Segmented clustering method for perforation of unconventional horizontal well

Abstract

The invention discloses an unconventional horizontal well perforation segmentation and clustering method, which relates to the field of geological engineering, and comprises the following steps of: performing competitive evaluation on an optimization result through a double optimization model, performing multi-scheme curve comparison, performing modular design on an algorithm, and performing component interface and modularization; comparing various segmentation and clustering results on a logging curve, and comparing and displaying the results on a graphical interface; calculating scores of yield, cost, safety and the like of a perforation division scheme through super-resolution image reconstruction based on a convolutional network, and performing adjustment according to the scores; and the evaluation module analyzes each scheme, gives evaluation in combination with engineering dessert and geological dessert calculation results, optimizes perforation cluster positions through an algorithm, and generates a final optimization scheme. According to the method, segmentation and clustering schemes can be continuously compared and evaluated, the design scheme is continuously optimized, finally, the adaptability of the method is wider, the calculation result is more accurate, field personnel can use the method more conveniently, and the timeliness and energy efficiency of field operation are improved.

Description

Technical Field

[0001] The present invention belongs to the field of geological engineering, and more specifically relates to a method for perforating and clustering in segments for unconventional horizontal wells. Background Art

[0002] By converting manual experience extraction into an artificial intelligence model and abstracting the process of engineering activities into a software functional process, the artificial intelligence method 1.0 for segmenting and clustering developed through in-depth integration of deep learning research in artificial intelligence systems fills the domestic gap and achieves a breakthrough from scratch.

[0003] However, the current intelligent segmenting and clustering method is for a single well type and cannot meet the requirements of "high adaptability", "high accuracy", and "high coverage" of the perforating design optimization software for the efficient development of complex oil and gas reservoirs. It cannot compare and evaluate the results of multiple schemes and continuously optimize the design scheme.

[0004] Chinese Patent Document with Publication No. CN111456709A and Publication Date of July 28, 2020 discloses a method for multi-stage fracturing and segmenting and clustering in horizontal wells based on logging curves. Aiming at the current problems of segmenting and clustering in horizontal wells, a process for segmenting and clustering in horizontal wells is established, and segmenting and clustering are integrated into the development of unconventional oil and gas reservoirs. This invention extracts constraint conditions such as reservoir type, number of segments, casing collar, and number of perforating clusters through segmenting and clustering principles, forms the basis for segmenting and clustering such as reservoir quality, engineering quality, completion quality, and collar position through logging curve evaluation, and optimizes the positions of fracturing segments and perforating clusters using the optimal range algorithm within a segment to form a result data table. The advantages of this invention are reducing the data calculation amount, improving the accuracy of segmenting and clustering, improving the efficiency of segmenting and clustering, and being able to meet the needs of the perforating - bridge plug continuous operation process in a timely manner.

[0005] Chinese Patent Document with Publication No. CN107229989A and Publication Date of October 3, 2017 discloses a method for optimizing the perforating cluster shooting scheme in horizontal well fracturing segments. Through this method, a comprehensive compressibility index model that simultaneously considers geological sweet spots and engineering sweet spots in the reservoir can be established. Using this model, the sweet spot position and brittle - plasticity of the reservoir are determined, and the optimal cluster length and number of clusters are comprehensively determined in accordance with the principle of random extension of the main fracture, combined with induced stress and hole friction. The present invention can reduce the risk of forming ineffective and inefficient fractures, can provide accurate guidance for the setting of the perforating cluster shooting scheme in horizontal well fracturing segments, and the optimization results can greatly increase the stimulation volume and significantly improve the construction effect, thereby obtaining the maximum economic benefits.

[0006] However, the above technical solutions cannot compare and evaluate the results of multiple segmenting and clustering schemes and continuously optimize the design scheme. Summary of the Invention

[0007] To solve the problems in the above-mentioned existing technologies, the present invention proposes an unconventional horizontal well perforation sectional and clustered method. Using this method can solve the problems that the current sectional and clustered methods are for single well types and cannot meet the requirements of "high adaptability", "high accuracy", and "high coverage" of the perforation design optimization software for the efficient development of complex oil and gas reservoirs, and cannot compare and evaluate the results of multiple schemes and continuously optimize the design scheme.

[0008] The present invention is realized by adopting the following technical solutions: An unconventional horizontal well perforation sectional and clustered method consists of the following steps: S1. Conduct a competitive evaluation of the optimization scheme through a double optimization model, compare the curves of various sectional and clustered schemes, and design the algorithm modularly with component interfaces and modularization; S2. Compare various sectional and clustered schemes in S1 on the logging curve, display the comparison of the schemes on the graphical interface, and form an intermediate sectional and clustered scheme.

[0009] S3. Calculate the scores of production, cost, safety, etc. of the sectional and clustered scheme through super-resolution image reconstruction based on a convolutional network (SR-CNN), and adjust according to the scores; including using the production, cost, safety, etc. of existing production wells (high-yield wells, problem wells) as the basic reference values, combining geological data, comparing the intermediate sectional and clustered scheme with the existing sectional and clustered scheme to form a perforation sectional and clustered result, and an artificial intelligence model generates a scoring optimization system therefrom to further optimize the design of subsequent schemes; S4. The evaluation module analyzes each perforation sectional and clustered result, combines the calculation results of engineering sweet spots and geological sweet spots, gives an evaluation, and optimizes the perforation cluster position through a perforation geological engineering sweet spot dimensionality reduction selection algorithm based on positive and negative ideal solutions to generate a final optimized scheme.

[0010] In S1, the double optimization model refers to an optimization model based on geological data and an optimization model based on geological and engineering data; The competitive evaluation of the optimization scheme refers to the comparison of the closeness between the optimization scheme and the existing sectional and clustered scheme; The comparison of multiple scheme curves refers to the comparison of the existing sectional and clustered scheme, the sectional and clustered scheme based on artificial experience, the result schemes generated by the optimization model based on geological data and the optimization model based on geological and engineering data, and the scheme generated by sweet spot calculation.

[0011] The sectional and clustered results in S2 include the sectional and clustered results of manual sectional and clustering, the sectional and clustered results of empirical formulas, and the sectional and clustered results formed by the optimization model based on geological data and the optimization model based on geological and engineering data.

[0012] In S3, comparing the intermediate segmented clustering scheme with the existing segmented clustering scheme includes the following steps: Extract and transform the intermediate segmented clustering scheme, control the data volume, compare it with the existing scheme through a convolutional neural network mapping, form an optimization system, and make a fitting adjustment to the existing segmented clustering scheme.

[0013] In S1, obtaining the optimization scheme includes the following steps: S11: According to the reservoir segmented clustering principle, extract and calculate the segmented type, average segment length, number of perforation clusters per segment, and distance from the joint collar to avoid, and quantify the constraint parameters; S12: Use the horizontal section logging curve data to evaluate the reservoir quality, engineering quality, and completion quality as the basis for segmented clustering, and complete the horizontal section evaluation; S13: Calculate the range to determine the bottom boundary position of the fracturing section; S14: Correct the bottom boundary position of the fracturing section to complete the segmentation process; S15: Determine the perforation cluster position to complete the optimization calculation of the perforation cluster setting; S16: Use the fracturing section, perforation cluster, and reservoir type curve data to form the segmented clustering result data.

[0014] In S12, the evaluation parameters of reservoir quality include natural gamma, well diameter, deep resistivity, shallow resistivity, formation density, acoustic travel time, and compensated neutron curves. The evaluation parameters of engineering quality include brittleness, in-situ stress, and fracture pressure. The evaluation parameters of completion quality include cementing quality and joint collar data.

[0015] In S13, the range calculation process is to calculate the minimum segment length and the maximum segment length according to the average segment, calculate the top and bottom depths of this segment in combination with the bottom boundary of the previous segment, and determine the optimal segmentation point between the top and bottom depth boundaries.

[0016] In S14, the method for correcting the bottom boundary position of the fracturing section is to use the initial data of the fracturing section, the segmentation principle, and the casing joint collar depth position data to adjust the bottom depth of the segment so that the bridge plug is located in the middle of the casing.

[0017] In S16, the segmented clustering result data includes the number of fracturing stages and fracturing sections, bridge plug sealing positions, number of clusters, perforation cluster center points, cluster spacing, segment spacing, and reservoir classification.

[0018] In S4, combining the calculation results of engineering sweet spots and geological sweet spots includes the following steps: S41: Establish a geological sweet spot index model and an engineering sweet spot index model for the reservoir; S42: By assigning weights to the geological sweet spot index model and the engineering sweet spot index model, establish a comprehensive compressibility index model for the reservoir; S43: Modify the weight coefficients of the weight distribution in S42 according to the actual exploration results, so as to obtain a modified comprehensive compressibility index model, and judge the sweet spot position and brittle-plasticity of the reservoir based on this; S44: Analyze the probability that the fracture initiation and extension are the main fractures. On the premise of ensuring that only one main fracture extends within a cluster, determine the upper limit of the cluster length in combination with the induced stress, and determine the lower limit of the cluster length in combination with the hole friction; S45: Determine the number of clusters per section according to the brittle-plasticity.

[0019] Compared with the prior art, the advantages of the present invention are as follows: 1. By adopting two different data combinations (an optimization model based on geological data and an optimization model based on geological and engineering data) and a neural network model, the present invention generates a neural network calculation result scheme, compares it with the data of the results of other schemes, combines the calculation and screening of geological data and engineering data, obtains engineering sweet spot and geological sweet spot data, conducts scoring, and optimizes the design scheme in reverse according to the scoring results. Finally, it is combined with the actual generated drilling results to continuously optimize the evaluation system, so that its adaptability is wider, the calculation results are more accurate, it is more convenient for on-site personnel to use, and the timeliness and energy efficiency of on-site operations are improved. Description of the Drawings

[0020] Figure 1 is the step flow chart of the method of the present invention. Detailed Embodiments

[0021] To make the technical solutions of the present invention clearer and more understandable, the following further details the present invention through embodiments.

[0022] Embodiment 1 An unconventional horizontal well perforation sectioning and clustering method consists of the following steps: S1: Conduct a competitive evaluation of the optimization scheme through a double optimization model, compare the curves of various sectioning and clustering schemes, and design the algorithm modularly, with component interfaces and modularization; S2: Compare various sectioning and clustering schemes of S1 on the well logging curve, display the comparison of the schemes on the graphical interface, and form an intermediate sectioning and clustering scheme.

[0023] S3: Calculate the scores of the sectioning and clustering scheme such as production, cost, and safety through super-resolution image reconstruction based on a convolutional network (SR-CNN), and adjust according to the scores; including using the production, cost, safety, etc. of existing production wells (high-production wells, problem wells) as basic reference values, combining geological data, comparing the intermediate sectioning and clustering scheme with the existing sectioning and clustering schemes to form a perforation sectioning and clustering result, and an artificial intelligence model generates a scoring optimization system from this to further optimize the design of subsequent schemes; S4. The evaluation module analyzes the perforation section clustering results for each section, combines the calculation results of engineering sweet spots and geological sweet spots, gives an evaluation, optimizes the perforation cluster positions through the perforation geological engineering sweet spot dimensionality reduction selection algorithm based on positive and negative ideal solutions, and generates the final optimized plan.

[0024] This embodiment is the most basic embodiment. Compared with the prior art, it can compare and evaluate the results of multiple solutions and continuously optimize the design solution, improving the timeliness and energy efficiency of on-site operations.

[0025] Embodiment 2 An unconventional horizontal well perforation section clustering method consists of the following steps: S1. Conduct a competitive evaluation of the optimized plan through a dual optimization model, compare the curves of various section clustering plans, and design the algorithm modularly with component interfaces and modularity; S2. Compare the various section clustering plans of S1 on the logging curve, display the comparison of the plans on the graphical interface, and form an intermediate section clustering plan.

[0026] S3. Calculate the scores of production, cost, safety, etc. of the section clustering plan through super-resolution image reconstruction based on convolutional networks (SR-CNN) and adjust according to the scores; including using the production, cost, safety, etc. of existing production wells (high-yield wells, problem wells) as basic reference values, combining geological data, comparing the intermediate section clustering plan with the existing section clustering plan to form the perforation section clustering results, and an artificial intelligence model generates a score optimization system from this to further optimize the design of subsequent plans; S4. The evaluation module analyzes the perforation section clustering results for each section, combines the calculation results of engineering sweet spots and geological sweet spots, gives an evaluation, optimizes the perforation cluster positions through the perforation geological engineering sweet spot dimensionality reduction selection algorithm based on positive and negative ideal solutions, and generates the final optimized plan.

[0027] In S1, the dual optimization model refers to an optimization model based on geological data and an optimization model based on geological and engineering data; The competitive evaluation of the optimized plan refers to comparing the closeness of the optimized plan with the existing section clustering plan; The comparison of multiple plan curves refers to comparing the existing section clustering plan, the manual experience section clustering plan, the results plans generated by the optimization model based on geological data and the optimization model based on geological and engineering data, and the plan generated by sweet spot calculation.

[0028] The section clustering results in S2 include manual section clustering results, empirical formula section clustering results, and section clustering results formed by the optimization model based on geological data and the optimization model based on geological and engineering data.

[0029] In S3, comparing the intermediate segmented clustering scheme with the existing segmented clustering scheme includes the following steps: Extract and transform the intermediate segmented clustering scheme, control the data volume, map and compare it with the existing scheme through a convolutional neural network to form an optimization system, and make a fitting adjustment to the existing segmented clustering scheme.

[0030] In S1, obtaining the optimization scheme includes the following steps: S11: According to the principles of reservoir segmented clustering, extract and calculate the segmented type, average segment length, number of perforation clusters per segment, and distance to avoid coupling, and quantify the constraint parameters; S12: Use the horizontal section logging curve data to evaluate the reservoir quality, engineering quality, and completion quality as the basis for segmented clustering, and complete the horizontal section evaluation; S13: Calculate the range to determine the bottom boundary position of the fracturing section; S14: Correct the bottom boundary position of the fracturing section to complete the segmentation process; S15: Determine the perforation cluster position to complete the optimization calculation of the perforation cluster setting; S16: Use the fracturing section, perforation cluster, and reservoir type curve data to form the segmented clustering result data.

[0031] In S12, the evaluation parameters of reservoir quality include natural gamma, well diameter, deep resistivity, shallow resistivity, formation density, acoustic travel time, and compensated neutron curve. The evaluation parameters of engineering quality include brittleness, in-situ stress, and fracture pressure. The evaluation parameters of completion quality include cementing quality and coupling data.

[0032] In S13, the range calculation process is to calculate the minimum segment length and the maximum segment length based on the average segment, and calculate the top and bottom depths of this segment in combination with the bottom boundary of the previous segment, and determine the optimal segmentation point between the top and bottom depth boundaries.

[0033] In S14, the method for correcting the bottom boundary position of the fracturing section is to use the initial data of the fracturing section, the segmentation principle, and the coupling depth position data of the casing to adjust the bottom depth of the segmentation so that the bridge plug is located in the middle of the casing.

[0034] In S16, the segmented clustering result data includes the number of fracturing stages and fracturing sections, bridge plug sealing position, number of clusters, center point of perforation cluster, cluster spacing, segment spacing, and reservoir classification.

[0035] In S4, combining the calculation results of engineering sweet spots and geological sweet spots includes the following steps: S41: Establish a geological sweet spot index model and an engineering sweet spot index model for the reservoir; S42: By assigning weights to the geological sweet spot index model and the engineering sweet spot index model, establish a comprehensive compressibility index model for the reservoir; S43: Modify the weight coefficients of the weight distribution in S42 according to the actual exploration results, so as to obtain a modified comprehensive compressibility index model, and judge the sweet spot position and brittle-plasticity of the reservoir based on this; S44: Analyze the probability that the fracture initiation and extension are the main fractures. On the premise of ensuring that only one main fracture extends within a cluster, determine the upper limit of the cluster length in combination with the induced stress, and determine the lower limit of the cluster length in combination with the hole friction; S45: Determine the number of clusters per section according to the brittle-plasticity.

[0036] The algorithm is modularly designed, which is convenient for updating and upgrading. The components are interfaced and modularized, improving the development and debugging efficiency, and facilitating the update and deployment of the neural network. Interfacing: Divide each component into blocks and connect them using standard interfaces; Modularization: The neural network module can be conveniently replaced, which is conducive to the update and deployment of the neural network.

[0037] In this embodiment, according to the input data such as the basic well conditions, logging data, well deviation data, casing data, and existing sectional and clustered schemes, different calculation modules such as manual experience, an optimization model based on geological data, and an optimization model based on geological and engineering data extract and calculate the sectional type, average section length, number of perforation clusters per section, and distance to avoid couplings according to the corresponding reservoir sectional and clustering principles, and quantify the requirements of constraint parameters. Using the logging curve data of the horizontal section, evaluate the reservoir quality, engineering quality, and completion quality as the basis for sectional and clustering, and complete the horizontal section evaluation; calculate the range, determine the bottom boundary position of the fracturing section; correct the bottom boundary position of the fracturing section to complete the sectional processing; determine the perforation cluster position to complete the optimization calculation of the perforation cluster setting; use the fracturing section, perforation cluster, and reservoir type curve data to generate the corresponding sectional and clustering design results respectively. The effect evaluation module combines the calculation results of engineering sweet spots and geological sweet spots, analyzes each scheme, gives evaluations, distributes weights to the geological sweet spot index model and the engineering sweet spot index model through the weight coefficients corrected according to the actual exploration results, and optimizes the perforation cluster position through the perforation geological engineering sweet spot dimensionality reduction selection algorithm based on positive and negative ideal solutions to generate the final optimized scheme.

[0038] By adopting two different data combinations (an optimization model based on geological data and an optimization model based on geological and engineering data) and a neural network model, generate the neural network calculation result scheme, compare it with the data of other scheme results, combine the calculation and screening of geological data and engineering data, obtain the engineering sweet spot and geological sweet spot data, conduct scoring, and optimize the design scheme in reverse according to the scoring results. Finally, combine it with the actual generated drilling results to continuously optimize the evaluation system, ultimately making it more adaptable, with more accurate calculation results, more convenient for on-site personnel to use, and improving the timeliness and energy efficiency of on-site operations.

Claims

1. An unconventional horizontal well perforation sectioning and clustering method, characterized in that, the steps are as follows: S1. Conduct a competitive evaluation of the optimization scheme through a dual optimization model, compare the curves of various sectioning and clustering schemes, design the algorithm modularly, and make the component interfaces modular; S2. Compare the various sectioning and clustering schemes in S1 on the logging curve, display the comparison of the schemes on the graphical interface, and form an intermediate sectioning and clustering scheme; S3. Calculate scores such as production, cost, and safety of the sectioning and clustering scheme through super-resolution image reconstruction based on a convolutional network, and make adjustments according to the scores; including using the production, cost, safety, etc. of existing production wells (high-production wells, problem wells) as basic reference values, combining geological data, comparing the intermediate sectioning and clustering scheme with the existing sectioning and clustering scheme to form a perforation sectioning and clustering result, and an artificial intelligence model generates a scoring optimization system to further optimize the design of subsequent schemes; S4. The evaluation module analyzes each perforation sectioning and clustering result, combines the calculation results of engineering sweet spots and geological sweet spots, gives an evaluation, and optimizes the perforation cluster position through a perforation geological engineering sweet spot dimensionality reduction selection algorithm based on positive and negative ideal solutions to generate a final optimized scheme.

2. An unconventional horizontal well perforation sectioning and clustering method according to claim 1, characterized in that: In S1, the dual optimization model refers to an optimization model based on geological data and an optimization model based on geological and engineering data; The competitive evaluation of the optimization scheme refers to comparing the closeness of the optimization scheme with the existing sectioning and clustering scheme; The comparison of multi-scheme curves refers to comparing the existing sectioning and clustering scheme, the manual experience sectioning and clustering scheme, the result schemes generated by the optimization model based on geological data and the optimization model based on geological and engineering data, and the scheme generated by sweet spot calculation.

3. An unconventional horizontal well perforation sectioning and clustering method according to claim 1, characterized in that: The sectioning and clustering results in S2 include manual sectioning and clustering results, sectioning and clustering results of empirical formulas, and sectioning and clustering results formed by the optimization model based on geological data and the optimization model based on geological and engineering data.

4. An unconventional horizontal well perforation sectioning and clustering method according to claim 1, characterized in that: The comparison of the intermediate sectioning and clustering scheme with the existing sectioning and clustering scheme in S3 includes the following steps: Extract and transform the intermediate sectioning and clustering scheme, control the data volume, map and compare it with the existing scheme through a convolutional neural network to form an optimization system, and make a fitting adjustment to the existing sectioning and clustering scheme.

5. An unconventional horizontal well perforation sectioning and clustering method according to claim 1, characterized in that: In S1, the acquisition of the optimization scheme includes the following steps: S11: According to the reservoir sectioning and clustering principle, extract and calculate the section type, average section length, number of perforation clusters per section, and distance from the collar to be avoided, and quantify the constraint parameters; S12: Use the horizontal section logging curve data to evaluate the reservoir quality, engineering quality, and completion quality as the basis for sectioning and clustering, and complete the horizontal section evaluation; S13: Calculate the range and determine the bottom boundary position of the fracturing stage; S14: Modify the bottom boundary position of the fracturing stage to complete the stage division process; S15: Determine the perforation cluster position and complete the optimization calculation of perforation cluster setting; S16: Use the data of the fracturing stage, perforation cluster, and reservoir type curve to form the stage and cluster division result data.

6. An unconventional horizontal well perforation stage and cluster division method according to claim 5, wherein: In S12, the evaluation parameters of reservoir quality include natural gamma, well diameter, deep resistivity, shallow resistivity, formation density, acoustic travel time, and compensated neutron curve; the evaluation parameters of engineering quality include brittleness, in-situ stress, and fracture pressure; and the evaluation parameters of completion quality include cementing quality and collar data.

7. An unconventional horizontal well perforation stage and cluster division method according to claim 5, wherein: In S13, the range calculation process is to calculate the minimum and maximum section lengths based on the average section, calculate the top and bottom depths of this section in combination with the bottom boundary of the previous section, and determine the optimal sectioning point between the top and bottom depth boundaries.

8. An unconventional horizontal well perforation stage and cluster division method according to claim 5, wherein: In S14, the method for modifying the bottom boundary position of the fracturing stage is to use the initial data of the fracturing stage, sectioning principle, and casing collar depth position data to adjust the bottom depth of the section so that the bridge plug is located in the middle of the casing.

9. An unconventional horizontal well perforation stage and cluster division method according to claim 5, wherein: In S16, the stage and cluster division result data includes the number of fracturing stages and fracturing sections, bridge plug sealing position, number of clusters, center point of the perforation cluster, cluster spacing, section spacing, and reservoir classification.

10. An unconventional horizontal well perforation stage and cluster division method according to claim 1, wherein: In S4, combining the calculation results of engineering sweet spots and geological sweet spots to give an evaluation, including the following steps: S41: Establish a geological sweet spot index model and an engineering sweet spot index model for the reservoir; S42: By assigning weights to the geological sweet spot index model and the engineering sweet spot index model, establish a comprehensive compressibility index model for the reservoir; S43: Modify the weight coefficient of the weight assignment in S42 according to the actual exploration results to obtain a modified comprehensive compressibility index model, and judge the sweet spot position and brittle-plasticity of the reservoir based on this; S44: Analyze the probability of the main crack extending as the starting crack. On the premise of ensuring that there is only one main crack extending in a cluster, determine the upper limit of the cluster length in combination with the induced stress, and determine the lower limit of the cluster length in combination with the hole friction; S45: Determine the number of clusters per section according to the brittle-plasticity.

Images