A method for predicting underwater distributary channel sand bodies based on seismic waveforms
By combining seismic waveform and well logging curve models, and utilizing gradient transformation and neural network analysis, the problems of low accuracy and multiple solutions in underwater distributary channel sand body prediction were solved, achieving higher accuracy sand body prediction and reducing exploration risks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- PETROCHINA CO LTD
- Filing Date
- 2024-03-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies for predicting sand bodies in underwater distributary channels of terrestrial lacustrine basins suffer from problems such as large errors, low efficiency, large computational load, multiple solutions, and low prediction accuracy.
Combining the lateral resolution advantage of seismic waveforms and the vertical resolution advantage of well logging curve models, and through gradient transformation, normalization correction, and neural network analysis, combined with geological sedimentary patterns, the system utilizes seismic waveform characteristics and well logging curve model constraints for inversion to achieve accurate prediction of underwater distributary channel sand bodies.
It improves the accuracy of underwater distributary channel sand body prediction, reduces drilling risks in exploration and development, and is easy to operate with low ambiguity.
Smart Images
Figure CN120703837B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of petroleum exploration and development technology, specifically relating to a method for predicting underwater distributary channel sand bodies based on seismic waveforms. Background Technology
[0002] The exploration and development of petroleum is a highly complex process that integrates fundamental theoretical knowledge from multiple disciplines. These disciplines support, complement, and corroborate each other. Ultimately, success depends on fundamental research and technological innovation in each discipline, and the success of the final research results is equally important for both. When facing common problems, we typically need diverse professional techniques and meticulous summaries to address the complex and ever-changing geological targets encountered in petroleum exploration and development. In this process, petroleum exploration and development workers need to continuously innovate technologies, integrate and develop technologies, and improve accuracy. Simultaneously, they need to conduct detailed analyses and cross-referencing of research results from various aspects such as well logging, geology, and seismic analysis to eliminate the adverse effects of technical shortcomings in individual disciplines. In continental lacustrine basins, numerous underwater distributary channel sand bodies are the main oil-producing layers, containing abundant oil and gas. Improving the prediction accuracy of underwater distributary channel sand bodies in continental lacustrine basins by utilizing well logging information, seismic technology, and geological sedimentary theories will help increase the success rate of exploration and development of this stratum.
[0003] Typical techniques for predicting underwater distributary channel sand bodies in terrestrial lacustrine basins fall into two categories. One involves extracting the "amplitude, frequency, and facies" attributes of earthquakes, or performing mathematical operations on these three parameters before extracting attributes and analyzing the correlation between the extracted results and well data to calculate the consistency rate, thereby reducing exploration risk. While this method is relatively objective, it can only predict lithological variations within the target stratigraphic segment on a planar scale. Due to the limitations and ambiguity of seismic wave theory, it can introduce significant errors in predicting underwater distributary channel sand bodies. The other category utilizes the vertical resolution advantage of well logging to perform well logging curve model-constrained inversion. Based on sedimentary patterns, a certain algorithm is selected to calculate the well-logging difference model as a constraint. At the same time, the interface information of the seismic reflection wave is used to perform deconvolution operations using various wavelets to obtain the inversion plane information of the target layer. Then, the well difference model and the inversion plane information are mathematically calculated according to certain weights to obtain the final result. These common practices or techniques are inefficient, computationally intensive, prone to multiple solutions, and have low prediction accuracy. Therefore, if the simple and easily calculated waveform gradient information can be organically combined with geological sedimentary patterns and then applied to the constraint conditions of well logging curve model constraint inversion, the above drawbacks will be greatly improved. Summary of the Invention
[0004] To address the aforementioned problems, this invention proposes a method for predicting underwater distributary channel sand bodies based on seismic waveforms, comprising the following steps:
[0005] S1. Analyze the oil-producing strata revealed by drilling to determine the location of the underwater distributary channel sand bodies on the seismic profile;
[0006] S2. Analyze the typical characteristics of seismic waveforms corresponding to different lithological combinations;
[0007] S3. Perform gradient transformation point by point on each seismic trace within the time window corresponding to the target layer on the seismic profile.
[0008] S4. Using the gradient changes of the seismic waveform characteristics corresponding to the n lithological combinations in step S2 as the input channel, normalize and correct the gradient change set in step S3 and perform neural network analysis.
[0009] S5. Perform well logging curve model constraint inversion on the target layer and extract the inversion plane distribution set of the target layer;
[0010] S6. Normalize the inversion planar map of the target layer extracted in step S5 with the planar distribution map of the similarity of seismic waveform features generated in step S4, and perform mathematical operations.
[0011] Furthermore, in step S1, the oil-producing strata revealed by drilling are analyzed, and the depositional range of the underwater distributary channel is determined through curve characteristics and lateral comparison analysis. Using the three-dimensional seismic data of this area, a seismic work area is established, and the well-seismic calibration determines the well-seismic time-depth relationship, further clarifying the position of the underwater distributary channel sand body on the seismic profile.
[0012] Further, in step S2, the typical characteristics of seismic waveforms corresponding to different lithological combinations are analyzed, and the number of types is recorded as n. Gradient transformation is performed on these seismic traces with different characteristics, i.e., G = dx / dy, where G is the gradient transformation value, dx is the amplitude of the next sample point minus the amplitude of the previous sample point, and dy is the time of the next sample point minus the time of the previous sample point. The ratio of the two is used as a description of the changes in the seismic waveform, and the sequence of this ratio is used as an array of typical characteristics of the seismic waveform. Each array is assigned an independent natural number identifier from 1 to n.
[0013] Furthermore, in step S3, gradient transformation is performed point by point on each seismic trace within the time window corresponding to the target layer on the seismic profile, i.e., G = dx / dy, where G is the gradient transformation value, dx is the amplitude of the next sample point minus the amplitude of the previous sample point, and dy is the time of the next sample point minus the time of the previous sample point, forming a set of gradient changes for each seismic trace, which is then used as an array in subsequent calculations.
[0014] Furthermore, in step S4, the gradient changes of the seismic waveform features corresponding to the n lithological combinations in step S2 are used as input channels to perform normalization correction and neural network analysis on the gradient change set in step S3. According to the n lithological combinations corresponding to the underwater distributary channels, they are classified and assigned values based on the similarity of the seismic waveform features, and are assigned incremental natural numbers 1-n, while maintaining consistency with the number of lithological combinations corresponding to the underwater distributary channels.
[0015] Furthermore, in step S5, based on the completion of well-seismic calibration and stratigraphic interpretation, well logging curve analysis is carried out according to the geological characteristics of underwater distributary channel sand body deposition, and a reasonable geological model is established. Well logging curve model constraint inversion is carried out on the target stratigraphic segment, and the inversion plane distribution set of the target stratigraphic segment is extracted.
[0016] Further, in step S6, the inversion plane map of the target segment extracted in step S5 is normalized with the seismic waveform feature similarity plane distribution map generated in step S4, and a mathematical operation is performed. The formula is z = x * y, where z is the final result, x is the normalized well logging curve model constrained inversion plane dataset, and y is the normalized seismic waveform feature similarity plane distribution dataset.
[0017] Furthermore, the final well logging curve model constraint inversion and the correlation plane distribution map of the underwater distributary channel are obtained, realizing the prediction of the sand body of the underwater distributary channel.
[0018] The beneficial effects of this invention are as follows: From the perspective of studying underwater distributary channel sand bodies in continental lacustrine basins, this invention reduces the problem of multiple solutions in seismic prediction of underwater distributary channel sand bodies while improving the accuracy of oil and gas exploration and development. This invention is characterized by its simple operation, flexibility, and low multiple solutions. It can make full use of the lateral resolution advantage of seismic logging as well as the vertical resolution advantage of well logging. Multiple disciplines can corroborate and complement each other, thereby improving the prediction accuracy of underwater distributary channel sand bodies in continental lacustrine basins and reducing the drilling risks of exploration and development. Attached Figure Description
[0019] Figure 1 This is a diagram showing the five different seismic waveform characteristic modes corresponding to five lithological combinations in the underwater distributary channel of this invention.
[0020] Figure 2 This is a planar diagram of the seismic trace waveform classification set corresponding to the lithological assemblage of the present invention;
[0021] Figure 3 This is a plane distribution diagram of the constrained inversion of the well logging curve model of the present invention;
[0022] Figure 4 This is a diagram showing the predicted thickness of sand bodies in underwater distributary channels based on seismic waveforms, as presented in this invention. Detailed Implementation
[0023] To make the technical means and objectives of this invention easier to understand, the invention is further described below with reference to specific embodiments. A method for predicting underwater distributary channel sand bodies based on seismic waveforms is described below. This method first determines the well-seismic time-depth relationship through well-seismic calibration to clarify the position of the underwater distributary channel sand body on the seismic profile; then, gradient transformation is performed point-by-point on each seismic trace within the target section to form a gradient change set; the gradient change set is normalized and corrected, and subjected to neural network analysis; based on the required quantity, the gradient change set is categorized according to the magnitude of gradient change to form the required number of seismic trace gradient change sets; each seismic trace gradient change set is assigned an independent... The process involves identifying the seismic waveform characteristics of different lithological combinations and using these different seismic traces as typical features. The gradients of these different seismic traces are calculated and compared with the gradient change sets of each seismic trace to determine the distribution range of different lithological combinations of the underwater distributary channel sand bodies in the target section. Based on well-seismic calibration, well logging curve model-constrained inversion is performed on the target section, and the inversion planar distribution set of the target section is extracted. Then, the planar distribution sets of waveform classification corresponding to different lithological combinations of the underwater distributary channel sand bodies are mutually constrained with the planar distribution sets of the well logging curve model-constrained inversion to perform mathematical operations, thereby achieving a more accurate and reliable prediction of the underwater distributary channel sand bodies.
[0024] This invention innovatively utilizes the lateral resolution advantage of seismic waveform plane variations, combined with the high accuracy of well logging curve model-constrained inversion at well points, to improve the prediction accuracy of underwater distributary channel sand bodies. The process mainly includes six steps:
[0025] (1) Analyze the oil-producing strata revealed by drilling, determine the deposition range of the underwater distributary channel through curve characteristics and lateral comparison analysis, establish the seismic work area using the three-dimensional seismic data of the area, determine the well-seismic time-depth relationship through well-seismic calibration, and further clarify the position of the underwater distributary channel sand body on the seismic profile.
[0026] (2) Analyze the typical characteristics of seismic waveforms corresponding to different lithological combinations and record the number of types as n. Perform gradient transformation on these seismic traces with different characteristics, i.e., G = dx / dy. In the formula, G is the gradient transformation value, dx is the amplitude of the next sample point minus the amplitude of the previous sample point, and dy is the time of the next sample point minus the time of the previous sample point. Use the ratio of the two as a description of the changes in seismic waveforms, and use the sequence of this ratio as an array of typical characteristics of the seismic waveforms. Assign each array an independent natural number identifier from 1 to n.
[0027] (3) Perform gradient transformation point by point for each seismic trace within the time window corresponding to the target layer on the seismic profile, i.e. G = dx / dy, where G is the gradient transformation value, dx is the amplitude of the next sample point minus the amplitude of the previous sample point, and dy is the time of the next sample point minus the time of the previous sample point, forming a set of gradient changes for each seismic trace, and using it as an array for subsequent calculations.
[0028] (4) Using the gradient changes of the seismic waveform features corresponding to the n lithological combinations in step (2) as the input channel, normalize and correct the gradient change set in step (3) and perform neural network analysis. According to the n lithological combinations corresponding to the underwater distributary channel, classify and assign values according to the similarity of the seismic waveform features, assign them incrementing natural numbers 1-n respectively, and keep them consistent with the number of lithological combinations corresponding to the underwater distributary channel.
[0029] (5) Based on the completion of well-seismic calibration and stratigraphic interpretation, according to the geological characteristics of underwater distributary channel sand body deposition, well logging curve interconnection analysis is carried out, a reasonable geological model is established, well logging curve model constraint inversion is carried out for the target section, and the inversion plane distribution set of the target section is extracted.
[0030] (6) Normalize the inversion plane map of the target segment extracted in step (5) with the plane distribution map of the similarity of seismic waveform features generated in step (4), and perform mathematical operations. The formula is z = x * y, where z is the final result, x is the normalized well logging curve model constraint inversion plane dataset, and y is the normalized seismic waveform feature similarity plane distribution dataset. The final plane distribution map of the correlation between well logging curve model constraint inversion and underwater distributary channel is obtained, thereby realizing the prediction of sand bodies in underwater distributary channels.
[0031] To make the key aspects, core technologies, implementation effects, and advantages of this invention clearer, the following will describe the invention in further detail with reference to the accompanying drawings, taking the Fuyu oil-bearing Quanquan No. 4 Sandstone Formation I in a seismic work area in the southern Songliao Basin as an example, following the specific steps of this invention.
[0032] (1) Select a typical seismic work area with a full coverage area of 150 square kilometers. Through well-seismic calibration and well-to-well comparison, determine the location of the main target layer q4-I underwater distributary channel on the seismic profile.
[0033] (2) Analysis shows that there are five lithological combinations in the q4-I underwater distributary channel, corresponding to five different typical seismic waveform characteristics. Gradient transformations are performed on these seismic traces with different characteristics, i.e., G = dx / dy, where G is the gradient transformation value, dx is the amplitude of the next sample point minus the amplitude of the previous sample point, and dy is the time of the next sample point minus the time of the previous sample point. The ratio of these two values is used as a description of the seismic waveform changes, and the sequence of this ratio is used as an array of typical seismic waveform characteristics, sequentially assigned the numbers 1, 2, 3, 4, 5, as shown below. Figure 1 As shown, Figure 1 This diagram illustrates the five different seismic waveforms corresponding to five different lithological combinations in underwater distributary channels. It serves as a model for neural network analysis.
[0034] (3) Perform gradient transformation point by point for each seismic trace within the time window corresponding to q4-I on the seismic profile, i.e. G = dx / dy, where G is the gradient transformation value, dx is the amplitude of the next sample point minus the amplitude of the previous sample point, and dy is the time of the next sample point minus the time of the previous sample point. Use the ratio of the two as a description of the seismic waveform change, and use the sequence of this ratio as an array of the seismic waveform characteristics to form a gradient change set for each seismic trace.
[0035] (4) Using the gradient changes of the typical seismic waveform features corresponding to the five lithological combinations in step (2) as input channels, normalize and correct the gradient change set of each seismic channel in step (3) and perform neural network analysis. Classify and assign values according to the similarity of the five typical seismic waveform features corresponding to the lithological combinations of the q4-I underwater distributary channel, assigning them the numbers 1, 2, 3, 4, and 5 respectively, and generate a planar map containing five colors, such as... Figure 2 As shown in the figure, this figure shows the planar distribution of five types of seismic waveforms obtained by summarizing and classifying seismic reflection waveforms and calculating them using neural network algorithms in the underwater distributary channel sedimentary environment of a certain layer in a certain work area of Songnan. Ultimately, these will be used as constraints to predict the distribution of underwater distributary channel sand bodies in the target layer.
[0036] (5) Based on the well-seismic calibration, well logging curve model-constrained inversion is performed on the target layer q4-I underwater distributary channel, and the inversion planar distribution set of the target layer is extracted, such as... Figure 3 As shown, this figure is a planar distribution map of the characteristic parameter curve values extracted by well logging curve model constraint inversion under the sedimentary environment of a certain underwater distributary channel in a certain work area of Songnan. It will eventually be used as the basic map of the distribution of sand bodies in the underwater distributary channel in the prediction process.
[0037] (6) The well logging curve model constraint inversion plane map extracted in step (5) is normalized with the seismic waveform feature similarity plane distribution map generated in step (4), and mathematical operations are performed. The formula is z = x * y, where z is the final result, x is the normalized well logging curve model constraint inversion plane dataset, and y is the normalized seismic waveform feature similarity plane distribution dataset. The final well logging curve model constraint inversion and underwater distributary channel correlation plane distribution map is obtained, realizing the prediction of underwater distributary channel sand bodies, such as... Figure 4 The image shown is a predicted distribution map of the final underwater distributary channel sand thickness.
[0038] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A method for predicting underwater distributary channel sand bodies based on seismic waveforms, characterized in that, Includes the following steps: S1. Analyze the oil-producing strata revealed by drilling to determine the location of the underwater distributary channel sand bodies on the seismic profile; S2. Analyze the typical characteristics of seismic waveforms corresponding to different lithological combinations; S3. Perform gradient transformation point by point on each seismic trace within the time window corresponding to the target layer on the seismic profile. S4. Using the gradient changes of the seismic waveform characteristics corresponding to the n lithological combinations in step S2 as the input channel, normalize and correct the gradient change set in step S3 and perform neural network analysis. S5. Perform well logging curve model constraint inversion on the target layer and extract the inversion plane distribution set of the target layer; S6. Normalize the inversion planar map of the target segment extracted in step S5 with the planar distribution map of the similarity of seismic waveform features generated in step S4, and perform mathematical operations. In step S6, the inversion plane map of the target segment extracted in step S5 is normalized with the seismic waveform feature similarity plane distribution map generated in step S4, and mathematical operations are performed. The formula is z=x1*y1, where z is the final result, x1 is the normalized well logging curve model constrained inversion plane dataset, and y1 is the normalized seismic waveform feature similarity plane distribution dataset. The final well logging curve model constraint inversion and the correlation plane distribution map of the underwater distributary channel are obtained, enabling the prediction of sand bodies in the underwater distributary channel.
2. The underwater distributary channel sand body prediction method based on seismic waveforms as described in claim 1, characterized in that, In step S1, the oil-producing strata revealed by drilling are analyzed. The depositional range of the underwater distributary channel is determined through curve characteristics and lateral comparison analysis. The three-dimensional seismic data of this area is used to establish a seismic work area. The well-seismic calibration determines the well-seismic time-depth relationship and further clarifies the position of the underwater distributary channel sand body on the seismic profile.
3. The underwater distributary channel sand body prediction method based on seismic waveforms as described in claim 1, characterized in that, In step S2, the typical characteristics of seismic waveforms corresponding to different lithological combinations are analyzed, and the number of types is recorded as n. Gradient transformation is performed on these seismic traces with different characteristics, i.e., G=dx / dy, where G is the gradient transformation value, dx is the amplitude of the next sample point minus the amplitude of the previous sample point, and dy is the time of the next sample point minus the time of the previous sample point. The ratio of the two is used as a description of the changes in seismic waveforms, and the sequence of this ratio is used as an array of typical characteristics of the seismic waveforms. Each array is assigned an independent natural number identifier from 1 to n.
4. The underwater distributary channel sand body prediction method based on seismic waveforms as described in claim 3, characterized in that, In step S4, the gradient changes of the seismic waveform features corresponding to the n lithological combinations in step S2 are used as input channels to perform normalization correction and neural network analysis on the gradient change set in step S3. According to the n lithological combinations corresponding to the underwater distributary channels, they are classified and assigned values based on the similarity of the seismic waveform features, and are assigned incremental natural numbers 1-n, while maintaining consistency with the number of lithological combinations corresponding to the underwater distributary channels.
5. The underwater distributary channel sand body prediction method based on seismic waveforms as described in claim 4, characterized in that, In step S5, based on the completion of well-seismic calibration and stratigraphic interpretation, well logging curve analysis is carried out according to the geological characteristics of underwater distributary channel sand body deposition, and a reasonable geological model is established. Well logging curve model constraint inversion is carried out on the target section, and the inversion plane distribution set of the target section is extracted.