Carbonate reservoir oil well parameter screening method and geological well selection method

Abstract

The invention relates to the technical field of oil and gas development, in particular to a carbonate reservoir oil well parameter screening method and a geological well selection method. The carbonate reservoir oil well parameter screening method comprises the steps that dependent variable parameters and a plurality of independent variable parameters are screened out, test values of all variable parameters of a plurality of sets of known oil wells are obtained, a dimensionless regression model is established, estimated values of dimensionless regression coefficients before all the independent variable parameters are calculated through a least square method, and the estimated values of the dimensionless regression coefficients before all the independent variable parameters are calculated. And evaluating the dimensionless regression model, determining dimensionless regression coefficients conforming to evaluation as influence factors, screening out key influence factors, and determining well selection key parameters. According to the method, key well selection parameters are rapidly screened out through the dimensionless regression model, the average level of variable parameters does not need to be found, processing is more convenient, a foundation is provided for rapid well selection, the well selection efficiency is effectively improved, and meanwhile the requirement for high recovery efficiency can be met.

Description

Technical Field

[0001] This invention relates to the field of oil and gas development technology, and in particular to a method for screening oil well parameters and a geological well selection method for carbonate reservoirs. Background Technology

[0002] Gas injection gravity drive is an important development method in the later stages of waterflooding for old oilfields, used to alter the oil-water relationship, replenish formation energy, and improve the utilization of remaining oil. Its effectiveness is closely related to parameters such as geological structure, sedimentary distribution, reservoir properties, crude oil properties, gas type, development measures, and injection-production processes. However, due to incomplete geological understanding of carbonate reservoirs, difficulties in delineating fractures and vugs and describing well-cavity relationships, understanding the distribution patterns of remaining oil, and recognizing the relationship between wellbore and fractures / vugs, a systematic well selection method has not been established, leading to insufficient demonstration of gas injection potential. Therefore, it is difficult to quickly select oil wells with gas injection potential; even if selected, it is time-consuming and labor-intensive, and it is difficult to meet the requirements of high recovery rates.

[0003] Regression analysis is a method used to analyze the specific forms of correlation between phenomena, filter causal relationships, replace them with numerical parameters, and represent these relationships using mathematical models. In regression analysis, if there are two or more independent variables, it is called multiple regression; if multiple independent variables have a linear relationship with the dependent variable, it is called multiple linear regression. Multiple linear regression, because it considers the simultaneous changes of multiple variables, can better reflect the global relationship between each parameter and the dependent variable. However, since the units of the independent variables are generally inconsistent, the magnitude of the coefficients before the independent variables in multiple linear regression using actual variables does not necessarily indicate the importance of that factor. Therefore, mathematically, the standard score method is needed to unify all parameters to the same unit before performing linear regression. The regression coefficients at this point can reflect the importance of the corresponding independent variable. Since all variables are converted to standard scores, a constant term is no longer used, because when all independent variables take the average level, the dependent variable should also take the average level, and the average level corresponds to a standard score of 0. When the variables on both sides of the linear regression equation are all zero, the constant term is also zero. However, in practical engineering applications, standard scores often cannot accurately determine the average level. That is, the method of directly using the standard scores of the multiple linear regression model cannot accurately find the average level of each parameter of the oil well, and it is time-consuming and laborious, thus making well selection difficult. Summary of the Invention

[0004] The purpose of this invention is to provide a method for screening oil well parameters and a geological well selection method for carbonate reservoirs, so as to solve the problem that the existing method of using the standard score of the multiple linear regression model cannot quickly select oil wells with gas injection potential.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] This invention discloses a method for screening oil well parameters in carbonate reservoirs, comprising:

[0007] From multiple geological parameters selected for well selection, dependent and independent variables are identified. Test values ​​of all variables for several known wells are obtained. A dimensionless regression model is then established to correlate the dependent and independent variables. The expression for the dimensionless regression model is as follows:

[0008]

[0009] In the formula, Let be the dimensionless variable values ​​of the dependent variable parameters of the i-th known oil well. Let c be the dimensionless variable value of the m-th independent variable parameter of the i-th known oil well. m Here, c0 is the dimensionless regression coefficient before the m-th independent variable parameter, and μ is the constant term. i For the i-th test point, this is the error term;

[0010] Based on the test values ​​of all variable parameters of several known oil wells, the least squares method is used to calculate the estimated values ​​of the dimensionless regression coefficients before each independent variable parameter in the dimensionless regression model.

[0011] After assigning the estimated values ​​of the dimensionless regression coefficients to the dimensionless regression model, the dimensionless regression model is evaluated for goodness of fit, confidence range, and significance test. The dimensionless regression coefficients that meet the evaluation are determined as influencing factors.

[0012] The determined influencing factors are graphically represented, and the key influencing factors are selected based on the degree of influence in the representation. The independent variable parameters corresponding to the key influencing factors are determined as the key parameters for well selection.

[0013] Optionally, the method for establishing a dimensionless regression model relating the dependent variable parameters and each of the independent variable parameters includes:

[0014] Based on the selected dependent variable parameters and the corresponding independent variable parameters, a multiple linear regression model is established. The expression of the multiple linear regression model is as follows:

[0015] y i =a0+a1x 1,i +a2x 2,i +...+a m x m,i

[0016] In the formula, y i Let x be the test value of the dependent variable parameter of the i-th known oil well. m,iLet a be the test value of the m-th independent variable parameter of the i-th known oil well group. m represents the regression coefficient before the m-th independent variable parameter, and a0 is the constant term;

[0017] The dimensionless method is used to convert the test values ​​of all variable parameters of each known oil well into dimensionless variable values. The multiple dimensionless variable values ​​are then substituted into the multiple linear regression model to reconstruct the dimensionless regression model.

[0018] Optionally, the method for calculating the estimated values ​​of the dimensionless regression coefficients before each of the independent variable parameters using the least squares method includes:

[0019] Construct a coefficient matrix for all variable parameters;

[0020] The least squares method is used to establish a simultaneous equation for all variable parameters. The expression of the simultaneous equation is as follows:

[0021]

[0022] In the formula, n is the number of known oil wells;

[0023] Substitute the test values ​​of all variable parameters of each known oil well into the simultaneous equations to solve the coefficient matrix, and estimate the estimated values ​​of the dimensionless regression coefficients of each independent variable parameter.

[0024] Optionally, the method for evaluating the goodness of fit of the dimensionless regression model includes:

[0025] Based on the dimensionless regression model with the assigned dimensionless regression coefficients, the test values ​​of the independent variable parameters of each known oil well are substituted into the dimensionless regression model to calculate the estimated values ​​of the dependent variable parameters of each group of known oil wells.

[0026] Based on the test and estimated values ​​of the dependent variable parameters for each group of known oil wells, the standard deviation of the dimensionless regression model is calculated. The goodness of fit of the dimensionless regression model is evaluated based on the magnitude of the difference in the standard deviation. The expression for the standard deviation is:

[0027]

[0028] In the formula, SE is the standard deviation. Let be the dimensionless estimate of the dependent variable parameter of the i-th known oil well, where n is the number of known oil wells. Let be the deviation between the tested and estimated values ​​of the dependent variable parameter of the i-th known oil well group.

[0029] Optionally, the method for evaluating the confidence range of the dimensionless regression model includes:

[0030] The degrees of freedom kh of the dimensionless regression model are determined based on the test values ​​of all variable parameters of all known oil wells;

[0031] Select a confidence level, and based on the degrees of freedom kh and the confidence level, look up t in a preset table of t-statistic values. p The value;

[0032] Based on the calculated standard deviation and the found t p The confidence interval of the dimensionless regression model is calculated, and the expression for the confidence interval is:

[0033]

[0034] In the formula, Pr represents the upper and lower limits of the confidence interval, k represents the number of all known well test values ​​corresponding to the single variable parameter, h represents the number of all variable parameters, and t represents the number of all variables. p The values ​​in the t-statistic table for the degrees of freedom kh;

[0035] The confidence range of the dimensionless regression model is evaluated based on the calculated confidence interval.

[0036] Optionally, the method for evaluating the significance of the dimensionless regression model includes:

[0037] The significance of the dimensionless regression model is tested using the standard deviations of the test and estimated values ​​of the dependent variable parameters of the known oil wells.

[0038] The significance of the dimensionless regression coefficient is tested using a t-test based on the test values ​​of all known oil wells corresponding to the single independent variable parameter.

[0039] The significance of the established dimensionless regression model and the calculated dimensionless regression coefficients is determined based on the results of the F-test and t-test.

[0040] Optionally, the method for performing an F-test on the significance of the dimensionless regression model includes:

[0041] Choose any two groups of known oil wells and obtain the test values ​​of all variable parameters for each of the two groups of known oil wells;

[0042] Calculate the standard deviation of the test and estimated values ​​of the dependent variable parameter for each of the two sets of known oil wells;

[0043] The significance of the dimensionless regression model is assessed using an F-test based on the two calculated standard deviations. The expression for the F-test is as follows:

[0044]

[0045] Wherein, SE and SE' are respectively the standard deviations of the measured values and estimated values of the dependent variable parameters of the two groups of the known oil wells.

[0046] Look up the F-table value in the preset F-statistic value table according to the degrees of freedom k - h and the confidence level. If the F-test result < F-table value, it is evaluated that there is no significant difference in the measured values of all variable parameters of the two groups of the known oil wells in the dimensionless regression model. If the F-test result > F-table value, it is evaluated that there is a significant difference in the measured values of all variable parameters of the two groups of the known oil wells in the dimensionless regression model.

[0047] Optionally, the method for performing a t-test on the significance of the dimensionless regression coefficient includes:

[0048] Set all the measured values of the known oil wells corresponding to a single independent variable parameter as the population, and randomly select several groups of the measured values of the known oil wells from the population as samples.

[0049] Select two of the independent variable parameters and establish a null hypothesis. The expression of the null hypothesis is:

[0050] H0: μ1 = μ2

[0051] Wherein, H0 is the null hypothesis, μ1 is the population mean corresponding to one of the independent variable parameters, μ2 is the population mean corresponding to the other independent variable parameter, and μ1 = μ2 means that there is no significant difference between the two population means.

[0052] Calculate the statistic value t according to the population mean and the sample mean n ;

[0053] Determine the degree of freedom d of a single independent variable parameter according to the capacity of the sample f , and according to the degree of freedom d f Look up the theoretical value t(d f ) in the preset t-statistic value table. The expression of the degree of freedom d <00000​​​​​​​​​​​​​​​​​A comparison is made, and at the given significance level, the statistical value t is inferred from the comparison results. n Exceeding or falling below the theoretical value t(d) f The probability P-value of ).

[0057] The inferred probability P-value is compared with a given significance level. If the probability P-value is less than the significance level, the null hypothesis is rejected and a significant difference is determined between the two population means. If the probability P-value is greater than the significance level, the null hypothesis is accepted and no significant difference is determined between the two population means.

[0058] Optionally, the step of calculating the statistic t based on the population mean and the sample mean... n The methods include:

[0059] The statistic t is used to evaluate the degree of difference between the mean of a sample in the population and the population mean. n The calculation formula is:

[0060]

[0061]

[0062] In the formula, X i The independent variable parameter represents the test value of the i-th known oil well, where n is the number of known oil well test values ​​included in the sample, and t is the number of test values ​​of the i-th known oil well. n For statistical values, Here, SE is the sample mean, and μ is the sample standard deviation.

[0063] To evaluate the degree of difference between the means of the two groups of samples, the statistic t... n The calculation formula is:

[0064]

[0065] In the formula, SE1 and SE2 are the variances of the two sets of samples, respectively, and n1 and n2 are the number of known oil well test values ​​included in the two sets of samples, respectively. These are the average values ​​of the two groups of samples, respectively.

[0066] This invention also discloses a geological well selection method, which includes well selection based on key well selection parameters screened by the above-mentioned carbonate reservoir well parameter screening method.

[0067] Compared with the prior art, the beneficial effects of the method for screening oil well parameters and the geological well selection method for carbonate reservoirs provided in this invention are as follows:

[0068] By first screening dependent and independent parameters from multiple geological parameters for well selection, and then converting the test values ​​of all parameters corresponding to several known oil wells into dimensionless quantities, a dimensionless regression model relating the dependent and independent parameters is constructed. The dimensionless regression coefficients before each independent parameter are estimated, and the model is then evaluated for goodness of fit, confidence range, and significance. From the dimensionless regression coefficients that meet the evaluation criteria, key influencing factors are quickly identified, and the independent parameters corresponding to these factors are determined as key parameters for well selection. This method eliminates the need to find the average level of each parameter, making it more convenient to process and providing a foundation for rapid well selection, effectively improving well selection efficiency while also achieving high recovery rates. Attached Figure Description

[0069] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0070] Figure 1 A schematic flowchart of the overall process for screening parameters of carbonate reservoir oil wells provided in this embodiment of the invention;

[0071] Figure 2 A radar chart representing the influence factor graphically as provided in an embodiment of the present invention. Detailed Implementation

[0072] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0073] This invention discloses a method for screening oil well parameters in carbonate reservoirs, such as... Figure 1 As shown, it includes:

[0074] S1. Select the dependent variable parameter and multiple independent variable parameters from multiple geological parameters of well selection, and obtain the test values ​​of all variable parameters of several known oil wells. Establish a dimensionless regression model relating the dependent variable parameter and each independent variable parameter. The expression of the dimensionless regression model is:

[0075]

[0076] In the formula, Let be the dimensionless variable values ​​of the dependent variable parameters of the i-th known oil well. Let c be the dimensionless variable value of the m-th independent variable parameter of the i-th known oil well. m Here, c0 is the dimensionless regression coefficient before the m-th independent variable parameter, and μ is the constant term. i For the i-th test point, this is the error term;

[0077] S2. Based on the test values ​​of all variable parameters of several known oil wells, the least squares method is used to calculate the estimated values ​​of the dimensionless regression coefficients of each variable parameter in the dimensionless regression model.

[0078] S3. After assigning the estimated values ​​of the dimensionless regression coefficients to the dimensionless regression model, evaluate the goodness of fit, confidence range, and significance of the dimensionless regression model, and determine the dimensionless regression coefficients that meet the evaluation as influencing factors.

[0079] S4. The determined influencing factors are graphically represented, and the key influencing factors are selected based on the degree of influence in the representation. The independent variable parameters corresponding to the key influencing factors are determined as the key parameters for well selection.

[0080] As mentioned above, the geological parameters selected for well selection include: geological structural morphology, sedimentary distribution, reservoir properties, crude oil properties, gas types, development measures, and injection-production technology. By analyzing the selected geological structural morphology, sedimentary distribution, reservoir properties, crude oil properties, gas types, development measures, and injection-production technology parameters, the causal relationships among the relevant parameters are analyzed, and the relevant parameters are selected as independent or dependent variables based on these causal relationships. The preferred method for selecting independent variables is to establish a correlation matrix between the various independent variables. Preferably, the geological parameters selected for well selection also include gas-oil exchange rate, constant volume, non-constant volume, water cut increase type, remaining oil type, energy assessment, well-reservoir relationship, producing interval location, and reservoir type.

[0081] Furthermore, methods for establishing dimensionless regression models that correlate dependent variable parameters with independent variable parameters include:

[0082] Based on the selected dependent variable parameters and their corresponding independent variable parameters, a multiple linear regression model is established. The expression for the multiple linear regression model is as follows:

[0083] y i =a0+a1x 1,i +a2x 2,i +...+a m x m,i

[0084] In the formula, y i Let x be the test value of the dependent variable parameter of the i-th known oil well. m,i Let a be the test value of the m-th independent variable parameter of the i-th known oil well group. m The regression coefficients before the m-th independent variable parameter, a0 is the constant term, i = 1, 2, ..., n, and n is the number of known oil wells;

[0085] The dimensionless method is used to convert the test values ​​of all variable parameters of each known oil well into dimensionless variable values. The multiple dimensionless variable values ​​are then substituted into the multiple linear regression model to reconstruct the dimensionless regression model.

[0086] As mentioned above, because the units of the independent variable parameters in a multiple linear regression model are inconsistent, the regression coefficients a1, a2, ..., a2 before each independent variable parameter are different. m The magnitude of a value does not necessarily indicate the importance of the corresponding independent variable parameter. Using the standard regression method, a common mathematical approach, to process the multiple linear regression model, unifying the independent variable parameters to the same unit, and then performing linear regression, we obtain the standard regression equation:

[0087]

[0088] Since all variable parameters in the standard regression equation are converted into standard scores, the constant terms a1, a2, ..., a are no longer used. m Because when all independent variable parameters are taken as average, the dependent variable parameter should also be taken as average, and the average corresponds exactly to a standard score of 0. When both sides of the equation have variable parameters of 0, the constant term is 0. However, for engineering applications, the standard score is sometimes difficult to determine the average level. Therefore, the dimensionless method is used, transforming all variable parameters (independent and dependent variable parameters) into dimensionless quantities, unifying the units to dimensionless, and then performing linear regression. In this case, the regression coefficients can reflect the importance of the corresponding independent variable parameters. That is, the above-mentioned dimensionless regression model is established, and the regression coefficients are also dimensionless regression coefficients.

[0089] Furthermore, methods for calculating the estimates of the dimensionless regression coefficients of each independent variable parameter using the least squares method include:

[0090] Construct a coefficient matrix for all variable parameters;

[0091] The least squares method is used to establish a simultaneous equation for all variable parameters. The expression of the simultaneous equation is:

[0092]

[0093] In the formula, n is the number of known oil wells;

[0094] Substitute the test values ​​of all variable parameters of each known oil well into the simultaneous equations to solve the coefficient matrix, and estimate the estimated values ​​of the dimensionless regression coefficients of each variable parameter.

[0095] As mentioned above, the simultaneous equations have a total of n+1 variables ((y i x 1,i x 2,i , ..., x m,iGiven a linear regression model with variables (i = 1, 2, ..., n), there are n+1 multivariate linear equations. By taking different values ​​for the parameters of each variable, the coefficient matrix is ​​solved using a simultaneous equation system to estimate the dimensionless regression coefficients before each variable parameter. Parameter estimation for the constructed dimensionless regression model generally involves two methods: ordinary least squares (OLS) and generalized least squares (GLS). OLS finds the optimal function by minimizing the sum of squared errors and solves for the coefficient matrix through matrix operations. Its expression is shown below:

[0096]

[0097] Generalized least squares is an extension of ordinary least squares, allowing for efficient coefficient estimates when heteroscedasticity, autocorrelation, or both exist in the error term. Its expression is shown below:

[0098]

[0099] In the formula, β′ is the least squares error, Ω is the covariance matrix of the residual term, and X T For the independent variable parameter x m,i The coefficient transpose of , where X is the coefficient matrix of the independent variable parameters, x i Let y be the parameter matrix of the independent variable. i To show the dependent variable parameter matrix, This is the transpose of the independent variable parameter matrix.

[0100] Furthermore, methods for evaluating the goodness of fit of dimensionless regression models include:

[0101] Based on the dimensionless regression model with assigned dimensionless regression coefficients, the test values ​​of the independent variable parameters of the known oil wells are substituted into the dimensionless regression model to calculate the estimated values ​​of the dependent variable parameters of each group of known oil wells.

[0102] Based on the tested and estimated values ​​of the dependent variable parameters for each group of known oil wells, the standard deviation of the dimensionless regression model is calculated. The goodness of fit of the dimensionless regression model is evaluated based on the magnitude of the difference in standard deviation. The expression for the standard deviation is:

[0103]

[0104] In the formula, SE is the standard deviation. Let be the dimensionless estimate of the dependent variable parameter of the i-th known oil well, where n is the number of known oil wells. Let be the deviation between the tested and estimated values ​​of the dependent variable parameter of the i-th known oil well group.

[0105] As mentioned above, the size of the standard deviation reflects the degree of dispersion of the dependent variable. A larger standard deviation indicates that the data is more dispersed, while a smaller standard deviation indicates that the data is less dispersed.

[0106] Furthermore, methods for evaluating the confidence range of dimensionless regression models include:

[0107] The degrees of freedom kh of the dimensionless regression model are determined based on the test values ​​of all variable parameters of all known oil wells;

[0108] Select the confidence level, and based on the degrees of freedom kh and the confidence level, find the t-statistic value in the preset t-statistic table. p The value;

[0109] Based on the calculated standard deviation and the found t p The confidence interval for the dimensionless regression model is calculated using the following formula:

[0110]

[0111] In the formula, Pr represents the upper and lower limits of the confidence interval, k represents the number of all known well test values ​​corresponding to the single variable parameter, h represents the number of all variable parameters, and t represents the number of all variables. p The values ​​in the t-statistic table for the degrees of freedom kh;

[0112] The confidence range of the dimensionless regression model is evaluated based on the calculated confidence interval.

[0113] As mentioned above, the confidence level can be chosen as 95% (e.g., 0.95) or 99% (e.g., 0.99). Based on the comparison results, assess the appropriateness of the confidence interval. If the confidence interval includes the actual value, then it can be considered reasonable; if the confidence interval does not include the actual value, then further consideration of the model's goodness of fit or other factors may be necessary.

[0114] Furthermore, methods for evaluating the significance of dimensionless regression models include:

[0115] Based on the standard deviations of the known dependent variable parameters of the oil wells, an F-test was performed to assess the significance of the dimensionless regression model.

[0116] Based on the test values ​​of all known oil wells corresponding to the single independent variable parameter, a t-test was performed to assess the significance of the dimensionless regression coefficient.

[0117] The significance of the established dimensionless regression model and the calculated dimensionless regression coefficients is determined based on the results of the F-test and t-test.

[0118] As mentioned above, significance testing is a method used to determine whether there is a difference between the effects of the experimental treatment group and the control group, or between two different treatments, and whether this difference is significant. The hypothesis to be tested is usually denoted as H0, called the null hypothesis, and the hypothesis opposing H0 is denoted as H1, called the alternative hypothesis.

[0119] (A) When the null hypothesis is true, the decision to abandon the null hypothesis is called a Type I error, and its probability of occurrence is usually denoted as α;

[0120] (B) When the null hypothesis is false, the decision not to abandon the null hypothesis is called a Type II error, and its probability of occurrence is usually denoted as β.

[0121] (C) α+β is not necessarily equal to 1.

[0122] Typically, only the maximum probability α of committing a Type I error is considered, without taking into account the probability β of committing a Type II error. This type of hypothesis testing is also called a significance test, and the probability α is called the significance level.

[0123] The most commonly used values ​​for α are 0.01, 0.05, and 0.10. Generally, depending on the research question, if abandoning the true hypothesis would result in significant losses, a smaller value for α is chosen to reduce such errors; conversely, a larger value for α is chosen if the research question is more complex.

[0124] Two statistical methods are generally used for testing: the F-test for the significance of the regression model and the t-test for the significance of the regression coefficients. When randomly drawing samples from two research populations and comparing them, it is necessary to first determine whether the variances of the two populations are the same, i.e., homogeneity of variance, to determine whether there is a significant difference in the precision of the two populations. If the variances of the two populations are equal, the t-test can be used directly; if they are not equal, the t' test, variable transformation, or rank-sum test can be used.

[0125] Furthermore, methods for conducting F-tests to assess the significance of dimensionless regression models include:

[0126] Choose any two known oil wells and obtain the test values ​​of all variable parameters for each of the two known oil wells;

[0127] Calculate the standard deviation of the test and estimated values ​​of the dependent variable parameter for the two sets of known oil wells, respectively;

[0128] The significance of the dimensionless regression model is tested using the two calculated standard deviations. The expression for the F-test is as follows:

[0129]

[0130] In the formula, SE and SE′ are the standard deviations of the test and estimated values ​​of the dependent variable parameters for two sets of known oil wells, respectively;

[0131] Look up the F-table value in the preset F-statistic value table according to the degrees of freedom k - h and the confidence level. If the F-test result < F-table value, it is evaluated that there is no significant difference in the test values of all variable parameters of the two groups of known oil wells in the dimensionless regression model. If the F-test result > F-table value, it is evaluated that there is a significant difference in the test values of all variable parameters of the two groups of known oil wells in the dimensionless regression model.

[0132] As described above, the F-statistic value table is shown in Table 1. Table 1 is the F-statistic value table with a confidence level of 95% (one-sided, α = 0.05, F > F α ).

[0133] Table 1 F-statistic value table with a confidence level of 95%

[0134]

[0135] Furthermore, the method for performing a t-test on the significance of the dimensionless regression coefficient includes:

[0136] Set the test values of all known oil wells corresponding to a single independent variable parameter as the population, and randomly select several groups of test values of known oil wells from the population as the sample;

[0137] Select two independent variable parameters and establish a null hypothesis. The expression of the null hypothesis is:

[0138] H0: μ1 = μ2

[0139] In the formula, H0 is the null hypothesis, μ1 is the population mean corresponding to one independent variable parameter, μ2 is the population mean corresponding to another independent variable parameter, and μ1 = μ2 means that there is no significant difference between the two population means;

[0140] Calculate the statistic value t according to the population mean and the sample mean n ;

[0141] Determine the degree of freedom d of a single independent variable parameter according to the sample size f , and according to the degree of freedom d f Look up the theoretical value t(d f ) in the preset t-statistic value table. The expression of the degree of freedom d f is:

[0142] d f = n - 1

[0143] In the formula, d f is the degree of freedom of a single independent variable parameter, and n is the number of test values of known oil wells included in the sample.

[0144] Compare the calculated statistic value t n with the found theoretical value t(df By comparing the results, and at a given significance level, inferring the statistical value t based on the comparison results. n Exceeding or falling below the theoretical value t(d) f The probability P-value of ).

[0145] The inferred probability P-value is compared with the given significance level. If the probability P-value is less than the significance level, the null hypothesis is rejected and a significant difference is determined between the two population means. If the probability P-value is greater than the significance level, the null hypothesis is accepted and no significant difference is determined between the two population means.

[0146] As mentioned above, the t-test is mainly used for normal distributions with small sample sizes and unknown population standard deviations. It is a method for testing the degree of difference between the means of two small samples. The t-distribution theory is used to infer the probability of a difference occurring, thereby determining whether the difference between the two means is significant. The theoretical significance level of the difference is given as 0.01 or 0.05, and the theoretical significance levels for different degrees of freedom are denoted as t(d). f )0.01 and t(d f 0.05;

[0147] Comparison statistic t n and theoretical value t(d) f To infer the probability of occurrence, we use the statistic t given in Table 2. n Make a judgment based on the significance of the differences.

[0148] Table 2. Statistical values ​​t n Table of Significance of Differences

[0149] <![CDATA[Statistic value t n > probability P value Significance of difference <![CDATA[t n >=t(d f )0.01]]> P<＝0.01 The difference is very significant. <![CDATA[t n >=tt(d f )0.05]]> P<＝0.05 Significant differences <![CDATA[t n <t(d f )0.05]]> P>0.05 The difference was not significant.

[0150] Furthermore, based on the population mean and the sample mean, the statistic t is calculated. n The methods include:

[0151] To evaluate the difference between the sample mean and the population mean in a population, the t-statistic is used. n The calculation formula is:

[0152]

[0153]

[0154] In the formula, X i Let t be the test value of the i-th known oil well corresponding to the single independent variable parameter, n be the number of known oil well test values ​​included in the sample, and t be the test value of the i-th known oil well. n For statistical values, Here, SE is the sample mean, and μ is the sample standard deviation.

[0155] The above statistical value t n The calculation method is the single-population t-test, which tests whether the difference between a sample mean and a known population mean is significant. When the population distribution is normal, the deviation statistic of the sample mean from the population mean follows a t-distribution.

[0156] To evaluate the difference between the means of two groups of samples, the statistic t is used. n The calculation formula is:

[0157]

[0158] In the formula, SE1 and SE2 are the variances of the two sample groups, respectively, and n1 and n2 are the number of known oil well test values ​​included in the two sample groups, respectively. These are the average values ​​of the two groups of samples, respectively.

[0159] The above statistical value t n The calculation method is the independent samples t-test of the two population t-test, which tests whether the difference between the means of two samples and the populations they represent is significant.

[0160] This invention also discloses a geological well selection method, which includes well selection based on key well selection parameters screened by the above-mentioned carbonate reservoir well parameter screening method.

[0161] The method for screening oil well parameters and geological well selection in carbonate reservoirs according to embodiments of the present invention quickly screens out key well selection parameters through a constructed dimensionless regression model, eliminating the need to find the average level of each variable parameter, making the process more convenient and providing a foundation for rapid well selection. This effectively improves well selection efficiency while also achieving high recovery rates. Further explanation is provided below with specific implementation details:

[0162] The gas-oil exchange rate was selected as the dependent variable parameter in the multiple linear regression model, while constant or non-constant volume, water cut rise type, remaining oil type, energy evaluation, well-reservoir relationship, producing zone location, and reservoir type were selected as the independent variable parameters in the multiple linear regression model. The gas-oil exchange rate (Y) was used in conjunction with constant / non-constant volume (X1), water cut rise type (X2), remaining oil type (X3), energy evaluation (X4), well-reservoir relationship (X5), producing zone location (X6), and reservoir type (X7) to construct a multiple linear regression model for the above parameters.

[0163] Using data from wells LG45, LG701, LG701-4, LG7-8, LN8, LG9-H5, LG9C, LG15-H32, LN11-H1, LG7-5-2, HA902, HA13-2, and JY8 in the Lungu oil reservoir as test point data, the parameters of constant or non-constant volume, water cut increase type, remaining oil type, energy evaluation, well-reservoir relationship, producing interval location, and reservoir type were all converted into dimensionless quantities. Table 3 is a table of dimensionless variable values ​​for all variable parameters corresponding to the test values ​​of the above wells, and a dimensionless regression model was constructed.

[0164] Table 3. Dimensionless variable values ​​of test values ​​from each well.

[0165]

[0166]

[0167] Based on the gas-oil exchange rate and the test values ​​of each independent variable parameter in Table 3, the dimensionless regression coefficients, i.e., the influencing factors, of each independent variable parameter are estimated using the least squares method, as shown in Table 4, which is a table of influencing factors for each independent variable parameter:

[0168] Table 4. Influence Factors of Independent Variable Parameters

[0169] Parameter name Impact Factor Fixed volume / non-fixed volume 2.2 Water content rise type 3.7 Residual oil type 4.5 Energy assessment 2 Well-storage relationship 4 Production segment location 3.5 Storage type 2

[0170] The influence factors in Table 4 are preferably characterized by being plotted as a radar chart (e.g.) Figure 2 As shown in Table 4, radar charts are used to visually represent the influencing factors corresponding to each independent variable parameter. Figure 2 It can be clearly seen that the influence factor of the remaining oil type is 4.5, which has the highest influence. The influence factor of the well-reservoir relationship is 4, followed by the water cut rise type and the location of the producing layer. Therefore, the remaining oil type and the well-reservoir relationship can be regarded as the main control factors for well selection, and the water cut rise type and the location of the producing layer can be regarded as the secondary control factors. If necessary, the main control factors and the secondary control factors can be combined for analysis at the same time.

[0171] As described above, this embodiment of the invention uses the gas-to-oil conversion rate as the dependent variable parameter in a multiple linear regression model, and constant volume, non-constant volume, water cut increase type, remaining oil type, energy assessment, well-reservoir relationship, producing interval location, and reservoir type as the independent variable parameters in the multiple linear regression model, thereby selecting appropriate well selection parameters for faster and more accurate well selection. However, the selection of relevant parameters is not limited to those mentioned above; any parameter that affects well selection can be applied to the method for processing, thereby selecting appropriate well selection parameters for faster and more accurate well selection.

[0172] It should be understood that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Those skilled in the art can modify the technical solutions described in the above embodiments, or make equivalent substitutions for some of the technical features; and all such modifications and substitutions should fall within the protection scope of the appended claims of the present invention.

Claims

1. A method for screening parameters of oil wells in carbonate reservoirs, characterized in that, The carbonate reservoir oil well parameter screening method comprises: From the selected multiple geological parameters, dependent variable parameters and multiple independent variable parameters are screened out, and test values of all variable parameters of a plurality of known oil wells are obtained, a dimensionless regression model related to the dependent variable parameters and each independent variable parameter is established, and an expression of the dimensionless regression model is: wherein, non-dimensional variable value of dependent variable parameter of the i-th known oil well, non-dimensional variable value of the m-th independent variable parameter of the i-th known oil well, cm m non-dimensional regression coefficient before the m-th independent variable parameter, c0 is a constant term, μ i error term at the i-th test point; According to the test values of all variable parameters of the plurality of known oil wells, an estimated value of the dimensionless regression coefficient before each independent variable parameter in the dimensionless regression model is calculated by using a least square method; After the estimated value of the dimensionless regression coefficient is assigned to the dimensionless regression model, the dimensionless regression model is evaluated in terms of goodness of fit, confidence range and significance test, the dimensionless regression coefficient meeting the evaluation is determined as an influence factor; The determined influence factor is graphically characterized, a key influence factor is screened out according to an influence degree in the characterization, and the independent variable parameter corresponding to the key influence factor is determined as a key parameter for well selection.

2. The carbonate reservoir well parameter screening method according to claim 1, characterized in that, The method for establishing the dimensionless regression model related to the dependent variable parameters and each independent variable parameter comprises: According to the screened dependent variable parameters and corresponding independent variable parameters, a multiple linear regression model is established, and an expression of the multiple linear regression model is: y i = a0+ a1x 1,i +a2x 2,i +...+a m x m,i where y i is the test value of the dependent variable parameter for the ith known oil well, x m,i is the test value of the mth independent variable parameter for the ith known oil well, a m is the regression coefficient before the mth independent variable parameter, and a0 is the constant term. Each test value of all variable parameters of each known oil well is converted into a dimensionless variable value by using a dimensionless method, a plurality of dimensionless variable values are substituted into the multiple linear regression model, and the dimensionless regression model is reconstructed.

3. The carbonate reservoir well parameter screening method of claim 1, wherein, The method for calculating the estimated value of the dimensionless regression coefficient before each independent variable parameter by using the least square method comprises: A coefficient matrix of all variable parameters is established; A simultaneous equation for all variable parameters is established by using the least square method, and an expression of the simultaneous equation is: In the formula, n is the number of known oil wells; Each test value of all variable parameters of each known oil well is substituted into the simultaneous equation to solve the coefficient matrix, and the estimated value of the dimensionless regression coefficient before each independent variable parameter is estimated.

4. The carbonate reservoir well parameter screening method of claim 1, wherein, The method for evaluating the goodness of fit of the dimensionless regression model comprises: Based on the dimensionless regression model with the assigned dimensionless regression coefficient, each test value of the independent variable parameter of the known oil well is substituted into the dimensionless regression model, and an estimated value of the dependent variable parameter of each known oil well is calculated; According to the test value and the estimated value of the dependent variable parameter of each known oil well, a standard deviation of the dimensionless regression model is calculated, the goodness of fit of the dimensionless regression model is evaluated according to the difference value of the standard deviation, and an expression of the standard deviation is: where SE is the standard deviation, is the dimensionless estimate of the dependent variable parameter of the i-th known oil well, and n is the number of known oil wells, is the deviation of the test value and the estimate of the dependent variable parameter of the i-th known oil well.

5. The carbonate reservoir well parameter screening method of claim 4, wherein, The method for evaluating the confidence range of the dimensionless regression model comprises: According to the test values of all variable parameters of all known oil wells, a degree of freedom k-h of the dimensionless regression model is determined; a confidence level is selected, and a t value is found in a pre-determined table of t values based on the degrees of freedom k-h and the confidence level p ; According to the calculated standard deviation and the found t p value, a confidence interval of the non-dimensional regression model is calculated, the expression of which is: where Pr is the upper and lower limit of the confidence interval, k is the number of all known well test values corresponding to a single variable parameter, h is the number of all variable parameters, and t is the number of degrees of freedom k-h p is the number in the t-statistic table for the degrees of freedom k-h. The confidence range of the dimensionless regression model is evaluated according to the calculated confidence interval.

6. The carbonate reservoir well parameter screening method of claim 5, wherein, The method for evaluating the significance test of the dimensionless regression model comprises: F-test the significance of the dimensionless regression model according to the standard deviation of the dependent variable test value and the estimated value of the known oil well; t-test the significance of the dimensionless regression coefficient according to the test value of all the known oil wells corresponding to a single independent variable parameter; Determine whether the established dimensionless regression model and the calculated dimensionless regression coefficient are significant according to the results of F-test and t-test.

7. The carbonate reservoir well parameter screening method of claim 6, wherein, The method for F-testing the significance of the dimensionless regression model comprises: Optionally, two groups of the known oil wells are selected, and all the variable parameter test values of the two groups of the known oil wells are obtained respectively; The standard deviations of the dependent variable test value and the estimated value of the two groups of the known oil wells are calculated respectively; F-test the significance of the dimensionless regression model according to the two calculated standard deviations, and the expression of the F-test is: In the formula, SE and SE' are the standard deviations of the dependent variable test value and the estimated value of the two groups of the known oil wells respectively; According to the degrees of freedom k-h and the confidence level, the F-table value is found in the preset F-statistic value table, if the F-test result is less than the F-table value, it is evaluated that there is no significant difference in all the variable parameter test values of the two groups of the known oil wells in the dimensionless regression model, if the F-test result is greater than the F-table value, it is evaluated that there is a significant difference in all the variable parameter test values of the two groups of the known oil wells in the dimensionless regression model.

8. The carbonate reservoir well parameter screening method of claim 6, wherein, The method for t-testing the significance of the dimensionless regression coefficient comprises: Set all the test values of the known oil wells corresponding to a single independent variable parameter as a population, and randomly extract a number of test values of the known oil wells from the population as a sample; Select two independent variable parameters and establish a null hypothesis, and the expression of the null hypothesis is: H0: μ1= μ2 In the formula, H0 is the null hypothesis, μ1 is the population mean value corresponding to one of the independent variable parameters, μ2 is the population mean value corresponding to the other independent variable parameter, and μ1= μ2 means that there is no significant difference between the two population mean values; According to the overall average and the sample average, a statistical value t is calculated n ; determining the degrees of freedom d of a single said independent variable parameter from the volume of said sample f and from said degrees of freedom d f looking for the theoretical value t(d f ) in a pre-established table of values of t-statistics, said degrees of freedom d f being expressed by the formula: d f = n - 1 where d f is the degree of freedom for a single said independent variable parameter, n is the number of said known well test values contained in the sample; The calculated statistical value t n is compared with the found theoretical value t(d f ), and from the comparison result, at the given significance level, the probability P value is inferred that the statistical value t n exceeds or is lower than the theoretical value t(d f ). Compare the inferred probability P value with the given significance level, if the probability P value is less than the significance level, the established null hypothesis is rejected, and it is determined that there is a significant difference between the two population mean values, if the probability P value is greater than the significance level, the established null hypothesis is accepted, and it is determined that there is no significant difference between the two population mean values.

9. The carbonate reservoir well parameter screening method of claim 8, wherein, The method of calculating a statistical value t according to the overall average value and the sample average value n includes: If the degree of difference between a group of sample means and the population mean is to be evaluated, the statistical value t n The formula for calculating t is: wherein X i is the test value of the ith known oil well corresponding to a single said independent variable parameter, n is the number of said known oil well test values contained in the sample, t n is the statistical value, is the sample mean, SE is the sample standard deviation, and μ is the population mean. If the difference between the average values of the samples of the two groups is evaluated, the statistical value t n The formula for calculating the statistical value t is: In the formula, SE1 and SE2 are variances of two groups of samples respectively, n1 and n2 are numbers of the known oil well test values contained in the two groups of samples respectively, are average values of the two groups of samples respectively.

10. A method of geoselecting a well, the method comprising: The geological well selection method comprises selecting wells by using the key well selection parameters screened by the carbonate reservoir oil well parameter screening method according to any one of claims 1-9.

Images