A full-cycle synergistic optimization method for high-viscosity reservoir chemical flooding

Abstract

The present application relates to a kind of high viscosity reservoir chemical flooding whole cycle synergistic optimization method, belong to oil and gas field development enhanced oil recovery technical field.It includes: step 1: determining high viscosity reservoir whole process water drive development scheme, and carry out reservoir numerical simulation calculation;Step 2: determine the dynamic control initial threshold value between different development stages of high viscosity reservoir after water drive to chemical flooding and the initial value of injection-production parameter of each stage, carry out reservoir numerical simulation calculation to high viscosity reservoir chemical flooding initial scheme;Step 3: with the maximum tonnage dose oil increment as the goal, using fixed point iteration method and particle swarm optimization algorithm, the dynamic control threshold value between different development stages is simultaneously optimized, and the injection-production parameter of each production stage is converted, then the updated chemical flooding scheme is carried out reservoir numerical simulation calculation;Step 4: repeat step 3 until iteration converges to the maximum tonnage dose oil increment.The present application maximizes tonnage dose oil increment, overcomes the limitation of step optimization.

Description

Technical Field

[0001] This invention relates to a full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs, belonging to the technical field of enhanced oil recovery in oil and gas field development. Background Technology

[0002] my country possesses abundant high-viscosity crude oil resources. Injecting oil displacement agents with strong adhesion-reducing properties into the formation can effectively lower the adhesion work of high-viscosity crude oil on the rock surface. This process significantly weakens the binding force between crude oil and the rock surface, making it easier for crude oil to peel off, thereby greatly improving oil recovery. This process exhibits significant stage-specific characteristics and needs to be scientifically divided into a water cut decline stage, a valley plateau stage, and a water cut recovery stage. Chinese patent document CN120667071A proposes an optimization method and device for a discontinuous chemical flooding variable cycle combination scheme. This method determines the target optimization algorithm with the highest adaptability based on the permeability variation coefficient of the target reservoir and an integrated optimization algorithm. Then, based on the benefit-related objective function and constraints, the parameter values ​​of the chemical agent combination variables are iteratively optimized to obtain the optimized parameter values. Meanwhile, Chinese patent document CN120649854A provides a method for optimizing the proportions of an efficient and balanced chemical flooding system. By collecting data, the chemical agent proportions are continuously optimized through experiments and numerical simulations. The optimized ratio can improve the oil-water interface characteristics, increase oil fluidity, and effectively enhance the oil displacement effect and oilfield recovery rate. Laboratory verification and small-scale field tests have ensured the effectiveness of the ratio in practical applications, taking into account the oil displacement effect, economic benefits and practicality.

[0003] Existing optimization techniques have shortcomings. Stage division relies on preset static thresholds, such as fixed time points or cumulative oil production, causing stage transitions to deviate from actual development dynamics. Simultaneously, injection-production parameter optimization is separated from stage division, typically optimized independently within fixed stage boundaries, neglecting the significant alteration of reservoir response by optimized parameters, which directly impacts the optimal stage threshold. This "stage division first, parameter optimization later" approach results in a two-way disconnect between stage boundaries and parameter configuration. Static thresholds render parameter optimization ineffective, while dynamic changes caused by parameters cannot be fed back to stage adjustments. The lack of collaborative optimization capabilities for the joint decision space of stage transition thresholds and injection-production parameters necessitates the development of a closed-loop mechanism for synchronously and dynamically optimizing thresholds and parameters, achieving a unified approach to adaptive development stage division and global optimization of injection-production strategies. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs, in order to solve the aforementioned problems.

[0005] The technical solution of this invention is as follows: A full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs includes: Step 1: Determine the waterdrive development scheme for the entire process of high-viscosity reservoirs and perform reservoir numerical simulation calculations; Step 2: Determine the initial threshold for dynamic control between different development stages of waterflooding followed by chemical flooding in high-viscosity reservoirs, as well as the initial values ​​of injection and production parameters for each stage, and perform reservoir numerical simulation calculations for the initial chemical flooding scheme in high-viscosity reservoirs. Step 3: Based on the calculated development dynamics, with the goal of maximizing the increase in oil yield per ton of injection, the fixed-point iteration method and particle swarm optimization algorithm are used to simultaneously and collaboratively optimize the dynamic control thresholds for the transition between different development stages and the injection and production parameters for each production stage. Then, reservoir numerical simulation calculations are performed on the updated chemical flooding scheme. The fixed-point iterative method for dynamic control thresholds is based on the water cut curve obtained from reservoir numerical simulation. The curve is divided into three linear segments by piecewise linear regression. The slope of each segment is fitted by the least squares method to minimize the sum of squared residuals. Candidate boundary values ​​for the transition of different development stages are determined based on the slope difference between adjacent segments. The dynamic control thresholds of each stage are updated in subsequent optimization iterations. The particle swarm optimization algorithm treats each solution as a particle. First, based on the oil increase per ton of agent calculated by reservoir numerical simulation, it updates the individual historical best position and the global best position of the entire swarm. Then, the particle swarm optimization algorithm integrates the information of the individual historical best solution and the global best solution of the swarm to update the position parameters of each particle. Step 4: Repeat step 3 until the iteration converges to the maximum increase in oil production per ton of agent.

[0006] According to a preferred embodiment of the present invention, in step 1, a water-drive development scheme for the entire process of high-viscosity reservoirs is formulated, and reservoir numerical simulation calculations are performed. The calculation is stopped when the water cut reaches 98%, and the cumulative oil production of the entire process of water-drive development is calculated.

[0007] In a further preferred embodiment, in step 1, formulating a waterflooding development plan for the entire process of high-viscosity reservoirs means: continuously injecting water into the injection wells at a preset water injection rate without injecting chemical agents, and controlling the production of the production wells according to a preset production volume. Numerical simulation of oil reservoirs refers to: dividing the target oil reservoir into grid cells; assigning porosity, permeability, initial oil saturation, initial pressure, and rock and fluid properties to each grid cell; setting the locations of injection wells and production wells, completion intervals, and control conditions; using the mass conservation equation, Darcy's flow equation, and oil-water two-phase flow relationship, solving for the pressure, oil saturation, and water saturation of each grid cell step by step; and obtaining the changes in water cut and oil production over time based on the oil production rate and water production rate of each production well. The calculation stops when the water cut reaches 98%. The cumulative oil production of the entire water-drive development process is obtained by summing up the oil production over time.

[0008] According to a preferred embodiment of the present invention, in step 2, when the water flooding reaches a water content of 90%, the process is switched to chemical flooding, and water flooding continues after the chemical flooding is completed.

[0009] According to a preferred embodiment of the present invention, in step 2, the chemical flooding process of high viscosity reservoirs is divided into a water cut decline stage, a valley plateau stage, and a water cut recovery stage; the dynamic control threshold for the transition between different development stages refers to two key points on the dynamic curve of chemical flooding development of high viscosity reservoirs that characterize the stage transition, including the first turning point of the transition from the water cut decline stage to the valley plateau stage, and the second turning point of the transition from the valley plateau stage to the water cut recovery stage. Based on the reservoir numerical simulation results of the initial chemical flooding scheme, the water cut and cumulative oil production at each time step were obtained in the order of the simulation time. First, the first and second inflection points and their corresponding two simulation times were determined on the water cut dynamic curve. Then, the cumulative oil production corresponding to the two simulation times was directly read from the reservoir numerical simulation results. The cumulative oil production corresponding to the first inflection point was assigned as the first dynamic control initial threshold, and the cumulative oil production corresponding to the second inflection point was assigned as the second dynamic control initial threshold. In the subsequent development process, when the cumulative oil production reaches the first dynamic control initial threshold, the development process transitions from the water cut decline stage to the valley plateau stage; when the cumulative oil production reaches the second dynamic control initial threshold, the development process transitions from the valley plateau stage to the water cut recovery stage.

[0010] According to a preferred embodiment of the present invention, in step 2, the initial values ​​of injection and production parameters for the water cut decline stage, the valley platform stage, and the water cut recovery stage are determined respectively. The initial values ​​of injection and production parameters include the chemical agent injection concentration and chemical agent injection rate for each stage. First, based on the experimental results of chemical agent performance evaluation under the target reservoir conditions, the concentration range of the chemical agent was determined to be 0–2.5 kg / m³. 3 And the range of chemical agent concentrations is used as the range of values ​​for the chemical agent injection concentration; Then, based on the water absorption capacity of the injection well, the allowable injection pressure of the formation, and the working capacity of the injection equipment, the chemical injection rate range was determined to be 30–100 m. 3 / d; Within the range of chemical agent injection concentration and injection rate, chemical agent injection concentrations and injection rates are selected for the water cut decline stage, valley platform stage, and water cut recovery stage, respectively. A set of chemical agent injection concentrations and injection rates selected for each stage are assigned as the initial values ​​of the injection and production parameters for the corresponding stage. Multiple sets of different initial parameter combinations are generated, and each set of parameter combinations corresponds to an initial chemical flooding scheme, maintaining a balance between injection and production.

[0011] According to a preferred embodiment of the present invention, step 2, performing reservoir numerical simulation calculations for the initial chemical flooding scheme of high-viscosity reservoirs, refers to: assigning the initial dynamic control thresholds (including the first and second initial dynamic control thresholds) between different development stages in the initial chemical flooding scheme of high-viscosity reservoirs, as well as the chemical agent injection concentration and chemical agent injection rate at each stage, to the reservoir numerical simulation calculations established in step 1; performing dynamic simulations of development while maintaining injection-production balance; calculating and outputting the water cut and cumulative oil production during the chemical flooding process; and stopping the reservoir numerical simulation calculations when the overall water cut reaches 98%.

[0012] According to a preferred embodiment of the present invention, in step 3, piecewise linear regression is used to divide the water content curve of the chemical flooding stage into three linear segments, as shown in equation (I): (I); In formula (Ⅰ), For the first Moisture content corresponding to each data point; For the first The development time after chemical flooding corresponding to each data point; For the first Error term for each data point; This represents the total number of data points on the moisture content curve. For data point sequence numbers; For the numbering of linear line segments, These respectively represent the stages of declining water content, the valley-to-plateau stage, and the water content recovery stage; and The first The intercept and slope of the fitted line at each stage. and The positions of the first and second candidate inflection points in the moisture content data sequence are respectively assigned, satisfying the following conditions: , and These represent the development times corresponding to the first and second candidate inflection points, respectively.

[0013] According to a preferred embodiment of the present invention, in step 3, the first The slope of each stage and intercept The least squares method is used for calculation, and the calculation formulas are shown in equations (II) and (III) respectively: (II); (III); In equations (II) and (III), For the first The set of indexes for data points included in each stage; For the first The average development time corresponding to each data point within each stage; For the first The average moisture content of each data point within each stage.

[0014] According to a preferred embodiment of the present invention, in step 3, after completing the piecewise linear regression, the sum of squared residuals for each stage and the total sum of squared residuals for the three stages are calculated, as shown in equations (IV) and (V): (Ⅳ); (V); In equations (Ⅳ) and (Ⅴ), For the first The sum of squared residuals for each stage is used to characterize the deviation between the actual moisture content and the fitted moisture content for that stage. This is the sum of squared residuals for the three stages.

[0015] According to a preferred embodiment of the present invention, in step 3, the stage transition boundary is determined according to the following condition: the slope of the water cut decline stage. The slope of the valley-bottom plateau phase satisfies The slope of the water-bearing recovery phase ;in, This indicates that the moisture content decreases over time. This indicates that the rate of change in water content during the valley plateau stage does not exceed 10% of the absolute value of the rate of change in water content during the water content decline stage, and is therefore determined to be the valley plateau stage. This indicates that the moisture content increases over time. The slope condition and the sum of squares of the total residuals will be satisfied. The smallest candidate inflection point combination is determined as the optimal inflection point combination. The time point corresponding to the first optimal inflection point is determined as the first stage transition boundary time point for the transition from the water level decline stage to the valley plateau stage. The time point corresponding to the second optimal inflection point is determined as the second stage transition boundary time point for the transition from the valley plateau stage to the water level recovery stage.

[0016] According to a preferred embodiment of the present invention, in step 3, after determining the first stage transition boundary time point and the second stage transition boundary time point, the cumulative oil production corresponding to the first stage transition boundary time point and the second stage transition boundary time point are read from the output results of the same reservoir numerical simulation, respectively. The cumulative oil production corresponding to the first stage transition boundary time point is determined as the first dynamic control candidate threshold, and the cumulative oil production corresponding to the second stage transition boundary time point is determined as the second dynamic control candidate threshold.

[0017] According to a preferred embodiment of the present invention, the update process for the dynamic control threshold in step 3 is as follows: First, calculate the gradient direction of the dynamic control threshold change. The calculation formula is shown in equation (VI): (VI); In formula (VI), The currently used dynamic adjustment threshold, To dynamically adjust the candidate threshold, To dynamically adjust the direction and amount of change of the candidate threshold relative to the currently used dynamic control threshold; Then, calculate the adaptive learning rate. The calculation formulas are shown in equations (VII) and (VIII): (VII); (VIII); In equations (VII) and (VIII), Adaptive learning rate; Base learning rate; To dynamically adjust the relative error between the candidate threshold and the currently used boundary value; This is a reference error value; This is the lower bound factor for scaling the learning rate; This is the scaling factor for the learning rate. Finally, the dynamically adjusted threshold is updated using the obtained gradient direction and adaptive learning rate, as shown in equation (IX): (IX); In formula (IX), This is the updated dynamic adjustment threshold.

[0018] According to a preferred embodiment of the present invention, in step 3, the formula for calculating the increase in oil volume per ton of agent is as shown in equation (X): (X); In formula (X), To increase oil production per ton of agent, For the cumulative oil production of chemical flooding, This represents the cumulative oil production from the entire water-drive process. The cumulative amount of chemical agent injected to strongly reduce adhesion. The formulas for calculating the cumulative oil production of chemical flooding, the cumulative oil production of water flooding throughout the entire process, and the cumulative injection amount of strong adhesion-reducing chemicals are as follows: (XI); (XII); (XIII); in, and Two schemes are respectively Oil production rate at each time step; This refers to the mass concentration of the chemical agent, expressed in kg / m³. The total mass of chemical agent injected, in tons (t). The ground volume injection rate is expressed in m³ / d. The time step duration is expressed in days (d).

[0019] According to a preferred embodiment of the present invention, in step 3, with the objective function of maximizing the increase in oil yield per ton of agent, a particle swarm optimization algorithm is used to optimize the chemical agent injection concentration and chemical agent injection rate during the water cut decline, valley plateau, and water cut recovery stages. The updated parameters are then used as input parameters for the next iteration. The parameter update formulas are shown in equations (XIV) and (XV). (XIV); (XV); In formula (XIV), and The first Next iteration particle and the first During iteration The first particle Speed ​​of dimensional parameters; Inertial weights; and It is the acceleration constant; and A random number that varies between 0 and 1; This refers to the best position the particle itself has ever experienced, i.e., the individual's historical best position. The best position experienced by all particles in the current swarm is the global optimal position of the entire swarm, which is the current optimal objective function value.

[0020] According to a preferred embodiment of the present invention, in step 4, during the optimization iteration process, if the calculated increase in oil per ton of agent remains unchanged for 10 consecutive iterations, the calculation process is terminated.

[0021] The beneficial effects of this invention are as follows: 1. The dynamic control threshold for development phase transitions and key injection-production parameters for each phase are treated as unified variables, and optimization is performed synchronously within the same optimization framework to construct a collaborative optimization space for development phase division and injection-production parameters. This mechanism, through closed-loop feedback, ensures that the optimal phase division boundary and the parameter configuration within the optimal phase match and work together to maximize the increase in oil production per ton of agent, overcoming the limitations of step-by-step optimization.

[0022] 2. An adaptive threshold update mechanism based on fixed-point iteration and piecewise linear regression of the cumulative oil production curve is proposed. This method automatically identifies points of significant change in curve slope as candidate thresholds and introduces an adaptive learning rate dynamically adjusted according to the relative error for updating, ensuring that the threshold can adaptively adjust to the optimal position as the optimization process progresses, significantly improving the objectivity and accuracy of stage division.

[0023] 3. A particle swarm optimization algorithm is employed to efficiently search for the optimal combination of injection and extraction parameters at each stage, and this is coordinated with the fixed-point iterative update of the dynamically adjusted threshold. This method sets explicit iteration termination conditions, ensuring the convergence of the optimization process and the repeatability and reliability of the results. Through mathematical modeling and automated processes, manual intervention is significantly reduced, improving the computational efficiency and intelligence level of the optimization design. Attached Figure Description

[0024] Figure 1 This is a flowchart of the full-cycle synergistic optimization method for chemical flooding of high-viscosity reservoirs according to the present invention.

[0025] Figure 2 This is a schematic diagram illustrating the iterative process of dynamically adjusting the threshold in a collaborative optimization method.

[0026] Figure 3 This diagram illustrates the comparison of cumulative oil production optimization iterations using the collaborative optimization method and the ordinary optimization method.

[0027] Figure 4 This is a comparison chart of the cumulative oil production of the ordinary optimization method and the collaborative optimization method.

[0028] Figure 5 A comparison chart of chemical agent injection amounts between the conventional optimization method and the collaborative optimization method.

[0029] Figure 6 A comparison chart of the increase in oil production per ton of agent between the initial scheme and the co-optimized scheme. Detailed Implementation

[0030] The present invention will be further defined below with reference to the accompanying drawings and embodiments, but is not limited thereto.

[0031] Terminology Explanation: High-viscosity oil reservoirs refer to oil reservoirs where the underground crude oil viscosity is 100–2000 mPa·s.

[0032] Example 1 A full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs, such as... Figure 1 As shown, it includes: Step 1: Determine the waterdrive development scheme for the entire process of high-viscosity reservoirs and perform reservoir numerical simulation calculations; Step 2: Determine the initial threshold for dynamic control between different development stages of waterflooding followed by chemical flooding in high-viscosity reservoirs, as well as the initial values ​​of injection and production parameters for each stage, and perform reservoir numerical simulation calculations for the initial chemical flooding scheme in high-viscosity reservoirs. Step 3: Based on the calculated development dynamics, with the goal of maximizing the increase in oil yield per ton of injection, the fixed-point iteration method and particle swarm optimization algorithm are used to simultaneously and collaboratively optimize the dynamic control thresholds for the transition between different development stages and the injection and production parameters for each production stage. Then, reservoir numerical simulation calculations are performed on the updated chemical flooding scheme. The fixed-point iterative method for dynamic control thresholds is based on the water cut curve obtained from reservoir numerical simulation. The curve is divided into three linear segments by piecewise linear regression. The slope of each segment is fitted by the least squares method to minimize the sum of squared residuals. Candidate boundary values ​​for the transition of different development stages are determined based on the slope difference between adjacent segments. The dynamic control thresholds of each stage are updated in subsequent optimization iterations. The particle swarm optimization algorithm treats each scheme as a particle. First, based on the oil production increase per ton of agent calculated by reservoir numerical simulation, it updates the individual historical best position of each particle (scheme) and the global best position of the entire swarm. Then, the particle swarm optimization algorithm integrates the information of the individual historical best solution and the global best solution of the swarm to update the position parameters (i.e. injection and production parameters) of each particle (scheme). Step 4: Repeat step 3 until the iteration converges to the maximum increase in oil production per ton of agent.

[0033] Example 2 The difference between the full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs described in Example 1 and the method described in Example 1 is as follows: In step 1, a comprehensive waterflooding development plan for the high-viscosity reservoir is formulated, and reservoir numerical simulation calculations are performed. The calculation stops when the water cut reaches 98%, and the cumulative oil production from the entire waterflooding development process is calculated. In this embodiment, the calculation stops when the water cut reaches 98%, and the cumulative oil production from the entire waterflooding process is 1330096 m³. 3 .

[0034] In step 1, formulating a full-process water drive development plan for high-viscosity reservoirs means: continuously injecting water into injection wells at a preset water injection rate without injecting chemical agents, and controlling the production of production wells according to a preset production volume. Reservoir numerical simulation refers to: dividing the target reservoir into grid cells; assigning porosity, permeability, initial oil saturation, initial pressure, and rock and fluid properties to each grid cell; setting the locations of injection and production wells, completion intervals, and control conditions; using the mass conservation equation, Darcy's flow equation, and oil-water two-phase flow relationship, solving for the pressure, oil saturation, and water saturation of each grid cell step by step; obtaining the changes in water cut and oil production over time based on the oil and water production rates of each production well; and implementing this calculation using existing commercial reservoir numerical simulation software.

[0035] The calculation stops when the water cut reaches 98%. The cumulative oil production of the entire water-drive development process is obtained by summing up the oil production over time.

[0036] In step 2, water flooding is carried out until the water cut reaches 90%, then chemical flooding is switched to water flooding. After the chemical flooding is completed, water flooding is carried out again.

[0037] In step 2, based on the dynamic characteristics of the development of the oilfield, the chemical flooding process of high viscosity reservoirs is divided into the water cut decline stage, the valley plateau stage, and the water cut recovery stage. The dynamic control threshold for the transition between different development stages refers to two key points on the dynamic curve of chemical flooding development of high viscosity reservoirs that characterize the stage transition, including the first turning point from the water cut decline stage to the valley plateau stage, and the second turning point from the valley plateau stage to the water cut recovery stage. Based on the reservoir numerical simulation results of the initial chemical flooding scheme, the water cut and cumulative oil production at each time step were obtained in the order of the simulation time. First, the first and second inflection points and their corresponding two simulation times were determined on the water cut dynamic curve. Then, the cumulative oil production corresponding to the two simulation times was directly read from the reservoir numerical simulation results. The cumulative oil production corresponding to the first inflection point was assigned as the first dynamic control initial threshold, and the cumulative oil production corresponding to the second inflection point was assigned as the second dynamic control initial threshold. In the subsequent development process, when the cumulative oil production reaches the first dynamic control initial threshold, the development process transitions from the water cut decline stage to the valley plateau stage; when the cumulative oil production reaches the second dynamic control initial threshold, the development process transitions from the valley plateau stage to the water cut recovery stage.

[0038] In step 2, the initial values ​​of injection and production parameters are determined for the water cut decline stage, valley platform stage, and water cut recovery stage, respectively. The initial values ​​of injection and production parameters include the chemical agent injection concentration and chemical agent injection rate for each stage. First, based on the experimental results of chemical agent performance evaluation under the target reservoir conditions, the concentration range of the chemical agent that can reduce crude oil adhesion and improve oil displacement efficiency was determined to be 0–2.5 kg / m³. 3 And the range of chemical agent concentrations is used as the range of values ​​for the chemical agent injection concentration; Then, based on the water absorption capacity of the injection well, the allowable injection pressure of the formation, and the working capacity of the injection equipment, the chemical injection rate range of 30–100 m³ / h is determined to be within the range that will not exceed the formation fracturing pressure and the upper limit of the equipment's working capacity. 3 / d; the lower limit is the injection rate that allows for stable and continuous metering injection from the injection well, and the upper limit is the maximum stable injection rate that does not exceed the formation allowable injection pressure, the wellhead allowable pressure, and the rated flow rate and rated pressure of the injection equipment. Currently, the chemical injection rate range in actual field development of high-viscosity reservoirs is 30–100 m / s. 3 / d.

[0039] Within the range of chemical agent injection concentration and injection rate, chemical agent injection concentrations and rates were selected for the water cut decline, valley plateau, and water cut recovery stages, respectively. A set of chemical agent injection concentrations and rates selected for each stage was assigned as the initial values ​​for the corresponding injection-production parameters. Multiple different initial parameter combinations were generated using the same method, each corresponding to an initial chemical flooding scheme, maintaining injection-production balance. These initial parameters were then used as the initial particles for subsequent particle swarm optimization calculations. Maintaining injection-production balance during chemical flooding means setting the production rate of production wells based on the chemical agent injection rates determined for each stage, ensuring that the total injection volume of each injection well is equal to the total produced volume of each production well.

[0040] Step 2, which involves performing reservoir numerical simulation calculations for the initial chemical flooding scheme in high-viscosity reservoirs, refers to: assigning the initial dynamic control thresholds (including the first and second dynamic control thresholds) between different development stages in the initial chemical flooding scheme for high-viscosity reservoirs, as well as the chemical agent injection concentration and injection rate at each stage, to the reservoir numerical simulation calculations established in Step 1; performing dynamic simulations of development while maintaining injection-production balance; calculating and outputting the water cut and cumulative oil production during the chemical flooding process (obtained through commercial reservoir numerical simulation software); and stopping the reservoir numerical simulation calculations when the overall water cut reaches 98%.

[0041] In this example, in step 2, the initial threshold for dynamic control during the transition between different development stages is determined based on the development dynamic curve, and the initial values ​​of key injection and production parameters for each stage are set to form a collaboratively optimized initial scheme for reservoir numerical simulation calculation. The calculation stops when the water cut reaches 98%. The initial thresholds and parameters are shown in Table 1. Table 1. Initial Thresholds and Parameters; In step 3, piecewise linear regression is used to divide the water cut curve of the chemical flooding stage into 3 linear segments, as shown in equation (I): (I); In formula (Ⅰ), For the first Moisture content corresponding to each data point; For the first The development time after chemical flooding corresponding to each data point; For the first Error term for each data point; This represents the total number of data points on the moisture content curve. For data point sequence numbers; For the numbering of linear line segments, These respectively represent the stages of declining water content, the valley-to-plateau stage, and the water content recovery stage; and The first The intercept and slope of the fitted line at each stage. and The positions of the first and second candidate inflection points in the moisture content data sequence are respectively assigned, satisfying the following conditions: , and These represent the development times corresponding to the first and second candidate inflection points, respectively. Because... and These are the data point indices for the first and second turning points, respectively. In the middle, when and Time, that is and Then, the time corresponding to the candidate turning point is .

[0042] In step 3, the first The slope of each stage and intercept The least squares method is used for calculation, and the calculation formulas are shown in equations (II) and (III) respectively: (II); (III); In equations (II) and (III), For the first The set of indexes for data points included in each stage; For the first The average development time corresponding to each data point within each stage; For the first The average moisture content corresponding to each data point within each stage. Therefore, based on the development time and moisture content data within each stage, the slope and intercept of the fitted straight line for that stage are calculated.

[0043] In step 3, after completing the piecewise linear regression, the sum of squared residuals for each stage and the total sum of squared residuals for the three stages are calculated using the formulas shown in equations (Ⅳ) and (Ⅴ): (Ⅳ); (V); In equations (Ⅳ) and (Ⅴ), For the first The sum of squared residuals for each stage is used to characterize the deviation between the actual moisture content and the fitted moisture content for that stage. This is the sum of squared residuals for the three stages.

[0044] In step 3, the stage transition boundary is determined according to the following condition: the slope of the water cut decline stage. The slope of the valley-bottom plateau phase satisfies The slope of the water-bearing recovery phase ;in, This indicates that the moisture content decreases over time. This indicates that the rate of change in water content during the valley plateau stage does not exceed 10% of the absolute value of the rate of change in water content during the water content decline stage, and is therefore determined to be the valley plateau stage. This indicates that the moisture content increases over time. The slope condition and the sum of squares of the total residuals will be satisfied. The smallest candidate inflection point combination is determined as the optimal inflection point combination. The time point corresponding to the first optimal inflection point is determined as the first stage transition boundary time point for the transition from the water level decline stage to the valley plateau stage. The time point corresponding to the second optimal inflection point is determined as the second stage transition boundary time point for the transition from the valley plateau stage to the water level recovery stage.

[0045] In step 3, after determining the first and second stage transition boundary time points, the cumulative oil production corresponding to the first and second stage transition boundary time points is read from the output of the same reservoir numerical simulation. The cumulative oil production corresponding to the first stage transition boundary time point is determined as the first dynamic control candidate threshold, and the cumulative oil production corresponding to the second stage transition boundary time point is determined as the second dynamic control candidate threshold. This is used for subsequent iterative updates of the dynamic control threshold.

[0046] In step 3, such as Figure 2 As shown, the update process for the dynamically adjusted threshold is as follows: First, calculate the gradient direction of the dynamic control threshold change. The calculation formula is shown in equation (VI): (VI); In formula (VI), The currently used dynamic adjustment threshold, To dynamically adjust the candidate threshold, The dynamic control threshold is used to adjust the direction and amount of change of the candidate threshold relative to the currently used dynamic control threshold; the boundary value here is the dynamic control threshold, which is determined by piecewise linear regression and least squares method.

[0047] Then, calculate the adaptive learning rate. The calculation formulas are shown in equations (VII) and (VIII): (VII); (VIII); In equations (VII) and (VIII), Adaptive learning rate; The base learning rate is set at 0.2. To dynamically adjust the relative error between the candidate threshold and the currently used boundary value; A baseline scale for the error was defined as a reference error value, set at 10%. The learning rate scaling factor is set to 0.5 to ensure fine-tuning even with small errors. This is a scaling factor for the learning rate to prevent oscillations caused by excessive errors; the value is set to 2.0. Finally, the dynamically adjusted threshold is updated using the obtained gradient direction and adaptive learning rate, as shown in equation (IX): (IX); In formula (IX), This is the updated dynamic adjustment threshold. It will be used in the next optimization iteration.

[0048] In step 3, the formula for calculating the increase in oil yield per ton of agent is shown in equation (X): (X); In formula (X), To increase oil production per ton of agent, For the cumulative oil production of chemical flooding, This represents the cumulative oil production from the entire water-drive process. The cumulative amount of chemical agent injected to strongly reduce adhesion. In formula (X), The cumulative oil production output from the reservoir numerical simulation of the chemical flooding scheme is obtained by the reservoir numerical simulation from the start of development to the set termination conditions. The water drive scheme was obtained through reservoir numerical simulation. The water drive scheme and the chemical flooding scheme use the same reservoir geological model, well pattern, initial state and simulation termination conditions. Specifically, it is the cumulative oil production output of the reservoir numerical simulation from the start of development to the set termination conditions.

[0049] The chemical flooding scheme is determined based on the chemical injection concentration, injection rate, and time step length at each time step. Specifically, the mass of chemical agent injected in each time step is calculated separately, and then the masses of chemical agent injected across all time steps are summed. The unit for chemical agent injection concentration is kg / m³, the unit for injection rate is m³ / d, and the unit for time step length is d. After summing, the units are converted to obtain the cumulative mass of chemical agent injected in tons (t).

[0050] The formulas for calculating the cumulative oil production of chemical flooding, the cumulative oil production of water flooding throughout the entire process, and the cumulative injection amount of strong adhesion-reducing chemicals are as follows: (XI); (XII); (XIII); in, and Two schemes are respectively Oil production rate at each time step; This refers to the mass concentration of the chemical agent, expressed in kg / m³. The total mass of chemical agent injected, in tons (t). The ground volume injection rate is expressed in m³ / d. The time step duration is expressed in days (d).

[0051] Figure 3 This diagram illustrates the comparison of cumulative oil production optimization iterations using the collaborative optimization method and the ordinary optimization method. Figure 4 This is a comparison chart of the cumulative oil production of the ordinary optimization method and the collaborative optimization method.

[0052] In step 3, with maximizing the increase in oil yield per ton of agent as the objective function, the particle swarm optimization algorithm is used to optimize the chemical agent injection concentration and injection rate during the water cut decline, valley plateau, and water cut recovery stages. The updated parameters are then used as input parameters for the next iteration. The parameter update formulas are shown in equations (XIV) and (XV). (XIV); (XV); In formula (XIV), and The first Next iteration particle and the first During iteration The first particle Speed ​​of dimensional parameters; The inertia weight is set to 0.7298. and The acceleration constant is 1.4692. and A random number that varies between 0 and 1; This refers to the best position the particle itself has ever experienced, i.e., the individual's historical best position. The best position experienced by all particles in the current swarm is the global optimal position of the entire swarm, which is the current optimal objective function value.

[0053] Figure 5 A comparison chart of chemical agent injection amounts between the conventional optimization method and the collaborative optimization method.

[0054] Specifically, in each iteration, the chemical flooding parameter scheme corresponding to the current position of each particle is input into the reservoir numerical simulation model, the oil production per ton of agent corresponding to the parameter scheme is calculated, and this is used as the objective function value of the particle. For the ... For each particle, the objective function value obtained in the current iteration is compared with the objective function values ​​obtained in previous iterations. If the current objective function value is greater than the objective function value corresponding to the particle's existing historical best position, the particle's position is updated to its existing historical best position; otherwise, the original historical best position is retained. This yields the [f1]... During the next iteration ;

[0055] After updating the individual historical best positions of all particles, the fuel yield per ton of agent corresponding to each particle's individual historical best position is compared. The individual historical best position with the largest fuel yield per ton of agent is determined as the current global best position, and based on this, the following is obtained: In the next iteration, each particle updates its velocity by simultaneously referencing its own historical best position and the global best position of the particle swarm, according to equation (IX). Figure 6 A comparison chart of the increase in oil production per ton of agent between the initial scheme and the co-optimized scheme.

[0056] In step 4, during the optimization iteration process, the calculation process is terminated when the calculated increase in oil per ton of agent remains unchanged for 10 consecutive iterations. The optimized dynamic control threshold and parameters for each stage are shown in Table 2.

[0057] Table 2 shows the optimized dynamic control thresholds and parameters for each stage.

[0058] In the iterative optimization of chemical flooding in high-viscosity reservoirs, the dynamic control threshold characterizing the transition of development stages and the injection parameters (chemical agent injection concentration and chemical agent injection rate) of each stage of chemical flooding are used as unified joint decision variables. Synchronous optimization is carried out under the same optimization framework, and a stage transition boundary-injection-production parameter collaborative optimization space is constructed.

[0059] In this example, in step 4, during the iterative optimization of chemical flooding, the dynamic control threshold characterizing the development stage transition and the injection parameters of each stage of chemical flooding are used as unified joint decision variables. Synchronous optimization is performed within the same optimization framework, constructing a collaborative optimization space for development stage division and injection-production parameter design. After adopting the full-cycle collaborative optimization method for high-viscosity reservoir chemical flooding proposed in this invention, the final optimized scheme achieves a cumulative oil production of 1,511,725 ​​m³. 3 The oil yield per ton of dosage was 54 tons / ton, which is 8% higher than that of the ordinary optimization method.

Claims

1. A full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs, characterized in that, include: Step 1: Determine the waterdrive development scheme for the entire process of high-viscosity reservoirs and perform reservoir numerical simulation calculations; Step 2: Determine the initial threshold for dynamic control between different development stages of waterflooding followed by chemical flooding in high-viscosity reservoirs, as well as the initial values ​​of injection and production parameters for each stage, and perform reservoir numerical simulation calculations for the initial chemical flooding scheme in high-viscosity reservoirs. Step 3: Based on the calculated development dynamics, with the goal of maximizing the increase in oil yield per ton of injection, the fixed-point iteration method and particle swarm optimization algorithm are used to simultaneously and collaboratively optimize the dynamic control thresholds for the transition between different development stages and the injection and production parameters for each production stage. Then, reservoir numerical simulation calculations are performed on the updated chemical flooding scheme. The fixed-point iterative method for dynamic control thresholds is based on the water cut curve obtained from reservoir numerical simulation. The curve is divided into three linear segments by piecewise linear regression. The slope of each segment is fitted by the least squares method to minimize the sum of squared residuals. Candidate boundary values ​​for the transition of different development stages are determined based on the slope difference between adjacent segments. The dynamic control thresholds of each stage are updated in subsequent optimization iterations. The particle swarm optimization algorithm treats each solution as a particle. First, based on the oil increase per ton of agent calculated by reservoir numerical simulation, it updates the individual historical best position and the global best position of the entire swarm. Then, the particle swarm optimization algorithm integrates the information of the individual historical best solution and the global best solution of the swarm to update the position parameters of each particle. Step 4: Repeat step 3 until the iteration converges to the maximum increase in oil production per ton of agent.

2. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 1, a water-drive development plan for the entire process of high-viscosity reservoirs is formulated, and reservoir numerical simulation calculations are performed. The calculation is stopped when the water cut reaches 98%, and the cumulative oil production of the entire process of water-drive development is calculated. In step 1, formulating a full-process water drive development plan for high-viscosity reservoirs means: continuously injecting water into injection wells at a preset water injection rate without injecting chemical agents, and controlling the production of production wells according to a preset production volume. Numerical simulation of oil reservoirs refers to: dividing the target oil reservoir into grid cells; assigning porosity, permeability, initial oil saturation, initial pressure, and rock and fluid properties to each grid cell; setting the locations of injection wells and production wells, completion intervals, and control conditions; using the mass conservation equation, Darcy's flow equation, and oil-water two-phase flow relationship, solving for the pressure, oil saturation, and water saturation of each grid cell step by step; and obtaining the changes in water cut and oil production over time based on the oil production rate and water production rate of each production well. The calculation stops when the water cut reaches 98%. The cumulative oil production of the entire water-drive development process is obtained by summing up the oil production over time.

3. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 2, when the water flooding reaches a water cut of 90%, the process is switched to chemical flooding. After the chemical flooding is completed, water flooding continues. In step 2, the chemical flooding process of high viscosity reservoirs is divided into the water cut decline stage, the valley plateau stage, and the water cut recovery stage. The dynamic control threshold for the transition between different development stages refers to two key points on the dynamic curve of chemical flooding development of high viscosity reservoirs that characterize the stage transition, including the first turning point from the water cut decline stage to the valley plateau stage, and the second turning point from the valley plateau stage to the water cut recovery stage. Based on the reservoir numerical simulation results of the initial chemical flooding scheme, the water cut and cumulative oil production at each time step are obtained in the order of simulation time. First, the first inflection point and the second inflection point and their corresponding two simulation times are determined on the water cut dynamic curve. Then, the cumulative oil production corresponding to the two simulation times is directly read from the reservoir numerical simulation results. The cumulative oil production corresponding to the first inflection point is assigned as the first dynamic control initial threshold, and the cumulative oil production corresponding to the second inflection point is assigned as the second dynamic control initial threshold. In the subsequent development process, when the cumulative oil production reaches the first dynamic control initial threshold, the development process will switch from the water cut decline stage to the valley platform stage. When the cumulative oil production reaches the second dynamic control initial threshold, the development process transitions from the valley platform stage to the water cut recovery stage.

4. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 2, the initial values ​​of injection and production parameters are determined for the water cut decline stage, valley platform stage, and water cut recovery stage, respectively. The initial values ​​of injection and production parameters include the chemical agent injection concentration and chemical agent injection rate for each stage. First, based on the experimental results of chemical agent performance evaluation under the target reservoir conditions, the concentration range of the chemical agent was determined to be 0–2.5 kg / m³. 3 And the range of chemical agent concentrations is used as the range of values ​​for the chemical agent injection concentration; Then, based on the water absorption capacity of the injection well, the allowable injection pressure of the formation, and the working capacity of the injection equipment, the chemical injection rate range was determined to be 30–100 m. 3 / d; Within the range of chemical agent injection concentration and chemical agent injection rate, chemical agent injection concentration and chemical agent injection rate are selected for the water cut decline stage, valley platform stage and water cut recovery stage, respectively, and a set of chemical agent injection concentration and chemical agent injection rate selected for each stage are assigned as the initial values ​​of the injection and production parameters for the corresponding stage. Multiple sets of different initial parameter combinations are generated, each set of parameter combinations corresponds to an initial chemical flooding scheme, and injection and production are kept in balance; Step 2, which involves performing reservoir numerical simulation calculations for the initial chemical flooding scheme in high-viscosity reservoirs, refers to: assigning the initial dynamic control thresholds (including the first and second dynamic control thresholds) between different development stages in the initial chemical flooding scheme for high-viscosity reservoirs, as well as the chemical agent injection concentration and injection rate at each stage, to the reservoir numerical simulation calculations established in Step 1; performing dynamic simulations of development while maintaining injection-production balance; calculating and outputting the water cut and cumulative oil production during the chemical flooding process; and stopping the reservoir numerical simulation calculations when the overall water cut reaches 98%.

5. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 3, piecewise linear regression is used to divide the water cut curve of the chemical flooding stage into 3 linear segments, as shown in equation (I): （Ⅰ）； In formula (Ⅰ), For the first Moisture content corresponding to each data point; For the first The development time after chemical flooding corresponding to each data point; For the first Error term for each data point; This represents the total number of data points on the moisture content curve. For data point sequence numbers; For the numbering of linear line segments, These respectively represent the stages of declining water content, the valley-to-plateau stage, and the water content recovery stage; and The first The intercept and slope of the fitted line at each stage. and The positions of the first and second candidate inflection points in the moisture content data sequence are respectively assigned, satisfying the following conditions: , and These are the development times corresponding to the first and second candidate inflection points, respectively. In step 3, the first The slope of each stage and intercept The least squares method is used for calculation, and the calculation formulas are shown in equations (II) and (III) respectively: （Ⅱ）； （Ⅲ）； In equations (II) and (III), For the first The set of indexes for data points included in each stage; For the first The average development time corresponding to each data point within each stage; For the first The average moisture content of each data point within each stage.

6. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 3, after completing the piecewise linear regression, the sum of squared residuals for each stage and the total sum of squared residuals for the three stages are calculated using the formulas shown in equations (Ⅳ) and (Ⅴ): （Ⅳ）； （Ⅴ）； In equations (Ⅳ) and (Ⅴ), For the first The sum of squared residuals for each stage is used to characterize the deviation between the actual moisture content and the fitted moisture content for that stage. The sum of squared residuals for the three stages; In step 3, the stage transition boundary is determined according to the following condition: the slope of the water cut decline stage. The slope of the valley-bottom plateau phase satisfies The slope of the water-bearing recovery phase ;in, This indicates that the moisture content decreases over time. This indicates that the rate of change in water content during the valley plateau stage does not exceed 10% of the absolute value of the rate of change in water content during the water content decline stage, and is therefore determined to be the valley plateau stage. This indicates that the moisture content increases over time. The slope condition and the sum of squares of the total residuals will be satisfied. The smallest candidate inflection point combination is determined as the optimal inflection point combination. The time point corresponding to the first optimal inflection point is determined as the first stage transition boundary time point for the transition from the water level decline stage to the valley plateau stage. The time point corresponding to the second optimal inflection point is determined as the second stage transition boundary time point for the transition from the valley plateau stage to the water level recovery stage.

7. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 3, after determining the first stage transition boundary time point and the second stage transition boundary time point, the cumulative oil production corresponding to the first stage transition boundary time point and the second stage transition boundary time point are read from the same reservoir numerical simulation output results, respectively. The cumulative oil production corresponding to the first stage transition boundary time point is determined as the first dynamic control candidate threshold, and the cumulative oil production corresponding to the second stage transition boundary time point is determined as the second dynamic control candidate threshold. In step 3, the update process for the dynamically adjusted threshold is as follows: First, calculate the gradient direction of the dynamic control threshold change. The calculation formula is shown in equation (VI): （Ⅵ）； In formula (VI), The currently used dynamic adjustment threshold, To dynamically adjust the candidate threshold, To dynamically adjust the direction and amount of change of the candidate threshold relative to the currently used dynamic control threshold; Then, calculate the adaptive learning rate. The calculation formulas are shown in equations (VII) and (VIII): （Ⅶ）； （Ⅷ）； In equations (VII) and (VIII), Adaptive learning rate; Base learning rate; To dynamically adjust the relative error between the candidate threshold and the currently used boundary value; This is a reference error value; This is the lower bound factor for scaling the learning rate; This is the scaling factor for the learning rate. Finally, the dynamically adjusted threshold is updated using the obtained gradient direction and adaptive learning rate, as shown in equation (IX): （Ⅸ）； In formula (IX), This is the updated dynamic adjustment threshold.

8. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 3, the formula for calculating the increase in oil yield per ton of agent is shown in equation (X): （X）； In formula (X), To increase oil production per ton of agent, For the cumulative oil production of chemical flooding, This represents the cumulative oil production from the entire water-drive process. The cumulative amount of chemical agent injected to strongly reduce adhesion. The formulas for calculating the cumulative oil production of chemical flooding, the cumulative oil production of water flooding throughout the entire process, and the cumulative injection amount of strong adhesion-reducing chemicals are as follows: (XI); (XII); (XIII); in, and Two schemes are respectively Oil production rate at each time step; This refers to the mass concentration of the chemical agent, expressed in kg / m³. The total mass of chemical agent injected, in tons (t). The injection rate is the volumetric rate at which water is injected into the ground, expressed in m³ / d. The time step duration is expressed in days (d).

9. The full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to claim 1, characterized in that, In step 3, with maximizing the increase in oil yield per ton of agent as the objective function, the particle swarm optimization algorithm is used to optimize the chemical agent injection concentration and injection rate during the water cut decline, valley plateau, and water cut recovery stages. The updated parameters are then used as input parameters for the next iteration. The parameter update formulas are shown in equations (XIV) and (XV). (XIV); （ⅩV）； In formula (XIV), and The first Next iteration particle and the first During iteration The first particle Speed ​​of dimensional parameters; Inertial weight; and It is the acceleration constant; and A random number that varies between 0 and 1; This refers to the best position the particle itself has ever experienced, i.e., the individual's historical best position. The best position experienced by all particles in the current swarm is the global optimal position of the entire swarm, which is the current optimal objective function value.

10. A full-cycle synergistic optimization method for chemical flooding in high-viscosity reservoirs according to any one of claims 1-9, characterized in that, In step 4, during the optimization iteration process, if the calculated increase in oil per ton of agent remains unchanged for 10 consecutive iterations, the calculation process is terminated.

Images