A method for identifying lithology subdivision of tight glutenite based on conventional logging curves and imaging characteristic parameters

Abstract

The application discloses a method for identifying and subdividing dense sandstone conglomerate lithology based on conventional logging curves and imaging characteristic parameters, and comprises the following steps: obtaining a lithology sensitive curve; reducing the resolution of the electric imaging processing interpretation parameter curve by using a matching filter method, so that the resolution corresponds to the longitudinal resolution of the conventional logging curve; obtaining a support vector machine lithology identification model; and realizing automatic subdivision identification and division of the dense sandstone conglomerate by using the lithology identification model. The application makes full use of abundant stratum information provided by the imaging logging data, determines the conglomerate content by using a full-coverage borehole wall imaging threshold segmentation technology, obtains a sorting coefficient representing the heterogeneity degree of the stratum by using resistivity spectrum analysis, and then realizes the division of the sandstone conglomerate, the sandstone containing conglomerate and the sandstone. The resolution of the electric imaging processing interpretation parameter curve is reduced to the level of the conventional logging curve by using the matching filter technology, and the automatic layering reading is realized, so that the conventional and imaging logging processing results are simultaneously used to identify the target of the complex lithology.

Description

Technical Field

[0001] This invention relates to the field of unconventional tight gas exploration and development technology, specifically to a method for identifying the lithology of tight sandstone and conglomerate based on conventional logging curves and imaging characteristic parameters. Background Technology

[0002] With the continuous advancement of natural gas exploration in major oilfields across China and the improvement of geological understanding and resource exploration, tight sandstone and conglomerate reservoirs have become important exploration targets and resource replacement areas as important unconventional natural gas resources.

[0003] Accurate identification and classification of tight sandstone and conglomerate lithology is a crucial foundation for evaluating tight gas sweet spots. However, in China, conventional cross-plot methods are commonly used for lithology identification, which have low discriminant accuracy and cannot characterize lithology down to the grain size, making effective subdivision of sandstone, conglomerate, gravelly sandstone, and fine-grained sandstone difficult. In recent years, new technologies such as imaging logging have become powerful tools for lithology identification, but the mismatch between their vertical resolution and conventional logging curves hinders the combined identification of lithology using imaging logging and conventional logging. Summary of the Invention

[0004] The purpose of this invention is to provide a method for fine-grained identification of tight sandstone and conglomerate lithology based on conventional well logging curves and imaging characteristic parameters. This method can characterize the sandstone and conglomerate down to the grain size, enabling fine-grained identification of tight sandstone and conglomerate lithology. This improves the precision and reliability of identification of complex sandstone and conglomerate lithology and lays the foundation for the selection of superior lithologies.

[0005] To achieve the above objectives, this application proposes a method for lithological subdivision and identification of tight sandstone and conglomerate based on conventional well logging curves and imaging characteristic parameters, including:

[0006] Obtain lithology-sensitive curves, including conventional logging curves and electrical imaging processing and interpretation parameter curves;

[0007] The resolution of the electrical imaging processing interpretation parameter curve is reduced by using matched filtering to make it correspond to the longitudinal resolution of the conventional logging curve.

[0008] Obtain a support vector machine lithology identification model;

[0009] The lithological identification model enables automatic subdivision and classification of dense sandstone and conglomerate.

[0010] Furthermore, the acquisition of the conventional logging curves specifically involves: performing conventional logging response characteristic analysis on the lithology of the target strata in the study area, and selecting four lithology-sensitive parameter curves—deep lateral resistivity Rt, natural gamma ray GR, longitudinal wave time DT, and compensated neutron porosity CNL—using the cross-plot method.

[0011] Furthermore, the acquisition of the electro-imaging processing interpretation parameters includes determining the gravel content curve Vgra using full-coverage wellbore imaging threshold segmentation technology: firstly, the resistivity image after LLS calibration of the shallow lateral resistivity curve is filled with blank stripes between the electro-imaging plates by the Filtersim algorithm, and then the threshold of high-resistivity gravel is determined by the conductivity histogram to obtain the gravel content curve Vgra.

[0012] Furthermore, the acquisition of the electro-imaging processing interpretation parameters also includes obtaining the sorting coefficient curve SORT using imaging resistivity spectrum analysis: The electro-imaging logging data is displayed in the form of a variable-density logging curve, and statistical analysis is performed on the imaging resistivity data to divide the resistivity spectrum into: a matrix component, a relatively high-resistivity component, and a relatively high-conductivity component, thereby obtaining the sorting coefficient curve SORT, which reflects the homogeneity of the reservoir.

[0013] SORT = (70P - 30P) / 50P

[0014] Wherein, 70P is the resistivity when the cumulative frequency reaches 70%, 30P is the resistivity when the cumulative frequency reaches 30%, and 50P is the resistivity when the cumulative frequency reaches 50%. Under the same lithology, the larger the sorting index, the worse the sorting of the reservoir, and vice versa.

[0015] Furthermore, the resolution of the electrical imaging processing interpretation parameter curve is reduced using matched filtering to correspond to the longitudinal resolution of the conventional logging curve. Specifically:

[0016] By using the high-resolution average resistivity curve Rb calibrated on the LLS scale of the shallow lateral resistivity curve and the matched filter factor F = f^(-1)(((G_lls)(ω)) / ((G_(R_b))(ω))), a low-resolution curve R_b^' = F*R_b is obtained that is consistent with the shallow lateral curve only in terms of longitudinal resolution, but differs in absolute value; where f^(-1) represents the inverse Fourier transform, (G_lls)(ω) and (G_(R_b))(ω) are the frequency domain longitudinal response functions of the shallow lateral and electrical imaging instruments, respectively, and * represents convolution; using the same matched filter factor, the gravel content curve V_gra^' = F*V_gra and the sorting coefficient curve SORT^' = F*SORT are obtained, corresponding to the longitudinal resolution of the conventional logging curve.

[0017] Furthermore, a support vector machine lithology identification model is obtained, specifically:

[0018] The deep lateral resistivity Rt, natural gamma ray GR, longitudinal wave transit time DT, compensated neutron porosity CNL, and gravel content V_gra^' and sorting coefficient 〖SORT〗^' with reduced resolution were obtained from rock type samples in the core well to establish a lithology identification training sample set.

[0019] The RBF kernel function is used, and the optimal penalty parameter c and kernel function parameter g are determined by cross-validation. The entire training sample set is then trained using the optimal penalty parameter c and kernel function parameter g to obtain the support vector machine lithology identification model.

[0020] Furthermore, the lithology identification model enables automatic subdivision and classification of tight sandstone and conglomerate. Specifically, it involves: automatically reading the lithology sensitivity curve in layers and establishing a prediction sample set; and using the support vector machine lithology identification model to predict the lithology of the prediction sample set, thereby achieving automatic subdivision and classification of tight sandstone and conglomerate.

[0021] Furthermore, the automatic stratified reading of the lithology-sensitive curve is specifically as follows:

[0022] The conventional logging curve is decomposed into a series of four basic forms: peaks, valleys, plateaus, and steps. The location of the formation interface between adjacent basic forms is determined by the location of the extreme points of the first derivative or the zero points of the second derivative, thus achieving automatic stratification. The values ​​of the formation parameters are taken according to the form, taking the maximum, minimum, or geometric mean (when multiple extreme points appear within the layer).

[0023] As a further step, the cubic spline function differentiation method is adopted, and adaptive filtering technology is used to reduce the drastic fluctuations in the first and second derivatives. After obtaining the first and second derivatives of the conventional logging curves, the lithology-sensitive curves are morphologically divided to determine the formation interfaces and parameter values ​​of each formation.

[0024] As a further step, the deep lateral resistivity curve Rt, natural gamma curve GR, longitudinal wave time difference curve DT, compensated neutron porosity curve CNL, and the gravel content curve V_gra^' and sorting coefficient curve〖SORT〗^' with reduced resolution are read in layers to obtain a prediction sample set.

[0025] Compared with existing technologies, the above-mentioned technical solutions adopted in this invention have the following advantages: This invention fully utilizes the rich formation information provided by imaging logging data, determines the gravel content using full-coverage wellbore imaging threshold segmentation technology, and obtains sorting coefficients characterizing the degree of formation heterogeneity using resistivity spectrum analysis, thereby achieving the differentiation of conglomerate, gravelly sandstone, and sandstone. Matched filtering technology reduces the resolution of the electrical imaging processing interpretation parameter curve to the level of conventional logging curves, and automatic layered readings achieve the goal of simultaneously identifying complex lithologies using both conventional and imaging logging processing results. Attached Figure Description

[0026] Figure 1 The diagram shown is a flowchart of the technical solution of this invention.

[0027] Figure 2 The figure shown is a cross plot of lithology-sensitive conventional logging parameters in an example of the present invention;

[0028] Figure 3 The image shown is a map illustrating the gravel content of a well determined using full-coverage wellbore imaging threshold segmentation technology, as illustrated in an example of this invention.

[0029] Figure 4 The figure shown is a sorting coefficient diagram calculated by resistivity spectrum analysis for a well in an example of the present invention;

[0030] Figure 5 The image shown is a lithology correction diagram, imaging, and core image of a well in an example of this invention.

[0031] Figure 6 The image shown is a lithology correction diagram, imaging, and core image of a well in an example of this invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit the application; that is, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0033] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0034] Taking the tight sandstone and conglomerate reservoirs of the D fault depression in the southern Songliao Basin as an example, and in conjunction with the accompanying drawings and tables, the present invention will be further described in detail. Figure 1As shown, a method for fine-grained identification of tight sandstone and conglomerate lithology based on conventional well logging curves and imaging feature parameters specifically includes:

[0035] Step S1: Obtain lithology-sensitive curves, which include conventional logging curves and electrical imaging processing interpretation parameter curves;

[0036] Specifically, the conventional logging curve acquisition method is as follows: Conventional logging response characteristic analysis is performed on five main lithologies (submerged volcanic breccia, sandstone, conglomerate, gravelly sandstone, sandstone, and mudstone) in the target stratigraphic interval of the study area (see Table 1). The cross-plot method is used to select the curves for four lithology-sensitive parameters: deep lateral resistivity Rt, natural gamma ray GR, P-wave transit time DT, and compensated neutron porosity CNL (see Table 1). Figure 2 ).

[0037] Table 1. Conventional logging response characteristics of major lithologies in the Ying 2 Member of the Southern Songliao Basin

[0038]

[0039] The electro-imaging processing interpretation parameter curves include the gravel content curve V determined by full-coverage wellbore imaging threshold segmentation technology. gra And the SORT curve, calculated from resistivity spectrum analysis;

[0040] The gravel content curve V determined by the full-coverage wellbore imaging threshold segmentation technique gra Specifically, the process involves: first, using the Filtersim algorithm to fill the blank bands between the electrical imaging plates in the resistivity image after LLS calibration of the shallow lateral resistivity curve; then, determining the threshold for high-resistivity gravel using the conductivity histogram; and finally, obtaining the gravel content curve V. gra ,like Figure 3 As shown;

[0041] Figure 3 The first channel is the dynamic image channel; the second channel is the depth scale channel; the third channel is the full coverage image channel; the fourth channel is the threshold segmentation image channel; and the fifth channel is the gravel content channel.

[0042] The specific method for calculating the sorting coefficient curve SORT using resistivity spectrum analysis involves displaying electrical imaging logging data in the form of a variable density logging curve, such as... Figure 4 As shown, statistical analysis was performed on the imaging resistivity data, dividing the resistivity spectrum into three parts: the matrix part, the relatively high resistivity part, and the relatively high conductivity part, thus obtaining the sorting coefficient curve SORT, which reflects the homogeneity of the reservoir.

[0043]

[0044] Wherein, 70P is the resistivity when the cumulative frequency reaches 70%, 30P is the resistivity when the cumulative frequency reaches 30%, and 50P is the resistivity when the cumulative frequency reaches 50%. Under the same lithology, the larger the sorting index, the worse the sorting of the reservoir, and vice versa.

[0045] Figure 4 The first channel is the depth scale channel; the second channel is the dynamic image channel; the third channel is the resistivity spectrum image channel; and the fourth channel is the sorting coefficient channel.

[0046] Step S2: Use matched filtering to reduce the resolution of the electrical imaging interpretation parameter curve so that it corresponds to the longitudinal resolution of the conventional logging curve.

[0047] Specifically, the high-resolution average resistivity curve R of the coin with shallow lateral LLS scale. b and matched filter factor A low-resolution curve R' can be obtained that is consistent with the shallow lateral curve only in terms of longitudinal resolution, but differs in absolute value. b =F*R b Among them, f -1 This represents the inverse Fourier transform. and These are the frequency domain longitudinal response functions of the shallow lateral and electrical imaging instruments, respectively, with * indicating convolution. Using the same matched filter factor, a gravel content curve V' with longitudinal resolution comparable to conventional logging curves can be obtained. gra =F*V gra The sorting coefficient curve SORT' = F*SORT.

[0048] Step S3: Obtain the support vector machine lithology identification model;

[0049] Specifically, the following measurements were obtained from five cored wells: deep lateral resistivity (Rt), natural gamma ray (GR), longitudinal wave transit time (DT), compensated neutron porosity (CNL), and gravel content (V') calculated with reduced resolution, based on core and thin section data to determine rock type. gra Using the sorting coefficient calculation value SORT', a lithology identification training sample set is established.

[0050] The RBF kernel function was used, and the optimal penalty parameter c = 32 and kernel function parameter g = 0.125 were determined by cross-validation. The entire training sample set was then trained with the optimal parameters to obtain the support vector machine lithology identification model. The accuracy rate of the 278 training sample set was 95.2%.

[0051] Step 4: Automatic subdivision and classification of tight sandstone and conglomerate is achieved using the lithology identification model;

[0052] Specifically, conventional logging curves are decomposed into four basic forms: peaks, valleys, plateaus, and steps. The formation interface is determined between adjacent basic forms by using the extreme points of the first derivative or the zero points of the second derivative. This enables automatic stratification. The values ​​of each formation parameter are taken according to the form, including the maximum, minimum, or geometric mean (when multiple extreme points appear within the layer).

[0053] To improve the calculation accuracy of the first and second derivatives, the cubic spline function differentiation method is adopted, and adaptive filtering technology is used to reduce the drastic fluctuations in the first and second derivatives. After obtaining the first and second derivatives of the conventional logging curves, the lithology-sensitive parameter curves are morphologically divided to determine the formation interfaces and parameter values ​​of each formation.

[0054] This method was used to analyze the deep lateral resistivity curve Rt, natural gamma curve GR, P-wave transit time curve DT, compensated neutron porosity curve CNL, and gravel content curve V' with reduced resolution within the target layer. gra The sorting coefficient curve SORT' is used to perform hierarchical readings and establish a prediction sample set.

[0055] The support vector machine lithology identification model in step S3 is used to predict the lithology of the prediction sample set. Lithology predictions were performed on the seven wells encountered in the study area, and compared with the calibrated whole-wellbore electrical imaging and core samples. The lithology accuracy rate reached over 90%.

[0056] Figure 5 (a) and Figure 6 In (a), the first channel is the lithology logging curve channel; the second channel is the depth channel; the third channel is the logging lithology channel; the fourth channel is the lithology channel predicted by this invention; the fifth channel is the electrical logging curve channel; the sixth channel is the porosity logging curve channel; and the seventh channel is the logging interpretation conclusion channel.

[0057] Figure 5 (b) and Figure 6 (b) is Figure 5 (a) and Figure 6 (a) shows the electro-optical imaging pattern corresponding to the depth range of the rectangular blocks, where the first track is the static image track, the second track is the depth scale track, and the third track is the dynamic image track.

[0058] Figure 5 (c) and Figure 6 (c) is Figure 5 (a) and Figure 6 (a) Core photograph at a certain depth within the rectangular block.

[0059] This embodiment takes the tight sandstone and conglomerate reservoir in the D fault depression in the southern Songliao Basin as the research object. Field case processing and analysis using the method of this invention show that it can effectively identify five typical lithologies in the study area (submerged volcanic breccia, sandstone and conglomerate, gravelly sandstone, sandstone and mudstone). The consistency rate with the calibrated whole-wellbore electrical imaging map and core is over 90%, which plays an important role in the subsequent evaluation of the three qualities of tight gas.

[0060] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. A method for detailed identification of tight sandstone and conglomerate lithology based on conventional well logging curves and imaging feature parameters, characterized in that, include: Obtain lithology-sensitive curves, including conventional logging curves and electrical imaging processing and interpretation parameter curves; The resolution of the electrical imaging processing interpretation parameter curve is reduced by using matched filtering to make it correspond to the longitudinal resolution of the conventional logging curve. Obtain a support vector machine lithology identification model; The support vector machine lithology identification model is used to achieve automatic subdivision, identification, and classification of dense sandstone and conglomerate. The acquisition of the electrical imaging processing interpretation parameters includes determining the gravel content curve Vgra using full-coverage wellbore imaging threshold segmentation technology and obtaining the sorting coefficient curve SORT using imaging resistivity spectrum analysis. The resolution of the electrical imaging interpretation parameter curve is reduced using matched filtering to correspond to the longitudinal resolution of the conventional logging curve. Specifically: High-resolution average resistivity curve of the coin cell R, calibrated using the LLS scale with shallow lateral resistivity profiles. b and matched filter factor A low-resolution curve was obtained that is consistent with the shallow lateral curve only in terms of longitudinal resolution, but differs in absolute value. in, This represents the inverse Fourier transform. and The frequency domain longitudinal response functions of the shallow lateral and electrical imaging instruments are shown, respectively, with * indicating convolution. Using the same matched filter factor, a reduced-resolution gravel content curve corresponding to the longitudinal resolution of conventional logging curves is obtained. And the sorting coefficient curve with reduced resolution ; Obtain conventional logging curves of rock type samples from core wells, as well as calculated gravel content and sorting coefficients with reduced resolution, and establish a lithology identification training sample set. The automatic subdivision and classification of tight sandstone and conglomerate is achieved through the support vector machine lithology identification model. Specifically, the lithology sensitivity curve is automatically read in layers and a prediction sample set is established. The lithology of the prediction sample set is then predicted using the support vector machine lithology identification model to achieve automatic subdivision and classification of tight sandstone and conglomerate.

2. The method for subdividing and identifying the lithology of tight sandstone and conglomerate based on conventional well logging curves and imaging characteristic parameters according to claim 1, characterized in that, The acquisition of the conventional logging curves is specifically as follows: conventional logging response characteristic analysis is performed on the lithology of the target layer in the study area, and the curves of four lithology-sensitive parameters, namely deep lateral resistivity Rt, natural gamma ray GR, longitudinal wave time DT, and compensated neutron porosity CNL, are selected by cross plot method.

3. The method for detailed identification of tight sandstone and conglomerate lithology based on conventional well logging curves and imaging characteristic parameters according to claim 1, characterized in that, The method for obtaining the gravel content curve Vgra is as follows: First, the resistivity image after LLS calibration of the shallow lateral resistivity curve is filled with blank stripes between the electro-imaging plates by the Filtersim algorithm. Then, the threshold of high-resistivity gravel is determined by the conductivity histogram to obtain the gravel content curve Vgra.

4. The method for lithological subdivision and identification of tight sandstone and conglomerate based on conventional well logging curves and imaging characteristic parameters according to claim 1 or 2, characterized in that, The sorting coefficient curve (SORT) obtained using imaging resistivity spectrum analysis is as follows: Electrical imaging logging data is displayed as a variable-density logging curve, and statistical analysis is performed on the imaging resistivity data. The resistivity spectrum is divided into three parts: the matrix component, the relatively high-resistivity component, and the relatively high-conductivity component, thus obtaining the sorting coefficient curve (SORT) reflecting the homogeneity of the reservoir. SORT = (70P - 30P) / 50P Wherein, 70P is the resistivity when the cumulative frequency reaches 70%, 30P is the resistivity when the cumulative frequency reaches 30%, and 50P is the resistivity when the cumulative frequency reaches 50%. Under the same lithology, the larger the sorting index, the worse the sorting of the reservoir, and vice versa.

5. The method for subdividing and identifying the lithology of tight sandstone and conglomerate based on conventional well logging curves and imaging feature parameters according to claim 1, characterized in that, The conventional logging curves include deep lateral resistivity measurement value Rt, natural gamma measurement value GR, P-wave transit time measurement value DT, and compensated neutron porosity measurement value CNL; The RBF kernel function is used, and the optimal penalty parameter c and kernel function parameter g are determined by cross-validation. The entire training sample set is then trained using the optimal penalty parameter c and kernel function parameter g to obtain the support vector machine lithology identification model.

6. The method for detailed identification of tight sandstone and conglomerate lithology based on conventional well logging curves and imaging characteristic parameters according to claim 1, characterized in that, The automatic stratified reading of the lithology-sensitive curve is as follows: The conventional logging curve is decomposed into a series of four basic forms: peaks, valleys, plateaus, and steps. The location of the formation interface between adjacent basic forms is determined by using the extreme points of the first derivative or the zero points of the second derivative, thus achieving automatic stratification. The values ​​of the formation parameters are taken according to the form, taking their maximum, minimum, or geometric mean values ​​respectively.

7. The method for detailed identification of tight sandstone and conglomerate lithology based on conventional well logging curves and imaging characteristic parameters according to claim 6, characterized in that, The cubic spline function differentiation method was adopted, and adaptive filtering technology was used to reduce the drastic fluctuations in the first and second derivatives. After obtaining the first and second derivatives of the conventional logging curves, the lithology-sensitive curves were morphologically divided to determine the formation interfaces and parameter values ​​of each formation.

8. The method for detailed identification of tight sandstone and conglomerate lithology based on conventional logging curves and imaging characteristic parameters according to claim 7, characterized in that, The deep lateral resistivity curve Rt, natural gamma curve GR, P-wave transit time curve DT, compensated neutron porosity curve CNL, and gravel content curve with reduced resolution were analyzed within the target layer. And the sorting coefficient curve with reduced resolution By performing stratified readings, a prediction sample set is obtained.

Images