Method for predicting initial production of directional well of heavy oil reservoir
Through rheology testing and integrated learning algorithms, combined with the starting pressure gradient and stress sensitivity effect, a mathematical model for initial production prediction of heavy oil reservoirs was established, which solved the problem of large prediction error of initial production directional wells in offshore heavy oil reservoirs, and achieved the More accurate and efficient predictions.
Patent Information
- Application Number
- CN202411105656.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-12
- Publication Date
- 2024-11-15
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology has large errors in the prediction of initial production of directional wells on offshore heavy oil reservoirs, and it is difficult to accurately evaluate the initial production of heavy oil reservoirs, which affects development efficiency and reserve upgrades.
Through rheology testing, the rheology mode of heavy oil fluid is analyzed, the motion equation is obtained, and the starting pressure gradient, stress sensitivity effect and epidermal coefficient is combined to establish a mathematical model of primary production prediction, and the model is trained using well logging curve data to construct the primary production diagram and Data sets, an integrated learning algorithm is used to build an intelligent prediction model and conduct initial production prediction.
It improves the accuracy and efficiency of the initial production forecast of directional wells in heavy oil reservoirs, reduces prediction errors, and can quickly and accurately evaluate the initial production of the same type of reservoir.
Smart Images

Figure CN118966071A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of oil reservoir output prediction, and in particular to a method for predicting initial production of a directional well in a heavy oil reservoir. Background Art
[0002] Offshore heavy oil reservoirs have large reserves, but there are many vertical layers and the reserves are dispersed. The economic benefits of using a single horizontal well type for development are low. In order to improve the oil displacement efficiency and recovery rate of offshore heavy oil reservoirs, a development method of combining horizontal wells with directional wells and irregular well patterns to produce multiple layers has been formed. This development method leads to a large error between the predicted value and the actual value of the initial production of directional wells in heavy oil reservoirs. The actual development effect of directional wells is far less than the pre-drilling prediction index, which brings great trouble to the formulation of development technology policies. Initial production evaluation is the most important link that restricts reserve upgrading and development and mobilization. How to effectively evaluate the initial production of directional wells in offshore heavy oil reservoirs is the key to achieving efficient development of offshore heavy oil reservoirs.
[0003] The main existing research methods for initial production prediction can be divided into four categories: (1) Initial production prediction method based on statistical analysis. However, this method has low applicability and is not suitable for early initial production evaluation and offshore heavy oil reservoirs with less production data. (2) Initial production analysis method based on empirical formula. This method uses relevant equations to infer future initial production based on development experience and production dynamic trends. The most widely used empirical analysis method is the modified Arps decline curve, but this method has large errors. (3) Analytical or semi-analytical initial production model method based on physical assumptions. This method has rigorous derivation ideas, but the existing model cannot reasonably characterize the actual seepage characteristics of heavy oil reservoirs and is difficult to apply in mines. (4) Methods based on numerical simulation usually use numerical simulation software to predict initial production. This method can consider the impact of multiple factors on initial production, but it relies on computing power, model accuracy and the accuracy of input data. This method is relatively cumbersome. Summary of the invention
[0004] The technical problem to be solved by the present invention is to provide a method for predicting the initial production of a directional well in a heavy oil reservoir.
[0005] The technical solution adopted by the present invention to solve the technical problem is: a method for predicting the initial production of a directional well in a heavy oil reservoir, comprising the following steps:
[0006] Analyze the rheological pattern of heavy oil fluids using rheological tests and obtain the equation of motion of heavy oil reservoirs;
[0007] Based on the motion equation combined with the starting pressure gradient, stress sensitivity effect and skin coefficient, a mathematical model for initial production prediction is obtained;
[0008] Using the logging curve data to train the initial production prediction mathematical model to obtain analysis results;
[0009] On the basis of the above analysis results, a preliminary production chart is constructed.
[0010] and / or,
[0011] Using the analysis results to establish an analysis data set; based on the analysis data set, using an integrated learning algorithm to build an initial production intelligent prediction model;
[0012] Obtain the heavy oil reservoir data to be predicted, and combine the initial production intelligent prediction model and / or the initial production chart to obtain the predicted initial production.
[0013] Preferably, the method of analyzing the rheological pattern of the heavy oil fluid by rheological testing to obtain the motion equation of the heavy oil reservoir includes:
[0014] Sampling the heavy oil fluid, and dehydrating the sample to obtain a heavy oil sample that can be used for rheological tests;
[0015] Using the rheological test to analyze the relationship between the shear stress and viscosity of the heavy oil sample at a formation temperature when the shear rate is 0.02 to 500 s-1;
[0016] According to the relationship between the shear stress and the viscosity, the rheological mode is obtained;
[0017] The equation of motion is determined based on the rheological model.
[0018] Preferably, the rheological mode is a power-law fluid; the equation of motion is a power-law fluid equation;
[0019] The mathematical model for initial production prediction is established based on the motion equation combined with the starting pressure gradient, the stress sensitivity effect and the skin coefficient, including:
[0020] Based on the power law fluid equation combined with the start-up pressure gradient and the stress sensitivity effect, a directional well plane radial flow equation is derived;
[0021] The directional well plane radial flow equation is simplified to obtain a multi-factor initial production equation;
[0022] The skin coefficient is introduced into the multi-factor first-birth equation to obtain the first-birth prediction mathematical model.
[0023] Preferably, the directional well plane radial flow equation is as follows:
[0024]
[0025] Where q is the production; λ is the starting pressure gradient; α k is the stress sensitivity coefficient; B ois the crude oil volume coefficient; N is the dimension conversion coefficient; K is the original permeability of the reservoir; μ eff is the effective viscosity of the power law fluid; w is the power law exponential coefficient; U is the seepage velocity; m is the formation pressure; R is the radial distance; h is the reservoir thickness, m i is the original formation pressure.
[0026] Preferably, the initial production prediction mathematical model includes a nonlinear seepage initial production formula for heavy oil power-law fluid; the nonlinear seepage initial production formula for heavy oil power-law fluid is as follows:
[0027]
[0028] Where Q is the initial production of the directional well; h j is the reservoir thickness at each measuring point; B o is the crude oil volume coefficient; K j is the reservoir permeability of each measuring point; μ eff is the effective viscosity of the power-law fluid; α kj is the stress sensitivity coefficient for each measuring point; j Start a pressure gradient for each measuring point; R e R is the formation supply radius; w is the wellbore radius; m w is the bottom hole pressure; m e is the boundary pressure; w is the power law exponent coefficient; S is the skin coefficient; J is the number of measuring points; q j The output for each measuring point.
[0029] Preferably, a core experiment is carried out on the heavy oil reservoir to obtain a stress sensitivity coefficient.
[0030] Preferably, the constructing of the initial production plate based on the analysis results comprises:
[0031] Acquire bottom hole flow pressure data and initial production data under different power law exponents in the analysis results; generate a first initial production chart using the bottom hole flow pressure data and the initial production data as horizontal and vertical coordinate axes;
[0032] Obtaining the effective thickness of the oil layer and the initial production data under different power law indices in the analysis results; generating a second initial production chart with the effective thickness of the oil layer and the initial production data as horizontal and vertical coordinate axes;
[0033] Obtaining the power law index and the initial production data at different oil layer thicknesses in the analysis results; generating a third initial production chart using the power law index and the initial production data as horizontal and vertical coordinate axes;
[0034] Obtaining the startup pressure gradient data and the initial production data under different production pressure differences in the analysis results; generating a fourth initial production chart using the startup pressure gradient data and the initial production data as horizontal and vertical coordinate axes;
[0035] The primary production plates include the first primary production plates, the third primary production plates, the second primary production plates and the fourth primary production plates.
[0036] Preferably, the method of constructing an intelligent prediction model for initial production by using an integrated learning algorithm based on the analysis data set includes:
[0037] Building a prediction model using the ensemble learning algorithm;
[0038] Dividing the analysis data set into a test set and a training set according to a preset ratio;
[0039] The prediction model is trained using the training set, and the prediction model is verified using the test set to obtain the first-production intelligent prediction model.
[0040] Preferably, the ensemble learning algorithm includes: random forest and gradient boosting algorithm.
[0041] Preferably, the analysis data set includes: porosity, permeability, crude oil volume coefficient, oil phase viscosity, reservoir thickness, bottom hole pressure, boundary pressure, wellbore radius, supply radius, power law exponent, skin coefficient, start-up pressure gradient, stress sensitivity coefficient, production pressure difference, and initial production of directional wells.
[0042] The implementation of the present invention has the following beneficial effects:
[0043] The present invention uses rheological testing to analyze the rheological pattern of heavy oil fluids to obtain the motion equation of the heavy oil reservoir; and based on the motion equation combined with the starting pressure gradient, stress sensitivity effect and skin coefficient, a mathematical model for initial production prediction is obtained. The logging curve data is combined with the mathematical model for initial production prediction to obtain an analysis result. According to the analysis result, an initial production plate and / or an analysis data set can be constructed; based on the analysis data set, an integrated learning algorithm is used to construct an initial production intelligent prediction model. The data of the heavy oil reservoir to be predicted is obtained, and the initial production is predicted in combination with the initial production intelligent prediction model and / or the initial production plate. By using multiple types of parameters and multiple algorithms, the present invention can achieve fast and accurate prediction when evaluating the initial production of directional wells in the same type of heavy oil reservoirs, thereby improving the accuracy of initial production prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments:
[0045] Figure 1 A flowchart of a method for predicting initial production of a directional well in one embodiment;
[0046] Figure 2 The relationship curves between bottom hole flowing pressure and initial production under different power law exponents;
[0047] Figure 3 The influence curves of effective oil layer thickness and initial production under different power law exponents;
[0048] Figure 4 It is the relationship curve between power law index and initial production under different oil layer thickness;
[0049] Figure 5 It is the relationship curve between the starting pressure gradient and the initial production under different production pressure differences;
[0050] Figure 6 Schematic diagram of the fitting accuracy of the intelligent prediction model for initial production of directional wells in heavy oil reservoirs based on ensemble learning. DETAILED DESCRIPTION
[0051] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described in detail with reference to the accompanying drawings.
[0052] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.
[0053] The embodiment of the present invention provides a method for predicting the initial production of a directional well in a heavy oil reservoir. Figure 1 As shown, the directional well initial production prediction method includes the following steps:
[0054] Rheological tests are used to analyze the rheological pattern of heavy oil fluids and obtain the motion equation of heavy oil reservoirs.
[0055] Rheological testing is an important means to analyze and understand the flow behavior of heavy oil fluids in reservoirs. Through rheological testing, the rheological properties of heavy oil fluids can be obtained, including their viscosity, plasticity, and behavior under different shear stresses and shear rates.
[0056] Rheological tests are usually performed using a rheometer, which can measure the rheological curve of heavy oil under different temperature and pressure conditions, that is, the relationship between shear stress and shear rate. Different temperature conditions and shear rates can be selected according to the differences in oil fields and surface temperatures.
[0057] According to the rheological test results, the rheological mode of heavy oil is analyzed. There are three rheological modes: Bingham type, power law type and generalized Bingham type. The rheological curves of the three rheological modes are as follows:
[0058] θ=θ 0 +α p y (1)
[0059] θ=sα w (2)
[0060] θ=θ 0 +sy w (3)
[0061] Where, θ is the shear stress, N / m2; θ 0 is the yield stress, y is the shear rate, s-1; α p is the plastic viscosity, s is the consistency coefficient; w is the power law index, which characterizes the degree of non-Newtonianity of the fluid.
[0062] Using the rheological test results, combined with the geological and engineering data of the reservoir, a mathematical model describing the flow of heavy oil in the reservoir, i.e., the equation of motion, can be established. For non-Newtonian fluids, a modified Darcy's law or a more complex nonlinear seepage model needs to be used. In one embodiment, the fluid type of the oil field reservoir is a power law type, and the corresponding equation of motion for the fluid is:
[0063]
[0064] μ b =μ eff U w-1
[0065] Where w is the power law exponent coefficient; μb is the apparent viscosity of the power law fluid; μ eff is the effective viscosity of the power-law fluid; U is the seepage velocity; λ is the starting pressure gradient; R is the radial distance; and m is the formation pressure.
[0066] Temperature increases usually lead to a decrease in the viscosity of heavy oil, while changes in pressure also affect the flow behavior of the fluid. The flow of heavy oil in the reservoir can be more accurately predicted by the state of the fluid at the formation pressure and the preset temperature (e.g., power-law fluid apparent viscosity, power-law fluid effective viscosity).
[0067] Through the above content, rheological testing can be used more comprehensively to analyze the rheological pattern of heavy oil fluids and obtain the motion equation of heavy oil reservoirs, providing a scientific basis for the effective development of heavy oil.
[0068] Based on the equation of motion combined with the starting pressure gradient, stress sensitivity effect and skin coefficient, a mathematical model for initial production prediction was obtained.
[0069] Specifically, the starting pressure gradient is the minimum pressure gradient that needs to be overcome when heavy oil flows. The stress sensitivity effect can be expressed by the stress sensitivity effect coefficient, which is used to describe the sensitivity of permeability to pressure changes. The skin coefficient is a parameter of the flow resistance near the well wall. The skin coefficient is added to the motion equation of the oil field to consider the flow resistance near the well wall. The initial production prediction mathematical model is a mathematical structure used to simulate the initial production of heavy oil reservoirs. It can be a set of equations, formulas, algorithms or statistical relationships.
[0070] By adding the starting pressure gradient, stress sensitivity effect and skin coefficient to the equation of motion, it can be ensured that it can accurately reflect the production performance of heavy oil reservoirs.
[0071] The initial production prediction mathematical model is trained using logging curve data to obtain analysis results.
[0072] It should be noted that the logging curve data are single-well logging curves. The characteristics of underground rocks, fluids and oil and gas reservoirs can be analyzed and interpreted through a variety of single-well logging curves.
[0073] Based on the analysis results, a first-time production chart is constructed, and / or, an analysis data set is established using the analysis results. Based on the analysis data set, an integrated learning algorithm is used to construct a first-time production intelligent prediction model.
[0074] In a feasible example, a first-birth chart is constructed based on the analysis results and an analysis data set is established using the analysis results. Based on the analysis data set, an integrated learning algorithm is used to construct a first-birth intelligent prediction model.
[0075] In one practical example, a preliminary production plate is constructed based solely on the analysis results.
[0076] In a feasible example, only the analysis results are used to establish an analysis data set, and an integrated learning algorithm is used to build an initial production intelligent prediction model based on the analysis data set.
[0077] Obtain the heavy oil reservoir data to be predicted, and combine it with the initial production intelligent prediction model and / or the initial production chart to obtain the predicted initial production.
[0078] Specifically, the initial production chart can quickly evaluate the production potential of oil wells and the dynamic characteristics of oil and gas reservoirs, verify and calibrate reservoir models, and quickly predict the initial production of reservoirs. The initial production intelligent prediction model can further improve the accuracy of predicting initial production based on the initial production chart.
[0079] The present invention constructs a new initial production model of directional wells in heavy oil reservoirs by comprehensively considering multiple factors such as heavy oil power-law fluid, vertical reservoir heterogeneity, starting pressure and stress sensitivity, thereby achieving rapid prediction and evaluation of initial production of directional wells and improving work efficiency and accuracy of initial production.
[0080] In some executable embodiments, rheological testing is used to analyze the rheological pattern of the heavy oil fluid to obtain the motion equation of the heavy oil reservoir, including:
[0081] The heavy oil fluid is sampled and the sample is dehydrated to obtain a heavy oil sample that can be used for rheological tests.
[0082] First, the rheological properties of heavy oil (such as viscosity and plasticity) are affected by the water content. Dehydration can ensure that the test results accurately reflect the true rheological behavior of the oil sample. Secondly, the water may be unevenly distributed in the heavy oil, affecting the homogeneity of the sample. Dehydration helps to obtain a more uniform sample, thereby improving the repeatability of the experimental results. Through dehydration, a purer and more stable heavy oil sample can be obtained, providing more accurate data for rheological testing.
[0083] Rheological testing is used to analyze the relationship between shear stress and viscosity of heavy oil samples at formation temperature when the test shear rate is 0.02 to 500 s-1.
[0084] Formation temperature refers to the temperature conditions of underground oil and gas reservoirs. This temperature gradually increases with the increase of underground depth, usually about 25℃ to 30℃ for every 100 meters of descent. This gradient is called geothermal gradient. By testing heavy oil samples at formation temperature, the production performance of oil wells can be predicted more accurately.
[0085] The rheological model is obtained based on the relationship between shear stress and viscosity.
[0086] Specifically, the rheological behavior of heavy oil can be analyzed by plotting a graph of shear stress and shear rate. If the curve passes through the origin, it indicates that the heavy oil exhibits Newtonian fluid properties and its viscosity does not change with shear rate. If the curve does not pass through the origin or shows a nonlinear relationship, it indicates that the heavy oil exhibits non-Newtonian fluid properties and its viscosity changes with shear rate.
[0087] Determine the equation of motion based on the rheological model.
[0088] It should be noted that for non-Newtonian fluids, a suitable rheological model is selected to describe its flow characteristics. Common models include power-law fluid model, Bingham plastic fluid model, Casson-Yates model, etc. According to the principle of conservation of mass, a continuity equation is established to describe the change in fluid mass. According to Newton's second law and the principles of fluid mechanics, a momentum equation is established to describe the change in fluid momentum, which may include Darcy's law or nonlinear seepage law. According to actual conditions, additional effects such as gravity, capillary pressure, temperature changes, etc. are considered, and the equation of motion is adjusted accordingly. For non-Newtonian fluids, especially heavy oil, it is necessary to introduce the concept of starting pressure gradient, that is, the minimum pressure difference that must be overcome before the fluid starts to flow. The above physical relationship is converted into a mathematical expression to form a complete equation of motion.
[0089] In some executable embodiments, the rheological mode is a power-law fluid; and the equation of motion is a power-law fluid equation.
[0090] Specifically, the heavy oil in some heavy oil reservoirs conforms to the description of the power law model. Therefore, the heavy oil motion equation corresponding to the heavy oil reservoir is the power law fluid equation. The power law fluid equation can accurately describe the non-Newtonian flow characteristics of heavy oil, that is, the law of viscosity change with shear rate. The power law equation flexibly reflects the shear thinning or shear thickening behavior of heavy oil through two parameters, the flow index and the viscosity coefficient. It is crucial to predict the flow of heavy oil in the reservoir and help optimize the recovery strategy. Secondly, the simplicity of the power law model makes it easy to apply in the initial production data model of the reservoir, which improves the calculation efficiency and practicality of the model.
[0091] Based on the equation of motion combined with the starting pressure gradient, stress sensitivity effect and skin coefficient, a mathematical model for initial production prediction is established, including:
[0092] The plane radial flow equation of directional well is derived based on the power law fluid equation combined with the starting pressure gradient and stress sensitivity effect.
[0093] Specifically, the radial flow equation of the directional well plane is as follows:
[0094]
[0095] Where: q is the production; λ is the starting pressure gradient; α k is the stress sensitivity coefficient; B o is the crude oil volume coefficient, dimensionless; N is the dimension conversion coefficient; K is the original permeability of the reservoir; m i is the original formation pressure.
[0096] The multi-factor initial production equation is obtained by simplifying the plane radial flow equation of directional wells.
[0097] Introduction but And order The directional well plane radial flow equation is further simplified to obtain:
[0098]
[0099] Where, ξ is the introduced pressure conversion coefficient; i The supply frontier is R = R e The corresponding pressure conversion coefficient.
[0100] The above formula is a typical first-order nonlinear differential equation. Solving this equation yields:
[0101]
[0102] Among them, α k is the stress sensitivity coefficient, λ is the starting pressure gradient, (α k λ) 2 <<1, the above formula can be simplified to
[0103]
[0104] Consider α k λ<<1, and the boundary condition R=R e hour, The constant term can be eliminated, resulting in:
[0105]
[0106] Then, the initial production equation of directional wells in heavy oil reservoirs considering various factors is as follows:
[0107]
[0108] Among them, q j Calculate the yield for each measuring point; λ j Start the pressure gradient for each measuring point; α kj is the stress sensitivity coefficient of each measuring point; m w is the bottom hole pressure of the directional well; B o is the crude oil volume coefficient; R e is the formation supply radius; e is the natural logarithm.
[0109] The skin coefficient was introduced into the multi-factor first-birth equation to obtain a mathematical model for first-birth prediction.
[0110] Introducing the skin effect can significantly improve the accuracy of oil well production prediction. The skin coefficient reflects the flow resistance in the wellbore and near-wellbore area, which may be caused by factors such as wax deposition, contamination or damage on the wellbore wall. Considering the skin effect helps to simulate the actual production of the oil well more realistically, optimize the completion and production strategies, and reduce unexpected production losses.
[0111] In some executable embodiments, the directional well plane radial flow equation is as follows:
[0112]
[0113] Where q is the production; λ is the starting pressure gradient; α k is the stress sensitivity coefficient; B o is the crude oil volume coefficient; N is the dimension conversion coefficient; K is the original permeability of the reservoir; μ eff is the effective viscosity of the power-law fluid; w is the power-law exponent coefficient; U is the seepage velocity; m is the formation pressure; and R is the radial distance.
[0114] In some executable embodiments, the initial production prediction mathematical model includes a nonlinear seepage initial production formula for heavy oil power-law fluid; the nonlinear seepage initial production formula for heavy oil power-law fluid is as follows:
[0115]
[0116] Where Q is the initial production of the directional well; h j is the reservoir thickness at each measuring point; B o is the crude oil volume coefficient; K j is the reservoir permeability of each measuring point; μ eff is the effective viscosity of the power-law fluid; α kj is the stress sensitivity coefficient for each measuring point; j Start a pressure gradient for each measuring point; R e R is the formation supply radius; w is the wellbore radius; m w is the bottom hole pressure; m e is the boundary pressure; w is the power law exponent coefficient; S is the skin coefficient; J is the number of measuring points; q j The output for each measuring point.
[0117] In some executable embodiments, core experiments are conducted on heavy oil reservoirs to obtain stress sensitivity coefficients.
[0118] It should be noted that the purpose of conducting core experiments on heavy oil reservoirs to determine the stress sensitivity coefficient is to evaluate the impact of formation pressure changes on the permeability of the oil layer. The stress sensitivity coefficient is a performance parameter of the stress sensitivity effect. By applying different pressure conditions to core samples in the laboratory and observing the changes in their permeability, the sensitivity of the oil layer to stress changes can be measured. The stress sensitivity coefficient is a key parameter used to adjust the dynamic behavior of permeability in the numerical model of the reservoir and reflect the impact of the drop in formation pressure on the permeability of the oil layer during the production process. Considering the impact of the drop in formation pressure on the permeability of the oil layer when predicting the initial production of the reservoir will help to more accurately evaluate the production potential and dynamics of the oil well, thereby reducing development risks and improving the return on investment.
[0119] In some executable embodiments, constructing a preliminary production plate based on the analysis results includes:
[0120] Obtain bottom hole flow pressure data and initial production data under different power law indices in the analysis results. Generate a first initial production chart using the bottom hole flow pressure data and initial production data as horizontal and vertical coordinate axes. In one embodiment, Figure 2 As shown in the figure, assuming that other parameters remain constant, the smaller the power law index n is, the lower the directional well production is and the faster the initial production decreases; the lower the bottom hole pressure is, the greater the production pressure difference is, and the higher the initial production is.
[0121] Obtain the effective thickness and initial production data of the oil layer under different power law indexes in the analysis results; and generate a second initial production chart with the effective thickness and initial production data of the oil layer as the horizontal and vertical axes. In one embodiment, Figure 3 As shown in the figure, when other parameters remain constant, the initial production of the directional well will increase with the increase of the power law index n; when the effective thickness of the oil layer increases, the initial production increases rapidly.
[0122] The power law index and initial production data under different oil layer thicknesses in the analysis results are obtained; and the third initial production chart is generated with the power law index and initial production data as the horizontal and vertical axes. In one embodiment, Figure 4 As shown in the figure, when other parameters are constant, at the same oil layer thickness, the smaller the power law index n is, the smaller the initial production of the directional well is; as the power law index increases, the greater the increase in the initial production of the directional well is, and the greater the oil layer thickness is, the greater the increase in the initial production of the directional well with thickness is.
[0123] The startup pressure gradient data and initial production data under different production pressure differences in the analysis results are obtained; and the fourth initial production chart is generated using the startup pressure gradient data and initial production data as the horizontal and vertical coordinate axes. In one embodiment, Figure 5 As shown in the figure, other parameters remain unchanged. Under the same production pressure difference, the initial production of the directional well decreases as the starting pressure gradient increases.
[0124] The first production plates include the first first production plate, the third first production plate, the second first production plate and the fourth first production plate.
[0125] In some executable embodiments, using the analysis results to establish an analysis data set includes:
[0126] Standardize the analysis results and well logging data to unify the formats and units of different data sources. According to the need to predict the initial production of heavy oil reservoirs, select appropriate data structures and database systems to store the analysis results and well logging data in an orderly manner. In addition, create indexes and metadata for the analysis data set to facilitate data retrieval and understand the context of the data. Finally, establish access and update mechanisms to ensure that the analysis data set can be continuously updated and improved as new data is acquired. This process improves the availability of data.
[0127] In some executable embodiments, an integrated learning algorithm is used to construct an initial production intelligent prediction model based on the analysis of the data set, including:
[0128] Use ensemble learning algorithms to build prediction models.
[0129] The analysis data set is divided into a test set and a training set according to a preset ratio.
[0130] The prediction model is trained using the training set and verified using the test set to obtain the first-production intelligent prediction model.
[0131] In one embodiment, 80% of the data is used as a training set, and the remaining 20% of the data is used as a test set to evaluate the predictive performance of the model. The mean square error of the initial production intelligent prediction model is calculated to be 16.38, and the coefficient of determination is 0.976. Among them, the coefficient of determination is an important statistical indicator in regression analysis, which is used to measure the degree of fit of the model to the data; the mean square error is an indicator to measure the difference between the model prediction value and the actual observation value. In this embodiment, the comparison scatter plot of the actual initial production value and the predicted initial production of the heavy oil reservoir is shown in the figure below: Figure 6 shown.
[0132] In some executable embodiments, the ensemble learning algorithm includes: random forest and gradient boosting algorithm.
[0133] Specifically, random forest improves accuracy by building multiple decision trees and integrating their predictions, naturally handles interactions between features, and has good resistance to overfitting. The gradient boosting algorithm minimizes the loss function by gradually adding weak prediction models, which can capture the data patterns involved in the initial production prediction of heavy oil reservoirs. It has strong adaptability and performs well in a variety of prediction tasks. Both can effectively process and analyze the data of the data set, thereby improving the performance of the model.
[0134] In some executable embodiments, the analysis data set includes: porosity, permeability, crude oil volume coefficient, oil phase viscosity, reservoir thickness, bottom hole pressure, boundary pressure, wellbore radius, supply radius, power law index, skin coefficient, starting pressure gradient, stress sensitivity coefficient, production pressure difference, and initial production of directional wells.
[0135] In an executable embodiment, the actual data of directional well W1 in a heavy oil reservoir of an oil field is analyzed. Directional well W1 produces three layers: Zone 1, Zone 2 and Zone 3. The vertical heterogeneity of these layers is obvious, which has a significant impact on the initial production of the oil well. The relevant reservoir parameters are as follows: density ρ is 0.968kg / m3, porosity Φ is 0.26, and the average permeability K is 638mD, of which the permeability of Zone 1 layer is 423mD, Zone 2 is 254mD, and Zone 3 is 738mD. Crude oil volume coefficient B o is 1.02, and the viscosity of crude oil is 43.2 MPa·s. The total effective thickness of the oil layer is 11.5 m, of which the effective thickness of Zone 1 is 3.5 m, the effective thickness of Zone 2 is 3.0 m, and the effective thickness of Zone 3 is 5.0 m. The production pressure difference is 7 MPa, the wellbore radius Rw is 0.1 m, the supply radius Re is 200 m, and the power law exponent w is 0.98.
[0136] Before any measures were taken to increase production, the daily production of the well was only 16.9m3 / d. After the use of the fixed tubing acidification and plugging removal technology, the daily production was significantly increased to 55.8m3 / d, and the production status was relatively stable. The initial production calculated using the conventional initial production prediction formula was 82.5m3 / d, which is quite different from the actual initial production. In order to accurately predict the initial production, the present invention uses an improved initial production prediction method based on the logging curve data, and the calculated initial production is 58.2m3 / d, with an error of about 4.3%, which is close to the actual initial production, and the error is significantly reduced, demonstrating its superiority and high accuracy.
[0137] In a feasible embodiment, in order to simulate the changes in different reservoir conditions, key parameters such as porosity, permeability, oil layer thickness, starting pressure gradient, stress sensitivity, production pressure difference, etc. can be systematically adjusted to generate multiple variable scenarios. Thereby evaluating the impact of changes in reservoir characteristics on initial production. The porosity is set in the range of 0.15 to 0.35, the permeability is from 200mD to 700mD, the oil layer thickness is from 2 meters to 35 meters, the starting pressure gradient is from 0 to 0.045MPa / m, the stress sensitivity coefficient is from 0 to 0.0010251 / MPa, and the production pressure difference is from 2 to 20MPa. Each parameter is reasonably divided according to the actual distribution. Through this method, 7500 independent sample points are calculated and generated. Each sample point includes a complete set of input parameters and the corresponding initial production calculation results. Part of the data set is shown in the following table, where Φ is the porosity; K is the original permeability of the reservoir; B o is the crude oil volume coefficient; μ o is the oil phase viscosity; h is the reservoir thickness; m w is the bottom hole pressure; m e Displacement boundary pressure; R w is the wellbore radius; R e is the formation supply radius; w is the power law exponential coefficient; S is the skin coefficient; λ is the starting pressure gradient; α k is the stress sensitivity coefficient; Δm is the production pressure difference; Q is the initial production of the directional well.
[0138]
[0139]
[0140] The above embodiments only express the preferred implementation modes of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the patent scope of the present invention. It should be pointed out that, for ordinary technicians in this field, the above technical features can be freely combined without departing from the concept of the present invention, and several deformations and improvements can be made, which all belong to the protection scope of the present invention. Therefore, all equivalent changes and modifications made to the scope of the claims of the present invention should belong to the coverage of the claims of the present invention.
Claims
1. A method for predicting initial production of a directional well in a heavy oil reservoir, characterized in that: The following steps are involved: Analyze the rheological pattern of heavy oil fluids using rheological tests and obtain the equation of motion of heavy oil reservoirs; Based on the motion equation combined with the starting pressure gradient, stress sensitivity effect and skin coefficient, a mathematical model for initial production prediction is obtained; Using the logging curve data to train the initial production prediction mathematical model to obtain analysis results; On the basis of the above analysis results, a preliminary production chart is constructed. and / or, Using the analysis results to establish an analysis data set; based on the analysis data set, using an integrated learning algorithm to build an initial production intelligent prediction model; Obtain the heavy oil reservoir data to be predicted, and combine the initial production intelligent prediction model and / or the initial production chart to obtain the predicted initial production.
2. The method for predicting initial production of a directional well according to claim 1, characterized in that: The method of analyzing the rheological pattern of the heavy oil fluid by rheological testing to obtain the motion equation of the heavy oil reservoir includes: Sampling the heavy oil fluid, and dehydrating the sample to obtain a heavy oil sample that can be used for rheological tests; Using the rheological test to analyze the relationship between the shear stress and viscosity of the heavy oil sample at a formation temperature when the shear rate is 0.02 to 500 s-1; According to the relationship between the shear stress and the viscosity, the rheological mode is obtained; The equation of motion is determined based on the rheological model.
3. The method for predicting initial production of a directional well according to claim 1, characterized in that: The rheological mode is a power-law fluid; the motion equation is a power-law fluid equation; The mathematical model for initial production prediction is established based on the motion equation combined with the starting pressure gradient, the stress sensitivity effect and the skin coefficient, including: Based on the power law fluid equation combined with the start-up pressure gradient and the stress sensitivity effect, a directional well plane radial flow equation is derived; The directional well plane radial flow equation is simplified to obtain a multi-factor initial production equation; The skin coefficient is introduced into the multi-factor first-birth equation to obtain the first-birth prediction mathematical model.
4. The method for predicting initial production of a directional well according to claim 3, characterized in that: The radial flow equation of the directional well plane is as follows: Where q is the production; λ is the starting pressure gradient; α k is the stress sensitivity coefficient; B o is the crude oil volume coefficient; N is the dimension conversion coefficient; K is the original permeability of the reservoir; μ eff is the effective viscosity of the power law fluid; w is the power law exponential coefficient; U is the seepage velocity; m is the formation pressure; R is the radial distance, h is the reservoir thickness, m i is the original formation pressure.
5. The method for predicting initial production of a directional well according to claim 3, characterized in that: The mathematical model for initial production prediction includes a nonlinear seepage initial production formula for heavy oil power-law fluid; the nonlinear seepage initial production formula for heavy oil power-law fluid is as follows: Among them, Q is the initial production of the directional well; h j is the reservoir thickness at each measuring point; B o is the crude oil volume coefficient; K j is the reservoir permeability of each measuring point; μ eff is the effective viscosity of the power-law fluid; α kj is the stress sensitivity coefficient of each measuring point; j Start a pressure gradient for each measuring point; R e R is the formation supply radius; w is the wellbore radius; m w is the bottom hole pressure of the directional well; m e is the boundary pressure; w is the power law exponent coefficient; S is the skin coefficient; J is the number of measuring points; q j The output for each measuring point.
6. The method for predicting initial production of a directional well according to claim 3, characterized in that: A core experiment is carried out on the heavy oil reservoir to obtain a stress sensitivity coefficient.
7. The method for predicting initial production of a directional well according to claim 1, characterized in that: The constructing of the initial production plate based on the analysis results includes: Acquire bottom hole flow pressure data and initial production data under different power law exponents in the analysis results; generate a first initial production chart using the bottom hole flow pressure data and the initial production data as horizontal and vertical coordinate axes; Obtaining the effective thickness of the oil layer and the initial production data under different power law indices in the analysis results; generating a second initial production chart with the effective thickness of the oil layer and the initial production data as horizontal and vertical coordinate axes; Obtaining the power law index and the initial production data at different oil layer thicknesses in the analysis results; generating a third initial production chart using the power law index and the initial production data as horizontal and vertical coordinate axes; Obtaining the startup pressure gradient data and the initial production data under different production pressure differences in the analysis results; generating a fourth initial production chart using the startup pressure gradient data and the initial production data as horizontal and vertical coordinate axes; The primary production plates include the first primary production plates, the third primary production plates, the second primary production plates and the fourth primary production plates.
8. The method for predicting initial production of a directional well according to claim 1, characterized in that: The method of constructing a first-birth intelligent prediction model based on the analysis data set using an integrated learning algorithm includes: Building a prediction model using the ensemble learning algorithm; Dividing the analysis data set into a test set and a training set according to a preset ratio; The prediction model is trained using the training set, and the prediction model is verified using the test set to obtain the first-production intelligent prediction model.
9. The method for predicting initial production of a directional well according to claim 1, characterized in that: The ensemble learning algorithm includes: random forest and gradient boosting algorithm.
10. The method for predicting initial production of a directional well according to claim 1, characterized in that: The analysis data set includes: porosity, permeability, crude oil volume coefficient, oil phase viscosity, reservoir thickness, bottom hole pressure, boundary pressure, wellbore radius, supply radius, power law index, skin coefficient, start-up pressure gradient, stress sensitivity coefficient, production pressure difference, and initial production of directional wells.
Citation Information
Patent Citations
Oil reservoir complex structure well test interpretation method and device, electronic equipment and medium
CN113657046A
Low-permeability reservoir polluted well productivity calculation method, electronic equipment and storage medium
CN114510847A
Cited By
Self-adaptive control system for oil field equipment
CN119717555A
Offshore common heavy oil reservoir horizontal well cold production exploitation well release boundary optimization method
CN122174750A
