Shale oil large-scale crack identification method based on depth domain earthquake

Abstract

The invention relates to the technical field of shale oil exploration, and discloses a shale oil large-scale crack identification method based on depth domain earthquake, which comprises the following steps: acquiring data such as depth domain three-dimensional earthquake in a research area and performing structural interpretation; performing preprocessing including speed model tomography inversion correction and fault enhancement on the seismic data; a machine learning algorithm based on geological constraints is utilized to preferably select a maximum positive curvature attribute body from the multiple attribute bodies; sequentially carrying out active and passive ant tracking, and carrying out directional noise reduction for multiple times by combining geological priori knowledge of an inclination angle and a trend; and finally, noise and small and medium-scale cracks are filtered according to a threshold value determined by actual drilling well data, and the distribution of large-scale cracks is accurately identified. According to the method, the advantages of depth domain seismic data are fully utilized, through multi-step collaborative noise reduction and optimization, the problems of low recognition precision and poor coincidence rate of large-scale cracks in a complex structure area are effectively solved, and a reliable basis is provided for optimized deployment and fracturing design of shale oil wells.

Description

Technical Field

[0001] This invention belongs to the field of shale oil exploration technology, specifically relating to a method for identifying large-scale fractures in shale oil based on depth-domain seismic data. Background Technology

[0002] Large-scale fractures refer to fractures or fault systems with an extension length generally less than 2 km and a width ranging from centimeters to meters. They do not show significant phase changes on seismic profiles and generally require specialized technical methods for identification. The development of large-scale fractures has a significant impact on the deployment, drilling, and fracturing operations of shale oil horizontal wells, thereby affecting shale oil well production. For example, the development of large-scale fractures may cause frequent well kicks and lost circulation during drilling, severely impacting drilling efficiency; during fracturing operations, the development of large-scale fractures may lead to cross-flow in horizontal well groups, affecting fracturing effectiveness and ultimately impacting the EUR (Earnings Per Hour) of shale oil wells.

[0003] Current technologies for identifying large-scale fractures in shale oil production primarily rely on post-stack time-migrating seismic data, using various seismic attributes for prediction. However, this approach has two main shortcomings: First, time-domain seismic data is affected by variations in layer velocity, leading to inaccurate seismic imaging and discrepancies between reflected phase axes and actual geological conditions. Second, fracture prediction is generally based on post-stack attributes such as curvature and variance, lacking a clear technical framework, and the prediction results exhibit poor regularity and low consistency with actual drilling findings. Therefore, current technologies lack a method for fracture identification that integrates depth-domain migration seismic data. Summary of the Invention

[0004] This invention aims to solve the technical problem of difficult identification of large-scale fractures in shale oil fields in the Huazhuang area of ​​the Gaoyou Depression in the Subei Basin. To address this problem, this invention provides a method for identifying large-scale fractures in shale oil fields based on depth-domain seismic data. The method starts with depth-domain seismic data, fully considers factors affecting fracture identification, and comprehensively utilizes multiple technical means to establish a method for identifying large-scale fractures in shale oil fields based on depth-domain seismic data. This method clarifies the distribution and characteristics of large-scale fractures, providing support for shale oil well location deployment, drilling, and fracturing design.

[0005] This invention provides a method for identifying large-scale fractures in shale oil based on depth-domain seismic data, specifically including the following steps: S1. Obtain basic data of the study area, including depth-domain three-dimensional seismic data volume, target layer structure map of the study area, and drilling and logging data of shale oil wells, and interpret the target layer structure of the depth-domain three-dimensional seismic data volume. S2. Preprocess the depth domain three-dimensional seismic data volume, including filtering and fault enhancement processing, to obtain the preprocessed seismic volume; S3. Based on the preprocessed seismic body, perform discontinuity detection, extract attribute slices of multiple attribute bodies along the target layer, and select the attribute body with the largest positive curvature. S4. Based on the maximum positive curvature attribute body, perform active ant tracking, and filter out interference with tilt angle less than a preset degree and noise within a preset angle range along the main line direction to obtain the ant body after the first noise reduction. S5. Based on the ant body after the first noise reduction, perform passive ant tracking to obtain the ant body after the second noise reduction. S6. Determine the value range of noise, small- and medium-scale fractures, and large-scale fractures based on actual drilling data. Perform value range filtering on the ant body after the second noise reduction to filter out noise and small- and medium-scale fractures, obtain high-definition ant bodies, and extract slices along the target layer.

[0006] A computer device includes at least: one or more processors; and a memory storing one or more computer programs; wherein the processor invokes the computer programs to implement the steps of the method for identifying large-scale fractures in shale oil based on depth-domain seismic data.

[0007] A computer storage device stores a computer program that is invoked by a processor to implement the steps of the method for identifying large-scale fractures in shale oil based on depth-domain seismic data.

[0008] The technical solution provided by this invention has the following beneficial effects: Based on the identification of large-scale fractures, this application fully leverages the advantages of depth-domain seismic data in imaging areas with complex structures to identify large-scale fractures in shale oil. Firstly, seismic data preprocessing is performed, prioritizing structural guidance filtering and fault enhancement processing to prepare data for the next step of fracture identification. Based on depth-domain seismic data, data preprocessing is conducted to improve accuracy. In discontinuity detection, various methods are compared, and machine learning is employed, combined with the actual conditions of the region, ultimately selecting curvature bodies. In the calculation of ant bodies, by analyzing the local structure and actual drilling conditions, the value range of large-scale fractures is determined through low-dip fault and mainline direction noise filtering, optimized combination of active and passive ant body applications, and analysis of the ant body value range in conjunction with actual drilling, ultimately achieving good results. This provides guidance in the fracturing construction design of HY3, reasonably avoiding large-scale fracture development sections and ensuring the successful fracturing of the well. Attached Figure Description

[0009] The present invention will be further described below with reference to the accompanying drawings and examples. In the accompanying drawings: Figure 1 This is a schematic diagram of the overall process of a method for identifying large-scale fractures in shale oil based on depth domain seismic data according to the present invention. Figure 2This is a comparison of seismic profiles before and after data preprocessing. Figure 2 (a) in the image is the original seismic profile. Figure 2 (b) in the image shows the seismic profile after guided filtering and fault enhancement. It can be seen that the signal-to-noise ratio of the seismic profile is improved after preprocessing, and the continuous or discontinuous characteristics of the seismic phase axis are more obvious. Figure 3 The seismic attributes of the HY1 well area are sliced ​​along the layers, from left to right: variance volume, chaotic volume, and maximum positive curvature volume. The maximum positive curvature volume is selected for the next step of calculation. Figure 4 A three-dimensional image of an ant body slice in the HY1 well area. Figure 4 (a) is a slice of the active ant body calculated based on the maximum positive curvature body. Figure 4 (b) is the preliminary result obtained by filtering low-dipping faults based on active ant bodies. Figure 4 (c) in the figure is the result of the first data denoising based on the preliminary results; Figure 5 Three-dimensional image of the second noise reduction ant body slice in the HY1 well area.

[0010] Figure 6 A high-resolution 3D image of an ant body slice in the HY1 well area.

[0011] Figure 7 This is an overlay image of large-scale fracture identification and horizontal well fracturing segmentation in the HY1 well area. The three horizontal wells are labeled HY1, HY2, and HY3 from left to right. The serial numbers marked on the horizontal well trajectories indicate the fracturing segments of each well.

[0012] Figure 8 This is a composite image of high-resolution ant bodies and depth-domain seismic profiles from well HY1. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0014] Please refer to Figure 1 This invention provides a method for identifying large-scale fractures in shale oil based on depth-domain seismic data. The main steps are as follows: S1. Obtain basic data of the study area, including depth-domain three-dimensional seismic data volume, target layer structure map of the study area, and drilling and logging data of shale oil wells, and interpret the target layer structure of the depth-domain three-dimensional seismic data volume. It should be noted that the basic data of the study area in step S1 also includes the target layer information.

[0015] S2. Preprocess the depth domain three-dimensional seismic data volume, including filtering and fault enhancement processing, to obtain the preprocessed seismic volume; It should be noted that the filtering process in step S2 includes tilt-guided filtering and velocity model correction, wherein the velocity model correction is combined with the regional velocity field to perform tomographic inversion correction on the depth domain data.

[0016] As one embodiment, the present invention first performs filtering processing on depth-domain seismic data.

[0017] For depth domain data, velocity model correction techniques are introduced to improve data accuracy. This can be achieved by combining regional velocity fields with tomographic inversion correction of depth domain data, reducing structural distortion. Specifically, this process includes: Establish an initial velocity model: Based on drilling and logging data and regional geological understanding of the study area, an initial layer velocity model is established.

[0018] Forward modeling: Using an initial velocity model, seismic records are synthesized through ray tracing or wave equation forward modeling methods.

[0019] Residual Calculation and Inversion Update: The travel time residuals or waveform residuals between the synthetic seismic record and the actual observed seismic data are calculated. Based on these residuals, the perturbation of the subsurface velocity structure is inferred using the tomographic inversion method. The core of tomographic inversion is solving a large system of linear equations, the basic formula of which can be expressed as: Δd = L · Δm; where Δd is the residual vector of the observed data, L is the sensitivity matrix (or Jacobian matrix), whose element L_ij represents the partial derivative of the i-th observed data with respect to the j-th model parameter, and Δm is the velocity model perturbation vector to be determined.

[0020] Model update and iteration: The obtained disturbance Δm is superimposed on the initial velocity model to update the velocity model. Steps 2 to 4 are repeated until the residual Δd is less than the preset threshold or the number of iterations reaches the upper limit, finally obtaining a high-precision corrected velocity model.

[0021] This corrected velocity model was used for re-imaging in the depth domain, effectively reducing structural distortion caused by inaccurate velocity and providing a more reliable data foundation for subsequent fracture identification. Filtering primarily involves denoising this model, removing random and coherent noise from the seismic data.

[0022] Filtering primarily removes noise, eliminating random and coherent noise from seismic data, especially near faults or fractures, which can affect the accurate identification of faults and fractures. Dip-guided filtering is a commonly used method, utilizing the dip and azimuth angles of the formation for directional filtering along the strata. It has directional, edge detection, and edge-protective directional filtering functions, which can improve the signal-to-noise ratio of seismic data, make the continuous or discontinuous characteristics of seismic phase axes more obvious, and help in the identification of minor faults (see reference). Figure 2 (a) Secondly, fault enhancement processing. Enhancement mainly involves strengthening information such as faults on the profile after filtering to obtain the final enhanced seismic profile effect, thereby improving fault identification efficiency. Figure 2 (b) in the middle.

[0023] S3. Based on the preprocessed seismic body, perform discontinuity detection, extract attribute slices of multiple attribute bodies along the target layer, and select the attribute body with the largest positive curvature. It should be noted that in step S3, discontinuity detection uses at least one of variance volume, chaotic volume, and maximum positive curvature volume, and the attribute weights are dynamically adjusted to select the attribute volume through a machine learning attribute fusion method based on geological target constraints.

[0024] As one example, after seismic data preprocessing, the resolution and clarity of faults are improved to some extent. However, some faults still suffer from unclear boundaries and weak continuity. Therefore, discontinuity detection technology will be used to enhance the continuity of faults and highlight fault boundaries. Widely applicable and effective fault boundary identification methods include variance volume, curvature volume, and chaotic volume. When using discontinuity detection technology, a machine learning attribute fusion method based on geological target constraints is used to automatically select the optimal attribute combination, dynamically adjusting attribute weights according to geological characteristics. Specifically, this process includes: Sample Construction: Within the study area, regions with known geological conditions (such as those near drilled wells) were selected. Attribute slice data along the target layer were extracted from various attribute volumes, including variance volume, chaotic volume, and maximum positive curvature volume, to serve as the feature set. Simultaneously, based on geological information such as faults and fractures revealed by drilling, corresponding binary label maps of "fracture development" and "non-fracture development" were created as training samples.

[0025] Feature importance assessment and weight initialization: Ensemble learning algorithms such as Random Forest or XGBoost are used to train the training samples. The algorithm outputs an importance score for each seismic attribute (feature). The initial attribute weights W_i^0 are assigned based on this importance score, with attributes of higher importance receiving larger initial weights.

[0026] Dynamic weight adjustment of geological feature constraints: Based on the initial weights, regional tectonic features are introduced as constraints for dynamic adjustment. Specifically, if it is known that the study area mainly develops faults with a certain orientation (e.g., near-east-west), then for attributes that are more sensitive to that orientation (e.g., curvature bodies enhanced by directional derivative filtering), a tectonic guidance factor α (α>1) greater than 1 is multiplied by their initial weights for weighted enhancement. The dynamic adjustment formula for the weights can be expressed as: W ifinal = W i0 × (1 + β · S i Among them, W ifinal W is the final weight of the i-th attribute. i0 It is its initial weight, S i It is a measure of the consistency between the attribute body and the known tectonic direction of the region (e.g., obtained by calculating the cosine similarity between the orientation of the attribute body and the orientation of the regional tectonic direction), and β is an adjustment coefficient used to control the strength of geological constraints.

[0027] Attribute fusion and optimization: based on the final weight W ifinal Multiple attribute volumes are weighted and fused to generate a comprehensive fracture probability volume. By comparing the fused result with the single attribute result, the attribute volume that contributed the most to this study (in this embodiment, the attribute volume with the largest positive curvature) can be intuitively selected for subsequent calculations, or the fused volume can be directly used for subsequent calculations.

[0028] Figure 3 In the diagram, 'a', 'b', and 'c' represent cross-sections along the target layer for the variance volume, chaotic volume, and maximum positive curvature volume, respectively. In comparison, the variance volume cross-section only shows larger-scale EW-trending faults; the chaotic volume cross-section shows EW-trending faults more clearly than the variance volume cross-section; while the maximum curvature volume cross-section provides a clearer and more detailed characterization of near-EW-trending faults. Based on these comparisons, it is concluded that the maximum positive curvature volume is more effective at identifying fault boundaries in the HY1 well area.

[0029] S4. Based on the maximum positive curvature attribute body, perform active ant tracking, and filter out interference with tilt angle less than a preset degree and noise within a preset angle range along the main line direction to obtain the ant body after the first noise reduction. It should be noted that the active ant tracking in step S4 includes calculating the active ant body attributes and filtering out interference with an inclination angle of less than 45 degrees based on the fracture characteristics of the study area.

[0030] It should be noted that in step S4, noise is filtered within an angle range of ±5° along the main line direction to reduce the interference of stratum tilt on ant tracking.

[0031] As one example, the active ant tracking algorithm helps to uncover faults. Passive tracking based on fully revealed faults can effectively suppress noise and filter faults. Figure 4 (a) In calculating the active ant body attributes, given that the study area is dominated by shovel-shaped extensional normal faults with generally large dip angles, interference from dip angles less than 45 degrees was filtered out when calculating the ant body to reduce the interference of stratum dip on ant tracking, resulting in preliminary results ( Figure 4 (b) shows obvious fracture characteristics. However, the preliminary results map shows a large amount of strong noise around the fault along the main line direction. Considering that the Huazhuang area mainly has near-east-west and northeast-east trending faults, the data is filtered within an angle range of ±5° along the main line direction while calculating the active ant body, thus performing a second noise reduction on the data to improve the fracture identification effect. Figure 4 (c in the text)

[0032] S5. Based on the ant body after the first noise reduction, perform passive ant tracking to obtain the ant body after the second noise reduction. As one example, passive ant tracking tends to track extremely strong signals and discard weaker signals, thus helping to suppress noise and reflect the trend of large-scale fractures. To enhance fault continuity, passive ant tracking is performed on top of active ant tracking, and then slices along the fault are extracted. Figure 5 The noise was further suppressed, and the faults and large-scale cracks became clearer.

[0033] S6. Determine the value range of noise, small- and medium-scale fractures, and large-scale fractures based on actual drilling data. Perform value range filtering on the ant body after the second noise reduction to filter out noise and small- and medium-scale fractures, obtain high-definition ant bodies, and extract slices along the target layer.

[0034] It should be noted that the value range of the noise in step S6 is -1.0 to 0, the value range of the small and medium scale cracks is 0 to 0.6, and the value range of the large scale cracks is 0.6 to 1.0.

[0035] As one example, the ant body attribute value range is -1 to -1. Combined with the horizontal well drilling situation in the HY1 well area and the detection results of fracturing construction tracer, the noise value in the study area is basically distributed between -1.0 and 0, the value of small and medium-scale fractures is distributed between 0 and 0.6, and the value of large-scale fractures is between 0.6 and 1.0. Based on the above conclusions, noise between -1 and 0.6 and small and medium-scale fractures are filtered out, and a third noise reduction is performed to obtain high-definition ant bodies. The recognition effect of large-scale fractures is good. Figures 6-8 ).

[0036] In summary, this invention identified four large-scale fractures in the horizontal section of well HY1, which are basically consistent with the fault locations revealed during actual drilling of the well. Figure 8Analysis of tracer tracking results during fracturing showed that the four large-scale fractures also developed in the HY2 well, consistent with the predicted results. The large-scale fracture planar prediction results of the target layer (…) Figure 7 Two large-scale fractures extend into the horizontal section of the HY3 well. Therefore, during the fracturing design of the HY3 well, reasonable avoidance of these two large-scale fracture development sections was carried out, and the design scheme was optimized to support the successful fracturing of the well.

[0037] A computer device includes at least: one or more processors; and a memory storing one or more computer programs; wherein the processor invokes the computer programs to implement the steps of the method for identifying large-scale fractures in shale oil based on depth-domain seismic data.

[0038] A computer storage device stores a computer program that is invoked by a processor to implement the steps of the method for identifying large-scale fractures in shale oil based on depth-domain seismic data.

[0039] The preferred embodiments of the present invention disclosed above are only for the purpose of illustrating the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation described herein. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can better understand and utilize the present invention.

Claims

1. A method for identifying large-scale fractures in shale oil based on depth-domain seismic data, characterized in that, Includes the following steps: S1. Obtain basic data of the study area, including depth-domain three-dimensional seismic data volume, target layer structure map of the study area, and drilling and logging data of shale oil wells, and interpret the target layer structure of the depth-domain three-dimensional seismic data volume. S2. Preprocess the depth domain three-dimensional seismic data volume, including filtering and fault enhancement processing, to obtain the preprocessed seismic volume; S3. Based on the preprocessed seismic body, perform discontinuity detection, extract attribute slices of multiple attribute bodies along the target layer, and select the attribute body with the largest positive curvature. S4. Based on the maximum positive curvature attribute body, perform active ant tracking, and filter out interference with tilt angle less than a preset degree and noise within a preset angle range along the main line direction to obtain the ant body after the first noise reduction. S5. Based on the ant body after the first noise reduction, perform passive ant tracking to obtain the ant body after the second noise reduction. S6. Determine the value range of noise, small- and medium-scale fractures, and large-scale fractures based on actual drilling data. Perform value range filtering on the ant body after the second noise reduction to filter out noise and small- and medium-scale fractures, obtain high-definition ant bodies, and extract slices along the target layer.

2. The method for identifying large-scale fractures in shale oil based on depth-domain seismic data according to claim 1, characterized in that, The basic data of the study area in step S1 also includes the target layer information.

3. The method for identifying large-scale fractures in shale oil based on depth-domain seismic data according to claim 1, characterized in that, The filtering process in step S2 includes tilt-guided filtering and velocity model correction, wherein the velocity model correction is combined with the regional velocity field to perform tomographic inversion correction on the depth domain data.

4. The method for identifying large-scale fractures in shale oil based on depth-domain seismic data according to claim 1, characterized in that, In step S3, discontinuity detection uses at least one of variance volume, chaotic volume, and maximum positive curvature volume, and the attribute weights are dynamically adjusted to select the attribute volume through a machine learning attribute fusion method based on geological target constraints.

5. The method for identifying large-scale fractures in shale oil based on depth-domain seismic data according to claim 1, characterized in that, Step S4 involves actively tracking ants, which includes calculating the physical attributes of the active ants and filtering out interference with an inclination angle of less than 45 degrees based on the fracture characteristics of the study area.

6. The method for identifying large-scale fractures in shale oil based on depth-domain seismic data as described in claim 1, characterized in that, In step S4, noise is filtered within an angle range of ±5° along the main line direction to reduce the interference of stratum tilt on ant tracking.

7. The method for identifying large-scale fractures in shale oil based on depth-domain seismic data as described in claim 1, characterized in that, The noise value range in step S6 is -1.0 to 0, the value range of the small- and medium-scale cracks is 0 to 0.6, and the value range of the large-scale cracks is 0.6 to 1.

0.

8. A computer device, characterized in that, It includes at least: one or more processors; a memory storing one or more computer programs; wherein the processor calls the computer programs to implement the steps of the method for identifying large-scale fractures in shale oil based on depth domain seismic data as described in any one of claims 1-7.

9. A computer storage device, characterized in that, A computer program is stored, which is invoked by a processor to implement the steps of the method for identifying large-scale fractures in shale oil based on depth-domain seismic data as described in any one of claims 1-7.

Images