Methods, apparatus, equipment and media for optimizing the trajectory of fractured-vuggy reservoir wells
By optimizing the well trajectory of fractured-vuggy reservoirs using the unsupervised machine learning algorithm KNN, the problems of low efficiency and poor adaptability in existing technologies are solved, enabling rapid and efficient well trajectory design that meets the requirements of drilling safety and high productivity.
Patent Information
- Application Number
- CN202510630990.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-16
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies for optimizing well trajectories in fractured-vuggy reservoirs suffer from low efficiency, difficulty in adapting to complex geological conditions, lack of real-time data support and intelligent algorithm applications, resulting in slow drilling design speed and low efficiency, making it difficult to meet the needs of high-efficiency optimization.
The unsupervised machine learning algorithm KNN is used to optimize well trajectory based on a three-dimensional geological model of fractured-vuggy reservoirs. The optimal well trajectory is quickly determined by combining objective functions of production capacity, well leakage, and drilling safety factors.
It achieves rapid and efficient well trajectory optimization, improving design efficiency by 24 times. It is highly adaptable, capable of quantitatively evaluating multiple objectives, and meets the requirements of drilling safety and high productivity.
Smart Images

Figure CN123116759A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of oil and gas development, and more specifically, to a method for optimizing the trajectory of fractured-vuggy reservoir wells, a device for optimizing the trajectory of fractured-vuggy reservoir wells, and equipment and computer-readable storage medium for implementing the method for optimizing the trajectory of fractured-vuggy reservoir wells. Background Technology
[0002] The optimization design of well trajectories in fractured-vuggy reservoirs is of great significance in oil and gas resource development. It must not only aim at gas well productivity but also consider drilling safety and speed. Currently, while various technologies and methods exist, they have significant limitations. For example, many focus on manual design using geological models and maps, or on optimizing and comparing multiple schemes using numerical simulation techniques. These methods suffer from drawbacks in practical applications, including a single objective, slow speed, and low efficiency. Existing methods often only consider reservoir engineering aspects, such as the length, location, and orientation of horizontal sections, neglecting the complexity of drilling operations and the uncertainty of geological conditions, making it difficult for schemes to adapt to complex geology. Furthermore, the technical means are mostly based on traditional geological analysis and numerical simulation, lacking real-time data support and intelligent algorithm applications, making it difficult to quickly generate and optimize well trajectories, and failing to meet the needs of efficient design and optimization. For example, in Peng Mengyun's paper "Research on Horizontal Well Trajectories Optimization in Fractured-Vuggy Carbonate Reservoirs," published on June 1, 2014, multiple geostress models were established based on finite element numerical simulation, requiring multiple manual parameter adjustments for comparison. Furthermore, this process requires multiple iterations, resulting in low simulation efficiency for complex fracture-vuggy combinations and making it difficult to meet the needs of rapid on-site decision-making. The article "Key Technologies for Exploration and Development of Deep Fractured-Vuggy Carbonate Condensate Gas Reservoirs—Taking the Tazhong No. 1 Gas Field in the Tarim Basin as an Example," published by Yang Haijun et al. on March 4, 2020, utilizes fracture-vuggy body sculpting and quantitative evaluation, which is a traditional manual design and cannot meet the requirements for efficient production. The article "Seismic Model and Well Trajectory Design for High-Yield Wells in Tight Channel Sandstone Gas Reservoirs in the Central Sichuan Basin," published by Zhu Xun on September 25, 2020, mainly relies on research based on physical simulation experiments and fracture extension laws of tight reservoir fracturing to ensure that the main extension direction of hydraulic fractures in tight reservoirs is consistent with the direction of the maximum horizontal principal stress; however, this still largely depends on human experience.
[0003] Therefore, there is a need to develop more efficient, intelligent and comprehensive optimization design technologies to improve the effectiveness of well trajectory optimization design for fractured-vuggy reservoirs. Summary of the Invention
[0004] The purpose of this invention is to address at least one of the aforementioned shortcomings of the prior art. For example, one objective of this invention is to provide a more efficient method for optimizing the trajectory of fractured-vuggy reservoir wells.
[0005] To achieve the above objectives, the present invention provides a method for optimizing the trajectory of fractured-vuggy reservoir wells.
[0006] The method for optimizing the trajectory of fractured-vuggy reservoir wells includes the following steps:
[0007] S1. Establish a three-dimensional geological model of fractured-vuggy reservoirs.
[0008] S2. Preprocess the three-dimensional geological model of the fractured-vuggy reservoir to establish a training dataset.
[0009] S3. Train an unsupervised machine learning model using the training dataset, and use the unsupervised machine learning model to identify the optimal well trajectory corresponding to the maximum cumulative value of the objective function. The objective function is an objective function based on reservoir attribute parameters and includes one or more of the following: production capacity factors, well leakage factors, and drilling safety factors.
[0010] In an exemplary embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention, the unsupervised machine learning model may be a KNN model, and the reservoir attribute parameters may include matrix porosity mΦ. i Matrix permeability mk i Porosity fvΦ i and the permeability of the fissure fvk i .
[0011] In an exemplary embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention, the K value of the KNN model can be 10 to 50.
[0012] In an exemplary embodiment of the well trajectory optimization method for fractured-vuggy reservoirs of the present invention, the three-dimensional geological model of the fractured-vuggy reservoir may include basic parameters, well trajectory data of the pilot well, and reservoir geological data.
[0013] The basic parameters may include a preset grid size.
[0014] The well trajectory data of the pilot well may include drilling depth coordinates d1 i Actual vertical depth coordinate z i East-west offset coordinate x i and north-south offset coordinates y i .
[0015] The reservoir geological data may include the grid length, width, and height of the reservoir in the work area, the formation coordinates, and the matrix porosity (mΦ). i Matrix permeability mk i Porosity fvΦ i And the permeability of the fissures, fvk.
[0016] In an exemplary embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention, step S2 may include:
[0017] S21. Confirm the center coordinates of the grid in the three-dimensional geological model of the fractured-vuggy reservoir.
[0018] S22. Confirm the correspondence between the center point coordinates of the grid and the reservoir attribute parameters, and confirm the geological data volume based on the correspondence. The geological data volume includes the center point coordinates of the grid and the reservoir attribute parameters.
[0019] S23. Interpolate the geological data volume to obtain the reservoir discrete grid model data volume.
[0020] S24. Establish a training dataset based on the reservoir discrete grid model data volume.
[0021] In an exemplary embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention, step S3 may include:
[0022] S31. Confirm the starting point coordinates and ending point coordinates of the target well, establish the straight line equation of the starting point coordinates and ending point coordinates, and confirm the well trajectory point data volume based on the straight line equation.
[0023] S32. Import the center point coordinates of the grid into the unsupervised machine learning model, use the well trajectory point data volume as the search point set to obtain the well trajectory, and calculate the cumulative value of the objective function corresponding to the well trajectory.
[0024] S33. Repeat steps S31 to S32 to obtain multiple well trajectories and multiple corresponding cumulative values of objective functions. Select the well trajectory corresponding to the maximum cumulative value of objective functions as the optimal well trajectory.
[0025] In an exemplary embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention, the linear equation may be:
[0026] X(t) = X1 + t·(X2 - X1)
[0027] Y(t) = Y1 + t·(Y2 - Y1)
[0028] Z(t) = Z1 + t·(Z2 - Z1)
[0029] Where (X(t),Y(t),Z(t)) are well trajectory points, t is a parameter, and the value range can be [0,1]. When t=0, (X1,Y1,Z1) are the starting point coordinates, and when t=1, (X2,Y2,Z2) are the ending point coordinates.
[0030] In another aspect, the present invention provides a well trajectory optimization device for fractured-vuggy reservoirs. The device comprises, in sequence, a three-dimensional geological model establishment module for fractured-vuggy reservoirs, a training dataset establishment module, and an optimal well trajectory confirmation module.
[0031] Among them, the module for establishing a three-dimensional geological model of fractured-vuggy reservoirs is configured to establish a three-dimensional geological model of fractured-vuggy reservoirs.
[0032] The training dataset creation module is configured to preprocess the three-dimensional geological model of the fractured-vuggy reservoir and create a training dataset.
[0033] The optimal well trajectory confirmation module is configured to train an unsupervised machine learning model using the training dataset, and use the unsupervised machine learning model to confirm the optimal well trajectory corresponding to the maximum cumulative value of the objective function. The objective function is an objective function based on reservoir attribute parameters and includes one or more of the following: production capacity factors, well leakage factors, and drilling safety factors.
[0034] In another aspect, the present invention provides a computer device, the computer device comprising:
[0035] Processor; memory storing a computer program that, when executed by the processor, implements the fractured-vuggy reservoir well trajectory optimization method as described above.
[0036] In another aspect, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the fractured-vuggy reservoir well trajectory optimization method as described above.
[0037] Compared with the prior art, the beneficial effects of the present invention include at least one of the following:
[0038] (1) The present invention is based on block grid construction, and the algorithm is fast and efficient and has strong adaptability to complex wellbore trajectories.
[0039] (2) The present invention encrypts the regular grid after spatial coordinate point positioning of field data such as well trajectory data and geological data.
[0040] (3) This invention uses an unsupervised machine learning algorithm to optimize well trajectory with the goal of combining production capacity and drilling safety.
[0041] (4) The regularized grid used in this invention has a precision of up to the meter level, which is 5-10 times higher than that of the geological model.
[0042] (5) Conventional methods require about 3-5 days to optimize the trajectory of a well, while this invention can complete the process in 1 hour, increasing efficiency by more than 24 times.
[0043] (6) Conventional methods can only qualitatively consider a single objective, while this invention can quantitatively evaluate multiple objectives, making it more widely applicable. Attached Figure Description
[0044] The above and other objects and / or features of the present invention will become clearer from the following description taken in conjunction with the accompanying drawings, in which:
[0045] Figure 1 A flowchart illustrating an exemplary embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention is shown.
[0046] Figure 2 A schematic diagram of the raw data of the Petrel geological model is shown as an embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention.
[0047] Figure 3 A schematic diagram of the regularized and refined preprocessing permeability model is shown in an embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention.
[0048] Figure 4 The diagram illustrates the optimization results and smoothing process of a directional wellbore trajectory according to an embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention.
[0049] Figure 5 A schematic diagram of directional well trajectory optimization (λ=0.1) is shown in an embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention.
[0050] Figure 6 A schematic diagram of directional well trajectory optimization (λ=0.5) is shown in an embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention.
[0051] Figure 7 The diagram illustrates the production prediction of corresponding well trajectories under different constraints in an embodiment of the well trajectory optimization method for fractured-vuggy reservoirs of the present invention. Detailed Implementation
[0052] In the following sections, the method, apparatus, equipment, and medium for optimizing the trajectory of fractured-vuggy reservoir wells of the present invention will be described in detail with reference to exemplary embodiments.
[0053] It should be noted that the terms "first," "second," "third," etc., and "S1," "S2," "S3," etc., used in this invention are for ease of description and distinction only, and should not be construed as indicating or implying relative importance or describing a specific order or sequence. The terms "connected" and "linked" should be interpreted broadly; for example, they can refer to a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.
[0054] To address the shortcomings of fractured and vulnerable carbonate gas reservoirs, where qualitative well trajectory design using geological maps or models suffers from limited objectives, slow speed, and low efficiency, this invention utilizes the unsupervised machine learning algorithm KNN to automatically acquire geological characteristic parameters of spatial data points. It establishes an objective function based on permeability and porosity, providing a new and more efficient auxiliary method for well trajectory optimization design, achieving goals such as high oil and gas well production and drilling safety. This invention can rapidly optimize the trajectories of highly deviated and horizontal wells, meeting different production targets and well leakage constraints, and is suitable for gas reservoirs with strong heterogeneity due to fracture and vulnerabilities, abnormally high pressure, and high drilling risks. To address the aforementioned problems, this invention provides a well trajectory optimization method for fractured and vulnerable reservoirs.
[0055] In a first exemplary embodiment of the fractured-vuggy reservoir well trajectory optimization method of the present invention, as follows: Figure 1 As shown, the trajectory optimization method for fractured-vuggy reservoir wells includes the following steps:
[0056] S1. Establish a three-dimensional geological model of fractured-vuggy reservoirs.
[0057] Optionally, the 3D geological model of a fractured-vuggy reservoir may include basic parameters, well trajectory data of the pilot well, and reservoir geological data. The basic parameters may include grid size, formation temperature, original formation pressure, viscosity and density of formation oil and water, and apparent molecular weight of gas. The well trajectory data of the pilot well may include drilling depth coordinates d1. i Actual vertical depth coordinate z i East-west offset coordinate x i and north-south offset coordinates y i Reservoir geological data may include the grid length, width, and height of the reservoir in the work area, as well as stratigraphic coordinates, matrix porosity, matrix permeability, fracture-vuggy porosity, and fracture-vuggy permeability.
[0058] S2. Preprocess the three-dimensional geological model of fractured-vuggy reservoirs to establish a training dataset.
[0059] Optionally, step S2 may include:
[0060] S21. Confirm the center coordinates of the grid in the three-dimensional geological model of the fractured-vuggy reservoir.
[0061] S22. Confirm the correspondence between the center point coordinates of the grid and the reservoir attribute parameters. Based on the correspondence, confirm the geological data volume, which includes the center point coordinates of the grid and the reservoir attribute parameters.
[0062] S23. Interpolate the geological data volume to obtain the reservoir discrete grid model data volume.
[0063] S24. Establish a training dataset based on the reservoir discrete grid model data volume.
[0064] S3. Train an unsupervised machine learning model using the training dataset, and use the unsupervised machine learning model to identify the optimal well trajectory corresponding to the maximum cumulative value of the objective function. The objective function is an objective function based on reservoir attribute parameters and includes one or more of the following: production capacity factors, well leakage factors, and drilling safety factors.
[0065] Optionally, the unsupervised machine learning model can be the KNN model.
[0066] Optionally, the K value of the KNN model can be 10 to 50, for example, 10, 20, 30, 40, or 50.
[0067] Optionally, reservoir property parameters may include matrix porosity mΦ i Matrix permeability mk i Porosity fvΦ i and the permeability of the fissure fvk i .
[0068] Optionally, step S3 may include:
[0069] S31. Confirm the starting point coordinates and ending point coordinates of the target well, establish the straight line equation of the starting point coordinates and ending point coordinates, and confirm the well trajectory point data volume based on the straight line equation.
[0070] Alternatively, the equation of the line can be:
[0071] X(t) = X1 + t·(X2 - X1)
[0072] Y(t) = Y1 + t·(Y2 - Y1)
[0073] Z(t) = Z1 + t·(Z2 - Z1)
[0074] Where (X(t),Y(t),Z(t)) are well trajectory points, t is a parameter, and the value range can be [0,1]. When t=0, (X1,Y1,Z1) are the starting point coordinates, and when t=1, (X2,Y2,Z2) are the ending point coordinates.
[0075] S32. Import the center point coordinates of the grid into the unsupervised machine learning model, use the well trajectory point data volume as the search point set to obtain the well trajectory, and calculate the cumulative value of the objective function corresponding to the well trajectory.
[0076] S33. Repeat steps S31 to S32 to obtain multiple well trajectories and multiple corresponding cumulative values of objective functions. Select the well trajectory corresponding to the maximum cumulative value of objective functions as the optimal well trajectory.
[0077] In another aspect, the present invention provides a second exemplary embodiment of a fractured-vuggy reservoir well trajectory optimization device.
[0078] The well trajectory optimization device for fractured-vuggy reservoirs includes a three-dimensional geological model building module for fractured-vuggy reservoirs, a training dataset building module, and an optimal well trajectory confirmation module, which are connected in sequence.
[0079] The module for establishing a 3D geological model of fractured-vuggy reservoirs is configured to create such a model. The module for establishing a training dataset is configured to preprocess the 3D geological model of the fractured-vuggy reservoirs and create a training dataset. The module for confirming the optimal well trajectory is configured to train an unsupervised machine learning model using the training dataset and then use the unsupervised machine learning model to confirm the optimal well trajectory corresponding to the maximum cumulative value of the objective function. The objective function is based on reservoir attribute parameters and includes one or more of the following: productivity factors, well leakage factors, and drilling safety factors.
[0080] In another aspect, the present invention provides a third exemplary embodiment of a computer device. The computer device includes a processor and a memory. The memory stores a computer program. The computer program is executed by the processor, causing the processor to execute the computer program of the fractured-vuggy reservoir well trajectory optimization method according to the present invention.
[0081] In another aspect, the present invention provides a fourth exemplary embodiment of a computer-readable storage medium storing a computer program. The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to execute the method for optimizing the trajectory of fractured-vuggy reservoir wells according to the present invention. The computer-readable recording medium is any data storage device capable of storing data readable by a computer system. Examples of computer-readable recording media include: read-only memory, random access memory, read-only optical disk, magnetic tape, floppy disk, optical data storage device, and carrier waves (such as data transmission via the Internet through wired or wireless transmission paths).
[0082] To better understand the exemplary embodiments of the present invention described above, further descriptions are provided below in conjunction with specific embodiments and accompanying drawings, but the examples given are not intended to limit the present invention.
[0083] Example 1
[0084] In this embodiment, the well trajectory optimization method for fractured-vuggy reservoirs can utilize KNN to automatically identify the characteristics of spatial data points to determine the well trajectory direction, mainly including the following four steps:
[0085] Step 1: Establish a three-dimensional geological model of the fracture-vuggy reservoir.
[0086] a) Basic parameters: Set the grid size l.
[0087] b) Obtain well trajectory data for the pilot well: Well trajectory data includes drilling depth coordinates d1 i Actual vertical depth coordinate zi East-west offset coordinate x i North-south offset coordinates y i The data contains a subscript i to indicate that there are multiple data points (i.e., well trajectory data coordinate points), and is stored in the computer as a list.
[0088] c) Reservoir geological data: Obtain the 3D geological modeling data volume of the work area (if the 3D geological modeling data volume of the work area is exported by PETREL software, its file format is .GRDECL). The data volume contains geometric parameters such as grid length, width, height, and formation coordinates of the reservoir in the current work area, as well as geological parameters such as matrix porosity, matrix permeability, fracture-vuggy porosity, and fracture-vuggy permeability.
[0089] Step 2: Further regularization and refinement preprocessing of the 3D pore-fracture-vuggy reservoir geological model using Python, specifically including:
[0090] Export reservoir geological data volumes (in .GRDECL format) of the study area from geological modeling software such as PETREL, and then use the Python-based PyGRDECL library to convert the .GRDECL data volumes into VTK data volumes (e.g., .vtu). It is important to note that both .GRDECL and .vtu file formats use corner coordinates for grid construction.
[0091] One hexahedral mesh contains 8 vertices (p1, p2, p3, p4, p5, p6, p7, p8), p i =(x i ,y i ,z i The center coordinates p of all grid cells can be calculated from the coordinates of the 8 vertices. ci :p ci = 1 / 8 × (p1 + p2 + p3 + p4 + p5 + p6 + p7 + p8). In the processed data volume, the coordinates of the center point p of the grid are... ci (x i ,y i ,z i ), matrix porosity mΦ i Matrix permeability mk i Porosity fvΦ i and the permeability of the fissure fvk i A one-to-one correspondence, meaning a center point coordinate p ci (x i ,y i ,z i ), corresponding to a set of attribute values (mΦ) i ,mk i,fvΦ i fvk i ).
[0092] The size of the reservoir discrete grid is determined by the grid size *l*, and regular grid data is created using the Python-based NumPy library. The preprocessed geological data volume (x...) i y i , z i ,mΦ i ,mk i ,fvΦ i fvk i Importing Python, the `griddata` function of the `scipy.interpolate` module is used for interpolation to transform irregular PETREL geological data volumes into regular grid volumes, resulting in a reservoir discrete grid model data volume (.CSV file, containing the interpolated x...). i y i , z i ,mΦ i ,mk i ,fvΦ i fvk i (Attribute data volume). This method is very flexible and can handle interpolation problems of multi-dimensional and spatially irregular data volumes. It only requires input data of a suitable shape and a target grid. At the same time, it increases the density of the data through interpolation, thereby increasing the density of PETREL geological data and improving the accuracy of the model.
[0093] Step 3: Set the initial well trajectory coordinates and use the KNN algorithm to search for neighboring grids in each direction to establish an objective function based on permeability and porosity.
[0094] a) Calculation of well trajectory point coordinates. Given the well length L, the coordinates of the horizontal well's endpoint (X2, Y2, Z2) within a 360-degree radius of the starting point (X1, Y1, Z1) can be calculated. Taking one endpoint coordinate as an example, given the starting point (X1, Y1, Z1) and endpoint (X2, Y2, Z2) of the horizontal well, the equation of the straight line between the two points can be fitted:
[0095] X(t) = X1 + t·(X2 - X1)
[0096] Y(t) = Y1 + t·(Y2 - Y1)
[0097] Z(t) = Z1 + t·(Z2 - Z1)
[0098] Where (X(t),Y(t),Z(t)) are the coordinates of the well trajectory points, and t is a parameter that can take values in the range of [0,1]. When t=0, (X1,Y1,Z1) are the coordinates of the starting point, and when t=1, (X2,Y2,Z2) are the coordinates of the ending point.
[0099] Based on the above linear equation, the well trajectory point data volume (X) in any direction can be calculated. i ,Y i Z i ).
[0100] b) Calculate the objective function for the grid attributes corresponding to the well trajectory using KNN. KNN (K-Nearest Neighbors) is a commonly used unsupervised machine learning algorithm for classifying or regressing data based on feature similarity. In the KNN algorithm, K represents the number of neighbors, and the K nearest data points are used for decision-making. The KNN algorithm determines the nearest neighbors by calculating the distance between data points; therefore, the search range is defined by the feature space of the data points. Typically, the search range cannot be directly specified; instead, it is controlled by choosing an appropriate value for K. Based on the characteristics of KNN, different values of K can be specified to determine the grid refinement range.
[0101] Based on the unsupervised machine learning KNN algorithm, the center point coordinates (x, y, y) of the reservoir discrete grid model data volume in the .CSV file obtained in the second step are used to... i y i , z i Importing data points into the KNN model, using well trajectory point data volume (X) i ,Y i Z i The search point set is set, and the K value can be set to a range of 10 to 50. The search results are the coordinate numbers of the 10 to 50 grid points that are closest to all well trajectory points, including the well trajectory points.
[0102] Establish the objective function F(mΦ) according to design requirements. i ,mk i ,fvΦ i fvk i It can establish a separate objective function for production capacity, or it can establish an objective function that includes well leakage and drilling safety.
[0103] Based on the search results obtained above, the objective function F of the selected grid is summed to obtain the cumulative value of the objective function F_SUM under the current well trajectory conditions.
[0104] Step 4: Iterate through multiple well trajectories, compare the objective function, and optimize the drilling trajectory design. Select a new well trajectory endpoint, i.e., repeat steps a) to b) in step 3 above, to obtain the objective function value F under different well trajectory endpoints. j (mΦ i ,mk i ,fvΦ i fvk i ) Find the sum F_SUM j Select the maximum value (max(F_SUM)) from these total objective function values. j The well trajectory at this point is the optimal well trajectory.
[0105] Step 5: Input the basic parameters such as formation temperature, original formation pressure, viscosity and density of formation oil and water, apparent molecular weight of gas, and well trajectory parameters into reservoir modeling software (such as CMG, ECL, etc.) for effect verification.
[0106] Example 2
[0107] This embodiment is an application of the fractured-vuggy reservoir well trajectory optimization method in Embodiment 1, including:
[0108] 1) Similar to the first step in Example 1, this example extracts geological data from Petrel software. This data is typically stored in grid form, providing a detailed description of the basic structure of the strata. To process this data, a specialized geological processing library, such as PyVista or VTK, is first used to load the data into the program. During loading, ensure that the coordinate system and units of the data are correct so that geological features can be accurately represented in subsequent visualizations. Using 3D visualization tools, the matrix structure can be rendered, and appropriate color mapping can be applied to help understand different regions and properties of the strata.
[0109] Next, data on the pore development zones are loaded. This data is typically a specific attribute model or mask that marks the locations of pore developments. This data is then mapped onto the already loaded matrix model and visualized by adjusting transparency or using different colors. This process makes the pore development zones clearly visible within the overall geological model, such as... Figure 2 As shown, this is the original data of the Petrel geological model, with dimensions of 15000m×15000m×500m, which facilitates identification and analysis.
[0110] 2) Similar to step two in Example 1, process the data from the solution-fracture model. This data requires fine meshing to accurately reflect its complex structural features. We map these characteristics onto a geological mesh and use different styles (such as lines and color contrast) to display the fracture structure. To enhance the visual effect, lighting and shadow effects in the view can be adjusted, such as... Figure 3 As shown, these features are highlighted to make them more apparent in the model.
[0111] 3) Similar to steps three and four in Example 1, in well trajectory optimization, the initial well trajectory coordinates are first set, and then the KNN algorithm is used to determine the neighboring grids in each direction, calculated using the objective function. Reservoir grid coordinate data is imported into the KNN model. Using the well trajectory coordinates as search points, the grid points closest to the well trajectory are found by setting a K value (e.g., a range of 10-50). The objective function F of these grid points is calculated, obtaining the cumulative objective function value F_SUM of the current well trajectory. By iterating through different well trajectories and comparing them with the objective function, the well trajectory is optimized, and the well trajectory corresponding to the maximum objective function value is selected as the optimal well trajectory. Due to the high grid accuracy, the optimized well trajectory is often an irregular twisted curve, which can be further smoothed by thinning, such as... Figure 4 As shown.
[0112] Based on the objectives of preventing well leakage and increasing production, an objective function F can be established with permeability and porosity as parameters as constraints. The objective function F can be defined as follows:
[0113]
[0114] k i =mk i +fvk i
[0115] φ i =mΦ i +fvΦ i
[0116] Where F is the objective function; mk i fvk represents matrix permeability. i The permeability of the fissure; mΦ i fvΦ represents matrix porosity. i For the porosity of the pores; L i This represents the distance between the well trajectory point and the reservoir grid obtained from the KNN search.
[0117] Due to the porosity and permeability (Φ) of the rock i k i The larger the value of Φ, the more prone well leakage is to occurring. Therefore, the product of maximum porosity and permeability (Φ) should be ruled out as the primary selection criterion.i k i ) max As a constraint. Choose (Φ) i k i While the minimum value takes into account drilling safety, it cannot achieve sufficient production capacity to meet the needs of oil and gas field development. To ensure drilling safety and maximize production capacity, a [specific value is needed]. The logarithm of permeability, λ, serves as a constraint to minimize well leakage while reducing production loss. The value of λ is crucial to the objective function F and can be determined empirically or through numerical simulation. λ acts as an optimization coefficient; a larger λ indicates higher rock permeability. When λ = 0.5, it represents the product of permeability and porosity, Φ. i k i The logarithmic median. Porosity and permeability play an absolute control role in both productivity and well leakage, (Φ i k i ) max The maximum permeability of the reservoir rocks in the reaction study area, (Φ i k i ) min The minimum permeability of the reservoir rock in the study area is considered. Empirically, λ is generally taken as 0.5, meaning that in the median case, both well leakage risk and production capacity are balanced. Of course, the power factor of 0.5 can be adjusted to other values to balance production capacity and drilling safety. Different well trajectories are obtained when λ is taken as 0.5 and 0.1 (e.g.,...). Figure 5 and Figure 6 After importing these two well trajectory curves into reservoir simulation software, the production curve under the corresponding well trajectory conditions can be calculated (e.g., ...). Figure 7 As shown, solid line a represents λ = 0.5, and dashed line b represents λ = 1. The well trajectory constraint corresponding to the solid line is to the power of 0.5, and the well trajectory constraint corresponding to the dashed line is to the power of 0.1.
[0118] Although the present invention has been described above in conjunction with exemplary embodiments and accompanying drawings, those skilled in the art should understand that various modifications can be made to the above embodiments without departing from the spirit and scope of the claims.
Claims
1. A method for optimizing the trajectory of fractured-vuggy reservoir wells, characterized in that, The method includes the following steps: S1. Establish a three-dimensional geological model of fractured-vuggy reservoirs; S2. Preprocess the three-dimensional geological model of the fractured-vuggy reservoir to establish a training dataset; S3. Train an unsupervised machine learning model using the training dataset, and use the unsupervised machine learning model to identify the optimal well trajectory corresponding to the maximum cumulative value of the objective function. The objective function is an objective function based on reservoir attribute parameters and includes one or more of the following: production capacity factors, well leakage factors, and drilling safety factors.
2. The method for optimizing the trajectory of fractured-vuggy reservoir wells according to claim 1, characterized in that, The unsupervised machine learning model is a KNN model, and the reservoir attribute parameters include matrix porosity mΦ. i Matrix permeability mk i Porosity fvΦ i and the permeability of the fissure fvk i .
3. The method for optimizing the trajectory of fractured-vuggy reservoir wells according to claim 2, characterized in that, The K value of the KNN model is 10 to 50.
4. The method for optimizing the trajectory of fractured-vuggy reservoir wells according to claim 1, characterized in that, The three-dimensional geological model of the fractured-vuggy reservoir includes basic parameters, well trajectory data of the pilot well, and reservoir geological data, among which... The basic parameters include the preset grid size; The well trajectory data of the pilot well includes drilling depth coordinates d1 i Actual vertical depth coordinate z i East-west offset coordinate x i and north-south offset coordinates y i ; The reservoir geological data includes the grid length, width, and height of the reservoir in the work area, formation coordinates, and matrix porosity (mΦ). i Matrix permeability mk i Porosity fvΦ i And the permeability of the fissures, fvk.
5. The method for optimizing the trajectory of fractured-vuggy reservoir wells according to claim 1, characterized in that, Step S2 includes: S21. Confirm the center coordinates of the grid in the three-dimensional geological model of the fractured-vuggy reservoir; S22. Confirm the correspondence between the center point coordinates of the grid and the reservoir attribute parameters, and confirm the geological data volume based on the correspondence. The geological data volume includes the center point coordinates of the grid and the reservoir attribute parameters. S23. Interpolate the geological data volume to obtain the reservoir discrete grid model data volume; S24. Establish a training dataset based on the reservoir discrete grid model data volume.
6. The method for optimizing the trajectory of fractured-vuggy reservoir wells according to claim 5, characterized in that, Step S3 includes: S31. Confirm the starting point coordinates and ending point coordinates of the target well, establish the straight line equation of the starting point coordinates and ending point coordinates, and confirm the well trajectory point data volume based on the straight line equation; S32. Import the center point coordinates of the grid into the unsupervised machine learning model, use the well trajectory point data volume as the search point set to obtain the well trajectory, and calculate the cumulative value of the objective function corresponding to the well trajectory. S33. Repeat steps S31 to S32 to obtain multiple well trajectories and multiple corresponding cumulative values of objective functions. Select the well trajectory corresponding to the maximum cumulative value of objective functions as the optimal well trajectory.
7. The method for optimizing the trajectory of fractured-vuggy reservoir wells according to claim 6, characterized in that, The equation of the straight line is: X(t) = X1 + t·(X2 - X1) Y(t) = Y1 + t·(Y2 - Y1) Z(t) = Z1 + t·(Z2 - Z1) Where (X(t),Y(t),Z(t)) are well trajectory points, t is a parameter, and the value range can be [0,1]. When t=0, (X1,Y1,Z1) are the starting point coordinates, and when t=1, (X2,Y2,Z2) are the ending point coordinates.
8. A trajectory optimization device for fractured-vuggy reservoir wells, characterized in that, The well trajectory optimization device for fractured-vuggy reservoirs includes a three-dimensional geological model building module for fractured-vuggy reservoirs, a training dataset building module, and an optimal well trajectory confirmation module, which are connected in sequence. The module for establishing a 3D geological model of fractured-vuggy reservoirs is configured to establish a 3D geological model of fractured-vuggy reservoirs. The training dataset creation module is configured to preprocess the three-dimensional geological model of the fractured-vuggy reservoir and create a training dataset. The optimal well trajectory confirmation module is configured to train an unsupervised machine learning model using the training dataset, and use the unsupervised machine learning model to confirm the optimal well trajectory corresponding to the maximum cumulative value of the objective function. The objective function is an objective function based on reservoir attribute parameters and includes one or more of the following: production capacity factors, well leakage factors, and drilling safety factors.
9. A computer device, characterized in that, The computer device includes: At least one processor; and A memory storing program instructions configured to be executed by the at least one processor, the program instructions including instructions for executing the fractured-vuggy reservoir well trajectory optimization method according to any one of claims 1 to 7.
10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the trajectory optimization method for fractured reservoir wells as described in any one of claims 1 to 7.
