Polymer flooding whole-process recoverable reserve prediction method

Abstract

The invention discloses a polymer flooding whole-process recoverable reserve prediction method, which comprises comprehensive water content prediction of a target block, and the comprehensive water content prediction method comprises the following steps: determining an analogy block of the target block, according to the water content-time curve of the analogy block, four stages of blank water drive, water content reduction, water content recovery and subsequent water drive are divided; taking the water content at the end of the blank water drive period of the analogy block as an initial water content starting point, utilizing a support vector machine algorithm, taking geological indexes including a comprehensive water content index and development indexes of the analogy block as training feature data, and obtaining an SVR prediction model through SVR regression; using the SVR prediction model to solve the comprehensive water content of the target block in the water content decline stage and the water content rise stage; the method effectively solves the problems that the mathematical model of the polymer underground non-power law type viscoelastic fluid is difficult to express and the recoverable reserves in the polymer flooding injection stage are difficult to accurately predict.

Description

Technical Field

[0001] The present disclosure relates to the field of oil extraction, and particularly to a method for determining recoverable reserves in the polymer flooding stage in the field of reservoir engineering. Background Art

[0002] The statements in this section only provide background information related to the present disclosure and do not constitute prior art.

[0003] The basic displacement principle of polymer flooding in oilfields is to inject a viscoelastic fluid with a relatively large mobility, thereby increasing the swept volume and oil displacement efficiency. In terms of calculating the recoverable reserves of polymer flooding, there are currently four main types of calculation methods:

[0004] One is the method of increasing oil production per ton of polymer used in the early stage. By analyzing the injection polymer dynamics results of adjacent blocks, analogizing the effect of increasing oil production per ton of polymer in adjacent blocks, and combining the analysis of the target polymer injection object, the additional recoverable reserves are estimated. This method has strong subjective experience and poor operability.

[0005] The second is the parallel recursion correction method of water drive characteristic curves to estimate the additional recoverable reserves of polymer flooding. This method requires using modified displacement curves at different stages of polymer injection, considering changes in relative permeability curves during polymer injection, etc. It relies on a large amount of work such as relative permeability processing and research, and the calculation method is relatively complex with poor intuitive practicality.

[0006] The third is the numerical simulation method. A numerical model of polymer drive dynamics needs to be established. After conducting historical fitting of water drive, the production indicators of polymer flooding are predicted to obtain the recoverable reserves of polymer flooding. The modeling and numerical simulation cycle is long, and the parameters required for simulation are numerous and complex, which is not conducive to the rapid calculation of the recoverable reserves of polymer flooding and cannot meet the needs of a large number of on-site polymer flooding and recoverable reserves calibration work.

[0007] The entire process of polymer flooding can usually be divided into four stages: blank water drive stage, water cut decline stage, water cut recovery stage, and subsequent water drive stage. In the blank water drive and subsequent water drive stages, experts and scholars generally believe that it conforms to the water drive law, and the water drive curve method can be used, that is, the recoverable reserves are predicted according to the change law of water drive indicators. However, polymers are non-Newtonian viscoelastic fluids, and their underground working state is different from that of water drive development. Obviously, the above prediction methods are no longer applicable. Therefore, it is necessary to develop a prediction method for recoverable reserves development indicators suitable for polymer flooding.

[0008] It should be noted that the information disclosed in the above background art section is only used to enhance the understanding of the background of the present disclosure, and thus may include information that does not constitute prior art. Summary of the Invention

[0009] In view of this, the present disclosure provides a method for predicting the recoverable reserves throughout the polymer flooding process, which solves the problem that the existing calculation method is no longer applicable when calculating the recoverable reserves of polymer flooding by using the variation law of water flooding indicators, but the polymer is a non-power-law viscoelastic fluid and its underground working state is different from that of water flooding development.

[0010] To achieve the above-mentioned invention purpose, the method for predicting the recoverable reserves throughout the polymer flooding process includes the comprehensive water cut prediction of the target block, and is characterized in that the method for the comprehensive water cut prediction includes:

[0011] Determine the analog block of the target block, and divide the blank water flooding, water cut decline, water cut rise and subsequent water flooding into four stages according to the water cut-time curve of the analog block;

[0012] According to the comprehensive water cut variation law of the analog block in the blank water flooding and subsequent water flooding stages, predict the comprehensive water cut of the target block in the corresponding stages by the water flooding curve method;

[0013] Taking the water cut at the end of the blank water flooding of the analog block as the initial water cut starting point, using the support vector machine algorithm, taking the geological indicators and development indicators including the comprehensive water cut index of the analog block as training feature data, through SVR regression, obtain the SVR prediction model, and use the SVR prediction model to solve the comprehensive water cut of the target block in the water cut decline and water cut rise stages.

[0014] In the present disclosure and possible embodiments, in the training feature data, take the comprehensive water cut index of the analog block as the dependent variable, and take the geological indicators and development indicators of the analog block as the independent variables.

[0015] In the present disclosure and possible embodiments, perform the SVR regression by using SPSSPRO or MATLAB.

[0016] In the present disclosure and possible embodiments, perform the SVR regression by selecting a kernel function and adjusting the corresponding parameters. The kernel functions include linear kernel function, polynomial kernel function, Gaussian kernel function and Sigmoid kernel function.

[0017] In the present disclosure and possible embodiments, use the comprehensive water cut results of the target block in the four stages, and according to the liquid production level of each stage given by the reservoir engineering analysis, obtain the oil production results of the target block in the four stages, and determine the calculation result of the recoverable reserves throughout the polymer flooding process according to the oil production profile.

[0018] In the present disclosure and possible embodiments, the comprehensive water cut calculation formulas of the target block in the blank water flooding and subsequent water flooding stages include:

[0019] Log(Wp) = a + b * Np;

[0020] Where: Wp - cumulative water production; Lp - cumulative liquid production; Np - cumulative oil production; a, b - constants;

[0021] F w甲型 ’ = N * b * f w *(1 - f w );

[0022] Where: Fw’ - water cut rising rate; N - geological reserves; fw - water cut;

[0023] From the type A water drive curve formula, the water cut rising rates in the blank water drive and subsequent water drive stages are obtained, and the comprehensive water cut in the blank water drive and subsequent water drive stages is calculated through the water cut rising rate.

[0024] The present disclosure has the following beneficial effects:

[0025] The polymer flooding whole-process recoverable reserves prediction method of the present disclosure, on the basis of clarifying the differences in recoverable reserves prediction between polymer flooding and water flooding development, through introducing the support vector machine method, makes full use of the characteristics of mathematical statistics, algorithm learning and high fitting prediction accuracy of this method, and establishes a whole-process polymer flooding recoverable reserves calculation method different from the water drive curve and decline curve methods based on statistical regression in conventional water flooding; specifically, using the support vector machine algorithm, taking geological indicators and development index data such as injection volume, polymer dosage, cumulative injection volume, etc. in the polymer injection stage of the target block as input training feature data, taking the water cut index as training result data, fitting the curve change trend of the standard analog block by selecting a suitable kernel function and adjusting the corresponding parameters, and solving the comprehensive water cut in the water cut decline and water cut rising stages of the target block; compared with methods such as multiple linear regression, the fitting accuracy is higher, and the accurate prediction of the comprehensive water cut in the water cut recovery and water cut decline stages of polymer flooding can be realized; the method of the present invention, using the method idea of polymer flooding whole-cycle recoverable reserves prediction of analogy - learning - prediction, solves the difficult problems in the research field of the whole-stage / cycle prediction method of polymer flooding, innovatively forms a polymer flooding stage-by-stage recoverable reserves prediction method with strong operability, high prediction accuracy and programmable implementation, has very strong practicability, and realizes the high-quality calculation and evaluation of the recoverable reserves of polymer flooding and even ASP flooding. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Through the description of the embodiments of the present disclosure with reference to the following drawings, the above and other objects, features and advantages of the present disclosure will become clearer. In the drawings:

[0027] Figure 1 is the flow chart of the polymer flooding recoverable reserves prediction method of the embodiment of the present disclosure;

[0028] Figure 2 It is the polymer flooding water cut - time variation curve of the standard analog block in Block A of the eastern oilfield in the embodiment of the present disclosure;

[0029] Figure 3 It is the SVM fitting parameter setting diagram implemented by programming in the embodiment of the present disclosure;

[0030] Figure 4 It is the polymer flooding water cut prediction of Block B in the eastern oilfield in the embodiment of the present disclosure;

[0031] Figure 5 It is the trend chart of the oil production change after fitting in Block B of the eastern oilfield in the embodiment of the present disclosure. Detailed implementation manners

[0032] The following is a description of the present disclosure based on embodiments. However, it should be noted that the present disclosure is not limited to these embodiments. In the following detailed description of the present disclosure, some specific details are described in detail. However, for the parts that are not described in detail, those skilled in the art can also fully understand the present disclosure.

[0033] In addition, those of ordinary skill in the art should understand that the provided drawings are only for illustrating the purpose, features, and advantages of the present disclosure, and the drawings are not actually drawn to scale. At the same time, unless the context clearly requires, the words "including", "comprising", and other similar words throughout the specification and claims should be interpreted as having an inclusive meaning rather than an exclusive or exhaustive meaning; that is, it is the meaning of "including but not limited to".

[0034] Figure 1 It is the flow chart of the polymer flooding recoverable reserve prediction method in the embodiment of the present disclosure; as shown by Figure 1 The polymer flooding recoverable reserve prediction method includes the following steps:

[0035] I. Basic data preparation:

[0036] According to the conventional technical means in the art and combined with expert experience, select the block with similar conditions to the target block as the standard analog block; obtain the geological development index data sets such as reservoir physical properties, development methods, well pattern well spacing, injection pressure, injection volume, etc. of the standard analog block and the comprehensive water cut as the dependent variable, and establish the dependent variable and independent variable index sets for support vector machine regression.

[0037] II. Prediction of comprehensive water cut and liquid production in stages:

[0038] The whole process of polymer flooding can be divided into four stages: blank water flooding stage, water cut decline stage, water cut recovery stage, and subsequent water flooding stage.

[0039] 1. Prediction of comprehensive water cut in stages. The specific prediction method is as follows:

[0040] (1) For the blank water flooding stage and the subsequent water flooding stage, experts and scholars generally believe that it conforms to the water flooding law. Therefore, the comprehensive water cut prediction can be carried out according to reservoir engineering methods such as conventional water flooding curves and decline curves. Specifically, based on the conventional technical means in this field, the water cut rising law is determined according to the data of the analog block, so as to analogously obtain the prediction results of the comprehensive water cut index in the blank water flooding stage and the subsequent water flooding stage of the target block.

[0041] Specifically, the method for obtaining the comprehensive water cut in the blank water flooding and subsequent water flooding stages is as follows:

[0042] These two stages conform to the law of water flooding displacement characteristic curves. Based on the analysis of the water cut change law of the analog block, the water cut rising rate of the target block is calculated by using the Type A or Type C water flooding curve. Given the initial water cut, the water cut change law in the blank water flooding and subsequent water flooding stages is obtained. Taking the Type A water flooding curve as an example, the specific calculation formulas include:

[0043] Log(Wp) = a + b * Np;

[0044] Where: Wp - cumulative water production; Lp - cumulative liquid production; Np - cumulative oil production; a, b - constants;

[0045] F w甲型 ’ = N * b * f w *(1 - f w );

[0046] Where: Fw’ - water cut rising rate; N - geological reserves; fw - water cut;

[0047] From the Type A water flooding curve formula, the water cut rising rates in the blank water flooding and subsequent water flooding stages are obtained, and the comprehensive water cut in the blank water flooding and subsequent water flooding stages is calculated through the water cut rising rate.

[0048] (2) For the water cut decline stage and the water cut recovery stage, since the underground working principle of the polymer is different from the water flooding development law, the reservoir engineering method based on the water flooding law is no longer applicable.

[0049] For these two stages of polymer flooding, the present invention introduces the support vector machine regression algorithm, and makes full use of the characteristics of data analysis, algorithm learning, and controllable and optimizable fitting prediction accuracy of this algorithm to establish a comprehensive water cut index prediction method for the water cut decline and water cut recovery stages of polymer flooding. Specifically, the polymer flooding parameter indicators of the analog block (including multiple geological indicators and development indicators such as comprehensive water cut) are used as the input training feature data. By selecting an appropriate kernel function and adjusting control parameters, using SPSSPRO or MATLAB software, the comprehensive water cut of the target block in the water cut decline and water cut rise stages is solved; then, given the total liquid production volume of the whole process, combined with the water cut prediction results of each stage, the cumulative oil production of the whole process is obtained, which is the recoverable polymer reserve of the target block.

[0050] Specifically, the process of solving the comprehensive water cut of the target block in the water cut decline and water cut rise stages in the embodiments of the present disclosure is as follows:

[0051] ① Establish a support vector machine regression data sample set:

[0052] According to the conventional technical means in the art, using the polymer flooding injection-production static and dynamic data tables of multiple analog blocks as the data basis, a support vector machine regression training sample set composed of geological indicators (such as effective thickness, saturation pressure, effective permeability, crude oil viscosity, etc.) and development indicators (such as recovery degree before polymer injection, well spacing, polymer molecular weight, polymer concentration, etc.) is established.

[0053] ② Support vector machine fitting preparation:

[0054] Taking the use of the common SPSSPRO support vector machine regression (SVR) module as an example, first select the comprehensive water cut as the dependent variable Y, and use the support vector machine regression training sample set established in step ① as the independent variable X for training. All variables are quantitative variables.

[0055] ③ Support vector machine fitting prediction:

[0056] Set the support vector machine regression model parameters. The most important one is the selection of the kernel function, which generally includes linear kernel function (linear), polynomial kernel function (poly), Gaussian kernel function (rbf), and Sigmoid kernel function. After selecting a certain kernel function, the software program starts data analysis operations and establishes a primary regression model.

[0057] Apply the established support vector machine (SVR) primary regression model to the training and test data to obtain model evaluation (usually MSE / RMSE / MAPE / R 2The results (etc.). Since Support Vector Machine (SVR) regression cannot obtain a definite equation like traditional models and usually evaluates the model by the prediction accuracy of test data, an optimized SVR model is obtained through trial calculations for different combinations of training sets and kernel functions. Although this SVR model is trained and tested using data from analogous blocks, due to the similarity between the analogous blocks and the target block, it is defaulted that the target block can also use this SVR model for comprehensive water cut prediction, thereby obtaining the comprehensive water cut in the water cut decline and water cut recovery stages of the target block.

[0058] The predicted results of the comprehensive water cut indicators in the above blank water flooding stage, water cut decline stage, water cut recovery stage, and subsequent water flooding stage are the predicted water cut results for each stage of the target block.

[0059] 2. Prediction of liquid production in stages:

[0060] According to the conventional techniques in this field, reservoir engineering analysis and injection-production dynamic analysis are carried out on the target block to obtain the liquid production levels in each stage. Based on the liquid production levels in each stage of the target block, the predicted results of the liquid production in each stage of the target block are obtained.

[0061] III. Output of recoverable reserves

[0062] Through the predicted results of the water cut in the target block and the predicted results of the liquid production in the target block, the oil production profiles in the four stages of the target block are obtained. Through this oil production profile, the cumulative oil production in each stage can be calculated, and the sum of the cumulative oil production in each stage is the recoverable reserves of the whole process of the target block.

[0063] Taking Block B of polymer flooding in an eastern oilfield of China as an example, the polymer flooding recoverable reserves prediction method of the present disclosure embodiment is specifically used to predict the recoverable reserves of Block B.

[0064] Specifically, the dynamic production data of Block A of polymer flooding in an eastern oilfield of China is selected as the standard analogous block. The geological conditions of Block B are similar to those of Block A, and the driving mode and the types of chemical agents used are the same. Therefore, the SVR method based on sample data learning can be adopted to realize the dynamic prediction of the whole process of polymer flooding. The specific process is as follows:

[0065] 1. According to the dynamic and static data of Block A, establish a support vector machine training sample set:

[0066] Due to the characteristics of data analysis, as many geological indicators and development index parameters of Block A as possible need to be obtained to establish a support vector machine regression training sample set composed of geological indicators and development indicators, as shown in Table 1. In addition, a water cut - time change curve can be drawn, such as Figure 2 shown, as a reference for the staged water cut change pattern of Block B.

[0067] Table 1 Geological and Development Index Sets of Block A in the Eastern Oilfield

[0068]

[0069] 2. Apply reservoir engineering methods and support vector machine regression methods to conduct water cut prediction throughout the process of Block B:

[0070] In the blank water flooding and subsequent water flooding stages, apply reservoir engineering methods (water flooding curve method), and refer to the water cut rising law of Block A to predict the water cut index of Block B (the water cut prediction in the subsequent water flooding stage reaches 98%).

[0071] In the polymer injection stage, that is, the water cut decline stage and the water cut recovery stage, based on the support vector machine regression training sample set of Block A, obtain the support vector machine regression prediction model. Take the support vector machine regression training sample set of Block A as the input training feature data (as Figure 3 shown), through kernel function selection and parameter adjustment, and obtain the optimized prediction model through multiple trials.

[0072] Then, use this optimized prediction model to conduct comprehensive water cut prediction for Block B, and the results are as Figure 4 shown.

[0073] 3. Calculation results of recoverable reserves under a given liquid volume scale:

[0074] According to the conventional technical means in this field, combine the water cut prediction results of each stage obtained in Step 2 with the liquid production volume scale of 4 stages in Block B obtained through reservoir engineering analysis and analogy. After obtaining the liquid production level and the comprehensive water cut change law in each stage, the oil production in each stage can be calculated. Finally, add up the cumulative oil production in each stage to obtain the recoverable reserves of the entire polymer flooding process in Block B. The specific results are shown in Table 2 and Figure 5 :

[0075] Table 2 Calculation Results of Stage-by-Stage Recoverable Reserves in Block B of the Eastern Oilfield

[0076]

[0077] As can be seen from the specific description of the embodiments of the present disclosure, the present method can achieve the overall fitting of the comprehensive water cut during the two important stages of the water cut decline and the water cut recovery in polymer flooding, and the fitting accuracy of this example is above 95%. First, it well solves the problems of the aforementioned method of increasing oil production per ton of polymer and the method of correcting the water drive curve, which lack a theoretical basis and cannot accurately give the recoverable reserves in each stage; second, compared with the aforementioned numerical simulation method, the data indicators required by the present method are much fewer, there is no need to establish a huge data model, and the SVR method has a fast operation speed and can quickly and accurately give the recoverable reserves in stages, effectively solving the problem of the long calculation cycle (generally 3-6 months) of modeling and history matching in the numerical simulation method, and is more suitable for on-site application in oil fields.

[0078] The above-described embodiments are only for expressing the implementation manners of the present disclosure, and the description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present disclosure. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present disclosure, several deformations, equivalent substitutions, improvements, etc. can be made, and these all belong to the protection scope of the present disclosure. Therefore, the protection scope of the patent of the present disclosure shall be subject to the appended claims.

Claims

1. A method for predicting the recoverable reserves throughout the polymer flooding process, including the comprehensive water cut prediction of the target block, characterized in that, The method for comprehensive water cut prediction includes: Determine the analogous block of the target block, and divide it into four stages: blank water drive, water cut decline, water cut rise, and subsequent water drive according to the water cut-time curve of the analogous block; According to the comprehensive water cut change law of the analogous block in the blank water drive and subsequent water drive stages, predict the comprehensive water cut of the target block in the corresponding stages by the water drive curve method; Taking the water cut at the end of the blank water drive of the analogous block as the initial water cut starting point, using the support vector machine algorithm, taking the geological indicators and development indicators including the comprehensive water cut index of the analogous block as training feature data, through SVR regression, obtain the SVR prediction model, and use the SVR prediction model to solve the comprehensive water cut of the target block in the water cut decline and water cut rise stages.

2. The method for predicting the recoverable reserves throughout the polymer flooding process according to claim 1, characterized in that: In the training feature data, take the comprehensive water cut index of the analogous block as the dependent variable, and take the geological indicators and development indicators of the analogous block as the independent variables.

3. The method for predicting the recoverable reserves throughout the polymer flooding process according to claim 1 or 2, characterized in that: Use SPSSPRO or MATLAB for the SVR regression.

4. The method for predicting the recoverable reserves throughout the polymer flooding process according to claim 3, characterized in that: Perform the SVR regression by selecting the kernel function and adjusting the corresponding parameters. The kernel functions include linear kernel function, polynomial kernel function, Gaussian kernel function, and Sigmoid kernel function.

5. The method for predicting the recoverable reserves throughout the polymer flooding process according to claim 1, 2 or 4, characterized in that: Using the comprehensive water cut results of the target block in the four stages, according to the liquid production level in each stage given by reservoir engineering analysis, calculate the oil production results of the target block in the four stages, and determine the calculation result of the recoverable reserves in the whole process of polymer flooding according to the oil production profile.

6. The method for predicting the recoverable reserves throughout the polymer flooding process according to claim 5, characterized in that, The comprehensive water cut calculation formula of the target block in the blank water drive and subsequent water drive stages includes: Log(Wp) = a + b * Np; Where: Wp - cumulative water production; Lp - cumulative liquid production; Np - cumulative oil production; a, b - constants; F w甲型 ’ = N * b * f w * (1 - f w ); Where: Fw’ - water cut rise rate; N - geological reserves; fw - water cut; From the Type A water drive curve formula, obtain the water cut rise rates in the blank water drive and subsequent water drive stages, and calculate the comprehensive water cut in the blank water drive and subsequent water drive stages through the water cut rise rate.

Images