CN-121638079-B - Parameter value adjusting method and device for oil and gas reservoir numerical model

CN121638079BCN 121638079 BCN121638079 BCN 121638079BCN-121638079-B

Abstract

The invention discloses a parameter value adjusting method and device of a numerical model of a hydrocarbon reservoir, relates to the technical field of hydrocarbon reservoir development, and mainly aims to improve the accuracy of the numerical model of the hydrocarbon reservoir. The method comprises the steps of obtaining a plurality of parameter value combinations of each parameter pair in an initial oil and gas reservoir numerical model, constructing a first sample matrix and a second sample matrix based on the parameter value combinations, constructing a mixed matrix of each parameter by combining the first sample matrix and the second sample matrix, obtaining respective error value vectors of the three matrices according to the initial oil and gas reservoir numerical model, calculating a first-order sensitivity index and a total sensitivity index of each parameter according to the error value vectors based on a first-order sensitivity calculation formula, obtaining an interaction effect according to the first-order sensitivity index and the total sensitivity index, combining objective function values of each parameter value combination with a parameter analysis model, obtaining a parameter influence interpretability analysis result, and finally adjusting parameter values of the initial model according to the result, the total sensitivity index and the interaction effect to obtain an optimized oil and gas reservoir numerical model.

Inventors

Bian Wanjiang
ZHOU BIHUI
ZHANG QIHUI
ZHOU DAIYU
WU CANGYUAN
ZHANG YONGBIN
Pei Xuehao

Assignees

中国石油天然气股份有限公司
新疆塔里木石油勘探开发指挥部有限公司

Dates

Publication Date: 20260512
Application Date: 20260205

Claims (9)

1. A method for adjusting a parameter value of a reservoir numerical model, the method comprising: Acquiring a plurality of parameter value combinations of each parameter pair in an initial oil and gas reservoir numerical model; respectively constructing a first sample matrix and a second sample matrix based on a plurality of parameter value combinations of each parameter pair, and constructing a mixing matrix corresponding to each parameter according to the first sample matrix and the second sample matrix; Obtaining error value vectors respectively corresponding to the first sample matrix, the second sample matrix and the mixed matrix according to the initial hydrocarbon reservoir numerical model, wherein the error value vectors are obtained by objective function values between simulation results and actual observation data corresponding to each parameter value combination in the matrix; calculating a first-order sensitivity index and a total sensitivity index corresponding to each parameter according to error value vectors respectively corresponding to the first sample matrix, the second sample matrix and the mixed matrix based on a first-order sensitivity calculation formula; Obtaining interaction effects between each parameter and other parameters according to the first-order sensitivity index and the total sensitivity index, wherein the interaction effects are used for representing the contribution of the interaction between each parameter and other parameters to the objective function value; according to the objective function values respectively corresponding to the parameter value combinations, combining the parameter analysis model to obtain a parameter influence interpretability analysis result; adjusting the parameter values in the initial reservoir numerical model according to the parameter influence interpretability analysis result, the total sensitivity index and the interaction effect to obtain an optimized reservoir numerical model; adjusting the parameter values in the initial reservoir numerical model according to the parameter influence interpretability analysis result, the total sensitivity index and the interaction effect to obtain an optimized reservoir numerical model, wherein the method comprises the following steps: according to the total sensitivity index corresponding to each parameter or a SHAP value distribution diagram in an interpretation analysis result influenced by the parameter, priority ordering is carried out among the parameters to obtain a parameter priority ordering table, wherein the SHAP value distribution diagram comprises the contribution degree of each parameter to the objective function value; identifying a target parameter pair with interaction strength exceeding a preset threshold according to the interaction effect between each parameter and other parameters; And adjusting the parameter values in the initial reservoir numerical model according to the priority ranking table and the target parameter pair with the interaction strength exceeding the preset threshold value, so as to obtain an optimized reservoir numerical model.
2. The method of claim 1, wherein constructing a mixing matrix corresponding to each parameter from the first sample matrix and the second sample matrix comprises: Extracting all parameter values of parameters in the first sample matrix for each parameter; Extracting all parameter values of other parameters except the current parameter from the second sample matrix aiming at each parameter; And constructing a mixing matrix corresponding to the parameters by combining all values of the parameters in the first sample matrix with values of other parameters in the second sample matrix for each parameter, wherein the dimension of the mixing matrix is consistent with the dimensions of the first sample matrix and the second sample matrix.
3. The method of claim 1, wherein calculating the first-order sensitivity index and the total sensitivity index for each parameter from the error value vectors for the first sample matrix, the second sample matrix, and the hybrid matrix, respectively, based on the first-order sensitivity calculation formula, comprises: calculating a first-order sensitivity index of each parameter based on a first-order sensitivity calculation formula by combining error value vectors corresponding to the first sample matrix, the second sample matrix and the mixed matrix, wherein the first-order sensitivity index is used for representing the contribution of each parameter to the objective function value independently; And calculating the total sensitivity index of each parameter based on error value vectors respectively corresponding to the first sample matrix and the mixed matrix, wherein the total sensitivity index is used for representing the sum of the main effect and all interaction effects of each parameter.
4. The method of claim 1, wherein adjusting parameter values in the initial reservoir numerical model based on the prioritized list and a target parameter pair having an interaction strength exceeding a preset threshold to obtain an optimized reservoir numerical model comprises: determining a high-sensitivity parameter set based on the parameter priority ranking table, wherein the high-sensitivity parameter set comprises parameters with total sensitivity indexes higher than a first preset threshold value or with the mean value of the absolute values of the SHAP values higher than a second preset threshold value in a SHAP value distribution diagram; Judging whether each parameter in the high-sensitivity parameter set is any parameter in the target parameter pair; If not, the target parameter pair and the high-sensitivity parameter set form a new target parameter pair, and a local joint parameter adjusting space between the new target parameter pair is constructed; In the local joint parameter adjustment space, carrying out iterative search on the parameter values of the new target parameter pair by adopting a self-adaptive optimization algorithm, and adjusting the search step length and the search direction according to the change of the target function value in each iteration; The parameter value combination of the new target parameter pair obtained after iteration convergence is used for replacing the corresponding original parameter value in the initial oil and gas reservoir numerical model, and a candidate optimization model is generated; And carrying out prediction performance evaluation on the candidate optimization model, and if the performance evaluation result meets a preset precision index, determining the candidate optimization model as the optimized oil and gas reservoir numerical model.
5. The method of claim 1, wherein obtaining a plurality of parameter value combinations for each parameter pair in the initial reservoir numerical model comprises: if the number of the parameters is at least three, selecting any two parameters to form a parameter pair, and independently executing Latin hypercube sampling in a parameter space corresponding to the parameter pair to generate a parameter value combination of the parameter pair; and repeating the process to generate different parameter pairs, and independently executing Latin hypercube sampling on the plurality of different parameter pairs to obtain a plurality of parameter value combinations of the plurality of parameter pairs.
6. The method of claim 1, wherein obtaining error value vectors for the first sample matrix, the second sample matrix, and the hybrid matrix, respectively, from the initial reservoir numerical model, comprises: determining an expression of an objective function value, wherein the objective function is constructed based on deviation between a simulation result of an initial reservoir numerical model and actual observed data, and the actual observed data comprises oil and gas yield, bottom hole pressure and reservoir temperature; Inputting parameter values in the first sample matrix, the second sample matrix and the mixed matrix into an initial oil and gas reservoir numerical model for simulation operation to obtain simulation results corresponding to the matrices; Determining the objective function value in each matrix by combining the expression of the objective function value according to the simulation result and the actual observation data in each matrix; and obtaining error value vectors corresponding to the matrixes according to the objective function values in the matrixes.
7. A parameter value adjustment device for a reservoir numerical model, the device comprising: the acquisition unit is used for acquiring a plurality of parameter value combinations of each parameter pair in the initial oil and gas reservoir numerical model; The construction unit is used for respectively constructing a first sample matrix and a second sample matrix based on a plurality of parameter value combinations of each parameter pair acquired by the acquisition unit, and constructing a mixing matrix corresponding to each parameter according to the first sample matrix and the second sample matrix; The simulation unit is used for obtaining error value vectors respectively corresponding to the first sample matrix, the second sample matrix and the mixed matrix constructed by the construction unit according to the initial oil and gas reservoir numerical model, wherein the error value vectors are obtained by objective function values between simulation results corresponding to each parameter value combination in the matrix and actual observation data; The calculation unit is used for calculating a first-order sensitivity index and a total sensitivity index corresponding to each parameter according to the error value vectors respectively corresponding to the first sample matrix, the second sample matrix and the mixed matrix obtained by the simulation unit based on a first-order sensitivity calculation formula; The interaction unit is used for obtaining interaction effects between each parameter and other parameters according to the first-order sensitivity index and the total sensitivity index calculated by the calculation unit, wherein the interaction effects are used for representing the contribution of interaction between each parameter and other parameters to the objective function value; The analysis unit is used for combining the parameter analysis model to obtain a parameter influence interpretability analysis result according to the respectively corresponding objective function values of the parameter value combinations; the adjusting unit is used for adjusting the parameter value in the initial reservoir numerical model according to the parameter influence interpretability analysis result obtained by the analyzing unit, the total sensitivity index and the interaction effect obtained by the interaction unit to obtain an optimized reservoir numerical model; the adjusting unit is specifically configured to: according to the total sensitivity index corresponding to each parameter or a SHAP value distribution diagram in an interpretation analysis result influenced by the parameter, priority ordering is carried out among the parameters to obtain a parameter priority ordering table, wherein the SHAP value distribution diagram comprises the contribution degree of each parameter to the objective function value; identifying a target parameter pair with interaction strength exceeding a preset threshold according to the interaction effect between each parameter and other parameters; And adjusting the parameter values in the initial reservoir numerical model according to the priority ranking table and the target parameter pair with the interaction strength exceeding the preset threshold value, so as to obtain an optimized reservoir numerical model.
8. A storage medium comprising a stored program, wherein the program, when run, controls a device in which the storage medium is located to perform the parameter value adjustment method of the reservoir numerical model according to any one of claims 1 to 6.
9. A processor for running a program, wherein the program is run to perform the parameter value adjustment method of the reservoir numerical model according to any one of claims 1 to 6.

Description

Parameter value adjusting method and device for oil and gas reservoir numerical model Technical Field The invention relates to the technical field of oil and gas reservoir development, in particular to a parameter value adjusting method and device of an oil and gas reservoir numerical model. Background The reservoir numerical model is used to predict and optimize the reservoir development process by modeling the subsurface fluid flow. By performing history fitting on the model, namely adjusting parameter values of key parameters such as permeability, porosity, fault conductivity and the like, the simulation result of the model can be more close to actual production data, so that the characteristics of the oil and gas reservoir can be more accurately described and the dynamic change of the oil and gas reservoir can be predicted. Currently, in parameter value adjustment, a common method is based on single parameter sensitivity analysis. The method evaluates the sensitivity of the parameter by calculating the influence degree of the parameter value change of the single parameter on an objective function (such as the square sum of errors between a simulation result and actual data), and further assists in parameter value adjustment of the parameter. For example, when analyzing the influence of the permeability, the parameter values of other parameters are fixed, only the permeability value is changed, and the change rate of the objective function is calculated, so that the influence degree of the objective function on the fitting effect is determined. Although this approach reduces the blindness of manual parameter tuning to some extent, there are significant limitations. It can only reflect the independent influence of a single parameter, and cannot take into account the coupling relationship and interaction between multiple parameters. In an actual hydrocarbon reservoir, the parameters are often interrelated and co-acting. Therefore, parameter optimization is difficult to accurately guide by a parameter adjustment mode based on single parameter sensitivity analysis, and fitting results possibly deviate from real geological conditions, so that prediction accuracy and application effects of the model are affected. Disclosure of Invention In view of the above, the present invention provides a method and a device for adjusting parameter values of a hydrocarbon reservoir numerical model, which are mainly aimed at improving the accuracy of the hydrocarbon reservoir numerical model. In order to solve the technical problems, the invention provides the following scheme: In a first aspect, the present invention provides a method for adjusting a parameter value of a reservoir numerical model, the method comprising: Acquiring a plurality of parameter value combinations of each parameter pair in an initial oil and gas reservoir numerical model; respectively constructing a first sample matrix and a second sample matrix based on a plurality of parameter value combinations of each parameter pair, and constructing a mixing matrix corresponding to each parameter according to the first sample matrix and the second sample matrix; Obtaining error value vectors respectively corresponding to the first sample matrix, the second sample matrix and the mixed matrix according to the initial hydrocarbon reservoir numerical model, wherein the error value vectors are obtained by objective function values between simulation results and actual observation data corresponding to each parameter value combination in the matrix; calculating a first-order sensitivity index and a total sensitivity index corresponding to each parameter according to error value vectors respectively corresponding to the first sample matrix, the second sample matrix and the mixed matrix based on a first-order sensitivity calculation formula; Obtaining interaction effects between each parameter and other parameters according to the first-order sensitivity index and the total sensitivity index, wherein the interaction effects are used for representing the contribution of the interaction between each parameter and other parameters to the objective function value; according to the objective function values respectively corresponding to the parameter value combinations, combining the parameter analysis model to obtain a parameter influence interpretability analysis result; and adjusting the parameter values in the initial reservoir numerical model according to the parameter influence interpretability analysis result, the total sensitivity index and the interaction effect, so as to obtain an optimized reservoir numerical model. In a second aspect, the present invention provides a parameter value adjustment device for a reservoir numerical model, the device comprising: the acquisition unit is used for acquiring a plurality of parameter value combinations of each parameter pair in the initial oil and gas reservoir numerical model; The construction unit is used for respectively