Thin reservoir prediction method based on frequency division phase control
Through frequency-dividing phase control technology, low-frequency models are established and geological statistical inversion are performed, the problem of fine characterization of thin-layer reservoirs is solved, efficient reservoir identification and development is achieved, and exploration risks are reduced.
Patent Information
- Application Number
- CN202310623431.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-30
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2043-05-30
AI Technical Summary
The existing thin reservoir prediction technology has shortcomings in fine characterization and high-resolution description, which cannot meet the fast-paced needs of reservoir development, and the narrow frequency band and low main frequency of earthquake data affect the reliability of reservoir prediction.
Using a thin reservoir prediction method based on frequency-dividing phase control, a low-frequency model driven by big data waveform is established by using quality control and consistency processing of well logging data, and combining variable γ pre-stack deterministic inversion and geological statistical inversion Exercise, accurately depict the internal details of the thin sand layer, and expand to the full-band geological model.
It realizes fine characterization and identification of thin-layer reservoirs, reduces exploration risks, is suitable for fast-paced reservoir development, and improves the reliability and accuracy of predictions.
Smart Images

Figure CN119064998A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of seismic reservoir prediction, and more particularly to a thin reservoir prediction method based on frequency division phase control. Background Art
[0002] At present, domestic and foreign scholars have conducted a lot of research on thin reservoir identification and prediction. Generally speaking, it can be divided into two categories: qualitative or semi-quantitative thin reservoir prediction technology based on geophysical attributes and thin reservoir prediction technology based on comprehensive geological description of single karst transformation.
[0003] The qualitative or semi-quantitative thin reservoir prediction technology based on geophysical attributes is based on the large amount of geological information implicit in the original seismic data. By extracting seismic attributes, the reservoir lithology and physical properties reflected by them are analyzed to predict the deposition and oil and gas content of the reservoir. Due to the wide variety of seismic attributes and the great differences in the extracted seismic attributes, how to extract and optimize the characteristic attributes has become a difficult problem in realizing the identification and prediction of thin reservoirs.
[0004] The single karst transformation thin reservoir prediction technology is mainly based on high-resolution seismic data processing and inversion. Since thin reservoir research needs to be carried out on the basis of higher resolution, the existing seismic data often have narrow bandwidth and low main frequency, which cannot meet the accuracy of fine reservoir description. In order to solve this kind of problem, it is generally carried out in two aspects. On the one hand, it is through frequency extension processing of the original data, but this method has uncertainty in amplitude preservation and fidelity, which affects the reliability of reservoir prediction; on the other hand, from the seismic data acquisition stage, high-density seismic data acquisition and processing are used to improve the accuracy of seismic data imaging, but the cycle is long and cannot meet the fast-paced needs of reservoir development.
[0005] In summary, each of the above thin reservoir prediction methods has certain feasibility and limitations. Therefore, with the continuous deepening of oil field exploration and development, it is urgent to form a set of thin reservoir and unconventional reservoir inversion technology suitable for the fast-paced oil reservoir development. Summary of the invention
[0006] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a thin reservoir prediction method based on frequency division phase control, which establishes a variable γ prestack deterministic inversion for thin reservoirs based on a big data waveform-driven low-frequency model, performs phase-controlled prestack geostatistical inversion combination, accurately depicts the internal details of thin sand layers, and solves the problem of fine characterization and identification of thin-layer oil reservoirs.
[0007] To achieve the above object, the present invention provides the following technical solution: a thin reservoir prediction method based on frequency division phase control, which comprises the following steps:
[0008] S1. Perform quality control and consistency processing on well logging data. Perform quality control and optimization processing on pre-stack seismic data to improve the accuracy of pre-stack seismic data.
[0009] S2. Establish a low-frequency model based on the data processed in S1.
[0010] S3, based on the low-frequency model and combined with the data processed in S1, an improved variable-γ prestack joint deterministic inversion is carried out to obtain the low-frequency and medium-frequency models. The formula is as follows:
[0011] V P (t) = V P (t0)exp[2∫0 t R p (τ)d τ ]
[0012] V s (t) = V s (t0)exp[2∫0 t R s (τ)d τ ]
[0013] V ρ (t) = V ρ (t0)exp[2∫0 t R ρ (τ)d τ ]
[0014] Where t is the time sampling point, t0 is the initial time, τ = τ0 (s-1), s is the parameter of medium inhomogeneity, V P is the longitudinal wave velocity, V s is the shear wave velocity, V ρ is the density, R p is the longitudinal wave velocity reflection coefficient, R s is the shear wave velocity reflection coefficient, R ρ is the density reflection coefficient.
[0015] R p , R s , R ρ , obtained by the following formula:
[0016] R pp (θ) = (1 + tan 2 θ)R p -8γsin 2 θR s +(1-4γsin 2 θ)R ρ
[0017] Where θ is the incident angle, γ is the ratio of longitudinal and transverse wave velocities, and Rpp (θ) is the longitudinal wave reflection coefficient that changes with angle, θ, γ and R pp (θ) are obtained based on the data processed by S1.
[0018] S4. Carry out sand body thickness characterization based on the inversion results and produce a plane sand-to-ground ratio map.
[0019] S5. Based on the inversion results of S3, geostatistical inversion is performed with the plane sand-to-ground ratio map as a constraint, and the inversion results are expanded to high frequencies to obtain a full-band geological model.
[0020] The present invention is further configured as follows: S2 specifically comprises the following steps: first, performing big data analysis on the seismic waveform in the S1 result to obtain low-frequency data. Then, based on the longitudinal wave impedance, longitudinal-short wave velocity ratio, and density at the well point in the S1 result, the physical properties and structural characteristics of the underground medium are characterized, the low-frequency data are limited, and a low-frequency model is obtained.
[0021] The present invention is further configured as follows: after the low-frequency model is completed in S2, it is necessary to analyze whether the low-frequency model is consistent with geological knowledge. If it is consistent, it proceeds to S3, and if it is not consistent, it returns to S1. The judgment of whether the low-frequency model is consistent with geological knowledge includes verification based on the results of S1 and checking the consistency of various parts of the model.
[0022] The present invention is further configured as follows: when S2 establishes a low-frequency model, it is necessary to infer unknown seismic attributes based on known seismic attributes, and the same type of seismic attributes of different wells are obtained through a spatial conversion formula, which is as follows:
[0023]
[0024] In the formula, S n and S i is the seismic attribute of different wells, S i Obtained from S1 results. W is the relationship function between wells, W is inferred from S1 results, and n is the number of wells.
[0025] The present invention is further configured as follows: when S2 establishes a low-frequency model, it is necessary to infer unknown seismic attributes based on known seismic attributes, and different types of seismic physical properties of the same well are obtained through a physical property conversion formula, which is as follows:
[0026]
[0027] Where, L j and L i are different types of seismic physical properties, L i Obtained from S1 results. ij is the attribute-related matching function, λ ij Obtained by inferring the results of S1, where n is the number of wells.
[0028] The present invention is further configured as follows: when S2 establishes a low-frequency model, it is necessary to infer unknown seismic attributes based on known seismic attributes, and seismic physical properties of different types in different wells are obtained through a conversion formula for physical property parameters of the entire area, which is as follows:
[0029]
[0030] In the formula, SL j and SL i are the seismic physical properties of different types in different wells, w ij For SL j and SL i The corresponding well relationship function, w ij Obtained by inference from S1 results. ij is the attribute-related matching function, λ ij Obtained by inferring the results of S1, where n is the number of wells.
[0031] The present invention is further configured as follows: the geostatistical inversion in S5 adopts a variogram, and the variogram formula is as follows:
[0032]
[0033] Where h is the given step size, N(h) is the number of point pairs whose distance is equal to h, and Z(x i ) is at point x i The measured value of the variable, Z(x i +h) is the point x i The measured value of the variable at the deviation h, r(h) is the inversion result.
[0034] The present invention is further configured as follows: the processing of the logging data in S1 includes logging curve error correction and multi-well consistency processing.
[0035] The present invention is further configured as follows: the steps of logging curve error correction are:
[0036] For the layers with obvious deformation in the wellbore, Castagna formula or Smiths formula is used for correction.
[0037] Castagna formula: Where p is the formation density. Vp is the P-wave velocity. a, b, and c are all constants, obtained by analyzing the logging data of the undeformed layer of the wellbore.
[0038] Smiths formula: S(v) = dR e , where S(v) is the acoustic velocity. R is the resistivity. d and e are constants, obtained from the logging data of the undeformed layer of the wellbore.
[0039] For the layers with obvious collapse in the wellbore, the volume percentage model of logging response is used for correction. The formula is:
[0040] Where Ci is the logging response value of the non-collapsed well section of the i-th rock component. Vi is the volume percentage of the i-th rock component. C is the forward logging response value.
[0041] The present invention is further configured as follows: the multi-well consistency processing formula is: Where Z(x, y) is the earthquake attribute under study, (x, y) is the coordinate of a point on the plane, is the trend value, and g is the correction value.
[0042] Perform regression analysis on the known Z(x, y) in the logging data to obtain the regression equation f(x, y), and calculate the trend value based on f(x, y) Compare Z(x,y) with Subtract them and find g.
[0043] The present invention is further configured as follows: the processing of pre-stack seismic data in S1 includes pre-stack random noise suppression and time difference correction.
[0044] The present invention is further configured as follows: the pre-stack random noise suppression uses the GeoEast system to perform three-dimensional pre-stack random noise attenuation.
[0045] The present invention is further configured as follows: in S1, a non-rigid matching method is used to perform time difference correction on pre-stack seismic data.
[0046] The present invention is further configured as follows: the plane sand-to-land ratio map is drawn using a grid interpolation method. First, the plane sand-to-land ratio map is divided into a series of small grids, and the sand content value of each sample point is assigned to the corresponding grid node. Then, according to the spatial position and sand content value of the known data point, the sand content value of each grid node is calculated using an interpolation method, and a contour map of the sand content distribution is drawn.
[0047] The present invention is further configured as follows: the grid interpolation method is a linear interpolation method for contour mapping based on triangular grids.
[0048] In summary, compared with the prior art, the present invention has the following beneficial effects: for thin reservoirs, the present invention establishes a variable-gamma prestack deterministic inversion based on a big data waveform-driven low-frequency model, performs a phase-controlled prestack geostatistical inversion combination, accurately depicts the internal details of thin sand layers, solves the problem of fine characterization and identification of thin-layer reservoirs, and combines multiple methods such as amplitude attributes and oil and gas detection for comprehensive qualitative and quantitative mutual verification, solves the geological problem of increasing production and speeding up, reduces exploration risks, and has universal applicability and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] Figure 1 It is a schematic diagram of the process of an embodiment;
[0050] Figure 2 It is a comparison of pre-stack seismic data before and after processing;
[0051] Figure 3 Fine calibration of well seismic data for Well TK235;
[0052] Figure 4 Driving geological framework models for big data;
[0053] Figure 5 This is the distribution map of CK1-1 sand body in the East 1st District of Sangtamu area in Tahe Oilfield;
[0054] Figure 6 This is the distribution map of CK5-2 sand body in East Area 1 of Sangtamu area in Tahe Oilfield;
[0055] Figure 7 This is a schematic diagram of the characterization of the TKC1-2H well in the Carboniferous thin reservoir in Tahe;
[0056] Figure 8 Schematic diagram of the comprehensive characterization results of the Carboniferous thin reservoir in Tahe. DETAILED DESCRIPTION
[0057] The technical solution of the present invention will be clearly described below in conjunction with the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention, and all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the invention.
[0058] Example
[0059] This embodiment was applied in the Sangta Woodworking Area, which is the most enriched block of clastic oil and gas reservoirs in the Tahe Oilfield. Many Ordovician transit wells show good oil and gas in the Carboniferous, but the development has not been seen on a large scale. The Carboniferous Karashayi Formation is a sea-continent transition zone with frequent interlayers of sandstone and mudstone, and the overall structure is mud-enclosed sand. The reservoir has the characteristics of rapid lateral changes and small impedance differences between sand and mud. There is no obvious response feature on the seismic profile, and it is difficult to describe the distribution characteristics of thin reservoirs, which restricts the effect of reservoir development.
[0060] like Figure 1 FIG. 1 is a flow chart of a preferred embodiment of the present invention, a thin reservoir prediction method based on frequency division phase control, which comprises the following steps:
[0061] S1. Perform quality control and consistency processing on logging data, including logging curve error correction and multi-well consistency processing. Figure 2As shown, the pre-stack seismic data are quality controlled and optimized to improve their accuracy, including pre-stack random noise suppression and time difference correction.
[0062] S2. Build a low-frequency model based on the data processed in S1. Based on the results of S1, the seismic waveforms, such as Figure 3 As shown in the figure, a big data analysis is performed to obtain low-frequency data; then, based on the P-wave impedance, P-wave velocity ratio, and density at the well point in the S1 result, a detailed interpretation of the standard layer and section is performed to characterize the physical properties and structural characteristics of the underground medium, and the low-frequency data is limited to obtain a low-frequency model, such as Figure 4 shown.
[0063] S3, based on the low-frequency model and combined with the data processed in S1, an improved variable-γ prestack joint deterministic inversion is carried out to obtain the low-frequency and medium-frequency models. The formula is as follows:
[0064] V P (t) = V P (t0)exp[2∫0 t R p (τ)d τ ]
[0065] V s (t) = V s (t0)exp[2∫0 t R s (τ)d τ ]
[0066] V ρ (t) = V ρ (t0)exp[2∫0 t R ρ (τ)d τ ]
[0067] Where t is the time sampling point, t0 is the initial time, τ = τ0 (s-1), s is the parameter of medium inhomogeneity, V P is the longitudinal wave velocity, V s is the shear wave velocity, V ρ is the density, R p is the longitudinal wave velocity reflection coefficient, R s is the shear wave velocity reflection coefficient, R ρ is the density reflection coefficient;
[0068] R p , R s , R ρ , obtained by the following formula:
[0069] R pp (θ) = (1 + tan2 θ)R p -8γsin 2 θR s +(1-4γsin 2 θ)R ρ
[0070] Where θ is the incident angle, γ is the ratio of longitudinal and transverse wave velocities, and R pp (θ) is the longitudinal wave reflection coefficient that changes with angle, θ, γ and R pp (θ) are obtained based on the data processed by S1.
[0071] S4. Carry out sand body thickness characterization based on the inversion results and make a plane sand-to-ground ratio map, such as Figure 5 , 6 The plane sand-to-land ratio map is drawn using the grid interpolation method; first, the plane sand-to-land ratio map is divided into a series of small grids, and the sand content value of each sample point is assigned to the corresponding grid node; then, according to the spatial position and sand content value of the known data points, the sand content value of each grid node is calculated using the interpolation method, and the contour map of the sand content distribution is drawn.
[0072] S5, such as Figure 7 , 8 As shown in the figure, according to the S3 inversion results, the geostatistical inversion is performed with the plane sand-to-ground ratio map as a constraint, and the inversion results are extended to high frequencies to obtain a full-band geological model. The geostatistical inversion uses a variogram, and the variogram formula is as follows:
[0073]
[0074] Where h is the given step size, N(h) is the number of point pairs whose distance is equal to h, and Z(x i ) is at point x i The measured value of the variable, Z(x i +h) is the point x i The measured value of the variable at the deviation h, r(h) is the inversion result.
[0075] Specifically, for the S1 result, it is necessary to determine whether it conforms to the well-seismic calibration, AVO law, and geological understanding. If it conforms, proceed to S2. If it does not conform, proceed to S1 again. For S2, it is necessary to analyze whether the low-frequency model conforms to the geological understanding. If it conforms, proceed to S3. If it does not conform, return to S1. The judgment of whether the low-frequency model conforms to the geological understanding includes verification based on the S1 result and checking the consistency of each part of the model. For the S5 result, it is necessary to analyze whether it meets the geological understanding. If it conforms, output the result. If it does not conform, return to S3.
[0076] Specifically, when S2 establishes a low-frequency model, it needs to infer unknown earthquake attributes based on known earthquake attributes.
[0077] The same type of seismic attributes of different wells are obtained through the spatial transformation formula, which is as follows:
[0078]
[0079] In the formula, S n and S i is the seismic attribute of different wells, S i Obtained according to the results of S1; W is the relationship function between wells, W is inferred by the results of S1, and n is the number of wells.
[0080] Different types of seismic physical properties of the same well are obtained through the physical property conversion formula, which is as follows:
[0081]
[0082] Where, L j and L i are different types of seismic physical properties, L i Obtained according to S1 results; ij is the attribute-related matching function, λ ij Obtained by inferring the results of S1, where n is the number of wells.
[0083] The seismic physical properties of different types in different wells are obtained through the conversion formula of the physical property parameters of the whole area. The formula is as follows:
[0084]
[0085] In the formula, SL j and SL i are the seismic physical properties of different types in different wells, w ij For SL j and SL i The corresponding well relationship function, w ij Obtained by inferring the results of S1, where n is the number of wells.
[0086] Specifically, the error correction of logging curves in S1 is that for the layer sections with obvious deformation of the wellbore, the gamma curve, resistivity curve, etc. are less affected by the wellbore collapse and alteration and have better quality, and the Castagna formula or Smiths formula is used for correction.
[0087] Castagna formula: Where p is the formation density. Vp is the P-wave velocity. a, b, and c are all constants, obtained by analyzing the logging data of the undeformed layer of the wellbore.
[0088] Smiths formula: S(v) = dR e , where S(v) is the acoustic velocity. R is the resistivity. d and e are constants, obtained from the logging data of the undeformed layer of the wellbore.
[0089] For the layer sections with obvious collapse of the wellbore, the volume model of the correction curve is first established. According to the components, fluid volume content and corresponding logging response values, the functional response relationship between the target curve and the reference curve is established. The volume percentage model of the logging response is used for correction. The formula is:
[0090] Where Ci is the logging response value of the non-collapsed well section of the i-th rock component. Vi is the volume percentage of the i-th rock component. C is the forward logging response value.
[0091] In S1, the multi-well consistency processing uses the trend surface analysis method to perform multi-well correction. By statistically analyzing the mean response of the standard layer of several key wells as the logging response standard at the well point, a "trend surface" is established to achieve the same mean response of the standard layer of multiple wells as the trend surface, eliminating the well-to-well anomalies caused by factors such as instrument system errors. The formula is as follows:
[0092] Where Z(x, y) is the earthquake attribute under study, (x, y) is the coordinate of a point on the plane, is the trend value, and g is the correction value.
[0093] Perform regression analysis on the known Z(x, y) in the logging data to obtain the regression equation f(x, y), and calculate the trend value based on f(x, y) Compare Z(x,y) with Subtract them and find g.
[0094] The prestack random noise suppression in S1 uses the GeoEast system to perform three-dimensional prestack random noise attenuation. The principle is that the three-dimensional prestack seismic data can be transformed into the frequency-space domain through Fourier transform. For each frequency f component S(i, j, k), a prediction operator U(l, j, k) can be designed using the least squares principle. Its prediction error energy Q(f) is expressed as the following formula:
[0095]
[0096]
[0097] is the conjugate of V(i,j,k), where L, M, and N are the lengths of the three-dimensional pre-stack seismic data spatial window (longitudinal survey line, transverse survey line, and shot offset).
[0098] Take partial derivatives of the above two formulas and let
[0099] In the formula, p = 1, 2, 3, ..., LOPx; q = 1, 2, 3, ..., LOPy; r = 1, 2, 3, ..., LOPz. LOPx, LOPy, LOPz are the lengths of the prediction operators in the three directions of space, and they are usually of the same length. Finally, we get a matrix equation:
[0100] H·U=R
[0101] Where H and R are the covariance matrices from 3D prestack seismic data; U is the 3D prediction operator.
[0102] For each frequency component, the prediction operator is obtained by solving the matrix equation, and then the prediction operator is used to perform prediction filtering on the seismic data. Finally, the inverse Fourier transform is used to transform back to the time-space domain to obtain the denoised result.
[0103] In S1, the non-rigid matching method is used to correct the time difference of pre-stack seismic data. The implementation process is as follows:
[0104] Firstly, the pre-stack seismic data are sorted by the channel number, the far-offset NMO distortion band is removed, and the multiple waves are removed and then stacked. The stacked data are used as the reference data of NRM, and the amplitude in the reference data is regarded as a three-dimensional pixel, denoted as Sref(x, y, z, T) (the variables x, y, z represent the spatial coordinates, and T represents the time shift between two blocks of acquired data).
[0105] Then, the displacement field is edited to delete abnormal high-frequency changes. Multi-solution gradient technology is applied to displacement field estimation. The displacement field calculates the position difference of pixels between two data volumes to achieve the basic consistency of pixel positions between the two data volumes.
[0106] Finally, all sample points of the pre-stack gather data are corrected using the optimized displacement field. The sample point values before and after correction satisfy Sm(x,y,z,T)=S(x+dx(x,y,z,T)+dz(x,y,z,T),T). In the formula, S(x,y,z,T) and Sm(x,y,z,T) represent the data before and after correction, respectively.
[0107] After the optimization processing of pre-stack gathers, the consistency of seismic data becomes better, the boundaries of amplitude attribute anomalies are clear, the detailed characteristics of amplitude attributes are rich, and the overall quality is improved.
[0108] pass Figure 2It can be seen that after the pre-stack gather optimization processing, the problems of low signal-to-noise ratio, uneven phase axis and inconsistent near- and far-track spectra of the original gathers have been effectively solved, laying a good foundation for providing high-quality data for pre-stack inversion, etc. The thin reservoir prediction results based on the optimized gathers are more consistent with the well information, which also verifies the rationality of this gather optimization processing.
[0109] Specifically, in S4, firstly, d=H i / H j Calculate, H i is the total thickness of sandstone, H j The sandstone thickness and formation thickness of each well are extracted by using the linear interpolation method of contour mapping based on triangular meshes, and the plane sand-to-ground ratio map is drawn through five steps: boundary search, triangulation, linear interpolation, contour algorithm, and Bezier curve smoothing.
[0110] Boundary search: The boundaries of data points can be read manually or identified by algorithms. For large amounts of data, the convex hull scanning algorithm invented by Professor Graham is used to improve efficiency.
[0111] Triangulation: Triangulation of triangular meshes is based on the Delaunay criterion. Under the above boundary control, the data nodes are meshed, and the vertices of the triangular meshes after triangulation are the locations of the data nodes.
[0112] Linear interpolation: Add interpolation points inside the original mesh through linear interpolation. Connect the interpolation points with the three vertices of the original mesh to obtain three new triangular units.
[0113] Contour line algorithm: Through the third step above, a small triangular mesh has been formed. For each contour line, all triangular meshes are searched to find the corresponding isovalue points of the contour line, and all isovalue points are connected to obtain an isovalue line. Repeat the operation to obtain all contour lines. Further obtain the positional relationship between the contour line and the triangular unit.
[0114] Bezier curve smoothing: The contour lines obtained in the above steps are composed of broken lines, and the contour lines need to be smoothed. The quadratic Bezier curve smoothing is defined as: B(t) = (1-t) 2 P0+2t(1-t)P1+t 2 P2, where t∈[0,1]. Through the above steps, the sand-to-soil ratio contour map of the area is obtained, which provides a trend for the spatial lateral distribution of the next high-resolution reservoir inversion.
[0115] In summary, this embodiment establishes a variable γ prestack deterministic inversion based on a big data waveform-driven low-frequency model for thin reservoirs, performs phase-controlled prestack geostatistical inversion combination, accurately depicts the internal details of thin sand layers, solves the problem of fine characterization and identification of thin-layer reservoirs, and combines multiple methods such as amplitude attributes and oil and gas detection for comprehensive qualitative and quantitative mutual verification, solves the geological problem of increasing production and speeding up, reduces exploration risks, and has universal applicability and broad application prospects.
[0116] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A thin reservoir prediction method based on frequency division phase control, characterized in that: The steps include: S1. Perform quality control and consistency processing on well logging data; perform quality control and optimization processing on pre-stack seismic data; S2, establishing a low-frequency model based on the data processed in step S1; S3, based on the low-frequency model and combined with the data processed in S1, an improved variable-γ prestack joint deterministic inversion is carried out to obtain the low-frequency and medium-frequency models. The formula is as follows: Where t is the time sampling point, t0 is the initial time, τ = τ0 (s-1), s is the parameter of medium inhomogeneity, V P is the longitudinal wave velocity, V s is the shear wave velocity, V ρ is the density, R p is the longitudinal wave velocity reflection coefficient, R s is the shear wave velocity reflection coefficient, R ρ is the density reflection coefficient; R P , R s , R ρ , obtained by the following formula: R pp (θ)=(1+an 2 i)R p -8γsin 2 θR s +(1-4sin 2 i)R ρ Where θ is the incident angle, γ is the ratio of longitudinal and transverse wave velocities, and R pp (θ) is the longitudinal wave reflection coefficient that changes with angle, θ, γ and R pp (θ) were obtained based on the data processed by S1; S4. Carry out sand body thickness characterization based on the inversion results and produce a plane sand-to-ground ratio map; S5. Based on the inversion results of S3, geostatistical inversion is performed with the plane sand-to-ground ratio map as a constraint, and the inversion results are expanded to high frequencies to obtain a full-band geological model.
2. A thin reservoir prediction method based on frequency division phase control according to claim 1, characterized in that: The specific steps of S2 are: first, perform big data analysis on the seismic waveforms in the S1 results to obtain low-frequency data; then, based on the longitudinal wave impedance, longitudinal and shear wave velocity ratio and density at the well point in the S1 results, characterize the physical properties and structural characteristics of the underground medium, limit the low-frequency data, and obtain a low-frequency model.
3. A thin reservoir prediction method based on frequency division phase control according to claim 2, characterized in that: After the S2 low-frequency model is completed, it is analyzed whether the low-frequency model is consistent with the geological understanding. If it is consistent, S3 is carried out. If it is not consistent, it returns to S1. The judgment of whether the low-frequency model is consistent with the geological understanding includes verification based on the S1 results and checking the consistency of each part of the model.
4. The thin reservoir prediction method based on frequency division phase control according to claim 2 is characterized in that: When S2 establishes a low-frequency model, the unknown seismic attributes are inferred based on the known seismic attributes. The same type of seismic attributes of different wells are obtained through the spatial conversion formula, which is as follows: In the formula, S n and S i is the seismic attribute of different wells, S i Obtained according to the results of S1; W is the relationship function between wells, W is inferred by the results of S1; n is the number of wells.
5. The thin reservoir prediction method based on frequency division phase control according to claim 2 is characterized in that: When S2 establishes a low-frequency model, the unknown seismic attributes are inferred based on the known seismic attributes. The seismic physical properties of different types in the same well are obtained through the physical property conversion formula, which is as follows: Where, L j and L i are different types of seismic physical properties, L i Obtained according to S1 results; ij is the attribute-related matching function, λ ij Obtained by inferring the results of S1; n is the number of wells.
6. The thin reservoir prediction method based on frequency division phase control according to claim 2 is characterized in that: When S2 establishes a low-frequency model, the unknown seismic attributes are inferred based on the known seismic attributes. The seismic physical properties of different types in different wells are obtained through the full-area physical property parameter conversion formula, which is as follows: In the formula, SL j and SL i are the seismic physical properties of different types in different wells, w ij For SL j and SL i The corresponding well relationship function, w ij Obtained by inference from S1 results; ij is the attribute-related matching function, λ ij Obtained by inferring the results of S1; n is the number of wells.
7. The thin reservoir prediction method based on frequency division phase control according to claim 1 is characterized by: The geostatistical inversion in S5 uses the variogram, and the variogram formula is as follows: Where h is the given step size, N(h) is the number of point pairs whose distance is equal to h, and Z(x i ) is at point x i The measured value of the variable, Z(x i +h) is the point x i The measured value of the variable at the deviation h, r(h) is the inversion result.
8. The thin reservoir prediction method based on frequency division phase control according to claim 1 is characterized by: The processing of logging data in S1 includes logging curve error correction and multi-well consistency processing.
9. A thin reservoir prediction method based on frequency division phase control according to claim 8, characterized in that: The steps of logging curve error correction are: For the wellbore with deformed layers, Castagna formula or Smiths formula is used for correction; Castagna formula: Where p is the formation density; Vp is the P-wave velocity; a, b, and c are all constants, obtained by analyzing the logging data of the undeformed layer of the wellbore; Smiths formula: S(v) = dR e , where S(v) is the acoustic wave velocity; R is the resistivity; d and e are constants, obtained by analyzing the logging data of the undeformed layer of the wellbore; For the collapsed layer section of the wellbore, the volume percentage model of the logging response is used for correction, and the formula is: Where Ci is the logging response value of the non-collapsed well section of the i-th rock component; Vi is the volume percentage of the i-th rock component; and C is the forward logging response value.
10. The thin reservoir prediction method based on frequency division phase control according to claim 8, characterized in that: The multi-well consistency processing formula is: Where Z(x, y) is the earthquake attribute under study, (x, y) is the coordinate of a point on the plane, is the trend value, g is the correction value; Perform regression analysis on the known Z(x, y) in the logging data to obtain the regression equation f(x, y), and calculate the trend value based on f(x, y) Compare Z(x,y) with Subtract them and find g.
11. The thin reservoir prediction method based on frequency division phase control according to claim 1, characterized in that: The processing of prestack seismic data in S1 includes prestack random noise suppression and time difference correction.
12. A thin reservoir prediction method based on frequency division phase control according to claim 11, characterized in that: Prestack random noise suppression The GeoEast system is used to perform three-dimensional prestack random noise attenuation.
13. The thin reservoir prediction method based on frequency division phase control according to claim 11, characterized in that: In S1, the non-rigid matching method is used to perform time difference correction on pre-stack seismic data.
14. The thin reservoir prediction method based on frequency division phase control according to claim 1, characterized in that: The plane sand-to-land ratio map is drawn using the grid interpolation method; first, the plane sand-to-land ratio map is divided into a series of small grids, and the sand content value of each sample point is assigned to the corresponding grid node; then, based on the spatial position and sand content value of the known data points, the sand content value of each grid node is calculated using the interpolation method, and the contour map of the sand content distribution is drawn.
15. A thin reservoir prediction method based on frequency division phase control according to claim 14, characterized in that: The grid interpolation method is a linear interpolation method for contour mapping based on triangular grids.
Citation Information
Patent Citations
Seismic frequency-broadening processing-based phase-controlled random inversion thin reservoir prediction method
CN106154323A
Prestack geostatistical inversion method under three-dimensional double control
CN109521474A
Waveform inversion reservoir prediction method based on frequency division
CN110907994A
Method and system for predicting effective reservoir under phased forward constraints
CN111090117A
Phase control type dolomite reservoir earthquake prediction method and device based on deposition parameters
CN111983679A
Cited By
Reservoir prediction method and device based on deposition and earthquake dual forward modeling combination
CN119916483A
