Wellhead valve, wellhead valve parameter optimization method based on intelligent algorithm and system thereof

Abstract

The present application provides a wellhead valve and a wellhead valve parameter optimization method and system based on an intelligent algorithm, and relates to the fields of wellhead valves and valve parameter optimization. A pressure relief hole is added to the back of the valve plate to reduce the flow area of the wellhead valve at each opening degree, increase the throttling effect, make the flow field temperature in the wellhead valve higher during the entire CO2 release process, and reduce the risk of valve freezing. The parameter optimization method uses Latin hypercube sampling to determine the corresponding structure parameters of each sample point, constructs a radial basis function proxy model and performs sensitivity analysis to identify the influence factors of each parameter on the optimization target, and finally uses a genetic algorithm to optimize and obtain the final optimization result, establish the valve structure corresponding to the final optimization structure, and better prevent the valve from freezing. The present application is suitable for parameter optimization of pressure relief valves during dense-phase CO2 release.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wellhead valves and valve parameter optimization, and particularly relates to a wellhead valve optimization method based on an intelligent algorithm. BACKGROUND

[0002] For large-scale carbon dioxide transportation, high-temperature and normal-pressure liquid-phase carbon dioxide pipeline transportation, also known as "dense-phase" transportation, is adopted. However, compared with hydrocarbon compound pipeline transportation such as natural gas, the industry experience of carbon dioxide pipeline transportation is much less, and the related risks should be fully understood and effectively managed.

[0003] During the transportation of carbon dioxide, there is a control requirement for the single phase state. Near the critical region, a small difference in pressure and temperature will cause the state characteristics of carbon dioxide to appear sharp floating. Therefore, the pressure of the transportation process needs to be kept higher than the critical pressure to avoid two-phase transportation. The current operating pressure of the transportation process is 8.5-12 MPa, which is located in the "smooth" region of the carbon dioxide phase diagram, that is, the state characteristics of carbon dioxide in this region are stable and will not change abruptly. However, during the pressure relief process, carbon dioxide will inevitably undergo sharp changes in characteristics. Due to unexpected failures (third-party construction damage or corrosion) or planned maintenance, the pipeline often needs to be pressure relieved. During the pressure relief process, the escaping gas will cool the pipeline. If the temperature decreases too much, below the ductile-brittle transition temperature, the pipe material will become brittle, causing brittle fracture and serious pipe damage. The formation of dry ice during the pressure relief process of carbon dioxide also poses a certain risk of freezing of the pipeline or valve. Therefore, the relief of the carbon dioxide pipeline needs to be controlled to ensure that the pipeline temperature is not lower than the design temperature.

[0004] Therefore, in order to prevent the wellhead valve from freezing, the structure of the wellhead valve needs to be optimized. SUMMARY

[0005] The present application provides a wellhead valve, by increasing a pressure relief hole on the back of the valve plate, thereby reducing the flow area of the wellhead valve at each opening degree, increasing the throttling effect, so that the temperature of the flow field in the wellhead valve is at a higher level during the entire CO2 relief process, to reduce the risk of valve freezing. At the same time, a wellhead valve parameter optimization method based on an intelligent algorithm is also proposed to optimize the size of the pressure relief hole and better prevent the risk of valve freezing.

[0006] To achieve the above object, the present application provides the following technical scheme:

[0007] The present application provides a wellhead valve, the valve comprising a valve plate, the back of the valve plate being provided with a pressure relief hole.

[0008] Further, there is a preferred embodiment that the diameter a of the pressure relief hole ranges from 55mm to 65mm.

[0009] The position b of the pressure relief hole ranges from 40mm to 60mm.

[0010] The width c of the pressure relief hole ranges from 14mm to 22mm.

[0011] The application also provides a wellhead valve parameter optimization method based on an intelligent algorithm, which is used for optimizing the parameters of the pressure relief hole.

[0012] Step S1: According to the diameter a, position b and width c of the pressure relief hole, sample each sample point by Latin hypercube.

[0013] Step S2: Construct a radial basis function proxy model according to each sample point.

[0014] Step S3: Sample the diameter a, position b and width c of the pressure relief hole by using a global sensitivity analysis method based on Monte Carlo, obtain different input combinations, calculate the output results of the radial basis function proxy model under different input combinations, and obtain the contribution degree of each input parameter to the output results.

[0015] Step S4: According to the contribution degree of each input parameter to the output results, optimize the diameter a, position b and width c of the pressure relief hole by using a genetic algorithm, and obtain the optimal parameters of the diameter a, position b and width c of the pressure relief hole.

[0016] Further, each sample point corresponds to the diameter a, position b and width c of the pressure relief hole.

[0017] Further, the output of the global sensitivity analysis method based on Monte Carlo is the temperature of the wellhead valve.

[0018] Further, by statistically analyzing the variance of the output results, the total variance of the temperature change is decomposed into the contribution of different input parameters, and the contribution degree of each input parameter to the output results is obtained.

[0019] Further, there is a preferred embodiment that the optimal diameter a of the pressure relief hole is 55.10mm, the position b is 55.72mm, and the width c is 18.89mm.

[0020] The wellhead valve parameter optimization method based on the intelligent algorithm can be realized by using computer software, and therefore, the application also provides a wellhead valve parameter optimization system based on the intelligent algorithm, which comprises a storage device for executing the wellhead valve parameter optimization method based on the intelligent algorithm.

[0021] The application further provides a computer readable storage medium, and a computer program is stored on the computer readable storage medium, and the computer program is executed by a processor to perform the intelligent algorithm-based wellhead valve parameter optimization method in any of the above aspects.

[0022] The application further provides a computer device, which comprises a memory and a processor, and the memory stores a computer program, and when the processor executes the computer program stored in the memory, the processor performs the intelligent algorithm-based wellhead valve parameter optimization method in any of the above aspects.

[0023] The application has the following beneficial effects:

[0024] 1. The application provides a wellhead valve, which increases a pressure relief hole on the back of a valve plate, thereby reducing the flow area of the wellhead valve at each opening degree, increasing the throttling effect, making the flow field temperature in the wellhead valve at a higher level during the entire CO2 release process, and reducing the risk of valve freezing.

[0025] 2. The application provides an intelligent algorithm-based wellhead valve parameter optimization method, which scientifically samples by using Latin hypercube, determines the corresponding structure parameters of each sample point and the corresponding results under each working condition, identifies the influence factors of each parameter on the optimization target by constructing a radial basis function surrogate model and performing sensitivity analysis, and finally obtains the final optimization result by genetic algorithm optimization, and establishes a valve structure corresponding to the final optimized structure, thereby better preventing the valve from freezing.

[0026] The application is suitable for parameter optimization of a pressure relief valve in a dense-phase CO2 release process. BRIEF DESCRIPTION OF DRAWINGS

[0027] In order to more clearly illustrate the specific embodiments of the application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the specific embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0028] Figure 1 is a structure diagram of the wellhead valve and the pressure relief hole according to the embodiments of the application;

[0029] Figure 2 is a 1 / 23D flow area diagram of the wellhead gate valve according to the embodiments of the application;

[0030] Figure 3 is a grid division schematic diagram of the wellhead gate valve according to the embodiments of the application;

[0031] Figure 4is a CO2 saturation temperature vs. saturation pressure corresponding graph according to an embodiment of the present invention;

[0032] Figure 5 is a Latin hypercube sampling distribution graph according to an embodiment of the present invention, wherein (a) is a width sampling distribution graph, (b) is a diameter sampling distribution graph, and (c) is a diameter sampling distribution graph;

[0033] Figure 6 is a pressure cloud graph of the wellhead gate valve pipeline at 25% opening according to an embodiment of the present invention;

[0034] Figure 7 is a velocity cloud graph of the wellhead gate valve pipeline at 25% opening according to an embodiment of the present invention;

[0035] Figure 8 is a liquid CO2 volume fraction cloud graph of the wellhead gate valve pipeline at 25% opening according to an embodiment of the present invention;

[0036] Figure 9 is a gaseous CO2 volume fraction cloud graph of the wellhead gate valve pipeline at 25% opening according to an embodiment of the present invention;

[0037] Figure 10 is a temperature cloud graph of the wellhead gate valve pipeline at 25% opening according to an embodiment of the present invention;

[0038] Figure 11 is a valve risk area of frozen plugging according to an embodiment of the present invention;

[0039] Figure 12 is a genetic algorithm flow chart according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The specific embodiments of the present application will be further described in conjunction with the accompanying drawings. The following embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any form. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present application, and these changes and improvements are within the scope of protection of the present application.

[0041] Embodiment one, see Figures 1 to 4 The present embodiment is described, and the present embodiment provides a wellhead valve. By increasing a pressure relief hole on the back of the valve plate, the flow area of the wellhead valve at each opening is reduced, the throttling effect is increased, the temperature of the flow field in the wellhead valve is at a higher level during the entire CO2 release process, and the risk of valve frozen plugging is reduced.

[0042] The specific structure of the pressure relief hole is as follows Figure 1wherein a is the diameter of the pressure relief hole, the original size is 65mm, the value range is 55mm-65mm; b is the relative position of the pressure relief hole relative to the bottom of the valve plate, the original parameter is 50mm, the value range is 40-60mm; c is the width of the pressure relief hole, the original size is 18mm, the value range is 14-18mm, the specific case is shown in Table 1.

[0043] Table 1

[0044] Design parameters Original size / mm Lower boundary / mm Upper boundary / mm Relief hole diameter a 65 55 65 Relief hole position b 50 40 60 Relief hole width c 18 14 22

[0045] Temperature is an important parameter for measuring the risk of wellhead valve freezing, in order to ensure the stability of the data and the convergence of the optimization, the area of the three-phase point region on the symmetric surface of the valve is selected as the optimization target S, that is

[0046]

[0047] In actual application, the present embodiment constructs a CO2 pressure relief process wellhead valve freezing risk analysis method, and optimizes the structure of the valve through the analysis method;

[0048] The method is:

[0049] Firstly, a 1 / 23-D wellhead gate valve model is built, and the flow basin is extracted, the grid is divided, and the solver is set; then the 1 / 23-D wellhead gate valve model after setting is subjected to CO2 property setting, and the model is verified to obtain a CO2 freezing risk numerical simulation model; finally, the CO2 freezing risk numerical simulation model is subjected to CFD simulation calculation for two kinds of opening degrees, and the valve freezing risk is analyzed according to the calculation results, so as to optimize the structure of the valve, that is, the above-mentioned pressure relief hole is added on the back of the valve plate.

[0050] Further, the above-mentioned 1 / 23-D wellhead gate valve model is built, and the flow basin is extracted, the grid is divided, and the solver is set, which is specifically:

[0051] The wellhead valve is a plate gate valve for releasing CO2, mainly including a valve stem, a valve cover, a valve plate, a valve seat, a valve body and the like. During the CO2 release process, by rotating the handwheel of the gate valve, the valve stem and the valve plate are gradually opened under the push of the lead screw, and the whole opening process lasts about 60 seconds, and then the valve reaches the maximum opening degree and performs a stable release process.

[0052] Based on the above-mentioned model of the wellhead gate valve, the flow basin is extracted in the model processing software. The structure of the wellhead gate valve is a plane symmetric structure, therefore, the simulation under different opening degrees of the wellhead gate valve is simulated by using a 1 / 23-D model, and the calculation flow basin is extracted and simplified as shown in 2, considering the actual working environment of the valve. Figure 2As shown, the most important throttling area of the wellhead gate valve is near the gate plate. During the opening process of the valve, the throttling area of the gate plate region changes from small to large. At this time, the liquid CO2 flows through the valve under the drive of high pressure, and the temperature, pressure and flow rate change dramatically, and phase change occurs. Therefore, the embodiments analyze the possible freezing problems that may occur during the opening and working process of the valve by calculating the detailed changes of the flow field of the valve at different openings.

[0053] Further, in order to ensure the simulation accuracy and convergence, a flow field of 5 times and 10 times the inlet diameter of the valve is needed to be added at the inlet and outlet of the valve to make the flow field before and after the valve fully developed, wherein the outlet and inlet boundary conditions of the flow field are specified according to the specific working condition.

[0054] After the calculation of the flow field extraction is completed, the embodiments divide the calculation grid according to the structural characteristics of the wellhead gate valve. The simplified three-dimensional wellhead gate valve cavity model structure is simple, but the flow characteristics, phase change heat and mass transfer characteristics in the valve cavity are quite complex. Therefore, a mixed grid division method is adopted to generate high-quality grids.

[0055] Further, considering the complex flow and heat and mass transfer phenomena inside the wellhead gate valve, the grid needs to be processed regionally during the grid division. In the area near the pressure relief slit of the wellhead gate valve, the flow condition is relatively complex, and local encryption processing of the grid is needed, while in the area far away from the pressure relief slit (which can be considered as most of the area in the wellhead gate valve cavity), the flow of the fluid is relatively smooth, so relatively sparse grids can be divided in this area. Such regional grid division can not only ensure the accuracy of the simulation results of the pipe flow field, but also appropriately reduce the number of grids, save computing resources and improve the simulation calculation speed.

[0056] Further, for CFD analysis, grid density is one of the key factors to ensure the simulation calculation accuracy and convergence. In order to obtain the optimal calculation grid model, the embodiments simulate the wellhead gate valve by using grids of different density levels, and explore the influence of grid density on simulation accuracy by comparing the different temperatures at the outlet of the valve plate after simulation of different grids. When the number of grids is greater than 2,673,022, the simulation result is independent of the grid density, that is, the calculation result converges to the calculation grid. Considering the calculation efficiency and cost, the embodiments select the grid generation scheme with the number of grids of 2,673,022 for subsequent simulation calculation, and the final grid scheme is as shown in Figure 3 .

[0057] After the grid is divided, the solver parameters need to be set in the CFD simulation calculation, including: solver parameters (Pressure-based, Transient Absolute velocity formulation 3-D symmetric); working medium (gas-liquid two-phase CO2); solving method (Coupled); spatial discrete format (second-order upwind); convergence standard (all convergence residuals are less than 1e-4); turbulence model (Realizable k-ε); time step (3e-6s).

[0058] In addition to the basic solver parameter setting, the two-phase flow parameters of CO2 need to be set before calculation, including: liquid phase CO2 parameters (density, viscosity, thermal conductivity); gas phase CO2 parameters (density, viscosity, thermal conductivity, standard state enthalpy, reference temperature).

[0059] The above 1 / 2 3-D wellhead gate valve model is set with CO2 physical properties, and the model is verified, and the CO2 freeze-out risk numerical simulation model is obtained as follows:

[0060] The working pressure of the dense phase carbon dioxide pipeline is maintained at tens of megapascals at room temperature, and the pressure drops sharply after being discharged through the wellhead gate valve, and the corresponding saturation temperature will decrease rapidly, so the dense phase carbon dioxide will change from liquid phase to gas phase (similar to flash evaporation phenomenon). The dense phase carbon dioxide physical property parameters will fluctuate in a large range during the phase change, and correct carbon dioxide physical property parameters can ensure the accuracy of the simulation results.

[0061] The phase diagram and pressure-temperature relationship of carbon dioxide are shown in Figure 4 , the three-phase critical point is obtained by Refprop, T lim＝ 30.92℃, P lim = 7.37 MPa. The inlet pressure of the wellhead gate valve is 30 MPa, the temperature is 300 K, and the outlet back pressure is 0.1 MPa. According to the inlet and outlet pressure of the wellhead gate valve, the carbon dioxide physical properties are divided into three sections: 0.1-0.6 MPa, 0.6-7.37 MPa, and 7.37-30 MPa.

[0062] Among them, in the 7.37-30 MPa section, that is, from the inlet pressure of the wellhead gate valve 30 MPa to the critical pressure of carbon dioxide 7.37 MPa. In this pressure drop range, carbon dioxide does not change phase, and the saturation temperature under different pressures is set to a constant value of 35℃ (the temperature in the calculation domain at the initial moment of simulation is 25℃).

[0063] In the 0.6-7.37 MPa section, the saturation temperature corresponding to the pressure will drop sharply in the process of the decrease of the carbon dioxide pressure, and thus the phase change process of the carbon dioxide will occur in the section. The Refprop software is used to search for the saturation carbon dioxide property parameters under the pressure in the section, and the results are shown in Table 2.

[0064] Table 2

[0065]

[0066]

[0067]

[0068] The discrete point data shown in Table 2 is subjected to polynomial fitting to obtain the relationship between the saturation temperature and the corresponding saturation pressure,

[0069] T = 0.241P 3 - 4.2P 2 + 31.46P - 68.15

[0070] In the formula, T is the saturation temperature of the carbon dioxide in ℃, and P is the saturation pressure of the carbon dioxide in MPa. The correlation coefficient R of the fitting formula is 0.9991, and the fitting accuracy is high. 2

[0071] In the 0.1-0.6 MPa pressure section, the saturation temperature of the carbon dioxide in the 0.1-0.6 MPa section is no longer subjected to the fitting formula in the 0.6-7.37 MPa section, and is subjected to re-fitting according to the triple point and the 0.1 MPa point data. The relationship formula after the fitting is reloaded into the evaporation and condensation model of the FLUENT.

[0072] The fitting formula is as follows:

[0073] T = -87.42P 2 + 124.96P + 176.53

[0074] In the formula, T is the saturation temperature of the carbon dioxide in ℃, and P is the saturation pressure of the carbon dioxide in MPa. The correlation coefficient R of the fitting formula is 0.9995. 2

[0075] Embodiment two, the embodiment provides a wellhead valve parameter optimization method based on an intelligent algorithm, the optimization method is used for optimizing the pressure reduction hole parameters in the above-mentioned embodiment one, so that the valve freezing risk is better prevented, and the optimization method comprises the following steps:

[0076] Step S1: according to the diameter a, the position b and the width c of the pressure reduction hole, sample is obtained through Latin hypercube sampling; ​​

[0077] Step S2: constructing a radial basis function surrogate model according to each sample point;

[0078] Step S3: using a global sensitivity analysis method based on Monte Carlo to sample the diameter a, position b and width c of the pressure relief hole, obtain different input combinations, calculate the output results of the radial basis function surrogate model under different input combinations, and obtain the contribution degree of each input parameter to the output result;

[0079] Step S4: according to the contribution degree of each input parameter to the output result, using a genetic algorithm to optimize the diameter a, position b and width c of the pressure relief hole, and obtaining the optimal parameters of the diameter a, position b and width c of the pressure relief hole.

[0080] Embodiment three, see Figure 5 This embodiment is a specific description of the wellhead valve optimization method based on intelligent algorithm described in Embodiment Two;

[0081] Step S1: according to the diameter a, position b and width c of the pressure relief hole, sampling by Latin hypercube, and obtaining each sample point;

[0082] Specifically,

[0083] The Latin hypercube design method is selected to sample the diameter a, position b and width c of the pressure relief hole. The Latin hypercube design method can reasonably control the distribution of training samples in the design space, prevent repeated sampling in the local space, and maximize the space filling and uniformity of the training sample distribution.

[0084] The principle of the Latin hypercube design method is as follows: according to the design dimension, the design space is divided into a plurality of small subintervals, and then a plurality of sampling subintervals are determined by random arrangement, and finally a plurality of sample points are randomly arranged in each sampling subinterval.

[0085] (1) Let N sample points be generated, and the design space is normalized into a unit hypercube Ω = {X e R N |0≤x i ≤1}.

[0086] (2) The ith dimension of the unit hypercube is divided into n subintervals (0, 1 / N), (1 / N, 2 / N), …, (1-1 / N, 1). The jth subinterval of the ith dimension is denoted as

[0087] (3) Randomly take d different (1, 2, …, N) permutations to form a random permutation matrix (ω ij ) d×N .

[0088] (4) generating a sampling sub-interval matrix:

[0089]

[0090] (5) generating a point in each sub-interval in each sampling sub-interval matrix according to uniform distribution, obtaining a sample matrix (x 1 ), x 2 ), …, x N ).

[0091] According to the above determined optimization structure and corresponding optimization parameters, 12 groups of sample points are extracted by Latin hypercube sampling, and the sample results are shown in Table 3, and the sample distribution is shown in Figure 5 .

[0092] Table 3

[0093]

[0094]

[0095] By establishing the above sample model, simulation calculation is carried out under two typical openings respectively, node data is derived, S corresponding to each working condition is calculated, optimization data is obtained, and the data basis for constructing the proxy model is obtained.

[0096] Step S2: constructing a radial basis function proxy model according to each sample point;

[0097] Specifically:

[0098] According to the above extracted sample points, a radial basis function proxy model is constructed, and the accuracy thereof is verified,

[0099] Among them, the radial basis function proxy model constructed by the embodiment is a kind of radial symmetric function, which is an interpolation method, and has the advantages of simple form, strong adaptability and high precision.

[0100] Suppose that a function y = f(x) is an n-dimensional real-valued function, m sample points are selected by using a test design method, and the set composed of these sample points is represented as: X = {x1, x2, …, xm}. m} T The response value set y = {f(x1), f(x2), …, f(x m )} T , the form of fitting function y by using radial basis function is as follows:

[0101]

[0102] In the formula, λ ithe undetermined coefficient before the i-th basis function, ||x-x i || represents the Euclidean norm between the prediction point x and the sample point x i .

[0103] For an n-dimensional design space φ represents the basis function form of the radial basis, and the commonly used basis function forms are shown in Table 4, where r = ||x-x i ||.

[0104] Table 4

[0105]

[0106] The above formula can be obtained by bringing m sample points and response values into the formula:

[0107]

[0108] The above formula has m equations and n unknowns. In order to facilitate expression, the above formula can be written in matrix form:

[0109] Aλ = y

[0110] where, λ = {λ1, λ2, …, λ m} T , y = {f(x1), f(x2), …, f(x m )} T If a set of sample points and response values are given, the undetermined coefficient λ can be obtained according to the least square method.

[0111] It can be seen that the radial basis function proxy model describes the complex implicit function relationship between the structure response and the structure parameters by linear combination of the basis functions. With the increase of the number of parameters to be corrected, the number of sample points required to solve the undetermined coefficient has a linear relationship with the number of parameters to be corrected, so the radial basis function proxy model has the advantage of saving the calculation cost when fitting unknown problems.

[0112] The accuracy of the radial basis function proxy model is verified as follows:

[0113] The radial basis function proxy model calculates the predicted value of the structure response by inputting the variable (generally using additional sample points), and then compares it with the true structure output response corresponding to the sample points. The error between them is used as the standard to evaluate the fitting accuracy of the radial basis function proxy model. The current standards for testing the accuracy of the proxy model mainly include: the coefficient of determination R 2Test, Root Mean Square Error (RMSE), Maximum Absolute Error (MAE), normality test of residual, mean of residual, etc.

[0114] (1) Coefficient of determination R 2 Test

[0115] Coefficient of determination R 2 Test is an evaluation criterion for evaluating the overall fitting performance of the radial basis function surrogate model in the design space. The principle is to evaluate the error between the predicted value and the true value of the radial basis function surrogate model at the test points. R 2 The value range of R i is from 0 to 1, and the closer the value is to 1, the higher the fitting accuracy of the radial basis function surrogate model. Let represent the calculated value of the i-th test sample point, y 2 represent the true value of the i-th test sample point, and R t is expressed as follows:

[0116]

[0117] In the formula, n 2 represents the number of test sample points, is the average value of the true values of all test sample points.

[0118] (2) Root Mean Square Error (RMSE)

[0119] Root Mean Square Error is the same as coefficient of determination R 2 , which is also used to evaluate the fitting performance of the radial basis function surrogate model in the entire design space. The value is a number greater than 0, and the closer to 0, the better the fitting performance of the radial basis function surrogate model. Its expression is:

[0120]

[0121] However, the amplitude of RMSE varies greatly for different test problems. In order to intuitively compare and analyze the RMSE values of multiple different test functions, the RMSE results are standardized, that is, the normalized root mean square error (NRMSE):

[0122]

[0123] According to the above formula and combined with sample data, after constructing the radial basis function surrogate model, the accuracy of the surrogate model is verified by R 2The R² value is 0.9857, the RMSE is 0.0012, and the NRMSE is 0.0392. These results indicate that the R² value of the radial basis function surrogate model is... 2 The three metrics, RMSE, and NRMSE, are all optimal, therefore, the constructed radial basis function surrogate model has higher accuracy.

[0124] Step S3: Using the Monte Carlo-based global sensitivity analysis method, the diameter a, position b, and width c of the pressure relief hole are sampled to obtain different input combinations. The output results of the radial basis function surrogate model under different input combinations are calculated to obtain the contribution of each input parameter to the output results.

[0125] In Monte Carlo-based global sensitivity analysis, it is essential to first understand the fundamental principles of the method and its application to the optimization problem of wellhead valve temperature. The Monte Carlo method is a numerical computation method based on random sampling, simulating and analyzing the behavior of complex systems through extensive random sampling. In global sensitivity analysis, the Monte Carlo method samples a large number of input parameters (such as position, width, and diameter) and calculates the output of the surrogate model under different input combinations, thereby evaluating the contribution of each input parameter to the output. Specifically, for the wellhead valve temperature objective, the surrogate model serves as an approximate expression of the system, taking the wellhead valve temperature as the output. By simulating changes in the input parameters (position, width, and diameter), the temperature response under different conditions is calculated. In this process, the Monte Carlo method can not only handle the single effect of parameters but also capture the complex interaction effects that may exist between input parameters, thus providing comprehensive sensitivity information for optimization design.

[0126] By employing Monte Carlo global sensitivity analysis, we can quantify the impact of three parameters—location, width, and diameter—on wellhead valve temperature. In implementation, the input space is first randomly sampled using a surrogate model, generating numerous input combinations, and the temperature output corresponding to each combination is calculated. Then, by statistically analyzing the variance of these outputs, the total variance of temperature change is decomposed into the contributions of different input parameters. Typically, the results of sensitivity analysis are presented as the proportion of variance contribution of each input parameter, revealing which parameters dominate the temperature change. For example, if the analysis shows that location contributes the most to temperature change, it indicates that location is the most critical factor affecting wellhead valve temperature, and location adjustment should be prioritized during optimization. If there is a significant interaction between width and diameter, it may be found that they jointly affect temperature change, and optimizing only one parameter may not achieve optimal temperature control. Through this global analysis, engineers can obtain more comprehensive and accurate sensitivity information, providing a basis for subsequent optimization processes.

[0127] The Monte Carlo method proposed in this embodiment has strong adaptability and can handle the nonlinearity and complex interaction effects of various input parameters. This feature is particularly important in the wellhead valve temperature optimization problem, because the change in wellhead valve temperature can be influenced by multiple parameters and their interactions. Through global sensitivity analysis, engineers and other staff can delve into these interactions and ensure that the impact of each parameter is fully considered during optimization. In multi-objective optimization, sensitivity analysis can help balance the contradictions and conflicts between different objectives, allocate computing resources reasonably, and avoid excessive optimization in a local area while ignoring global performance. In summary, the Monte Carlo method based on global sensitivity analysis not only provides an intuitive ranking of parameter importance, but also helps identify potential interactions between parameters, thereby providing strong support for further optimization design and decision-making.

[0128] Table 5 below shows the results of parameter sensitivity verification, where X1 is the diameter of the pressure relief hole, X2 is the width of the pressure relief hole, and X3 is the longitudinal position of the pressure relief hole. Through sensitivity analysis of the diameter ((x_1)), width ((x_2)), and position ((x_3)) of the pressure relief hole, the results show that the position ((x_3)) has the most significant impact on the wellhead valve temperature, with the highest sensitivity index. This indicates that the position is a key parameter affecting temperature changes, and optimizing this parameter can significantly improve the performance of the wellhead valve. Therefore, in further optimization, priority should be given to adjusting the diameter value to achieve better temperature control.

[0129] Table 5

[0130] X1 X2 X3 0.2786 0.2901 0.4312

[0131] Step S4: According to the contribution of each input parameter to the output result, a genetic algorithm is used to optimize the diameter a, position b, and width c of the pressure relief hole, obtaining the optimal parameters of the diameter a, position b, and width c of the pressure relief hole.

[0132] When solving the wellhead valve temperature optimization problem using a genetic algorithm (GA), the focus is on the three control parameters: position, width, and diameter. Genetic algorithms are optimization methods that simulate natural selection and genetic mechanisms, gradually approaching optimal solutions through the evolutionary process of populations. In this process, position, width, and diameter are considered important variables in the genetic chromosome, and each individual represents a potential solution in the solution space. These individuals are gradually producing more advantageous offspring through operations such as crossover, mutation, and selection. Specifically, the genetic algorithm first initializes an initial population consisting of multiple individuals (i.e., parameter combinations), and the fitness of each individual is evaluated by a proxy model that can predict the wellhead valve temperature under given parameters. The establishment of the proxy model relies on historical data or experimental data, and the relationship between temperature and control parameters is fitted through machine learning methods (such as neural networks or regression analysis), providing temperature prediction information for the genetic algorithm.

[0133] As the genetic algorithm evolves, individuals in the population gradually exhibit better fitness, i.e., lower valve temperatures under specific parameter settings. Through crossover operations, genetic algorithms can combine the genes of excellent individuals into new individuals, further improving the solution quality in the search space; mutation operations help avoid local optimal solutions and explore new possible solutions through certain randomness. In each generation, individuals are selected through a selection mechanism to ensure that the optimal solution is constantly propagated in the population. The advantage of genetic algorithms is their global search capability, which can effectively avoid local optimal solutions and handle complex, nonlinear optimization problems. Therefore, in the optimization of wellhead valve temperature, genetic algorithms can search for an optimal solution that minimizes or optimizes the wellhead valve temperature by adjusting the three parameters of position, width, and diameter. The final optimization results are shown in Table 6.

[0134] Table 6

[0135] X1 X2 X3 55.10 18.89 55.72

[0136] As can be seen from Table 5, the optimal pressure relief hole diameter a is 55.10 mm, the position b is 55.72 mm, and the width c is 18.89 mm.

[0137] Embodiment Four, see Figures 6 to 11 This embodiment is a verification explanation of the results obtained by optimizing the diameter a, position b, and width c of the pressure relief hole using a genetic algorithm as described in Embodiment Three.

[0138] The genetic algorithm optimization result corresponds to a pressure relief hole diameter of 55.10 mm, a width of 18.89 mm, and a relative position of 55.72 mm from the bottom of the valve plate. According to the simulation calculation results of the above structure and calculation method, Figures 3 to 8shown.

[0139] wherein, Figure 6 The pressure cloud chart of the wellhead gate valve pipeline at 25% opening degree can be seen that the entire CO2 pressure level is at the highest position at the inlet of the valve, and the pressure cloud chart changes greatly when CO2 flows through the valve plate area, forming a large pressure gradient, which is due to the fact that the valve has not been fully opened, and the small flow area of the valve plate forms throttling; similarly, when CO2 flows out of the valve plate area, a more obvious pressure gradient is also formed, and CO2 forms secondary pressure reduction here, and finally the pressure of CO2 gradually approaches atmospheric pressure.

[0140] Figure 7 The velocity cloud chart of the wellhead gate valve pipeline at 25% opening degree can be seen from the Bernoulli principle that the pressure of the fluid will decrease and the flow rate will increase when the fluid flows through the slit, so from the velocity cloud chart, it can be seen that the flow rate of the fluid increases significantly when CO2 flows through the valve plate area, which is consistent with the change of the pressure cloud chart. At this time, the velocity reaches nearly 400 m / s, which is a supersonic state. Due to the compressibility of CO2 gas, in this supersonic state, a strong compression wave similar to a Mach disk will be formed in the flow field after passing through the valve plate area, and a velocity gradient change will be formed before and after the compression wave, which is specifically manifested as a slight decrease in velocity before the compression wave and an increase in velocity to about three times the speed after the compression wave.

[0141] Figure 8 and Figure 9 The volume fraction cloud charts of liquid and gaseous CO2 in the wellhead gate valve pipeline at 25% opening degree are shown in FIGS. 12 and 13, respectively. It can be seen that the volume fraction of liquid CO2 in the pipeline at the inlet is large, about 20%, and the gas holdup is 80%. Near the valve plate, due to the large change in temperature and pressure of CO2, CO2 undergoes a violent phase change, and the gas holdup here increases significantly compared with the inlet pipeline, and can reach 100% at most. In a small section of the pipeline after CO2 flows out of the valve area, due to the decrease in temperature of CO2, a potential frozen plugging area is formed, and the gas holdup of CO2 decreases compared with that inside the valve. In the latter half of the outlet pipeline, due to the decrease in pressure, CO2 is no longer in a liquid state, and the gas holdup is about 100% at this time.

[0142] Figure 10For the temperature cloud chart of the wellhead gate valve pipeline at 25% opening, it can be seen that the CO2 temperature is relatively high at the inlet pipeline of the valve, maintained at about 288K, and the CO2 temperature drops sharply when the CO2 flows out of the valve plate area, at which time the minimum temperature is formed. According to the standard phase diagram of CO2, the triple point temperature of CO2 is 217K, and it can be seen that after the CO2 flows out of the valve plate area, the CO2 in the pipeline in some areas is lower than 217K, at which time the CO2 will undergo a large phase change under the great change of pressure and temperature, and at the same time, due to the temperature being lower than the triple point, the CO2 has the possibility of freezing, that is, at this time, the wellhead gate valve has the risk of freezing.

[0143] Figure 11 For the area of the temperature cloud chart on the symmetry plane below the triple point temperature in the area where the valve may have the risk of freezing, it can be known that the larger the area, the greater the risk of freezing of the wellhead gate valve in the use process. It is calculated that at 25% opening, the area below the triple point temperature accounts for 0.3108.

[0144] According to the simulation results, it is calculated that under the genetic algorithm optimization, the S of the freezing risk area is 0.3108, which is smaller than the S of 0.3249 in the original structure, that is, the structure under the genetic algorithm optimization has a certain effect on reducing the freezing risk of the wellhead valve.

[0145] Embodiment six, this embodiment is a specific description of the principle of the global sensitivity analysis method based on Monte Carlo described in the above-mentioned embodiments;

[0146] Consider a square-integrable function y=f(x) with n independent input variables defined in the unit hypercube [0,1] n The Sobol decomposition of f(x) is represented as the sum of a set of functions with increasing dimensions:

[0147]

[0148] Where f0 is a constant, and each summation term The integral of each summation term over any of its independent variables is zero, i.e.:

[0149]

[0150] The number of summation terms in the above formula is:

[0151]

[0152] The constant f0 is the average value of the function:

[0153]

[0154] where, for simplicity, dx denotes dx1,...,dx n In the following sense, the summands are orthogonal to each other:

[0155]

[0156] According to the above assumptions, the decomposition in the equation is unique as long as f(x) is integrable over. Moreover, the summands in the decomposition can be analyzed. Indeed, the univariate summands are:

[0157]

[0158] In this expression, denotes the integration over all variables except x i and x j . Similarly, the bivariate summands are:

[0159]

[0160] where, denotes the integration over all variables except x i and x j . In general, the notation “~” in the equations denotes “complement”. Following this structure, any summand can be written as a difference of a multidimensional integral and a lower-order summand.

[0161] Now consider that the input parameters are independent random variables uniformly distributed in the interval [0,1]:

[0162] Therefore, the model response Y = f(X) is a random variable whose variance D (also called total variance) is:

[0163]

[0164] By integrating the square of the equation

[0165]

[0166] Using the above formulas in combination, the total variance can be decomposed as follows:

[0167]

[0168] where the partial variances appearing in the above expansion are:

[0169]

[0170] The Sobol global sensitivity index is defined as:

[0171]

[0172] By definition, combining the above equations, they satisfy:

[0173]

[0174] Therefore, each index is a measure of sensitivity that describes how much of the total variance is due to the uncertainty in the set of input parameters {i1,..., in}. The first order sensitivity index S s gives the effect of each parameter individually on the output, while the higher order sensitivity indices account for possible synergistic effects of the parameters on the output. i

[0175] The total sensitivity indices are further defined in order to assess the total effect of the input parameters. They are defined as the sum of all sensitivity indices containing parameter i:

[0176]

[0177] It is easy to see that:

[0178]

[0179] where S ～i denotes the sum of all indices not containing parameter i.

[0180] The Sobol global sensitivity indices are usually computed using Monte Carlo simulations. According to equations and, estimates of the mean of the response, the total variance and the partial variances can be obtained using N sim samples:

[0181]

[0182] In the last equation, x m = (x 1m , x 2m ,..., x nm ), and

[0183]

[0184] Moreover, the superscripts (1) and (2) in the equations denote that two different samples are generated and mixed, respectively. Similar expressions allow to estimate the total sensitivity indices

[0185]

[0186] ​​

[0187] Embodiment seven, see Figure 12 This embodiment is described in the principle of the genetic algorithm described above, which is a specific description of the genetic algorithm;

[0188] The core part of the standard genetic algorithm consists of five parts of individual coding, population initialization, fitness function, genetic operation and control parameter. The genetic operation of genetic algorithm includes selection, crossover and mutation. In this section, the specific workflow of standard genetic algorithm will be introduced. Before that, first of all, the individual coding, fitness function and genetic operation are briefly introduced.

[0189] Individual coding: in genetic algorithm, "individual" or "chromosome" represents a solution to the optimization problem. Genetic algorithm cannot directly search the set of feasible solutions of actual optimization problem, so it is necessary to convert the feasible solution of actual optimization problem into the type that can be recognized by genetic algorithm through coding. Common coding methods mainly include binary coding, permutation coding, integer coding, etc. The individual coding method mainly depends on the actual problem to be solved, so it can also be defined based on the optimization problem to be solved.

[0190] Fitness function: in genetic algorithm, fitness function is used to evaluate the good and bad of individuals in population (solutions to optimization problem), that is, fitness function is designed based on the objective function to be optimized in optimization problem. In the selection stage of genetic algorithm, the fitness function is used to evaluate the advantages and disadvantages of individuals, so as to select excellent individuals to participate in evolution.

[0191] The genetic operation of standard genetic algorithm consists of three operations of selection, crossover and mutation:

[0192] Selection: selection operation simulates the "survival of the fittest" principle in nature, which is an important part of genetic algorithm. The commonly used selection strategies mainly include roulette wheel selection and tournament selection. The idea of roulette wheel selection is that the probability of being selected is positively related to the size of the individual fitness value, that is, the larger the individual fitness value, the greater the probability of being selected. The calculation formula of the probability of being selected is shown in formula (6.60), wherein f(a i ) represents the fitness value of the ith individual, and N represents the total number of individuals in the population.

[0193]

[0194] The above roulette selection strategy is based on probability, and the tournament selection strategy is different from it. The tournament selection directly selects by comparing the fitness function values. The specific steps are as follows: first, a certain number of individuals are randomly selected from the population (sampling with replacement), and then the individual with the maximum fitness value is selected to join the offspring population. Repeat the above steps until the size of the offspring population reaches the size of the original population.

[0195] Crossover and mutation: In genetic algorithm, crossover operation simulates the crossing over of chromosomes in the process of biological reproduction, and its purpose is to generate new individuals. Through crossover operation, the excellent genes of parent individuals can be inherited to offspring individuals, which is an important way for genetic algorithm to obtain excellent individuals. In the crossover operation, offspring individuals are generated by exchanging gene fragments of two parent individuals. Specific crossover strategies include single-point crossover, two-point crossover, multi-point crossover, etc. Appropriate crossover strategies can be selected according to different types of coding methods. Mutation operation simulates gene mutation in the process of biological evolution, and changes the gene fragments of some parts of individuals in a random way to form new individuals. Mutation operation can bring new gene fragments to the population, to a certain extent, maintain the diversity of individuals in the population, and avoid the genetic algorithm from falling into local optimum in the search process. In summary, crossover and mutation operations are effective guarantees for genetic algorithm to search for optimal solutions. As described above, the specific workflow framework of the standard genetic algorithm is shown in Figure 12 .

[0196] The above only describes the embodiments of the present application, and is not limited to the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the scope of the claims of the present application.

Claims

1. A method for optimizing wellhead valve parameters based on intelligent algorithms, characterized in that, The wellhead valve comprises a valve plate, and a pressure relief hole is arranged on the back surface of the valve plate; the diameter a of the pressure relief hole ranges from 55 mm to 65 mm; the position b of the pressure relief hole ranges from 40 mm to 60 mm; the width c of the pressure relief hole ranges from 14 mm to 22 mm; the optimization method is to optimize the parameters of the pressure relief hole in the wellhead valve, and the method is: S1: according to the diameter a, position b and width c of the pressure relief hole, Latin hypercube sampling is performed to obtain each sample point; S2: a radial basis function proxy model is constructed according to each sample point; S3: a global sensitivity analysis method based on Monte Carlo is adopted to sample the diameter a, position b and width c of the pressure relief hole, obtain different input combinations, calculate the output results of the radial basis function proxy model under different input combinations, and obtain the contribution degree of each input parameter to the output result; S4: according to the contribution degree of each input parameter to the output result, a genetic algorithm is used to optimize the diameter a, position b and width c of the pressure relief hole, and the optimal parameters of the diameter a, position b and width c of the pressure relief hole are obtained.

2. The smart algorithm based wellhead valve parameter optimization method as claimed in claim 1, wherein, Each sample point corresponds to the diameter a, position b and width c of the pressure relief hole.

3. The smart algorithm based wellhead valve parameter optimization method as claimed in claim 1, wherein, The output of the global sensitivity analysis method based on Monte Carlo is the temperature of the wellhead valve.

4. The smart algorithm based wellhead valve parameter optimization method as claimed in claim 1, wherein, By statistically analyzing the variance of the output result, the total variance of the temperature change is decomposed into the contribution of different input parameters, and the contribution degree of each input parameter to the output result is obtained.

5. The smart algorithm based wellhead valve parameter optimization method as claimed in claim 1, wherein, The optimal diameter a of the pressure relief hole is 55.10 mm, the optimal position b of the pressure relief hole is 55.72 mm, and the optimal width c of the pressure relief hole is 18.89 mm.

6. A wellhead valve optimization system based on intelligent algorithms, characterized by, The optimization system comprises a storage device, which is used to execute the intelligent algorithm-based wellhead valve parameter optimization method and steps of claim 1.

7. A computer readable storage medium characterized by The computer readable storage medium stores a computer program, and the computer program is executed by the processor to execute the intelligent algorithm-based wellhead valve parameter optimization method of any one of claims 1-5.

8. A computer device, comprising: The device comprises a memory and a processor, and the memory stores a computer program; when the processor executes the computer program stored in the memory, the processor executes the intelligent algorithm-based wellhead valve parameter optimization method of any one of claims 1-5.

Images