Deep low-permeability compact clastic rock-based favorable reservoir identification method

Abstract

The invention discloses a favorable reservoir identification method based on deep low-permeability compact clastic rock, which comprises the following steps of: firstly, establishing a geophysical lithophasor plate, then carrying out seismic lithofacies prediction, converting seismic data into a sensitive attribute body through an inversion technology, converting the value of the sensitive attribute body into geophysical lithofacies through the projection of the geophysical lithophasor plate, and finally identifying the favorable reservoir of the deep low-permeability compact clastic rock. And identification of compact clastic rocks is realized. According to the method, specific quantification is carried out through a probability theory method by utilizing the correlation between seismic data and lithofacies, lithofacies prior information is determined from existing logging information, the required posterior probability is calculated by utilizing the prior information, and the prior probability and the likelihood function of the lithofacies are obtained through known logging data by adopting a Bayesian classification technology, so that the probability of the lithofacies is obtained. According to a uniform distribution hypothesis, calculating a probability value of a geophysical lithofacies of each sample point, quantitatively describing a relationship between the lithofacies and an earthquake by using the probability values of the lithofacies, and establishing a hypotonic compact clastic rock geophysical lithofacies version by using the relationship.

Description

Technical Field

[0001] The present invention relates to the field of seismic exploration, and in particular to a method for identifying favorable reservoirs based on deep low-permeability dense clastic rocks. Background Art

[0002] Tight clastic gas reservoirs hold enormous reserves, but their geology is complex, making exploration and development challenging. Finding high-quality sandstone is crucial for achieving industrial production capacity. Lithofacies, a geological term, generally refers to the morphological and structural characteristics of rocks, which are used to describe and classify them. Geophysical lithofacies is a combination of geophysics and petrography, utilizing a combination of geophysical methods such as seismic waves, electromagnetic waves, microwaves, millimeter waves, and radiography to study the physical properties and structural characteristics of rocks and subsequently classify reservoirs. Geophysical lithofacies analysis allows for the precise identification of clastic reservoirs. In the exploration of low-permeability, tight clastic rocks, identifying favorable lithofacies in clastic rocks allows for the prediction of high-quality reservoirs and improves the success rate of oil and gas exploration. Therefore, the development of geophysical lithofacies quantitative plates is of great significance. At present, there are great difficulties in producing geophysical lithofacies of deep low-permeability and dense clastic rocks, mainly due to the following reasons: (1) deep reservoirs are low-permeability and dense, and the difference in geophysical response is small, which makes it difficult to identify sandstones, restricting the prediction of clastic reservoirs; (2) the vertical lithofacies of clastic rocks are complex. Due to the superposition of multiple lithofacies, the seismic reflection characteristics are usually the comprehensive reflection of multiple lithofacies, making it difficult to find the accurate lithofacies position in seismic data interpretation. At present, there are few relevant literatures on the production of geophysical lithofacies of deep low-permeability and dense clastic rocks. Usually, rock physics experiments are conducted to analyze the rock physics laws, clarify the main factors of the reservoir, establish a rock physics plate, and use the rock physics plate for analysis to obtain lithofacies information. It is feasible to use this idea and process to carry out lithofacies analysis of well logging. However, when the above idea is applied to geophysical lithofacies analysis, due to the great difference in resolution between seismic and well logging, the vertical lithofacies changes of the seismic lithofacies results obtained by directly using the rock physics plate of well logging are very frequent and unstable. Furthermore, due to the limited number of wells logged in a given area, it is difficult to accurately determine the planar distribution of the reservoir. Furthermore, due to vertical instability, it is also difficult to determine the planar distribution of the reservoir. Therefore, it is necessary to develop a geophysical lithofacies plate for deep, low-permeability, dense clastic rocks, and use this plate to predict seismic lithofacies.

[0003] The main purpose of the geophysical lithofacies database is to perform lithofacies classification on each data point in the seismic data of the target layer, identify favorable lithofacies in clastic rocks, establish a relationship between lithofacies and seismic data, and thus determine the distribution of favorable reservoirs. Lithofacies classification is generally difficult to perform with seismic data, but the attribute volume derived from linear changes in seismic data has practical lithologic significance and is somewhat correlated with lithofacies. Therefore, a new method for identifying favorable reservoirs in deep, low-permeability, dense clastic rocks is urgently needed to address this issue. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a method for identifying favorable reservoirs in deep, low-permeability, dense clastic rocks. This method effectively addresses the drawbacks of previous methods, such as the large discrepancies between well logging and seismic data, which led to the direct use of rock physics plates to convert lithofacies, resulting in frequent and unstable vertical variations and unclear lateral patterns. This invention can effectively reduce the difficulty of identifying favorable reservoirs and improve the accuracy of predicting favorable clastic reservoirs. It is suitable for identifying dense clastic rocks and has excellent application prospects for the exploration and development of dense clastic rocks.

[0005] To solve the above technical problems, the present invention provides a method for identifying favorable reservoirs based on deep low-permeability dense clastic rocks, comprising the following steps:

[0006] Establish geophysical lithofacies volume version;

[0007] Carry out seismic lithofacies prediction and convert seismic data, well logging and geological data into sensitive attribute bodies through inversion technology;

[0008] Converting the value of the sensitive attribute body into geophysical lithofacies by projecting the geophysical lithofacies quantity plate;

[0009] Find the location of favorable sandstone in the geophysical lithofacies profile and identify favorable dense clastic rocks.

[0010] The beneficial effects of the present invention are:

[0011] The present invention obtains geophysical lithofacies prior probability information from well logging curve data based on the Bayesian classification method. Under the condition of clearly studying the lithologic type, the well logging lithofacies classification results are used as labels to calculate the geophysical lithofacies prediction probability. Finally, the geophysical lithofacies quantitative version is established using three types of information: sensitive attributes, lithofacies types, and lithofacies probability values. Through geophysical lithofacies quantitative version mapping, seismic attribute data is geophysical lithofacies classified to obtain a favorable lithofacies distribution of clastic rocks. This solves two previous problems: the difficulty in identifying small sandstones due to the differences in geophysical responses of deep low-permeability and dense clastic rocks, and the difficulty in finding accurate lithofacies positions due to complex vertical lithofacies changes. The corresponding geophysical lithofacies results obtained by using geophysical lithofacies quantitative version mapping are compared with actual earthquakes and well logging, further confirming the advantages and application potential of the geophysical lithofacies quantitative version of deep low-permeability and dense clastic rocks.

[0012] There are currently few relevant literatures on how to establish a geophysical lithofacies quantitative plate. Since the seismic response characteristics of deep low-permeability dense clastic rocks are slightly different, it is difficult to identify favorable reservoirs. The seismic response reflects the comprehensive reflection of multiple lithofacies, resulting in the inability to accurately identify the exact location of favorable reservoirs. The present invention innovatively produces a geophysical lithofacies quantitative plate for deep low-permeability dense clastic rocks, and identifies favorable reservoirs through the relationship between the plate and lithofacies. The deep learning network is used to enhance the feature learning ability of the network, and the corresponding seismic features are obtained through well logging data; the reflection coefficient characteristics of well logging and seismic are used to avoid the annihilation of the pinch-out point features by the comprehensive seismic response, and a clear indication of the pinch-out point is provided. This invention can effectively reduce the difficulty of identifying favorable reservoirs, improve the prediction accuracy of favorable clastic reservoirs, and identify favorable dense clastic rocks, which is of great significance for the exploration and development of dense clastic rocks. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a flow chart of the method for identifying favorable reservoirs of deep low-permeability dense clastic rocks based on geophysical lithofacies volume of the present invention;

[0014] Figure 2 It is a flow chart for establishing geophysical lithofacies volume version;

[0015] Figure 3 It is a schematic diagram of the intersection analysis of geophysical sensitive attributes. DETAILED DESCRIPTION

[0016] The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more precise definition of the protection scope of the present invention.

[0017] See also Figure 1 , embodiments of the present invention include:

[0018] A method for identifying favorable reservoirs based on deep low-permeability dense clastic rocks includes the following steps:

[0019] Establish geophysical lithofacies volume version;

[0020] Carry out seismic lithofacies prediction and convert seismic data, well logging and geological data into sensitive attribute bodies through inversion technology;

[0021] Converting the value of the sensitive attribute body into geophysical lithofacies by projecting the geophysical lithofacies quantity plate;

[0022] Find the location of favorable sandstone in the geophysical lithofacies profile and identify favorable dense clastic rocks.

[0023] For details, see Figure 2 The specific steps to establish the geophysical lithofacies version include:

[0024] S101: Clarify the geophysical lithofacies classification type of clastic rocks through understanding the geological data of the study area;

[0025] Typical clastic reservoirs can usually be divided into four types of lithofacies according to the characteristics of well logging curves: mudstone facies, poor sandstone facies (water-containing), favorable sandstone facies, and tight sandstone facies. For the convenience of calculation and production of quantitative plates, the corresponding geophysical lithofacies types are represented by the numbers 1, 2, 3, and 4, respectively.

[0026] S102: By performing cross-analysis on well logging attributes (P-wave velocity, S-wave velocity, density, GR, shale content, resistivity, water saturation, etc.), one or two geophysical lithofacies sensitive attributes can be identified;

[0027] S103: Using the logging curve data, perform lithofacies calibration on the logging to obtain the geophysical lithofacies calibration results of the logging; mainly based on the multiple logging curves (P-wave velocity, S-wave velocity, density, GR, mud content, resistivity, water saturation, etc.), the target layer is calibrated according to the characteristics analyzed in steps S101 and S102. The four corresponding geophysical lithofacies of mudstone, poor sandstone (water-containing), favorable sandstone, and tight sandstone are marked with numbers 1, 2, 3, and 4, respectively.

[0028] S104: Perform statistical analysis based on the lithofacies type of the target layer to calculate the prior probability of each type of lithofacies;

[0029] Using the calibrated logging results, the number of each lithofacies was counted and the proportion of each lithofacies was calculated, thus obtaining the prior probability of each of the four geophysical lithofacies.

[0030] S105: Under a certain lithofacies condition, the probability of selecting the logging sample is calculated using the non-parametric estimation kernel function method to calculate the likelihood function;

[0031] S106: Calculate the geophysical lithofacies prediction probability of each sampling point based on the Bayesian classification principle;

[0032] Based on the prior probability, likelihood function and probability density (assuming uniform distribution) of the lithofacies, the geophysical lithofacies prediction probability of each sampling point is calculated using the Bayesian classification method.

[0033] S107: Select sensitive attributes and lithofacies, and use the predicted probability values ​​to create a geophysical lithofacies quantitative version through mapping, so as to realize the production of a geophysical lithofacies quantitative version of deep low-permeability dense clastic rock. The specific steps include:

[0034] The lithofacies type and sensitive attributes are used as the coordinate axes of the two-dimensional geophysical lithofacies quantitative plate, and the results of each sample point are put into the quantitative plate in a mapping manner. The threshold is set according to experience, and the probability equipotential line is outlined according to the threshold with the 100% accurate position of the lithofacies as the center. The threshold is generally set to 10%, 50% and 90% according to experience, which means that the probability of error in judging this lithofacies is 10%, 50% and 90% respectively, and finally the production of the geophysical lithofacies quantitative plate of deep low-permeability dense clastic rock is realized.

[0035] Here we use a typical application case of tight sandstone lithofacies prediction to illustrate the principle of this method. The principle is to first use well logging and geological data, combined with geological knowledge to divide geophysical lithofacies. Tight clastic rocks can generally be divided into four lithofacies: mudstone facies, poor sandstone facies (water-bearing), favorable sandstone facies, and tight sandstone facies. Based on this division, we use the intersection analysis of multiple lateral curve data of the target layer to identify 1-2 sensitive attributes, such as Figure 3 As shown, the intersection analysis using the Lame constant * density and Poisson's ratio attributes reveals clear boundaries between the four lithofacies, with distinct characteristics for each. For example, favorable sandstone reservoirs exhibit low Lame constant * density and low Poisson's ratio, while mudstone facies exhibit low Lame constant * density and low Poisson's ratio. This intersection analysis demonstrates that these two attributes can effectively distinguish the four geophysical lithofacies. Lame constant * density and Poisson's ratio can serve as sensitive attributes for geophysical lithofacies identification. After identifying the sensitive attributes, well logging parameters can be used to calibrate the lithofacies. If the calibration results are inconsistent with geological understanding, the sensitive attributes must be reanalyzed, returning to steps S101 and S102.

[0036] The geophysical lithofacies version mainly uses Bayesian classification technology to predict seismic lithofacies. Bayesian classification technology is a supervised machine learning technology, and its formula is:

[0037]

[0038]

[0039] Among them, X represents the data sample point, c i For a certain lithofacies, p(c i |X) is a sample point classified as lithofacies c i The posterior probability density, p(X|c i ) is the likelihood function, p(c i ) is lithofacies c i The prior probability density of , its size is equal to the proportion of this lithofacies in all lithofacies. p(X) is the marginal probability density, which plays a normalization role, u i is the mean value of the lithofacies i data points, σ i is the standard deviation of the lithofacies i data sample point.

[0040] After completing the calibration of the well logging geophysical lithofacies, the proportion of each can be statistically analyzed to calculate the proportion of each lithofacies. This is mainly done by dividing the number of samples of each lithofacies by the total number of samples. The statistical results here are p(c1) = 36.64%, p(c2) = 21.23%, p(c3) = 14.07%, and p(c4) = 28.06%. These respectively indicate that the proportions of mudstone facies, poor sandstone facies (water-bearing), favorable sandstone facies, and tight sandstone facies are 36.64%, 21.23%, 14.07%, and 28.06%, respectively. This gives the prior probability p(c4) of each geophysical lithofacies. i ). In order to obtain the posterior probability density, we need to first calculate p(X|c i ), the non-parametric kernel function method is generally used to calculate p(X|c i ). Among them, the one-dimensional kernel density estimation formula is:

[0041]

[0042] Where n is the data point size, h is the smoothing amount, and K is the selected kernel function. Due to the complexity of lithofacies characteristics, using only a single parameter for lithofacies classification is often not ideal. Using multiple parameters for lithofacies classification requires the use of high-dimensional kernel density estimation, the formula is:

[0043]

[0044] Where D is the number of selected parameters. i ) is the x and sample point X in D-dimensional space i distance.

[0045] Generally, the kernel function multivariate Gaussian function is used as the kernel function, but the Gaussian function cannot decay to zero at a distance from the center, which affects the calculation efficiency. Therefore, the Epanechnikov kernel function is selected:

[0046]

[0047] The geophysical lithofacies quantity plate was established using two sensitive attributes, namely the Lame constant*density and Poisson's ratio, in order to classify and distinguish the lithofacies as a whole.

[0048] To verify the reliability of the template, data from a validation well were used for validation. The well was first calibrated, and the lithofacies of each well logging data point were determined as labels. Each data point from the validation well was then projected onto the established template, and the accuracy was calculated. Table 1 (Data from the validation template using well logging data) shows the four lithofacies of tight sandstone on the left: mudstone, poor sandstone, favorable sandstone, and tight sandstone. The right side of the table shows the geophysical lithofacies prediction results using the geophysical lithofacies template. For mudstone prediction, the first number is 89.56%, indicating an 89.56% accuracy rate for predicting mudstone when it is actually mudstone. Similarly, the probability of predicting a poor sandstone when it is actually mudstone is 4.72%, the probability of predicting a favorable sandstone when it is actually mudstone is 0.07%, and the probability of predicting a tight sandstone when it is actually mudstone is 5.65%. The values ​​on the diagonal corners of the table indicate that the probabilities of the actual lithofacies being consistent with the predicted lithofacies are 89.56%, 77.73%, 93.26% and 89.84% respectively, indicating that the accuracy of lithofacies classification using geophysical quantities is relatively high.

[0049] Table 1 Data used to verify the quantitative version using well logging data

[0050]

[0051] Based on the geophysical quantitative plate, we further developed seismic lithofacies prediction. Through inversion, we converted seismic data into sensitive attribute volumes, and then converted the values ​​of these sensitive attribute volumes into geophysical lithofacies through quantitative plate projection. Compared to the difficulty of interpreting favorable tight sandstone reservoirs on seismic sections, geophysical lithofacies sections clearly display favorable sandstone reservoirs, accurately locate favorable sandstones, and identify favorable tight clastic rocks.

[0052] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention's description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for identifying favorable reservoirs based on deep low-permeability dense clastic rocks, characterized by: The following steps are involved: Establish geophysical lithofacies volume version; Carry out seismic lithofacies prediction and convert seismic data, well logging and geological data into sensitive attribute bodies through inversion technology; Converting the value of the sensitive attribute body into geophysical lithofacies by projecting the geophysical lithofacies quantity plate; Find the location of favorable sandstone in the geophysical lithofacies profile and identify favorable dense clastic rocks.

2. The method according to claim 1, characterized in that The specific steps to establish the geophysical lithofacies version include: S101: Clarify the geophysical lithofacies classification type of clastic rocks through understanding the geological data of the study area; S102: Identify 1-2 geophysical lithofacies sensitive attributes through well logging attribute cross-analysis; S103: performing lithofacies calibration on the well logging using the well logging curve data to obtain geophysical lithofacies calibration results of the well logging; S104: Perform statistical analysis based on the lithofacies type of the target layer to calculate the prior probability of each type of lithofacies; S105: Under a specific lithofacies condition, a likelihood function is calculated using a non-parametric kernel function estimation method; S106: Calculate the geophysical lithofacies prediction probability of each sampling point based on the Bayesian classification principle; S107: Select sensitive attributes and lithofacies, and use the predicted probability values ​​to establish a geophysical lithofacies quantitative version through mapping, so as to realize the production of a geophysical lithofacies quantitative version of deep low-permeability dense clastic rock.

3. The method according to claim 2, characterized in that In step S103, the well logging curve data is used to perform lithofacies calibration on the well logging. The specific steps of obtaining the geophysical lithofacies calibration results of the well logging include: The target layer is calibrated according to the characteristics analyzed in step S101 and step S102 based on the multiple logging curves, and the four corresponding geophysical lithofacies, namely mudstone facies, water-poor sandstone facies, favorable sandstone facies, and tight sandstone facies, are calibrated respectively.

4. The method according to claim 2, characterized in that In step S104, the specific steps of performing statistical analysis based on the lithofacies type of the target layer and calculating the prior probability of each type of lithofacies include: Using the calibrated logging results, we conducted quantitative statistics for each lithofacies and calculated the proportion of each lithofacies to obtain the prior probabilities of the four geophysical lithofacies: mudstone, poor water-bearing sandstone, favorable sandstone, and tight sandstone.

5. The method according to claim 2, characterized in that In step S106, according to the Bayesian classification principle, the method for calculating the geophysical lithofacies prediction probability of each sampling point is: Among them, X represents the data sample point, c i For a certain lithofacies, p(c i |X) is a sample point classified as lithofacies c i The posterior probability density, p(X|c i ) is the likelihood function, p(c i ) is lithofacies c i The prior probability density of , p(X) is the marginal probability density.

6. The method according to claim 2, characterized in that In step S107, sensitive attributes and lithofacies are selected, and a geophysical lithofacies quantitative plate is established by mapping using the predicted probability values. The specific steps for producing a geophysical lithofacies quantitative plate for deep low-permeability dense clastic rocks include: The lithofacies type and sensitive attributes are used as the coordinate axes of the two-dimensional geophysical lithofacies quantitative plate, and the results of each sample point are mapped into the quantitative plate. The threshold is set according to experience, and the probability equipotential line is drawn according to the threshold with the 100% accurate position of the lithofacies as the center, finally realizing the production of the geophysical lithofacies quantitative plate of deep low-permeability dense clastic rock.

7. The method according to claim 6, characterized in that The thresholds are set to 10%, 50% and 90%, respectively indicating that the probability of error in determining the lithofacies is 10%, 50% and 90%.

Images