High-efficiency parallel optimization method and device considering actual engineering constraints of horizontal well

Abstract

The present application relates to a kind of efficient parallel optimization method and device considering the actual engineering constraint of horizontal well, method includes the following steps: S1 constructs water drive reservoir production optimization mathematical control model;S2 based on horizontal well engineering application condition and reservoir numerical model, the constraint judgment method of horizontal well is constructed;S3 based on the population in Latin hypercube sampling initialization algorithm, using the constraint judgment method of horizontal well is handled, database is constructed;S4 using the excellent individual in database forms temporary population, based on temporary population, Gaussian model is constructed;S5 based on differential evolution algorithm with different strategies obtains three sub-populations, and using the constraint judgment method of horizontal well is handled;S6 based on the pseudo-update strategy of temporary population, the potential individual of sub-population is obtained;S7 based on Euclidean distance, select the individual closest to current optimal individual, and carry out parallel numerical simulation;S8 updates database.

Description

Technical Field

[0001] This invention relates to an efficient parallel optimization method and apparatus that takes into account the actual engineering constraints of horizontal wells, belonging to the field of oil and gas field development technology. Background Technology

[0002] Horizontal wells are an important tool for developing complex oil reservoirs. Compared to vertical wells, the horizontal section of a horizontal well extends into the reservoir, providing a larger contact area with the oil-bearing rock. This effectively increases oil production and efficiency, significantly improving economic benefits. However, the development cost of horizontal wells is very high, and the risks are substantial. Improper deployment of horizontal wells can lead to significant economic losses.

[0003] Therefore, the deployment of horizontal wells is one of the most important decisions in oilfield development. First, the well spacing must be close enough to help create as much reservoir volume as possible for increased production; however, the well spacing must also be wide enough to reduce fracture interference (inter-well interference and "fracturing shock") and over-investment in oilfield development. Furthermore, the horizontal section of the well should extend along the better and thicker reservoir, perpendicular to the direction of maximum principal stress, in order to traverse more fracture zones and expand reservoir connectivity to achieve higher production capacity.

[0004] Since the 1990s, scholars have successively proposed optimization methods for horizontal well locations. With the development of numerical computing technology and modern optimization algorithms, the effectiveness and efficiency of optimization have been improved to a certain extent, and the methods for horizontal well location optimization have been further developed. The Differential Evolutionary Algorithm (DEA), proposed in 1997 by Rainer Storn and Kenneth Price based on evolutionary ideas such as genetic algorithms, is essentially a multi-objective (continuous variable) optimization algorithm (MOEAs) used to solve for the overall optimal solution in a multi-dimensional space. Compared to genetic algorithms, DEA shares similarities in that both randomly generate an initial population and use the fitness value of each individual in the population as the selection criterion. Both processes mainly include three steps: mutation, crossover, and selection. The difference lies in that genetic algorithms control parent crossover based on fitness values, and the probabilities of offspring selected after mutation are determined by the offspring's fitness value. In maximization problems, individuals with higher fitness values ​​have a higher probability of being selected. In contrast, the mutation vector in DEA is generated from the parent's difference vector and cross-referenced with the parent's individual vector to generate a new individual vector, which then directly selects from its parent. Clearly, the differential evolution algorithm has a more significant approximation effect than the genetic algorithm.

[0005] Horizontal well location optimization requires constraints because horizontal wells cannot intersect, and the well spacing cannot be too close to meet the practical application requirements of oilfields. Therefore, horizontal well location optimization is a constrained optimization problem. Differential evolutionary algorithm (DEA) is an intelligent global optimization method with very low requirements on the properties of the function itself. It often only requires that the objective function value be computable, without requiring it to have continuity, differentiability, or other analytical properties. Furthermore, it is based on population evolution, thus DEA can be used to solve constrained optimization problems. The key to using DEA to solve constrained optimization problems lies in how to effectively handle constraints, that is, how to effectively balance the search between feasible and infeasible regions.

[0006] Furthermore, numerical simulations of actual oilfield models require extremely expensive computational resources, and optimization is time-consuming. Parallel computing can effectively reduce the computational burden of numerical simulations. Therefore, there is a need to provide a parallel algorithm for expensive horizontal well constraint optimization. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides an efficient parallel optimization method and apparatus that considers the actual engineering constraints of horizontal wells. This method uses a surrogate model instead of an expensive numerical simulator, and uses a parallel optimization algorithm to significantly reduce the computation time during actual evaluation. The algorithm incorporates constraints between horizontal well locations, enabling large-scale horizontal well location optimization.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] An efficient parallel optimization method considering practical engineering constraints of horizontal wells includes the following steps:

[0010] S1 Constructs a mathematical control model for optimizing water-drive reservoir production;

[0011] S2 constructs a constraint judgment method for horizontal wells based on the engineering application conditions of horizontal wells and reservoir numerical models;

[0012] S3 uses the population in the Latin hypercube sampling initialization algorithm, performs constraint processing using the constraint judgment method of horizontal wells, calculates the NPV in the mathematical model for production optimization of water-drive reservoirs, and constructs a database.

[0013] S4 uses the best individuals in the database to form a temporary population, and then constructs a Gaussian model based on the temporary population.

[0014] S5 uses the differential evolution algorithm to obtain three subpopulations with different strategies, and uses the horizontal well constraint judgment method for constraint processing.

[0015] S6 uses a pseudo-update strategy based on a temporary population to obtain potential individuals from the subpopulation;

[0016] S7 uses Euclidean distance to select the individual closest to the current best individual for parallel numerical simulation;

[0017] S8 updates the database. If the iteration stops when the condition is met, the optimal individual is output; otherwise, return to step S4.

[0018] The efficient parallel optimization method, preferably, includes step S1 specifically comprising:

[0019] Based on reservoir development needs and with the objective of maximizing the net present value (NPV), a mathematical control model for optimizing waterflood reservoir production is constructed, as shown below:

[0020]

[0021] In the formula, x is the control variable of the production optimization problem; ψ(x) represents the objective function value, i.e., the economic benefit NPV; T is the total time step of the numerical simulation; Δt n The step size for simulating the nth iteration is d; d is the number of control variables; ω is the annual decay rate. and These represent the oil production rate, water production rate, and water injection rate of the i-th production well at time step n, respectively; e o d w d wi These represent the economic benefits per unit volume of crude oil, the cost of treating per unit volume of wastewater, and the cost of injecting per unit volume of water, respectively; NP and NI represent the total number of production wells and water injection wells in the reservoir, respectively.

[0022] The efficient parallel optimization method, preferably, includes step S2 specifically comprising:

[0023] S21 extracts the horizontal segment trajectory features of each horizontal well in the reservoir numerical model, approximating the horizontal segment trajectory of the horizontal well as a line segment on the plane. For any horizontal well k, it extracts the grid coordinates corresponding to the two endpoints of its horizontal segment in the reservoir numerical model. and As its horizontal segment trajectory identifier Tr k ;

[0024] S22 constructs a set of horizontal segment trajectory identifiers, Tr={Tr1,Tr2,...,Tr...} s ,...,Tr s+t}, where s is the number of new horizontal wells whose optimized locations need to be determined, and t is the number of existing horizontal wells in the reservoir model;

[0025] S23 For any two newly added horizontal wells or any newly added horizontal well and existing horizontal wells a and b in the horizontal segment trajectory identifier set Tr, use their corresponding horizontal segment trajectory identifiers Tr a With Tr b Preprocessing before constraint is performed; for horizontal wells a and b, the result of the preprocessing is used to form two rectangles whose diagonals are the lines connecting the two endpoints of the horizontal segments of horizontal wells a and b.

[0026] S24 determines whether the rectangles formed by the two horizontal wells a and b overlap to determine whether the constraint conditions are met.

[0027] The efficient parallel optimization method, preferably, includes step S3 specifically comprising:

[0028] S31 uses the Latin hypercube sampling method to generate NP horizontal well location schemes S = [x1; x2; ...; x...]. NP ];

[0029] S32 performs horizontal well location constraint determination, and the determination method is the same as step S2;

[0030] S33 If there are individuals that do not meet the constraints, then Latin hypercube sampling is performed again until all individuals meet the constraints.

[0031] S34 uses a numerical simulator to perform parallel computation on all individuals to obtain the NPV response value y = [y1; y2; ...; y] in the mathematical control model for production optimization of water-driven reservoirs. NP Use S and y to construct the initial database DB.

[0032] The efficient parallel optimization method, preferably, includes step S4 specifically comprising:

[0033] S41 Select the τ individuals with the highest NPV response values ​​from the database DB to construct a temporary population P. temp ;

[0034] S42 utilizes P temp Constructing a Gaussian proxy model M G .

[0035] The efficient parallel optimization method, preferably, includes step S5 specifically comprising:

[0036] S51 generates three subpopulations P based on three different strategies using the differential evolution algorithm. i (i = 1, 2, 3);

[0037] S52 pairs of three subpopulations P i (i = 1, 2, 3) are subject to constraint processing, and the processing method is the same as steps S31 to S33.

[0038] The efficient parallel optimization method, preferably, includes step S6 specifically comprising:

[0039] S61 utilizes the Gaussian model M G Calculate the target response value in each subpopulation and calculate the fitness value based on the confidence lower bound (LCB) criterion.

[0040] S62 For each subpopulation, select the individual with the best response value. Deposit Q i In (i = 1, 2, 3), update it to P. temp In, but not updating M G ;

[0041] S63 when Q i If the number of individuals stored in (i = 1, 2, 3) reaches q, then proceed to the next step; otherwise, use the updated P. temp Return to step S5.

[0042] The efficient parallel optimization method, preferably, includes step S7 specifically comprising:

[0043] S71 calculates Q i From individuals in (i = 1, 2, 3) to the current best individual x best Given the Euclidean distance, select the m individuals S with the smallest distance. new = [x1; x2; ...; x m Parallel numerical simulations were performed to obtain the corresponding NPV response value y. new = [y1; y2; ...; y m ];

[0044] The number of iterations for S72 is increased by m.

[0045] A second aspect of the present invention provides a highly efficient parallel optimization device that considers the actual engineering constraints of horizontal wells, comprising:

[0046] The first processing unit is used to construct a mathematical control model for optimizing water-drive reservoir production.

[0047] The second processing unit is used to construct a constraint judgment method for horizontal wells based on the engineering application conditions of horizontal wells and the reservoir numerical model.

[0048] The third processing unit is used to perform constraint processing based on the population in the Latin hypercube sampling initialization algorithm, using the constraint judgment method of horizontal wells, and to calculate the NPV in the mathematical model for production optimization of water-drive reservoirs and build a database.

[0049] The fourth processing unit is used to form a temporary population using the best individuals in the database, and to build a Gaussian model based on the temporary population.

[0050] The fifth processing unit is used to obtain three subpopulations using different strategies based on the differential evolution algorithm, and to perform constraint processing using the constraint judgment method of horizontal wells;

[0051] The sixth processing unit is used to obtain potential individuals from the subpopulation based on a pseudo-update strategy of the temporary population.

[0052] The seventh processing unit is used to select the individual closest to the current best individual based on Euclidean distance and perform parallel numerical simulations.

[0053] The eighth processing unit is used to update the database. When the iteration stopping condition is met, it outputs the optimal individual; otherwise, it returns to step S4.

[0054] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described efficient parallel optimization method considering the actual engineering constraints of horizontal wells.

[0055] A fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described efficient parallel optimization method considering the actual engineering constraints of horizontal wells.

[0056] The present invention has the following advantages due to the adoption of the above technical solutions:

[0057] 1. This invention proposes a novel horizontal well location optimization constraint algorithm, which treats the horizontal segment of the horizontal well as the diagonal of a rectangle to determine whether the rectangles overlap. While ensuring that the horizontal wells do not intersect, it also ensures the distance between the horizontal wells, reducing inter-well interference and fracturing impact, which meets actual needs and provides useful guidance for oilfield operations.

[0058] 2. This invention utilizes different differential evolutionary algorithms to generate subpopulations, balancing the exploration of new directions and response values ​​in the search space, thereby finding a better global optimal solution and effectively improving algorithm performance.

[0059] 3. The sampling method of this invention is more reasonable, and the location of the optimal solution can be quickly located. Only a few dozen iterations are needed to obtain a suitable horizontal well location scheme.

[0060] 4. The optimization method of the present invention is not prone to getting trapped in local optima, and provides a horizontal well location scheme that can obtain the maximum economic benefits while ensuring good stability.

[0061] 5. This invention enables parallel optimization of horizontal well locations, reduces the computation time of numerical simulation, significantly improves the optimization speed, and is beneficial for practical field applications. Attached Figure Description

[0062] Figure 1 This is a flowchart of a horizontal well constraint optimization method based on a parallel algorithm provided in an embodiment of the present invention;

[0063] Figure 2 This is an optimized curve diagram provided in this embodiment of the present invention;

[0064] Figure 3 This is a comparison diagram of the remaining oil saturation of the reservoir model before and after optimization provided in this embodiment of the invention, wherein... Figure 3 'a' represents the state before optimization. Figure 3 b represents the optimized version;

[0065] Figure 4 This is a schematic diagram of the optimized horizontal well location provided in this embodiment of the present invention. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention are described clearly and completely below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0067] This invention addresses the issue that numerical simulation of existing oilfield models requires expensive computational resources and consumes a significant amount of time during optimization. Parallel computing can effectively reduce the computational burden of numerical simulation. Therefore, this invention proposes an efficient parallel optimization method that considers the actual engineering constraints of horizontal wells. This method treats the horizontal segments of a well as the diagonals of rectangles to determine whether the rectangles overlap. While ensuring that horizontal wells do not intersect, it also maintains the distance between them, reducing inter-well interference and fracturing impact, thus meeting practical needs and providing valuable guidance for oilfield operations.

[0068] like Figure 1 As shown, the efficient parallel optimization method for considering the actual engineering constraints of horizontal wells involved in this invention includes the following steps:

[0069] S1. Construct a mathematical control model for optimizing water-drive reservoir production. The specific method is as follows:

[0070] Based on reservoir development needs, and with the objective of maximizing the net present value (NPV), a mathematical control model for optimizing waterflood reservoir production is constructed, as shown below:

[0071]

[0072] In the formula, x is the control variable of the production optimization problem; ψ(x) represents the objective function value, i.e., the economic benefit NPV; T is the total time step of the numerical simulation; Δt n The step size for the nth iteration is in days; d is the number of control variables; ω is the annual decay rate. and These represent the oil production rate, water production rate, and water injection rate of the i-th production well at the nth time step, respectively, with units of STB / D; e o d w d wi The figures represent the economic benefits per unit volume of crude oil, the cost of treating per unit volume of wastewater, and the cost of injecting per unit volume of water, all in USD / STB. NP and NI represent the total number of production wells and injection wells in the reservoir, respectively.

[0073] S2. Based on the engineering application conditions of horizontal wells, a constraint judgment method for horizontal wells is constructed. The specific method is as follows:

[0074] Based on the constraints of "the horizontal trajectories of any two wells do not intersect" and "the positions of any two horizontal wells should be a certain distance apart" in the deployment of horizontal wells, necessary features are extracted for constraint judgment of horizontal wells that need to be optimized in well location deployment.

[0075] The first step is to extract the horizontal segment trajectory features of each horizontal well in the reservoir numerical model (a known reservoir numerical model). The horizontal segment trajectory of the horizontal well is approximated as a line segment on the plane. For any horizontal well k, the grid coordinates corresponding to the two endpoints of its horizontal segment in the reservoir model are extracted. and As its horizontal segment trajectory identifier Tr k .

[0076] The second step is to construct a set of horizontal segment trajectory identifiers, Tr = {Tr1, Tr2, ..., Tr...} s ,...,Tr s+t}, where s is the number of new horizontal wells whose optimized locations need to be determined, and t is the number of existing horizontal wells in the reservoir numerical model.

[0077] The third step involves using the corresponding horizontal segment trajectory identifier Tr for any two newly added horizontal wells or any newly added horizontal well and existing horizontal wells a and b from the horizontal segment trajectory identifier set Tr. a With Tr b Preprocessing before constraints are applied. Where Tr a The coordinates of the two endpoints of the horizontal segment of horizontal well a in the reservoir numerical model and Tr b The coordinates of the two endpoints of the horizontal segment of the corresponding horizontal well b in the reservoir numerical model and

[0078] For Tr a Preprocess the coordinates in the data:

[0079] For Tr b The coordinates in the data are preprocessed.

[0080] For horizontal well a, the reservoir model takes coordinates [A] in the x-axis direction. xmin A xmax ], take [A] in the y-axis direction coordinate. ymin A ymax Within the range of [ ], a rectangle is formed whose diagonal is the line connecting the two endpoints of the horizontal segment of horizontal well a. Similarly, for horizontal well b, there also exists a rectangle formed by B. xmin B xmax B ymin B ymax This forms a rectangle whose diagonal is the line connecting the two endpoints of the horizontal segment of the horizontal well b.

[0081] The fourth step is to determine whether the rectangles formed by the two horizontal wells, a and b, overlap. If the two rectangles do not overlap, then the two horizontal wells are deemed to meet the constraints of "the horizontal trajectories of any two wells do not intersect" and "the positions of any two horizontal wells should be a certain distance apart" in the engineering deployment. This ensures that the two horizontal wells do not intersect, while also guaranteeing a certain well spacing, reducing interference between wells and the impact of fractures. The specific determination method is as follows:

[0082] max(A xmin B xmin )≤min(A xmax B xmax (2)

[0083] max(A ymin B ymin )≤min(A ymax B ymax (3)

[0084] If both of the above equations are satisfied, then the two rectangles are determined not to overlap, and the corresponding two horizontal well locations satisfy the constraint conditions; otherwise, the two horizontal well locations are determined not to satisfy the constraint conditions.

[0085] S3. Based on the population in the Latin hypercube sampling initialization algorithm, constraints are applied to construct the database. The specific method is as follows:

[0086] The first step is to use the Latin hypercube sampling method to generate NP horizontal well location schemes S = [x1; x2; ...; x NP ];

[0087] The second step is to determine the well location constraints of the horizontal well, and the determination method is the same as in step two.

[0088] The third step is to resample individuals that do not meet the constraints, and then perform Latin hypercube sampling again until all individuals meet the constraints.

[0089] The fourth step involves using a numerical simulator to perform parallel computations on all individuals to obtain the NPV response values ​​y = [y1; y2; ...; y] in the mathematical control model for production optimization of water-driven reservoirs. NP ]; Construct the initial database DB using S and y, and set the current iteration number to NP.

[0090] S4. Use the best individuals from the database to form a temporary population, and construct a Gaussian model based on the temporary population. The specific method is as follows:

[0091] The first step is to select the τ individuals with the highest NPV response values ​​from the database to construct a temporary population P. temp ;

[0092] The second step is to use P temp Constructing a Gaussian proxy model M G .

[0093] S5. Based on the differential evolution algorithm, three subpopulations are obtained using different strategies, and constraints are applied. The specific method is as follows:

[0094] The first step involves generating three subpopulations P based on three different strategies using the differential evolution algorithm. i (i = 1, 2, 3), the three strategies are as follows:

[0095] v i =x r1 +F(x r2 -x r3 (4)

[0096] v i =x best +F(x r1 -x r2 (5)

[0097] v i =x i +F(x best -xi )+F(x r1 -x r2 (6)

[0098] In the formula: v i x represents the decision variable generated by the i-th mutation strategy; r1 ,x r2 ,x r3 These are distinct individuals randomly selected from the population; x best It is the individual with the best objective function value in the current population; F is the mutation operator, which is a constant and F∈(0,2]; x i This refers to the i-th individual in the population. After mutation, a crossover operation is required. For each dimension of the decision variable, j = 1, 2, ..., d, a random number rand(0, 1) in the interval [0, 1] needs to be generated. If the random number is greater than the crossover probability CR, no crossover operation is performed, and the value of this dimension remains unchanged. If the random number is less than the crossover probability CR, a crossover operation is required to determine the value u of this dimension. j The rules are as follows:

[0099]

[0100] The second step is to perform constraint processing, which is the same as in step three.

[0101] S6. A population-based pseudo-update strategy is used to obtain potential individuals in a subpopulation. The specific method is as follows:

[0102] The first step is to use the Gaussian model M. G Calculate the objective function value for each subpopulation. The objective function value is calculated using the lower confidence bound (LCB) criterion, which is as follows:

[0103]

[0104] In the formula: x is the predicted vector; Represents the predicted value; is the standard deviation; w is a constant, and w∈(0,2).

[0105] Obtain the target response value in each subpopulation.

[0106] The second step is to select the individual with the best response value for each subpopulation. Deposit Q i In (i = 1, 2, 3), update it to P. temp In, but not updating M G ;

[0107] Third step, if Q iIf the number of individuals stored in (i = 1, 2, 3) reaches q, then proceed to the next step; otherwise, use the updated P. temp Now, return to step five.

[0108] S7. Based on Euclidean distance, select the individual closest to the current best individual and perform parallel numerical simulation. The specific method is as follows:

[0109] First step, calculate Q i From individuals in (i = 1, 2, 3) to the current best individual x best Given the Euclidean distance, select the m individuals S with the smallest distance. new = [x1; x2; ...; x m Parallel numerical simulations were performed to obtain the corresponding NPV response value y. new = [y1; y2; ...; y m ].

[0110] The second step is to increase the number of iterations by m.

[0111] S8. Update the database. If the iteration stopping condition is met, output the optimal individual; otherwise, return to step four. The specific method is as follows:

[0112] Using S new and y new Update the database DB. If the number of iterations reaches the set maximum number of iterations, stop the iteration and output the optimal individual x. best Otherwise, return to step four.

[0113] The technical solution of the present invention will be described below with reference to specific embodiments.

[0114] This embodiment primarily studies and experiments the aforementioned efficient parallel optimization method that considers the actual engineering constraints of horizontal wells to verify its effectiveness. The experimental model is a reservoir model defined by Eclipse, with a grid distribution of 101*101*1, and each grid cell is 100*100*50 feet in size. This model has 9 production wells and 4 injection wells.

[0115] Figure 2 The graph shows that the optimization method of this invention converges very quickly and can find the location of the global optimum in a short time. Figure 3 This is a comparison chart of the remaining oil saturation in the reservoir model before and after optimization. It can be seen that the newly added horizontal wells effectively extracted the remaining oil in the reservoir. Figure 4 This is a schematic diagram of the optimized horizontal well locations. All three horizontal wells are located in areas with high remaining oil content to increase oil production.

[0116] A second aspect of the present invention provides a highly efficient parallel optimization device that considers the actual engineering constraints of horizontal wells, comprising:

[0117] The first processing unit is used to construct a mathematical control model for optimizing water-drive reservoir production.

[0118] The second processing unit is used to construct a constraint judgment method for horizontal wells based on the engineering application conditions of horizontal wells and the reservoir numerical model.

[0119] The third processing unit is used to perform constraint processing based on the population in the Latin hypercube sampling initialization algorithm, using the constraint judgment method of horizontal wells, and to calculate the NPV in the mathematical model for production optimization of water-drive reservoirs and build a database.

[0120] The fourth processing unit is used to form a temporary population using the best individuals in the database, and to build a Gaussian model based on the temporary population.

[0121] The fifth processing unit is used to obtain three subpopulations using different strategies based on the differential evolution algorithm, and to perform constraint processing using the constraint judgment method of horizontal wells;

[0122] The sixth processing unit is used to obtain potential individuals from the subpopulation based on a pseudo-update strategy of the temporary population.

[0123] The seventh processing unit is used to select the individual closest to the current best individual based on Euclidean distance and perform parallel numerical simulations.

[0124] The eighth processing unit is used to update the database. When the iteration stopping condition is met, it outputs the optimal individual; otherwise, it returns to step S4.

[0125] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described efficient parallel optimization method considering the actual engineering constraints of horizontal wells.

[0126] A fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described efficient parallel optimization method considering the actual engineering constraints of horizontal wells.

[0127] This invention is described based on flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to specific embodiments. It should be understood that each block of the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the flowcharts and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0128] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0129] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An efficient parallel optimization method considering practical engineering constraints of horizontal wells, characterized in that, Includes the following steps: S1 Construct a mathematical control model for optimizing water-drive reservoir production; S2 Based on the engineering application conditions of horizontal wells and the numerical model of reservoirs, a constraint judgment method for horizontal wells is constructed. S3 uses the population in the Latin hypercube sampling initialization algorithm, performs constraint processing using the constraint judgment method of horizontal wells, calculates the NPV in the mathematical model for production optimization of water-drive reservoirs, and constructs a database. S4 uses the best individuals in the database to form a temporary population, and builds a Gaussian model based on the temporary population; S5 uses the differential evolution algorithm to obtain three subpopulations with different strategies, and uses the horizontal well constraint judgment method for constraint processing. S6 uses a pseudo-update strategy based on a temporary population to obtain potential individuals from the subpopulation; S7 uses Euclidean distance to select the individual closest to the current best individual for parallel numerical simulation; S8 Update the database. If the iteration stopping condition is met, output the optimal individual; otherwise, return to step S4. Step S1 specifically includes: Based on reservoir development needs and with the objective of maximizing the net present value (NPV), a mathematical control model for optimizing waterflood reservoir production is constructed, as shown below: In the formula, For control variables in production optimization problems; This represents the objective function value, i.e., the economic benefit NPV; This represents the total time step of the numerical simulation; For the first n The number of iterations simulates the span of the step size; d It refers to the number of control variables; This represents the annual decay rate. , and Representing the first i The first production well n The oil production rate, water production rate, and the first time step in the time step i The injection well in the first n The water injection rate in each time step; , , These are the economic benefits per unit volume of crude oil, the cost required to treat per unit volume of wastewater, and the cost of injecting per unit volume of water, respectively. NP and NI These are the total number of production wells and water injection wells in the oil reservoir; Step S2 specifically includes: S21 extracts the horizontal segment trajectory characteristics of each horizontal well in the reservoir numerical model, approximating the horizontal segment trajectory of the horizontal well as a line segment on the plane. For any horizontal well... Extract the grid coordinates of the two endpoints of the horizontal segment in the reservoir numerical model. and As its horizontal segment trajectory identifier ; S22 Construct a set of horizontal segment trajectory identifiers. ,in, To determine the number of new horizontal wells to be added for optimal well location, This represents the number of existing horizontal wells in the reservoir model. S23 Targeting the set of horizontal segment trajectory identifiers Any two newly added horizontal wells or any newly added horizontal well and an existing horizontal well and Use its corresponding horizontal segment trajectory identifier and Pre-processing before constraint; for horizontal wells and Using the results of the preprocessing, two diagonal wells are formed. and A rectangle formed by connecting the two endpoints of the horizontal segment; S24 for two horizontal wells and The overlapping of the rectangles is checked to determine whether the constraints are met. Step S3 specifically includes: S31 is generated using the Latin hypercube sampling method. Horizontal well location plan ; S32 Perform horizontal well location constraint judgment, the judgment method is the same as step S2; S33 If there are individuals that do not meet the constraints, then Latin hypercube sampling is performed again until all individuals meet the constraints. S34 uses a numerical simulator to perform parallel computation on all individuals to obtain the NPV response value in the mathematical control model for production optimization of water-drive reservoirs. , using S and y Construct the initial database DB.

2. The efficient parallel optimization method according to claim 1, characterized in that, Step S4 specifically includes: S41 Select the database DB with the highest NPV response value. Individuals construct temporary populations ; S42 utilize Constructing a Gaussian proxy model .

3. The efficient parallel optimization method according to claim 2, characterized in that, Step S5 specifically includes: S51 generates three subpopulations based on three different strategies using the differential evolution algorithm. ; S52 pairs of three subpopulations Perform constraint processing, using the same method as steps S31 to S33.

4. The efficient parallel optimization method according to claim 3, characterized in that, Step S6 specifically includes: S61 uses the Gaussian model Calculate the target response value in each subpopulation and calculate the fitness value based on the confidence lower bound (LCB) criterion. S62 For each subpopulation, select the individual with the best response value. Deposit In, and update it to In the middle, but not updated. ; S63 when The individuals stored in it have all reached If one is found, proceed to the next step; otherwise, utilize the updated [system / method / etc.]. Return to step S5.

5. The efficient parallel optimization method according to claim 4, characterized in that, Step S7 specifically includes: S71 Calculation From the middle individual to the current best individual Given the Euclidean distance, select the m individuals with the smallest distance. Parallel numerical simulations were performed to obtain the corresponding NPV response values. ; S72 Iteration count increased m .

6. A highly efficient parallel optimization device considering the practical engineering constraints of horizontal wells, characterized in that, include: The first processing unit is used to construct a mathematical control model for optimizing water-drive reservoir production; the first processing unit specifically includes: Based on reservoir development needs and with the objective of maximizing the net present value (NPV), a mathematical control model for optimizing waterflood reservoir production is constructed, as shown below: In the formula, For control variables in production optimization problems; This represents the objective function value, i.e., the economic benefit NPV; This represents the total time step of the numerical simulation; For the first n The number of iterations simulates the span of the step size; d It refers to the number of control variables; This represents the annual decay rate. , and Representing the first i The first production well n The oil production rate, water production rate, and the first time step in the time step i The injection well in the first n The water injection rate in each time step; , , These are the economic benefits per unit volume of crude oil, the cost required to treat per unit volume of wastewater, and the cost of injecting per unit volume of water, respectively. NP and NI These are the total number of production wells and water injection wells in the oil reservoir; The second processing unit is used to construct a constraint judgment method for horizontal wells based on the engineering application conditions of horizontal wells and the reservoir numerical model; the second processing unit specifically includes: S21 extracts the horizontal segment trajectory characteristics of each horizontal well in the reservoir numerical model, approximating the horizontal segment trajectory of the horizontal well as a line segment on the plane. For any horizontal well... Extract the grid coordinates of the two endpoints of the horizontal segment in the reservoir numerical model. and As its horizontal segment trajectory identifier ; S22 Construct a set of horizontal segment trajectory identifiers. ,in, To determine the number of new horizontal wells to be added for optimal well location, This represents the number of existing horizontal wells in the reservoir model. S23 Targeting the set of horizontal segment trajectory identifiers Any two newly added horizontal wells or any newly added horizontal well and an existing horizontal well and Use its corresponding horizontal segment trajectory identifier and Pre-processing before constraint; for horizontal wells and Using the results of the preprocessing, two diagonal wells are formed. and A rectangle formed by connecting the two endpoints of the horizontal segment; S24 for two horizontal wells and The overlapping of the rectangles is checked to determine whether the constraints are met. The third processing unit is used to perform constraint processing based on the population in the Latin hypercube sampling initialization algorithm, using the constraint judgment method of horizontal wells, and to calculate the NPV in the mathematical model for water-drive reservoir production optimization, and to construct a database; the third processing unit specifically includes: S31 is generated using the Latin hypercube sampling method. Horizontal well location plan ; S32 Perform horizontal well location constraint judgment, the judgment method is the same as step S2; S33 If there are individuals that do not meet the constraints, then Latin hypercube sampling is performed again until all individuals meet the constraints. S34 uses a numerical simulator to perform parallel computation on all individuals to obtain the NPV response value in the mathematical control model for production optimization of water-drive reservoirs. , using S and y Construct the initial database DB; The fourth processing unit is used to form a temporary population using the best individuals in the database, and to build a Gaussian model based on the temporary population. The fifth processing unit is used to obtain three subpopulations using different strategies based on the differential evolution algorithm, and to perform constraint processing using the constraint judgment method of horizontal wells; The sixth processing unit is used to obtain potential individuals from the subpopulation based on a pseudo-update strategy of the temporary population. The seventh processing unit is used to select the individual closest to the current best individual based on Euclidean distance and perform parallel numerical simulations. The eighth processing unit is used to update the database. When the iteration stopping condition is met, it outputs the optimal individual; otherwise, it returns to step S4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the efficient parallel optimization method for considering the actual engineering constraints of horizontal wells as described in any one of claims 1-5.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the efficient parallel optimization method for considering the actual engineering constraints of horizontal wells as described in any one of claims 1-5.

Images