A method and system for reservoir fracture prediction

Abstract

The present application relates to a kind of reservoir fracture prediction method and system, belong to oil and gas field exploration and development technical field.The reservoir fracture prediction method of the present application, by establishing virtual well, eliminates the influence of borehole collapse on post-casing logging data, then the logging data at the collapse position in the borehole of each logging target layer is corrected, and finally the porosity and shale content of each logging target layer are calculated, which can improve the accuracy of the calculated porosity and shale content, and further improve the accuracy of fracture prediction.

Description

TECHNICAL FIELD

[0001] The present application relates to a reservoir fracture prediction method and system, belonging to the technical field of oil and gas field exploration and development. BACKGROUND

[0002] At present, there are many methods for predicting fractures using seismic data, among which the widely used methods mainly include three categories: post-stack P-wave attribute extraction technology, pre-stack full-azimuth anisotropy technology and S-wave splitting double scanning. Compared with conventional pre-stack fracture prediction methods, pre-stack full-azimuth fracture prediction can provide more accurate data and is beneficial to the prediction of azimuthal anisotropy by full-azimuth or wide-azimuth acquisition, and by inputting the formation characteristics of porosity and shale content of well logging into the rock physics forward model to forward the data of the measured curve of different marker layers. Therefore, well logging curve correction is a necessary work in the early stage. According to the theoretical parameters of well logging response, the reliable curve of wellbore rules is referred, and then the density repair is carried out in the layer section of wellbore collapse or density record error, and the overall comprehensive correction processing is carried out on various well logging data, so as to eliminate various system errors in the well logging processing process, so that the well logging curves in the study area have a unified scale standard, which is a necessary means to provide reliable well logging curves for subsequent seismic inversion work.

[0003] AVO (Amplitude versus Offset) analysis technology has the potential to identify changes in pore fluid and can be used to estimate the saturation and pressure of pore fluid. Therefore, AVO analysis technology can be applied to the identification of fractured oil or gas reservoirs. In the existing published technology, for example: Chinese patent document CN111045083A discloses a reservoir gas detection method, which belongs to the field of seismic exploration reservoir oil and gas prediction. It uses pre-stack seismic trace data for AVO analysis to solve the density change rate; since density can directly reflect the change of geological body, when the reservoir is saturated with gas, it will cause the decrease of formation density, and the change of density is more obvious than the parameters such as velocity and frequency, therefore, based on the above-mentioned published document, the change of density can directly reflect the change of geological body to complete the identification of fractured oil reservoir. However, due to the differences in measurement environment and wellbore environment, the measured well logging curves often have differences, and there are often problems such as local missing of well section and lack of consistency between multiple wells. This leads to large errors in density, velocity body and other data when constructing a pre-stack framework model, resulting in large errors in the process of AVO analysis to solve the density change rate.

[0004] In view of the problem that accurate P and S wave velocities are required for pre-stack inversion and AVO analysis, and that there are missing problems in the measured logging curves, a large amount of work has been carried out in China. The main methods are divided into two categories: one is a statistical fitting method, that is, according to a large amount of existing P and S wave, density and other information, an empirical formula of P and S wave and other parameters is established. The other is a rock physics forward model, that is, by assuming the pore shape and other information, a rock physics forward model is constructed to predict P and S wave, density and other information. In the construction of the rock physics forward model to predict P and S wave and other information, accurate information of shale content, porosity and other information needs to be obtained. However, due to the influence of measurement environment and wellbore environment, different periods and different measurement devices, there is a large error in the information of the measured curve. The patent literature CN112415632 discloses a method and system for predicting the S wave velocity of a low-porosity and low-permeability tight sandstone reservoir, but does not provide a solution to the above problems. The patent literature CN111751878 discloses a method and device for predicting S wave velocity, which improves the accuracy of S wave prediction, but does not provide a solution to the processing of logging curves of multiple periods and multiple measurement sequence devices.

[0005] In addition, the traditional low-frequency model is often obtained by interpolation and extrapolation of logging, and the low-frequency model often reflects the entire sedimentary background and sedimentary rule of the study area, so the accuracy of the low-frequency model will also affect the results of the entire pre-stack inversion. The patent literature CN109283574 discloses a low-frequency model construction method and computer readable storage medium, which uses seismic attributes to determine geological information and guide low-frequency modeling, but there is a problem that the accuracy of the seismic attributes is relatively low and the accuracy of the attribute description of the geological body is not high.

[0006] Therefore, there is an urgent need for a reservoir fracture prediction method that can eliminate the problem of low accuracy of pre-stack full azimuth fracture prediction caused by errors in logging curves due to wellbore environment and instrument measurement. SUMMARY

[0007] The purpose of the present application is to provide a reservoir fracture prediction method that can solve the problem of low accuracy of pre-stack full azimuth fracture prediction caused by errors in logging curves due to wellbore environment and instrument measurement.

[0008] Another purpose of the present application is to provide a reservoir fracture prediction system.

[0009] In order to achieve the above purpose, the technical scheme adopted by the reservoir fracture prediction method of the present application is as follows:

[0010] A reservoir fracture prediction method, comprising the following steps:

[0011] (1) obtaining the casing-free natural gamma data of each logging target layer in the study area;

[0012] (2) based on the depth value of the un-collapsed position in the borehole of each well purpose layer and the casing natural gamma data, the casing natural gamma data of the virtual well purpose layer is established; the casing natural gamma data of a position A point in the virtual well is equal to the casing natural gamma data of a position B point in the borehole of a certain well which has the same depth as the position A point and does not collapse;

[0013] (3) taking the virtual well as a standard well, the casing natural gamma data of the standard well purpose layer is used to standardize the casing natural gamma data of each well purpose layer, then the correction coefficient of the collapsed position in the borehole of each well purpose layer is calculated, the logging data of the collapsed position in the borehole of each well purpose layer is corrected by using the correction coefficient, and finally the porosity and shale content of each well purpose layer are calculated by using the corrected logging data of each well purpose layer; the logging data includes density, neutron, acoustic time difference and natural gamma;

[0014] (4) the porosity and shale content of each well purpose layer and the geological stratification information of the logging are used to construct a rock physics forward model, and then rock physics forward is carried out to obtain the P-wave velocity, S-wave velocity and density of each well purpose layer;

[0015] (5) based on the P-wave velocity, S-wave velocity and density of each well purpose layer in the research area, a framework model is constructed, and then the P-wave velocity, S-wave velocity and density in the framework model are spatially interpolated under the constraint of the geological database to obtain the P-wave velocity body, S-wave velocity body and density body of the research area purpose layer;

[0016] (6) the P-wave velocity body, S-wave velocity body and density velocity body of the low frequency part are selected from the P-wave velocity body, S-wave velocity body and density velocity body of the research area purpose layer to obtain a background geological model;

[0017] (7) pre-stack full azimuth fracture prediction of the research area purpose layer is carried out.

[0018] The reservoir fracture prediction method of the application eliminates the influence of borehole collapse on the casing logging data by establishing a virtual well, then the logging data of the collapsed position in the borehole of each well purpose layer is corrected, and finally the porosity and shale content of each well purpose layer are calculated, which can improve the accuracy of the calculated porosity and shale content, and further improve the accuracy of fracture prediction.

[0019] Preferably, the method for carrying out pre-stack full azimuth fracture prediction of the research area purpose layer comprises the following steps: under the constraint of the background geological model, AVAZ inversion is carried out by using full azimuth common reflection angle gather to predict the fracture density and direction of the research area purpose layer.

[0020] Preferably, the method for calculating the correction coefficient of the collapse position in the borehole of a certain well logging is as follows: assuming that the casing natural gamma of the collapse position A in the borehole of a certain well logging is C, and the casing natural gamma of the position B in the virtual well having the same depth as the collapse position A is D, then the correction coefficient of the collapse position A in the borehole of the well logging is (D-C) / D.

[0021] Preferably, the method for correcting the well logging data at the collapse position in the borehole of each well logging target layer by using the correction coefficient is as follows: the well logging data of the collapse position A in the borehole of a certain well logging target layer is E, then the corrected well logging data of the collapse position A is E* [1+(D-C) / D].

[0022] Preferably, the geological database comprises field outcrop photos, geological deposition patterns and three-dimensional geological models.

[0023] The technical scheme of the reservoir fracture prediction system of the present application is as follows:

[0024] A reservoir fracture prediction system comprises a processor and a memory, and a computer program stored on the memory and running on the processor, the processor is coupled with the memory, and the processor implements the reservoir fracture prediction method as described above when executing the computer program.

[0025] The reservoir fracture prediction system of the present application eliminates the influence of borehole collapse on casing logging data by establishing a virtual well, then corrects the well logging data at the collapse position in the borehole of each well logging target layer, and finally calculates the porosity and shale content of each well logging target layer, which can improve the accuracy of the calculated porosity and shale content, and further improve the accuracy of fracture prediction. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 The figure is a schematic diagram of the reservoir fracture prediction system of Example 1;

[0027] Figure 2 The figure is a flowchart of the reservoir fracture prediction method of Example 2;

[0028] Figure 3 The figure is a distribution histogram of the casing natural gamma curve of each well logging target layer before standardization in Example 2;

[0029] Figure 4 The figure is a distribution histogram of the casing natural gamma curve of each well logging target layer after standardization in Example 2;

[0030] Figure 5 The figure is a distribution histogram of the density curve of each well logging target layer before correction in Example 2;

[0031] Figure 6Density curve distribution histogram of each logging target layer after correction in Example 2;

[0032] Figure 7 Acoustic traveltime curve distribution histogram of each logging target layer before correction in Example 2;

[0033] Figure 8 Acoustic traveltime curve distribution histogram of each logging target layer after correction in Example 2;

[0034] Figure 9 BP neural network model schematic diagram in Example 2;

[0035] Figure 10 Schematic diagram of the framework model constructed in Example 2; Figure 10 In the figure, Bellow all represents below the target layer, above all represents above the target layer, s3x-smooth, s4s-smooth, s4d-smooth represent the names of different target layers;

[0036] Figure 11 Velocity body model schematic diagram of the target layer in the research area in Example 2; Figure 11 In the figure, Bellow all represents below the target layer, above all represents above the target layer, s3x-smooth, s4s-smooth, s4d-smooth represent the names of different target layers;

[0037] Figure 12 Longitudinal and transverse wave velocity ratio plan view of the second member of the second horse camp of the Mesozoic in the research area in Example 2;

[0038] Figure 13 The third top surface structure and fracture orientation prediction superimposed map of the second horse camp of the Mesozoic in the research area in Example 2. DETAILED DESCRIPTION

[0039] The technical solutions of the present application will be further described below in combination with specific embodiments.

[0040] Example 1

[0041] The reservoir fracture prediction system of the present embodiment comprises a processor and a memory, and the memory has stored thereon a computer program, which is run on the processor to realize the prediction of the reservoir fracture;

[0042] Specifically, the schematic diagram of the system is as shown in Figure 1As shown, the device comprises a gamma logging device, a communication unit, a first storage unit, a processing device, an analysis module, a lattice model establishing module, a spatial data model establishing module, a screening module, a background geological model establishing module and a second storage unit; the gamma logging device is used to obtain post-casing natural gamma data, the communication unit is used to transmit the post-casing natural gamma data to the processing device, the first storage unit is used to store early logging data, and the second storage unit is used to store a geological expert database; the processing device comprises a first processing unit, a second processing unit, a comparison unit, a machine learning system, a third storage unit and a rock physics forward model establishing module, the third storage unit is used to store data in the processing device, the rock physics forward model establishing module is used to construct a rock physics forward model by using porosity, shale content and geological information and seismic parameters of each logging target layer, and the machine learning system has a first neural network unit, a second neural network unit, a third neural network unit and a fourth neural network unit;

[0043] The first processing unit is connected to the machine learning system, and after receiving the post-casing gamma data, the first processing unit is used to normalize the post-casing natural gamma data of each logging target layer and calculate correction coefficients at the collapse positions in the boreholes of each logging target layer; the second processing unit is used to correct the logging data at the collapse positions in the boreholes of each logging target layer and obtain the porosity and shale content of each logging target layer.

[0044] The first processing unit comprises the first neural network unit and the second neural network unit, the first neural network unit is used to normalize the post-casing natural gamma data of each logging target layer, and the second neural network unit is used to calculate the correction coefficients at the collapse positions in the boreholes of each logging target layer.

[0045] The second processing unit comprises the third neural network unit and the fourth neural network unit, the third neural network unit is used to correct the logging data (including density, neutron, acoustic time difference and natural gamma) at the collapse positions in the boreholes of each logging target layer by using the correction coefficients, and the fourth neural network unit is used to obtain the porosity and shale content of each logging target layer.

[0046] The analysis module is configured to perform petrophysical forward analysis according to a petrophysical forward model to obtain petrophysical forward analysis data (including P-wave velocity, density and S-wave velocity of each well target layer); the framework model establishing module is configured to construct a framework model based on the P-wave velocity, S-wave velocity and density of each well target layer in the study area; the spatial data model establishing module is configured to perform spatial interpolation on the P-wave velocity, S-wave velocity and density in the framework model by Kriging or sequential Gaussian algorithm to obtain a velocity volume of the target layer in the study area under the constraint of the geological database; the screening module is configured to screen out a low-frequency P-wave velocity volume, a low-frequency S-wave velocity volume and a low-frequency density volume from the P-wave velocity volume, the S-wave velocity volume and the density volume of the target layer in the study area; and the background geological model establishing module is configured to establish a background geological model by using the low-frequency P-wave velocity volume, the low-frequency S-wave velocity volume and the low-frequency density volume.

[0047] Embodiment 2

[0048] Taking a certain area as a study area, the reservoir fracture prediction method of the embodiment, as shown in Figure 2 , specifically includes the following steps:

[0049] (1) Obtain the casing natural gamma data of the target layer in the study area:

[0050] First, select wells in the study area (the more the number of wells, the better, and try to select wells with different logging instrument sequences and long time from now to facilitate standardization correction), then place 3 sets of gamma logging devices of the same manufacturer and model along the well wall of the wellbore of each well at intervals of 2m in the target layer (use the same manufacturer and model of logging instrument to carry out re-measurement of the casing gamma data of the target layer, which can eliminate the system error caused by multiple logging sequence instruments), obtain the casing natural gamma data of the target layer (the average value of 3 sets of casing natural gamma data) from the gamma logging device, and then send the casing gamma data to the processing device through the communication unit;

[0051] (2) After receiving the casing natural gamma data, the processing device establishes the casing natural gamma data of the virtual well target layer based on the depth value and the casing natural gamma data of the uncollapsed position in the wellbore of each well target layer, and the casing natural gamma data of a certain position A in the virtual well is equal to the casing natural gamma data of a position B in the wellbore of a certain well which has the same depth as the position A and has not collapsed; since the natural gamma record error or loss is prone to occur at the collapsed position of the wellbore, this step excludes the influence of the data by establishing a virtual standard well;

[0052] Due to the relatively small research area and the relatively stable sedimentary environment, the casing behind natural gamma data at the same depth of the uncollapsed position in the wellbore of each well logging is relatively close; in specific implementation, the casing behind natural gamma data of the well logging with the least collapsed position in the wellbore can be selected as the basis data, then the casing behind natural gamma data at the uncollapsed position in the wellbore of the well logging is taken as the casing behind natural gamma data of the position point with the same depth in the virtual standard well, and the missing casing behind natural gamma data in the virtual standard well is searched from other well loggings;

[0053] The position or section in the wellbore of a certain well logging without collapse can be determined by the wellbore curve;

[0054] (3) The neural network learning system in the processing device processes the casing behind natural gamma data of each well logging target layer and the well logging data at the collapsed position in the wellbore of each well logging target layer to obtain the porosity curve and the shale content curve of each well logging target layer; the processing method is as follows:

[0055] The virtual well in step (2) is taken as a standard well, the casing behind natural gamma data of the well logging target layer is used to standardize the casing behind natural gamma data of each well logging target layer (the standardization method is histogram method), then the correction coefficient of the collapsed position in the wellbore of each well logging target layer is calculated, and the logging data (including density, neutron, acoustic time difference and natural gamma) at the collapsed position in the wellbore of each well logging target layer is corrected by using the correction coefficient, thereby completing the correction of the well logging curve of each well logging target layer, and finally the porosity and shale content of each well logging target layer are calculated and obtained by using the corrected well logging curve of each well logging target layer;

[0056] The calculation method of the correction coefficient of the collapsed position in the wellbore of a certain well logging is as follows: assuming that the casing behind natural gamma of the collapsed position A point in the wellbore of a certain well logging is C, and the casing behind natural gamma of the position B point with the same depth as the collapsed position A point in the virtual well is D, then the correction coefficient of the collapsed position A point in the wellbore of the well logging = (D-C) / D;

[0057] The method for correcting the well logging data at the collapsed position in the wellbore of each well logging target layer by using the correction coefficient is as follows:

[0058] Taking the corrected density as an example, assuming that the density logging data of the collapsed position A point in the wellbore of a certain well logging target layer is E, then the corrected density logging data of the collapsed position A point = E×[1+(D-C) / D];

[0059] The distribution histogram of the casing behind natural gamma curve of each well logging target layer before standardization and the distribution histogram of the casing behind natural gamma curve of each well logging target layer after standardization are respectively as shown in Figures 3-4 ; Figures 3-4In the specific embodiment, the distribution histogram of the natural gamma ray curve of each well-logging target layer after casing is represented by different colors.

[0060] The distribution histogram of the density curve of each well-logging target layer before correction and the distribution histogram of the density curve of each well-logging target layer after correction are shown in Figs. 4 and 5, respectively. Figures 5-6 Figures 5-6 In the specific embodiment, the distribution histogram of the density curve of each well-logging target layer is represented by different colors.

[0061] The distribution histogram of the interval transit time curve of each well-logging target layer before correction and the distribution histogram of the interval transit time curve of each well-logging target layer after correction are shown in Figs. 6 and 7, respectively. Figures 7-8 Figures 7-8 In the specific embodiment, the distribution histogram of the interval transit time curve of each well-logging target layer is represented by different colors.

[0062] In the specific implementation, the processing method is completed by a neural network learning system in a processing device, and the neural network learning system includes a first neural network, a second neural network, a third neural network, and a fourth neural network. The first neural network, the second neural network, the third neural network, and the fourth neural network are BP neural networks, and a model schematic diagram is shown in Fig. 8. Figure 9

[0063] The first neural network is a BP neural network, and the neural network is provided with an input layer, an output layer, and two hidden layers. The first neural network is used to perform standardization processing on the casing natural gamma ray data of each well-logging target layer. First, the casing natural gamma ray data of each well-logging target layer and the casing natural gamma ray data of a standard well are taken as inputs of the neural network, and the casing natural gamma ray data after standardization processing is taken as an output of the neural network. Then, the network is trained until the fitness value of an individual (an individual refers to a well) in a group (a group refers to all wells) is equal to the inverse of a training error (the greater the error between a prediction result and an actual training target value, the lower the fitness of the individual). The calculation method of the training error is shown in Formula 1.

[0064] ||error||=(|Y1-T1| 2 +|Y2-T2| 2 +…+|Y n -T n | 2 ) 1 / 2 Formula 1

[0065] In Formula 1, Y is a prediction result of a BP operator; T is an actual training target value; and n is the number of training samples, that is, the number of all wells.

[0066] ​​​The second neural network is a BP neural network, which is provided with an input layer, an output layer and two hidden layers, and is used to calculate the correction coefficient of the collapse position in the wellbore of each logging target layer; the normalized post-casing natural gamma ray curve is taken as the input of the neural network, and the correction coefficient of the collapse position in the wellbore of each logging target layer is taken as the output of the neural network, and then the network is trained until the fitness value of the individual in the population is equal to the inverse of the training error; the calculation method of the training error in the second neural network is the same as that in the first neural network;

[0067] The third neural network is a BP neural network, which is provided with an input layer, an output layer and two hidden layers, and is used to correct the logging data (including density, neutron, acoustic time difference and natural gamma ray) at the collapse position in the wellbore of each logging target layer by using the correction coefficient; the correction coefficient of the collapse position in the wellbore of each logging target layer and the logging curves (including the density logging curve, the neutron logging curve, the acoustic time difference logging curve and the natural gamma ray logging curve) of each logging target layer are taken as the input of the neural network, and the corrected logging curves of each logging target layer are taken as the output of the neural network, and then the network is trained until the fitness value of the individual in the population is equal to the inverse of the training error; the calculation method of the training error in the third neural network is the same as that in the first neural network;

[0068] The fourth neural network is a BP neural network, which is provided with an input layer, an output layer and two hidden layers, and is used to obtain the porosity and shale content of each logging target layer; the corrected logging curves (acoustic time difference logging curve and natural gamma ray logging curve) of each logging target layer are taken as the input of the neural network, and the porosity curve and the shale content curve are taken as the output of the neural network, and then the network is trained until the fitness value of the individual in the population is equal to the inverse of the training error; the calculation method of the training error in the fourth neural network is the same as that in the first neural network;

[0069] The porosity curve and the shale content curve of different horizons in the target layer are obtained through the machine learning system; the core of the machine learning system is a neural network model, which can correct the data, and can eliminate the error caused by filtering by iteratively training the collected gamma ray energy spectrum by setting fitness as the attention mechanism; the result is more accurate, wherein the fitness can be obtained by analyzing the difference between the actual gamma ray energy spectrum sample and the traditional smooth filtering of the measured gamma ray energy spectrum;

[0070] In the learning process of the neural network model, the fitness is only used as a general feature, and when the learning of a new task is carried out, the neural network model can extract the result of the iterative training after adding the general feature (fitness), and compare and analyze the gamma ray spectrum before the training to obtain the fitness of the new task. Therefore, after a large number of new task training, the fitness can be stabilized in a relatively accurate interval.

[0071] (4) Constructing a rock physics forward model by using the porosity, shale content and geological layering information of each well target layer obtained in step (3), and then carrying out rock physics forward to obtain rock physics forward data (including the P-wave velocity, density and S-wave velocity of each well target layer);

[0072] In order to carry out AVO analysis, a rock physics forward model needs to be constructed first. In seismic AVO analysis, the situation of lacking S-wave slowness logging curve is often encountered, so it is necessary to synthesize a reliable acoustic curve with less error compared with the measured curve, and to investigate the rock elastic response characteristics under different fluid saturation states. The rock physics forward model is constructed by using the logging curve, which is a means to establish the relationship between the reservoir characteristics and the rock elastic parameters. The focus is on establishing the influence relationship between the rock physical parameters and the response elastic parameters (Poisson's ratio and other parameters), and further establishing the corresponding relationship between the elastic parameter curve in the entire depth domain from the ground to the underground and the seismic response (frequency, amplitude and other information of P-wave and S-wave). In the area without data (P-wave velocity, S-wave velocity and density data) points, the rock physics forward model is used to obtain a curve (P-wave and S-wave velocity curve) that not only conforms to the basic rock physics principle but also is consistent with the logging formation evaluation model. Logging formation evaluation and rock physics modeling are an organic whole with geophysical forward work, which can make up for the shortcomings of logging, solve the problems of missing data in seismic AVO analysis and seismic inversion, and lay a solid foundation for the interpretation of seismic inversion parameters; comparing the forward curve (porosity and shale content curve) with the measured curve (measured porosity and shale content curve), if the two curves have consistent rhythm change rules, it can be proved that the established rock physics forward model is relatively reliable;

[0073] In addition, the rock physics template crossplot analysis can also prove that the curve rule after forward is good;

[0074] (5) Constructing a grid model to obtain the P-wave velocity body, S-wave velocity body and density body of the study area:

[0075] ① Based on the P-wave velocity, S-wave velocity and density of each well target layer in the study area, a grid model is constructed, as shown in FIG. 1; Figure 10 Figure 10 ​In the figure, the horizontal coordinate is position coordinate, and the vertical coordinate is time, with unit of ms, and each color band in the figure represents stratum with different depth and deposition difference;

[0076] 2) Under the constraint of the geological database, the longitudinal wave velocity, transverse wave velocity and density in the framework model are spatially interpolated by sequential Gaussian algorithm to obtain the longitudinal wave velocity body, transverse wave velocity body and density body of the target layer in the study area, and the velocity body model is shown in Figure 11 ; Figure 11 In the figure, the horizontal coordinate is position coordinate, and the vertical coordinate is time, with unit of ms, and the black curve in the figure is geological interface, and different colors represent different velocity; the geological database includes field outcrop photo, geological deposition mode and three-dimensional geological model;

[0077] (6) Construction of background geological model:

[0078] The longitudinal wave velocity body, transverse wave velocity body and density velocity body of low frequency part are selected from the longitudinal wave velocity body, transverse wave velocity body and density velocity body of the target layer in the study area to obtain the background geological model;

[0079] (7) Preparing for pre-stack full-azimuth fracture prediction:

[0080] Under the constraint of the background geological model, AVAZ inversion is carried out by using full-azimuth common reflection angle gather to predict the fracture density and direction of the target layer in the study area;

[0081] The pre-stack full-azimuth fracture prediction is based on full-azimuth seismic data acquisition, according to the inversion ray tracing method of explosion reflection surface refraction model to decompose and image the seismic wave field, using full-azimuth common reflection angle gather to carry out AVAZ inversion to predict the fracture density and direction; compared with the conventional pre-stack fracture prediction direction, the pre-stack full-azimuth fracture prediction requires full-azimuth or wide-azimuth acquisition, high resolution of acquired data and full azimuth information, which is beneficial to the prediction of azimuth anisotropy;

[0082] The principle of the pre-stack full-azimuth fracture prediction is as follows: common reflection migration imaging is used to reconstruct the imaging gather in the common reflection angle domain, and ray tracing imaging is carried out from the imaging point to the ground, so that the seismic information of the ground is projected to the local angle domain underground, and each imaging point contains azimuth information of the stratum, then the full-azimuth common reflection gather is obtained by continuously migrating all seismic data in the local angle domain underground; this method can overcome the false appearance in the common offset gather, and effectively improve the accuracy of seismic imaging; the residual moveout of the full-azimuth reflection angle gather in different directions can obtain accurate and effective HTI anisotropy parameters and fracture information; because of the advantages of full-azimuth acquisition and imaging, this gather can provide real amplitude reflection coefficient and obtain real amplitude anomaly, so as to reflect the information of underground velocity and lithology change, which is more conducive to seismic attribute analysis and azimuth information prediction of fracture development.

[0083] The P-wave and S-wave velocity ratio plane of the second member of the Ermaqing Formation in the Mesozoic in the Pubei area is obtained by AVAZ inversion, as shown in FIG. 2; Figure 12 ( Figure 12 The horizontal and vertical coordinates in FIG. 2 are geodetic coordinates, and the color coordinates represent the P-wave and S-wave velocity ratio). According to the reference Figure 11 , the A and B blocks with the best fracture development in the study area show low P-wave and S-wave velocity ratio, which is basically consistent with the P-wave and S-wave velocity ratio characteristics of the previous inversion Figure 12 , and can better depict the fracture distribution characteristics;

[0084] The Pubei area is obtained by AVAZ inversion, as shown in FIG. 3; Figure 13 ( Figure 13 The horizontal and vertical coordinates in FIG. 3 are geodetic coordinates, and the color coordinates represent the fracture development degree). It can be seen from Figure 13 that the main fractures and structural traces are basically consistent, mainly in the north-east direction; the fracture density is large in the target area to the east and south of the B block.

Claims

1. A method of reservoir fracture prediction, characterized by, The method comprises the following steps: (1) obtaining casing-after natural gamma data of each well logging target layer in the research area; (2) based on the depth value of the un-collapsed position in the wellbore of each well logging target layer and the casing-after natural gamma data, the casing-after natural gamma data of the virtual well logging target layer is established; the casing-after natural gamma data of a position A in the virtual well is equal to the casing-after natural gamma data of a position B in the wellbore of a certain well logging which has the same depth as the position A and has not collapsed; (3) taking the virtual well as a standard well, the casing-after natural gamma data of the standard well target layer is used to standardize the casing-after natural gamma data of each well logging target layer, then the correction coefficient of the collapsed position in the wellbore of each well logging target layer is calculated, the logging data of the collapsed position in the wellbore of each well logging target layer is corrected by using the correction coefficient, and finally the porosity and shale content of each well logging target layer are calculated by using the corrected logging data of each well logging target layer; the logging data includes density, neutron, acoustic time difference, natural gamma; (4) the porosity and shale content of each well logging target layer obtained in step (3) and the geological stratification information of the well logging are used to construct a rock physics forward model, and then rock physics forward is carried out to obtain the P-wave velocity, S-wave velocity and density of each well logging target layer; (5) based on the P-wave velocity, S-wave velocity and density of each well logging target layer in the research area, a grid model is constructed, and then the P-wave velocity, S-wave velocity and density in the grid model are spatially interpolated under the constraint of the geological database to obtain the P-wave velocity body, S-wave velocity body and density body of the target layer in the research area; (6) the P-wave velocity body, S-wave velocity body and density body of the low frequency part are selected from the P-wave velocity body, S-wave velocity body and density body of the target layer in the research area to obtain a background geological model; (7) pre-stack full azimuth fracture prediction of the target layer in the research area is carried out.

2. The reservoir fracture prediction method of claim 1, wherein, The method for carrying out the pre-stack full azimuth fracture prediction of the target layer in the research area comprises the following steps: under the constraint of the background geological model, AVAZ inversion is carried out by using full azimuth common reflection angle gather to predict the fracture density and direction of the target layer in the research area.

3. The reservoir fracture prediction method of claim 1, wherein, The calculation method of the correction coefficient of the collapsed position in the wellbore of a certain well logging is as follows: assuming that the casing-after natural gamma of the collapsed position A in the wellbore of a certain well logging is C, and the casing-after natural gamma of the position B in the virtual well which has the same depth as the collapsed position A is D, then the correction coefficient of the collapsed position A in the wellbore of the well logging is (D-C) / D.

4. The reservoir fracture prediction method of claim 3, wherein, The method for correcting the logging data of the collapsed position in the wellbore of each well logging target layer by using the correction coefficient is as follows: the logging data of the collapsed position A in the wellbore of a certain well logging target layer is E, then the corrected logging data of the collapsed position A is E×[1+(D-C) / D].

5. A reservoir fracture prediction system characterized by, The reservoir fracture prediction method comprises a processor and a memory, and a computer program stored on the memory and running on the processor, the processor is coupled with the memory, and the processor implements the reservoir fracture prediction method according to any one of claims 1-4 when executing the computer program.

Images