

[0012] It can be seen that the conventional injection and production parameter design is to take the reservoir as a whole or a well group as a unit and manually set multiple groups of time and injection and production volume combinations.

[0013] Among these solutions, the one with the best effect is selected and applied to actual oil fields.

[0014] However, since the values ​​of parameters in the scheme are greatly affected by human factors, they are arbitrary and blind.

[0015] In addition, due to the limited number of enumeration schemes, the multiple schemes developed through the limited enumeration method may not be optimal, making it difficult to achieve the best development effect of the oil field.

[0016] This method is less applicable to reservoirs with complex connectivity and poor well pattern regularity.

[0017] The reservoir development and production optimization control technology solves these problems well. This technology is based on the fitted reservoir geological model and designs the injection and production plan for a single well. Its goal is to maximize the economic benefits of the reservoir and describe the control of the reservoir production system as an optimization problem. By optimizing the injection and production parameters of oil and water wells, the production control plan for each stage is automatically customized for the oil and water wells to improve the development effect. Conventional gradient algorithms are difficult to meet the calculation requirements of large-scale reservoir production optimization.

[0018] Therefore, it is urgent to propose an efficient and reliable technical system to solve the problem of oil reservoir production optimization and improve oil reservoir recovery rate and economic benefits. Summary of the invention

[0019] The purpose of the present invention is to propose an oil reservoir injection and production optimization method based on ensemble Kalman filtering in view of the deficiencies in the prior art.

[0020] It is difficult to design parameters through enumeration schemes for reservoir models with complex connectivity and poor well network regularity, and gradient-based optimization algorithms are difficult to apply to large reservoirs and gradient solutions are very difficult, facing the problems of cumbersome processes, time-consuming and labor-intensive. The present invention solves the problem of reservoir production optimization, uses a program to automatically modify the well control degree, and solves the problem of optimal injection and production well control degree with the goal of maximizing the economic net present value, thereby improving the economic benefits of the reservoir and the prediction of the reservoir state field.

[0021] The technical solution is as follows:

[0022] A reservoir injection and production method based on ensemble Kalman filtering comprises the following steps:

[0023] a1. Use well control degree and maximum NPV to construct state variables;

[0024] a2. Run the reservoir numerical simulator to obtain the NPV under the current well control degree;

[0025] a3. Calculate the Kalman gain matrix;

[0026] a4. Perform Kalman filter update;

[0027] a5. When the number of updates reaches the maximum number of sets, the optimal variable and NPV are output.

[0028] Furthermore, the NPV formula is as follows:

[0029]

[0030] Among them, r o ,r w ,r wi ——Crude oil price, water production cost price, water injection price, $ / STB;

[0031] ——The average oil production rate of the j-th production well at time n, the average water production rate of the j-th production well at time n, the average water injection volume of the ith injection well at time n, STB / d; L is the total control time step.

[0032] Furthermore, the construction of the state vector includes dividing the production time into M regulated time steps, and the production system of each time step is set as the predicted state vector:

[0033] U=[mc] T ,c=[c 1 ,c 2 …c M ] (2).

[0034] Furthermore, the step a2 includes writing a production system file according to the input well control degree, running a commercial reservoir numerical simulator CMG, obtaining the simulated results, and forming a new state vector:

[0035] U j-1,i =f(U o j-1,i ) (3)

[0036] Where i is 1 to the number of set members (N), U represents the predicted state vector, and Uo represents the analysis step state vector.

[0037] Further,.

[0038] Furthermore, the step a2 constructs state variables according to different throttle settings, and calculates the total cumulative oil production or NPV value.

[0039] Furthermore, the correlation between the state vector and the measured vector is expressed by the following formula:

[0040] D=HU (4)

[0041] Where H is the matrix for selecting measurement variables from the variables.

[0042] Furthermore, the Kalman gain matrix K in step a3 is calculated as follows:

[0043] K j-1 =P j-1 f H T (HP j f H T ) -1 (5)

[0044] Among them, the matrix P n f is an approximation of the model error covariance matrix.

[0045] Furthermore, the matrix P n f The formula is as follows:

[0046] P f i =L f i (L f i ) T (6)

[0047] Where L f i Depend on

[0048]

[0049] in is the mean of the set.

[0050] Furthermore, the step a4 specifically includes:

[0051] The data (mj-1) and the state variables of the production system settings (cj-1) in the previous iteration step j-1 are then run on each set member (i) to generate mj and cj:

[0052]

[0053] Where DOj-1 is calculated from the known NPV:

[0054] D o j-1 =max(m j-1 +std(m j-1 )(9)

[0055] Where std represents standard deviation.

[0056] The beneficial effects of the present invention are:

[0057] The present invention establishes a collective Kalman filter state variable, takes the economic net present value as a parameter, and designs an injection and production plan based on a fitted reservoir geological model with a single well as the object. The internal files of the simulator are automatically modified by a program, and the collective Kalman filter is used to perform automatic injection and production parameter optimization design to obtain the optimal single-well injection and production plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is the ENKF optimization update calculation flow chart;

[0059] Figure 2 The permeability model of Example 3;

[0060] Figure 3 This is the algorithm optimization result of Example 3;

[0061] Figure 4 The oil saturation field before optimization of Example 3;

[0062] Figure 5 This is the oil saturation field after optimization in Example 3. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0064] Embodiment 1:

[0065] An oil reservoir injection and production optimization technology based on ensemble Kalman filtering has an optimization variable of a single well control degree on a control step, and includes the following optimization steps:

[0066] 1. Establish initial reservoir numerical simulation model

[0067] 2. Parameter assignment, setting the initial control variables of a single well

[0068] 3. Run the INSIM connection unit reservoir numerical simulator to calculate the initial maximum NPV.

[0069] 4. Calculate the Kalman gain matrix, perform iterative calculations of state variables, and update the injection and production plan for a single well.

[0070] 5. Repeat steps 2-4.

[0071] 6. Output the final calculated state variables.

[0072] Embodiment 2:

[0073] A reservoir injection and production optimization method based on ensemble Kalman filtering, wherein the optimization variable is the control degree of a single well on a control step, includes the following optimization process:

[0074] 1. Establish initial reservoir numerical simulation model;

[0075] 2. Parameter assignment, setting the initial control variables of a single well;

[0076] 3. Run the numerical simulator to calculate the initial state variable NPV;

[0077] 4. Calculate the Kalman gain matrix and perform iterative calculation of state variables;

[0078] 5. Repeat steps 2-4;

[0079] 6. Output the final calculated state variables.

[0080] Embodiment three:

[0081] A reservoir injection and production method based on ensemble Kalman filtering comprises the following steps:

[0082] Step 1:

[0083] Calculate the maximum NPV for Kalman filter optimization

[0084]

[0085] Among them, r o ,r w ,r wi ——Crude oil price, water production cost price, water injection price, $ / STB;

[0086] ——The average oil production rate of the j-th production well at time n, the average water production rate of the j-th production well at time n, the average water injection volume of the i-th injection well at time n, STB / d; L is the total control time step; Step 2:

[0087] In the standard ensemble Kalman filter, the state vector consists of spatially distributed data at a specific time. Here, the well control degree at different time steps and the maximum NPV in step 1 are used as the ensemble setting to construct the state variable. The production time is divided into M regulated time steps. The production regime within these time steps is constant. The production regime at each time step is set as the vector

[0088] U=[mc] T ,c=[c 1 ,c 2 …c M ] (2)

[0089] According to the input well control degree, the production system file is written, and the connection unit reservoir numerical simulator INSIM is run. This software takes the connection unit body between wells as the research basis, and obtains the well connection relationship model that can quickly simulate the oil and water production dynamics between wells based on the reservoir fluid material balance equation and Berkeley displacement equation. The reservoir data file with the changed well control degree is input into INSIM for numerical simulation calculation, and the simulated results are obtained to form a new state vector.

[0090] U j-1,i =f(U o j-1,i ) (3)

[0091] Where i is 1 to the number of set members (N), U represents the predicted state vector, and Uo represents the analysis step state vector

[0092] In this method, different choke settings are considered to construct state variables and the total cumulative oil production or NPV value is calculated.

[0093] Step 3:

[0094] Considering the matching relationship of selecting the measured variables from the state variables to update the state variables, the correlation between the state vector and the measured vector is expressed by the following formula.

[0095] D=HU (4)

[0096] Among them, H is the matrix for selecting measurement variables from variables, and D is the measurement variable.

[0097] Calculate the Kalman gain matrix K,

[0098] K j-1 =P j-1 f H T (HP j f H T ) -1 (5)

[0099] Among them, the matrix P n f is an approximation of the model error covariance matrix, calculated by

[0100] P f i =L f i (L f i ) T (6)

[0101] Where L f i Depend on

[0102]

[0103] in is the mean of the set.

[0104] Perform ensemble Kalman filter update. The specific process is as follows: update the data (m j-1 ) and production system settings (c j-1 ) state variables. Then, the Kalman filter is run on each member of the set (i) to generate m j and c j .

[0105]

[0106] in is calculated from the known NPV.

[0107] D o j-1 =max(m j-1 +std(m j-1) (9)

[0108] Where std represents standard deviation.

[0109] Embodiment 4:

[0110] Based on the INSIM reservoir numerical simulation software of the connected unit system, reservoir case optimization calculation was performed according to the ensemble Kalman filter algorithm.

[0111] This embodiment establishes a conceptual model of a water-drive reservoir with five injections and four productions. The model is highly heterogeneous, with the maximum connection unit conductivity being 100m3.d-1.MPa-1 and the minimum being 5.744m3.d-1.MPa-1.

[0112] An oil-water two-phase simulation was carried out, with the actual control range of the model being 2500×1200ft and the production time being 2 years.

[0113] The model was optimized for 20 steps in total. The calculation results of the model are as follows: Figure 3 As shown in the figure, it can be seen that the NPV after the ensemble Kalman filter optimization is significantly higher than the initial NPV; and Figure 4 and Figure 5 It can be seen that after the well control degree of a single well at each control time step, the reservoir water injection displacement is more uniform and the overall oil saturation decreases, which once again demonstrates the superiority of this method.

[0114] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims.

Claims

1. A reservoir injection and production method based on ensemble Kalman filtering, It is characterized in that The steps include: a1. Use well control degree and maximum NPV to construct state variables; a2. Run the reservoir numerical simulator to obtain the NPV under the current well control degree; a3. Calculate the Kalman gain matrix; a4. Perform Kalman filter update; a5. When the number of updates reaches the maximum number of sets, the optimal variable and NPV are output.

2. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 1, It is characterized in that The NPV formula is as follows: Among them, r o ,r w ,r wi ——Crude oil price, water production cost price, water injection price, $ / STB; ——The average oil production rate of the j-th production well at time n, the average water production rate of the j-th production well at time n, the average water injection volume of the ith injection well at time n, STB / d; L is the total control time step.

3. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 2, It is characterized in that The construction of the state vector includes dividing the production time into M regulated time steps, and the production system of each time step is set as the predicted state vector: U=[m c] T ,c=[c 1 ,c 2 …c M ] (2)。 4. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 3, It is characterized in that The step a2 includes writing the production system file according to the input well control degree, running the commercial reservoir numerical simulator CMG, obtaining the simulated results, and forming a new state vector: IN j-1,i =f(U o j-1,i ) (3) Where i is 1 to the number of set members (N), U represents the predicted state vector, and Uo represents the analysis step state vector.

5. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 4, It is characterized in that 。 6. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 5, It is characterized in that The step a2 constructs state variables according to different throttle settings, and calculates the total cumulative oil production or NPV value.

7. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 6, It is characterized in that The correlation between the state vector and the measured vector is expressed by the following formula: D=HU (4) Where H is the matrix for selecting measurement variables from the variables.

8. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 7, It is characterized in that The Kalman gain matrix K in step a3 is calculated as follows: K j-1 =P j-1 f H T (HP j f H T ) -1 (5) Among them, the matrix P n f is an approximation of the model error covariance matrix.

9. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 8, It is characterized in that The matrix P n f The formula is as follows: P f i =L f i (L f i ) T (6) Where L f i Depend on in is the mean of the set.

10. The oil reservoir injection and production method based on ensemble Kalman filtering according to claim 9, It is characterized in that The step a4 specifically includes: The data (m j-1 ) and production system settings (c j-1 ) state variables. Then, the Kalman filter is run on each member of the set (i) to generate m j and c j : Where D O j-1 It is calculated from the known NPV: D o j-1 =max(m j-1 +std(m j-1 ) (9) Where std represents standard deviation.

Images