Method for predicting river sand body by dual constraint of dense well pattern geological model and seismic waveform

Abstract

The method for predicting river sand body by using geological pattern and seismic waveform dual constraints in dense well pattern relates to the field of reservoir geophysics, and solves the problem of uncertainty in interwell prediction caused by strong heterogeneity of reservoir and interference of thin layer in the process of seismic reservoir prediction. The method comprises the following steps: S1, analyzing and evaluating the quality of seismic data to determine the tuning thickness of the target layer; S2, establishing a geological-seismic knowledge base of river sand body; S3, classifying the geological response mode of reservoir based on the established geological-seismic knowledge base of river sand body; S4, establishing a geological-seismic response model of reservoir based on the classification of the geological response mode of reservoir; S5, selecting sensitive seismic attributes of different types of sand body; S6, analyzing and evaluating the prediction effect. The method improves the prediction accuracy of interwell river sand body.

Description

Technical Field

[0001] This invention relates to the field of reservoir geophysics, and in particular to a method for seismic prediction of channel sand bodies under the dual constraints of a dense well network geological model and seismic waveform. Background Technology

[0002] As old oilfields in eastern China enter the late stage of development with ultra-high water cut, the remaining oil is more scattered. Adjusting development measures and tapping potential requires accurate prediction of the boundaries, orientation, and connectivity of inter-well channel sand bodies. Existing seismic reservoir prediction mainly includes two methods. One is the qualitative prediction technology of reservoir based on seismic dynamic attributes. This technology reflects the lateral changes of the reservoir by the lateral relative changes of the dynamic attributes of seismic reflection waves. It is a simple and practical technology for reservoir prediction. See (1) Wan Xiaolong, Liu Ruijing, et al., "Prediction of Lacustrine Gravity Flow Sedimentary Tight Sandstone Reservoirs Based on Intelligent Fusion of Seismic Attributes" (Petroleum Science Bulletin, Vol. 8, No. 1, 2023); (2) Peng Zuolei, Chen Ming, "Improving the Prediction Accuracy of Thin Reservoirs by Multi-element Seismic Attribute Analysis" (Fault Block Oil and Gas Field, Vol. 27, No. 3, 2020); (3) Wang Yanguang, Li Hao, et al., "A Composite Seismic Attribute for Estimating the Thickness of Thin and Thin Interbedded Sand Bodies" (Petroleum Geophysical Exploration, Vol. 55, No. 1, 2020). Second, the reservoir quantitative prediction technology with well-seismic inversion as the core, characterized by the high vertical resolution of the reservoir prediction results, is a conventional application technology in the fine description of reservoirs in the target area of ​​developed oilfields. See (4) Zhang Xiuli, Jiang Yan, et al., “Random seismic inversion under dense well network conditions and its application in channel sand body prediction” (Petroleum Geophysical Exploration, No. 49, No. 5, 2014); (5) Yin Xingyao, Pei Song, et al., “Multi-scale rapid matching and tracking multi-domain joint seismic inversion method” (Chinese Journal of Geophysics, Vol. 63, No. 9, 2020).

[0003] The above (1) adopts the machine learning method of support vector machine (SVR), first selects the frequency band and then selects the attribute, establishes the nonlinear mapping relationship between the frequency attribute and the sand body thickness interpreted by well logging, and realizes the quantitative prediction of tight sandstone; the above (2) proposes a method for quantitative characterization and prediction of thin reservoirs based on principal component analysis-support vector machine algorithm (PCA-SVR), and predicts the thickness of the Pu I4 layer sandstone in the well area; the above (3) proposes a composite seismic attribute composed of amplitude and frequency, and estimates the cumulative thickness of sand body through theoretical model and practical application; the above (4) studies the random seismic inversion method under dense well network conditions and its application in channel sand body prediction, and guides the development and application to achieve practical results; the above (5) introduces low frequency model constraints in the fast matching pursuit algorithm, which effectively improves the convergence accuracy and makes the inversion results have rich high and low frequency information. The above technologies all use well point data as hard data and seismic data as soft constraints between wells, and use linear or nonlinear algorithms for prediction. They have been widely used in oil and gas exploration and development at home and abroad, especially in the identification of sand bodies in sedimentary systems such as deltaic distributary plains and meandering rivers, and have achieved good development and application results.

[0004] However, the above two reservoir prediction methods still have the following drawbacks: 1. Due to insufficient vertical resolution of seismic data, when using seismic attribute analysis to predict thin reservoirs, the interference of thin interbedded seismic waveforms results in low accuracy for predicting the distribution of sand bodies in other types of reservoirs, except for some "mud-encased sand" reservoirs. 2. Seismic inversion has high vertical resolution, but it is more dependent on the initial model, leading to multiple solutions in the prediction results. 3. Detailed dissection results of dense well network channel sand bodies show that the width, thickness, and other characteristics of channel sand bodies of the same type deposited in the same sedimentary environment have certain regularities. Existing methods lack the constraints of known geological information or models for inter-well prediction, relying entirely on changes in seismic information to extrapolate or interpolate through algorithms. Furthermore, the accuracy of narrow and thin reservoir prediction is low due to the influence of factors such as the quality of seismic data and well network density. Summary of the Invention

[0005] This invention addresses the uncertainty in inter-well prediction results caused by strong reservoir heterogeneity and thin-layer interference in the background technology of seismic reservoir prediction. It provides a seismic prediction method for channel sand bodies using a close-well geological model and seismic waveform dual-constraint approach. This method enhances the certainty of inter-well prediction results and improves the prediction accuracy of channel sand bodies between wells.

[0006] The present invention solves its problem through the following technical solution: the well network geological model-seismic waveform dual-constraint method for seismic prediction of channel sand bodies includes the following steps:

[0007] S1: Analyze and evaluate the quality of seismic data to determine the target layer tuning thickness;

[0008] S2: Establish a geological and seismic knowledge base for river sand bodies;

[0009] S3: Based on the established geological-seismic knowledge base of channel sand bodies, the reservoir geological response patterns are classified;

[0010] S4: Based on the classification of reservoir geological response patterns, establish reservoir geological-seismic response models; and select the seismically sensitive attributes of different types of sand bodies.

[0011] S5: Based on the seismically sensitive attributes of different types of sand bodies, different methods are used to perform pattern recognition reservoir prediction for different types of sand bodies;

[0012] S6: Analyze and evaluate the prediction results.

[0013] Furthermore, step S1 determines the target layer tuning thickness through seismic data quality analysis and evaluation. The specific method is as follows:

[0014] S11. Through spectral analysis, determine the bandwidth and dominant frequency information of the seismic data of the target layer; bandwidth and dominant frequency together affect the resolution of subsequent thin reservoir prediction.

[0015] S12. Based on the dominant frequency information and the actual strata velocity in the study area, the target layer tuning thickness is calculated according to the Wides quarter-wavelength formula.

[0016] The Wides-quarter seismic wavelength formula:

[0017]

[0018] In the formula: h is the tuning thickness, λ is the seismic wavelength, v is the reservoir velocity, and f is the dominant seismic frequency.

[0019] Furthermore, the method for establishing the channel sand body geological-seismic knowledge base in step S2 includes:

[0020] S21. Under the guidance of high-resolution sequence stratigraphy theory, combined with core data from the study area and the three-dimensional detailed dissection results of single sand bodies in a dense well network, we analyzed and established superposition patterns of different types of channel sand bodies, and statistically analyzed the width, thickness, and number of sandstone and mudstone layers of the sand bodies.

[0021] S22. Based on the actual logging curves of the oilfield, statistically analyze the velocity, density, and acoustic impedance parameters of sandstone and mudstone, and establish a geological-seismic knowledge base for the target layer reservoir in the study area.

[0022] The different types of river sand bodies are stacked in the following ways: cut-overlap, superposition, and isolation.

[0023] Furthermore, the reservoir geological response mode classification method in step S3 includes:

[0024] S31. Using reservoir geology-seismic knowledge base information, statistically analyze the characteristics of three parameters: scale, purity, and brightness of different types of sand bodies;

[0025] S32. Analyze the variation characteristics of reservoir three parameters using a relative coordinate system triangular diagram: take scale, purity, and brightness as three end elements, draw a triangular diagram to classify reservoir three end element types, normalize the scale, purity, and brightness data of different types of sand bodies, and project them onto the triangular diagram.

[0026] S33. Based on the characteristics of scale, purity, and brightness variation, different types of sand bodies are classified: From the perspective of seismic identification capability, different types of sand bodies are divided into three types: superimposed type (including main thick layer and isolated thin layer), transitional type, and thin interbedded type.

[0027] Furthermore, the change in statistical scale, namely the ratio of statistical reservoir thickness to target layer tuning thickness;

[0028] The change in statistical purity is the ratio of the product of the number of sand layers and the thickness of sandstone (A) to the sum of the product of the number of sand layers and the thickness of sandstone (A) and the product of the number of mudstone layers and the thickness of mudstone (B) (C); that is, A / C.

[0029] The change in the statistical brightness parameter is the ratio of the acoustic impedance of the statistical sandstone to that of the mudstone.

[0030] Furthermore, step S4, based on the classification of reservoir geological response modes, establishes a reservoir geological-seismic response model, including the following methods:

[0031] Based on the classification of reservoir geological response modes, a three-dimensional geological model is established in conjunction with well data; forward modeling is performed using the convolution theory F(t)=S(t)*R(t), and a three-dimensional geological-seismic response model is established using a self-excitation and self-absorption method.

[0032] Furthermore, the convolution theory model is as follows:

[0033] F(t) = S(t) * R(t);

[0034] Wherein: F(t) is the synthetic seismic record; S(t) is the seismic wavelet, the dominant frequency of which is consistent with the dominant frequency information in step S11; R(t) is the reflection coefficient, which is calculated from the velocity and density logging curves.

[0035] Furthermore, the method for selecting the seismically sensitive attributes of different types of sand bodies in step S4 is as follows:

[0036] S41. Based on the variation of the target reservoir thickness, select an appropriate time window to extract various seismic attributes such as amplitude, frequency, and statistical attributes from the three-dimensional geological-seismic response model, and normalize the attributes.

[0037] S42. Based on the cross-plot of normalized seismic attributes and wellpoint sandstone thickness or the comparison method of seismic attributes and well-based sedimentary facies maps, select the seismic sensitive attributes of different types of sand bodies established in step S2.

[0038] Furthermore, step S5, which describes a method for pattern recognition reservoir prediction using different methods for different types of sand bodies, includes:

[0039] S51, Overlapping river channels: Prediction is performed using wave group feature pattern recognition method;

[0040] S52, Transitional Channels: Prediction is performed using the model-constrained principal component parameter prediction method;

[0041] S53, Thin interbedded channels: The reservoir prediction method is performed using model-constrained characteristic curve simulation.

[0042] Furthermore, the specific method for predicting the superimposed river channel using wave group feature pattern recognition is as follows:

[0043] First, based on the good correspondence between superimposed channel sand bodies and seismic waveforms, correlation analysis technology was used to establish seismic waveform models for wellpoint channels and non-channel sand bodies.

[0044] Secondly, by combining information such as width and thickness in the established geological model library, the optimization of key parameters such as well-side passage number, search radius, and prediction correlation coefficient is guided by the intersection of three parameters: well-side passage number, search radius, and prediction correlation coefficient. Finally, based on the selected key parameters of different types of sand bodies, the thickness of the target layer channel sand body is predicted in plane.

[0045] Furthermore, the specific method for predicting the transitional river channel using the model-constrained principal component parameter prediction method is as follows:

[0046] First, based on the sensitive attributes obtained from the forward modeling results, sensitive seismic attributes such as amplitude, spectrum, and statistics within the time window of the target layer are extracted;

[0047] Secondly, based on the basic principles of principal component analysis, the relationship between the thickness of sandstone near the well and its components is analyzed in three-dimensional space to obtain the weight coefficients of the sandstone thickness and the three principal components.

[0048] The components include a first principal component, a second principal component, and a third principal component;

[0049] Finally, by utilizing the above-mentioned correspondence between sandstone thickness and components, the multi-attribute seismic data is reduced in dimensionality and denoised, thereby enabling effective planar prediction of the thickness of transitional channel sand bodies.

[0050] Furthermore, the method for predicting reservoirs in thin interbedded channels using model-constrained characteristic curve simulation includes: First, due to the complex sand body combination structure and thin-layer seismic interference factors in thin interbedded channels, the seismic attributes of the target layer contain reservoir information of the upper and lower layers. Through well-seismic matching unit analysis, the correlation between the seismic attributes of the target layer and the sandstone thickness of the target layer and the upper and lower units is analyzed using the Pearson correlation coefficient method. Based on the unit information with the strongest well-seismic correlation, a reasonable prediction unit in the vertical direction is determined, and an initial framework model is established based on the prediction unit division results.

[0051] Secondly, the large overlap of impedance curves in thin interbedded sandstone and mudstone leads to low reservoir prediction accuracy. By interpreting lithological constraints through well logging, and integrating multiple curves such as spontaneous potential and resistivity, curve reconstruction is performed using multiple regression or neural network algorithms to obtain sensitive characteristic curves that can distinguish between sandstone and mudstone.

[0052] Finally, the Markov chain Monte Carlo stochastic simulation algorithm was used to perform three-dimensional fine prediction of the sensitive characteristic curves of thin interlayered channel sand bodies.

[0053] Compared with the above-mentioned background technology, the present invention has the following beneficial effects:

[0054] This invention presents a seismic prediction method for channel sand bodies based on a dual constraint of a dense well network geological model and seismic waveform. It takes into account oilfield core data, detailed dissection results and understanding of dense well network reservoirs, and fully leverages the constraint effect of the geological model between wells based on the classification of different types of reservoirs and seismic response analysis. This reduces the ambiguity of traditional prediction methods and enables accurate prediction of the distribution characteristics of channel sand bodies between wells.

[0055] Compared with the prior art, the advantages of the present invention are as follows:

[0056] 1. By analyzing the variation characteristics of three parameters—scale, purity, and brightness—a semi-quantitative relationship was established between geological models of different types of sand bodies and seismic responses;

[0057] 2. By using the known geological model and seismic waveforms of the dense well network to make dual constraints on the prediction of sand bodies between wells, the prediction results can not only meet the understanding of the geological laws of sand body deposition, but also reduce the uncertainty of the prediction results caused by the low resolution of seismic data. Attached Figure Description

[0058] Figure 1 This is a flowchart of the seismic prediction method for channel sand bodies using a dense well network geological model and seismic waveform dual constraints, as presented in this invention.

[0059] Figure 2 This is a triangular diagram illustrating the reservoir three-terminal element type classification in an embodiment of the present invention;

[0060] Figure 3 This is a comparison chart of the prediction effects of the superimposed, transitional, and thin interlayered river channels according to embodiments of the present invention (a1 and a2 in the figure are the conventional superimposed type and the superimposed type of the present invention, respectively; b1 and b2 are the conventional transitional type and the transitional type of the present invention, respectively; c1 and c2 are the conventional thin interlayered type and the thin interlayered type of the present invention, respectively). Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0062] like Figure 1 As shown, a method for seismic prediction of channel sand bodies using a dual-constraint system of dense well network geological models and seismic waveforms includes the following steps:

[0063] Step 1: Analyze and evaluate the quality of seismic data to determine the target layer tuning thickness;

[0064] Through spectral analysis, the bandwidth and dominant frequency information of the target layer seismic data are determined, the tuning thickness of the target layer is determined, and the resolution of the seismic data is clarified. Step 1 specifically includes the following two points:

[0065] (1) By analyzing the spectrum, the bandwidth and dominant frequency information of the seismic data of the target layer are determined. The bandwidth and dominant frequency together affect the resolution of subsequent thin reservoir prediction and provide the dominant frequency information of the forward wavelet in the subsequent step 4.

[0066] (2) Based on the dominant frequency information and the actual strata velocity in the study area, the Wides quarter-wavelength formula is applied: In the formula, h is the tuning thickness, λ is the seismic wavelength, v is the reservoir velocity, and f is the seismic dominant frequency, and then the tuning thickness of the target layer is calculated.

[0067] Step 2: Establish a geological and seismic knowledge base for river channel sand bodies;

[0068] Based on core data, detailed 3D dissection results of dense well network reservoirs, and reservoir distribution characteristics in the study area, the superposition patterns of different types of sand bodies are analyzed. The width, thickness, number of sand and mud layers, and variation characteristics of parameters such as velocity, density, and wave impedance of different types of sand bodies are statistically analyzed to establish a reservoir geology-seismic knowledge base. The specific procedures for step 2 include:

[0069] (1) Under the guidance of high-resolution sequence stratigraphy theory, combined with the core data of the study area and the three-dimensional fine dissection results of single sand bodies in dense well network, we analyzed and established stacking patterns such as cutting, superposition, and isolation of different types of channel sand bodies, and counted the width, thickness, and number of sand and mudstone layers of the sand bodies.

[0070] (2) Based on the actual logging curves of the oilfield, the velocity, density and acoustic impedance parameters of sandstone and mudstone were statistically analyzed, and a geological-seismic knowledge base of the target layer reservoir in the study area was established.

[0071] Step 3: Based on the established geological-seismic knowledge base information of channel sand bodies, classify the reservoir geological response patterns;

[0072] Based on the reservoir geology-seismic knowledge base information from step 2, the characteristics of three parameters—scale, purity, and brightness—of different types of sand bodies are statistically analyzed. The variation characteristics of these three parameters are analyzed using a relative coordinate system triangular diagram. From a seismic perspective, based on the characteristics of the three end-members, different types of sand bodies are classified into three types: superimposed, transitional, and thin interbedded. Scale is the ratio of reservoir thickness to the target layer tuning thickness; purity is the ratio of the product of the number of sand layers and sandstone thickness to (the product of the number of sand layers and sandstone thickness + the product of the number of mudstone layers and mudstone thickness); brightness is the ratio of the acoustic impedance of sandstone to mudstone. The specific procedures in step 2 include:

[0073] (1) Using the reservoir geology-seismic knowledge base information in step 2, firstly, the change in scale is statistically analyzed, that is, the ratio of the reservoir thickness to the target layer tuning thickness obtained in step 1; secondly, the change in purity is statistically analyzed, that is, the ratio of the product of the number of sand layers and the thickness of sandstone to (the product of the number of sand layers and the thickness of sandstone + the product of the number of mudstone layers and the thickness of mudstone); finally, the change in brightness parameters is statistically analyzed, that is, the ratio of the acoustic impedance of sandstone to that of mudstone.

[0074] (2) Using scale, purity, and brightness as three end-members, draw a triangular diagram of reservoir three-end-member types, normalize the scale, purity, and brightness data of different types of sand bodies, and project them onto the triangular diagram.

[0075] (3) Based on the characteristics of scale, purity and brightness variation, from the perspective of seismic identification capability, different types of sand bodies are divided into three types: superimposed type (including main thick layer and isolated thin layer), transitional type and thin interlayered type.

[0076] Step 4: Based on the classification of reservoir geological response patterns, establish reservoir geological-seismic response models; and select the seismically sensitive attributes of different types of sand bodies.

[0077] Based on convolution theory and the classification in step 3, a three-dimensional geological-seismic response model is established using forward modeling. For the target layer, an appropriate time window is selected to extract various seismic attributes from the forward modeling data. Using the cross-plot method of seismic attributes and sandstone thickness, or the comparison method of seismic attributes and well-based sedimentary facies maps, the seismically sensitive attributes of different types of sand bodies established in step 2 are optimized. The specific methods of step 3 include:

[0078] (1) Based on the classification in step 3, a three-dimensional geological model is established in conjunction with well data; forward modeling is performed using the convolution theory F(t) = S(t) * R(t); where F(t) is the synthetic seismic record; S(t) is the seismic wavelet, and the dominant frequency of the seismic wavelet is consistent with the dominant frequency information in step S11; R(t) is the reflection coefficient, which is calculated from the velocity and density logging curves. A three-dimensional geological-seismic response model is established using a self-excitation and self-reception method.

[0079] (2) Based on the change in the thickness of the target reservoir, select an appropriate time window to extract various seismic attributes such as amplitude, frequency and statistics from the three-dimensional geological-seismic response model, and normalize the attributes.

[0080] (3) Based on the intersection diagram of normalized seismic attributes and well point sandstone thickness or the comparison method of seismic attributes and well-based sedimentary facies maps, select the sensitive seismic attributes of different types of sand bodies established in step 2.

[0081] Step 5: Based on the selected seismically sensitive attributes of different sand body types, different methods are used for pattern recognition and reservoir prediction of different sand body types; the specific methods include the following steps:

[0082] (1) Overlapping river channels are predicted using wave group feature pattern recognition methods; specific methods include:

[0083] First, based on the good correspondence between superimposed channel sand bodies and seismic waveforms, correlation analysis technology was used to establish seismic waveform models for wellpoint channels and non-channel sand bodies.

[0084] Secondly, combining the width, thickness and other information in the geological model library established in step 2, the optimization of key parameters such as well-side passage number, search radius and prediction correlation coefficient is guided by the intersection of three parameters: well-side passage number, search radius and prediction correlation coefficient. Finally, based on the selected key parameters of different types of sand bodies, the thickness of the target layer channel sand body is predicted in plane.

[0085] (2) Transitional channels are predicted using the model-constrained principal component parameter prediction method; specific methods include:

[0086] First, based on the sensitive attributes obtained from the forward modeling results in step 4, sensitive seismic attributes such as amplitude, spectrum, and statistics within the time window of the target layer are extracted;

[0087] Secondly, based on the basic principles of principal component analysis, the correspondence between the sandstone thickness near the well and the first, second, and third principal components is analyzed in three-dimensional space to obtain the weight coefficients of the sandstone thickness and the three principal components.

[0088] Finally, by utilizing the above relationships, the earthquake's multiple attributes are reduced in dimensionality and denoised, thereby enabling effective planar prediction of the thickness of transitional channel sand bodies.

[0089] (3) Thin interbedded channels are predicted using a model-constrained characteristic curve simulation method for reservoir prediction; specific methods include:

[0090] First, due to the complex structure of thin interbedded channel sand bodies and the influence of thin-layer seismic interference factors, the seismic attributes of the target layer contain reservoir information of the upper and lower layers. Through well-seismic matching unit analysis, the correlation between the seismic attributes of the target layer and the sandstone thickness of the target layer and the upper and lower units is analyzed using the Pearson correlation coefficient method. Based on the unit information with the strongest well-seismic correlation, a reasonable prediction unit in the vertical direction is determined, and an initial framework model is established based on the prediction unit division results. Second, the large overlap area of ​​the impedance curves of thin interbedded sandstone and mudstone leads to low reservoir prediction accuracy. Well logging interpretation of lithological constraints is used, and multiple curves such as spontaneous potential and resistivity are integrated. Multiple regression or neural network algorithms are used to reconstruct the curves to obtain sensitive characteristic curves that can distinguish between sandstone and mudstone. Finally, the Markov chain Monte Carlo stochastic simulation algorithm is used to perform three-dimensional fine prediction of the sensitive characteristic curves of thin interbedded channel sand bodies.

[0091] Step 6: Prediction Effect Analysis and Evaluation

[0092] The prediction effect is analyzed and evaluated based on the sandstone thickness interpreted by post-well or new well logging, and the prediction results are exported in data format or vector format.

[0093] Example 1

[0094] To make the objectives, technical solutions, and advantages of this invention clearer, the following description, using the Daqing Changyuan Oilfield's dense well network A block as an example, will be further detailed with reference to the accompanying drawings.

[0095] I. Background of the Study on Seismic Prediction Method for Channel Sandbody under the Dual Constraint of Well Network Geological Model and Seismic Waveform:

[0096] The Changyuan Oilfield has entered the late stage of ultra-high water cut, with the remaining oil distribution highly scattered. Statistical results from typical blocks indicate that 22.3% of the remaining oil is directly related to reservoir sand bodies. Accurately predicting the distribution characteristics of inter-well sand bodies is a crucial prerequisite and foundation for accurately understanding the remaining oil and guiding personalized potential tapping. In recent years, a well-seismic combined reservoir prediction technology, centered on seismic sedimentology and seismic stochastic inversion, has been developed, improving the accuracy of inter-well channel sand body prediction and providing effective guidance for development and application.

[0097] However, due to the strong heterogeneity of fluvial-deltaic reservoirs and the influence of thin-layer seismic interference, existing reservoir prediction methods rely on changes in seismic information during inter-well prediction, using linear or nonlinear algorithms for extrapolation or interpolation. This fails to fully utilize the known detailed geological analysis results and understanding of dense well networks, resulting in some prediction results still exhibiting uncertainty and contradicting existing geological knowledge. Therefore, taking Block A of the dense well network in the Daqing Changyuan Oilfield as an example, this study investigates a seismic prediction method for channel sand bodies under the dual constraints of dense well network geological model and seismic waveform.

[0098] II. A method for predicting seismic events in channel sand bodies using a dense well network geological model and dual-constraint seismic waveforms, including the following steps:

[0099] This invention first establishes a reservoir geology-seismic knowledge base (Table 1) based on seismic data quality analysis, core data from the study area, detailed three-dimensional dissection results of dense well network reservoirs, and reservoir characteristics, representing different types of sand body stacking patterns. Secondly, it statistically analyzes the variation characteristics of the three parameters (scale, purity, and brightness) of different types of sand bodies using relative coordinate system trigonometric diagrams. From a seismic perspective, based on the characteristics of the three end-members, different types of sand bodies are classified into three types: stacked, transitional, and thin interbedded. Figure 2 Then, based on convolution theory, a three-dimensional geological-seismic response model is constructed using forward modeling to extract seismic attributes of the target layer. Sensitive attributes are then optimized using either the cross-plot method of seismic attributes and sandstone thickness or the comparison method of seismic attributes with well-based sedimentary facies maps. Finally, based on the optimized sensitive attributes, wave group feature patterns are used for identification of superimposed channels, transitional channels, and thin interbedded channel sand bodies. Figure 3 a1, a2), mode-constrained principal component analysis parameter prediction ( Figure 3 Simulation of b1, b2) and mode constraint characteristic curves ( Figure 3 The c1 and c2 reservoir prediction methods were used to improve the prediction accuracy of inter-well channel sand bodies.

[0100] Table 1 Geological-Seismic Knowledge Base for Different Types of Reservoirs

[0101]

[0102] III. Comparison of Results between the Invention and Traditional Methods

[0103] (1) Traditional seismic attribute analysis and seismic inversion reservoir prediction methods

[0104] This traditional method primarily relies on fine well-seismic calibration results to determine seismic standard layers near the target layer, establish a stratigraphic framework, and perform seismic attribute analysis and seismic inversion reservoir prediction. Seismic attribute analysis involves selecting an appropriate time window near the target layer to extract various seismic attributes such as amplitude and frequency, and then optimizing the linear or nonlinear relationship between seismic attributes and well-interpreted sandstone thickness to achieve planar prediction of sandstone thickness. Seismic inversion utilizes specialized inversion software such as Jason, employing key steps or parameter settings such as lithology-sensitive curve analysis and variogram analysis to perform three-dimensional reservoir prediction.

[0105] Reservoir prediction using this method suffers from uncertainty and low accuracy due to the strong heterogeneity of the reservoir and the interference of thin layers. Furthermore, it contradicts existing geological knowledge.

[0106] (2) The present invention provides a method for predicting seismic events in channel sand bodies based on a well network geological model and seismic waveform dual constraints.

[0107] Compared with traditional seismic attribute analysis and seismic inversion reservoir prediction methods, this invention applies dual constraints to the prediction of inter-well sand bodies using known geological models and seismic waveforms in a dense well network. This ensures that the prediction results not only meet the geological laws of sand body deposition but also reduce the uncertainty caused by low seismic data resolution.

[0108] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the implementation methods of the present invention, and should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of the present invention.

Claims

1. A method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints, characterized in that: Includes the following steps: S1: Analyze and evaluate the quality of seismic data to determine the target layer tuning thickness; S2: Establish a geological and seismic knowledge base for river sand bodies; S3: Based on the established geological-seismic knowledge base of channel sand bodies, the reservoir geological response patterns are classified; S4: Based on the classification of reservoir geological response patterns, establish reservoir geological-seismic response models; and select the seismically sensitive attributes of different types of sand bodies. S5: Based on the seismically sensitive attributes of different types of sand bodies, different methods are used to perform pattern recognition reservoir prediction for different types of sand bodies; S6: Analyze and evaluate the prediction results.

2. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 1, characterized in that: Step S1 involves analyzing and evaluating the quality of seismic data to determine the target layer tuning thickness. The specific method is as follows: S11. Through spectral analysis, determine the bandwidth and dominant frequency information of the seismic data of the target layer; bandwidth and dominant frequency together affect the resolution of subsequent thin reservoir prediction. S12. Based on the dominant frequency information of the target layer seismic data and the velocity of the actual strata in the study area, the tuning thickness of the target layer is calculated according to the Wides quarter-wavelength formula. The Wides-quarter seismic wavelength formula: In the formula: h is the tuning thickness, λ is the seismic wavelength, v is the reservoir velocity, and f is the dominant seismic frequency.

3. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 2, characterized in that: The method for establishing the geological-seismic knowledge base of the channel sand body in step S2 includes: S21. Under the guidance of high-resolution sequence stratigraphy theory, combined with core data from the study area and the three-dimensional detailed dissection results of single sand bodies in a dense well network, we analyzed and established superposition patterns of different types of channel sand bodies, and statistically analyzed the width, thickness, and number of sandstone and mudstone layers of the sand bodies. S22. Based on the actual logging curves of the oilfield, statistically analyze the velocity, density, and acoustic impedance parameters of sandstone and mudstone, and establish a geological-seismic knowledge base for the target layer reservoir in the study area. The different types of river sand bodies are stacked in the following ways: cut-overlap, superposition, and isolation.

4. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 3, characterized in that: The reservoir geological response mode classification method in step S3 includes: S31. Using reservoir geology-seismic knowledge base information, statistically analyze the characteristics of three parameters: scale, purity, and brightness of different types of sand bodies; S32. Analyze the variation characteristics of reservoir three parameters using a relative coordinate system triangular diagram: take scale, purity, and brightness as three end elements, draw a triangular diagram to classify reservoir three end element types, normalize the scale, purity, and brightness data of different types of sand bodies, and project them onto the triangular diagram. S33. Based on the characteristics of scale, purity, and brightness variation, different types of sand bodies are classified: From the perspective of seismic identification capability, different types of sand bodies are divided into three types: superimposed type, transitional type, and thin interbedded type. The stacking type includes a main body thick layer stacking and an isolated thin layer stacking.

5. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 4, characterized in that: The change in statistical scale, i.e., the ratio of statistical reservoir thickness to the target layer tuning thickness; To calculate the change in purity, first calculate the product of the number of sand layers and the thickness of sandstone (A), and the product of the number of mudstone layers and the thickness of mudstone (B); then calculate the sum of the product of the number of sand layers and the thickness of sandstone (A) and the product of the number of mudstone layers and the thickness of mudstone (B) (C); finally, calculate the ratio of A to C, which represents the change in purity. The change in the statistical brightness parameter is the ratio of the acoustic impedance of the statistical sandstone to that of the mudstone.

6. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 5, characterized in that: Step S4, based on the classification of reservoir geological response modes, establishes a reservoir geological-seismic response model, including the following methods: Based on the classification of reservoir geological response modes, a three-dimensional geological model is established in conjunction with well data; forward modeling is performed using convolution theory, and a three-dimensional geological-seismic response model is established using a self-excitation and self-absorption method.

7. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 6, characterized in that: The convolution theory model is as follows: F(t) = S(t) * R(t); Wherein: F(t) is the synthetic seismic record; S(t) is the seismic wavelet, the dominant frequency of which is consistent with the dominant frequency information in step S11; R(t) is the reflection coefficient, which is calculated from the velocity and density logging curves.

8. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 6, characterized in that: The method for selecting the seismically sensitive attributes of different types of sand bodies in step S4 is as follows: S41. Based on the variation of the target reservoir thickness, select an appropriate time window to extract amplitude, frequency, and statistical seismic attributes from the three-dimensional geological-seismic response model, and normalize the attributes. S42. Based on the cross-plot of normalized seismic attributes and wellpoint sandstone thickness or the comparison method of seismic attributes and well-based sedimentary facies maps, select the seismic sensitive attributes of different types of sand bodies established in step S2.

9. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 8, characterized in that: Step S5, which describes a method for pattern recognition reservoir prediction using different methods for different types of sand bodies, includes: S51, Overlapping river channels: Prediction is performed using wave group feature pattern recognition method; S52, Transitional Channels: Prediction is performed using the model-constrained principal component parameter prediction method; S53, Thin interbedded channels: The reservoir prediction method is performed using model-constrained characteristic curve simulation.

10. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 9, characterized in that: The specific method for predicting the superimposed river channel using wave group feature pattern recognition is as follows: First, based on the good correspondence between superimposed channel sand bodies and seismic waveforms, correlation analysis technology was used to establish seismic waveform models for wellpoint channels and non-channel sand bodies. Secondly, by combining the width and thickness information in the established geological model library, the optimization of key parameters such as well-side passage number, search radius, and prediction correlation coefficient is guided by the intersection of three parameters: well-side passage number, search radius, and prediction correlation coefficient. Finally, based on the selected key parameters of different types of sand bodies, the thickness of the target layer channel sand body is predicted in plane.

11. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 9, characterized in that: The specific method for predicting the transitional river channel using the model-constrained principal component parameter prediction method is as follows: First, based on the sensitive attributes obtained from the forward modeling results, we extract the amplitude, spectrum, and statistical sensitive seismic attributes within the time window of the target layer. Secondly, based on the basic principles of principal component analysis, the relationship between the thickness of sandstone near the well and its components is analyzed in three-dimensional space to obtain the weight coefficients of the sandstone thickness and the three principal components. The components include a first principal component, a second principal component, and a third principal component; Finally, by utilizing the above-mentioned correspondence between sandstone thickness and components, the multi-attribute seismic data is reduced in dimensionality and denoised, thereby enabling effective planar prediction of the thickness of transitional channel sand bodies.

12. The method for seismic prediction of channel sand bodies using a dense well network geological model and seismic waveform dual constraints as described in claim 9, characterized in that: The aforementioned thin interbedded channel reservoir prediction method using model-constrained characteristic curve simulation includes: First, due to the complex structure of thin interbedded channel sand bodies and the influence of thin-layer seismic interference factors, the seismic attributes of the target layer contain reservoir information of the upper and lower layers. Through well-seismic matching unit analysis, the correlation between the seismic attributes of the target layer and the sandstone thickness of the target layer and the upper and lower units is analyzed using the Pearson correlation coefficient method. Based on the unit information with the strongest well-seismic correlation, a reasonable prediction unit in the vertical direction is determined, and an initial framework model is established based on the prediction unit division results. Secondly, the large overlap of impedance curves in thin interbedded sandstone and mudstone leads to low reservoir prediction accuracy. By interpreting lithological constraints through well logging, combining spontaneous potential and resistivity curves, and using multiple regression or neural network algorithms to reconstruct the curves, sensitive characteristic curves that can distinguish between sandstone and mudstone can be obtained. Finally, the Markov chain Monte Carlo stochastic simulation algorithm was used to perform three-dimensional fine prediction of the sensitive characteristic curves of thin interlayered channel sand bodies.

Images