A method for fine division of fracture-vuggy reservoir rock facies

Abstract

The present invention discloses a method for fine division of fracture-vug reservoir rock facies, including: obtaining single-well data of a region to be measured and performing preprocessing; performing depth alignment on required core physical property data; performing principal component analysis on the single-well data to obtain dimension-reduced and noise-reduced fusion characterization data; dividing it into a clustering data set and a test data set; using the elbow method to select the optimal classification cluster for the clustering data set; using a clustering algorithm improved by fuzzy C-means to classify the data and establishing an identification model for fracture-vug reservoir rock facies; using the silhouette coefficient to evaluate the identification model for fracture-vug reservoir rock facies; combining the test data set to test the identification model for fracture-vug reservoir rock facies; using the test identification model for fracture-vug reservoir rock facies to output a test clustering result and performing sensitivity analysis using a rank correlation matrix; using a sensitivity curve to perform crossplotting to obtain an identification chart for fracture-vug reservoir rock facies.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas reservoir exploration and development, and in particular to a method for finely dividing rock facies of a fracture-vuggy reservoir. Background Art

[0002] In recent years, with the continuous deepening of exploration, great progress has been made in the research and understanding of microbial mound shoals. However, issues such as how high-quality reservoir facies belts are distributed and the reservoir-forming mechanism still need to be further explored, which severely restricts the further efficient development of oilfields. As the basis of the mound shoal structure, rock facies is a concentrated manifestation of the macroscopic physical property changes and microscopic pore structure characteristics of rocks. The pore-fracture configuration relationships of different rock facies vary greatly, resulting in large differences in the reservoir-seepage capacity and strong heterogeneity within the mound shoal. Accurately identifying rock facies is of great significance for subsequent research on mound shoal gas reservoirs. Therefore, it is necessary to conduct fine identification and division of the rock facies in the block, and implement the vertical and horizontal distribution laws of rock facies, so as to provide technical support for accurate well placement, oil well productivity evaluation, prediction of favorable remaining oil zones, and evaluation of water injection development potential in the next step.

[0003] Among them, rock facies is a comprehensive concept, which refers to a reservoir genetic unit with certain rock physical properties and capable of reflecting sedimentation, diagenesis, and late tectonic transformation. Macroscopically, it is manifested as the change of reservoir rock physical properties, and microscopically, it is manifested as the heterogeneous characteristics of the reservoir pore structure. Its related research has broad application prospects in the process of oil and gas exploration and development, improves the success rate of exploration wells, and can also evaluate and predict the fluid mobility, providing a geological basis for exploration and development targets. Since the concept of rock facies was proposed, geologists have proposed many rock facies identification methods based on basic data and research practice. They can be roughly classified into the following categories:

[0004] First, the lithofacies identification method based on geological theory: According to the principles of sedimentation and diagenesis, through field outcrop surveys, the sedimentary facies environment is roughly divided based on sedimentary interfaces and diagenetic environments, and then the lithofacies types are further subdivided according to the cuttings and core thin sections of each sedimentary facies. Based on the microscopic analysis of rock thin sections in the target reservoir area, the relationship between well logging identification templates and seismic data volumes is established to identify and predict dolomite reservoirs. For example, the publicly disclosed technology with the publication number CN106908856A and the name "A seismic prediction method for lacustrine thin-layer dolomite reservoirs", and another example is the publicly disclosed technology with the publication number CN109577962A and the name "A lithofacies analysis method for continental fine-grained sedimentary rocks". This technology classifies lithofacies according to lithology or lithology combination methods, organic carbon content, silica content, and calcium content, and establishes lithofacies classification rules. There are also some geologists who identify based on core descriptions and combine the characteristics of lithology and physical properties changes to establish lithofacies classification and naming rules, divide lithofacies types, and thus clarify favorable lithofacies and their distribution ranges, such as the publicly disclosed technology with the publication number CN105201490A and the name "A lithofacies analysis method for shale intervals". The advantage of this type of method is that the theoretical basis is relatively sufficient, but the disadvantage is that it can only roughly divide lithofacies macroscopically, with low division accuracy for formations with strong heterogeneity and many lithofacies types, unable to accurately depict the distribution characteristics of reservoir interbeds, and difficult to meet the requirements of fine reservoir description.

[0005] Second, the division method based on cores and related data: For example, the publicly disclosed technology with the publication number CN115078434A and the name "A method for identifying facies-controlled karstification in the early diagenetic stage of carbonate rocks based on the coupling of petrology and geochemistry". This method analyzes the relationship between rock fabric, karstification, and geochemical characteristics, classifies karst intensity, and establishes an identification model for facies-controlled karstification in the early diagenetic stage. Another example is the publicly disclosed technology with the publication number CN111814879A and the name "A classification method and device for reservoir rock physical facies, electronic equipment, and storage medium". This method divides unimodal reservoirs and / or bimodal reservoirs according to the peak values of pore throat radius distributions of multiple capillary pressure curves and a preset interval, determines multiple J-function curves for the unimodal reservoirs and / or the bimodal reservoirs; establishes a relationship between the multiple J-function curves and water saturation to determine the classification of reservoir rock physical facies. It should be noted that the advantage of this type of method is that it utilizes reliable core analysis and testing data, and the division basis is also relatively sufficient. However, the disadvantage of this type of technology is that if you want to finely depict flow units in the entire oil reservoir, continuous coring experiments need to be carried out on a large number of wells. Restricted by factors such as drilling engineering technology, coring conditions, and high economic costs, the vast majority of oil fields do not have such conditions, so this method is not convenient for extensive application research within the region.

[0006] Third, a method for classifying lithofacies by mathematical methods: In recent years, the advantages of machine learning in solving complex classification problems have been gradually recognized. This method is currently widely used in major oilfields. The overall implementation idea is to first select macroscopic and microscopic features reflecting rock structure and texture, rock physical properties and fluid properties through cores, thin sections and various analytical tests, establish lithofacies classification criteria, logging response characteristics of standard lithofacies, and select suitable artificial intelligence algorithms to establish a lithofacies recognition model to classify lithofacies types. For example, the publicly disclosed technology with the publication number CN201910889252.7 and the name "A reservoir classification and recognition method based on support vector machine algorithm" solves the non-linear problem between lithofacies types of reservoirs and multiple factors to a certain extent. Another example is the publicly disclosed technology with the publication number CN113820754A and the name "A deep tight sandstone reservoir evaluation method for identifying diagenetic facies of reservoirs based on artificial intelligence". For deep tight sandstone reservoirs, it first distinguishes diagenetic facies based on lithology classification results, and uses logging data of determined lithofacies to train a BP neural network to identify diagenetic facies types of each well in the study area. This type of method is a refinement of the second type of method (the method based on cores and related data). Its advantage is that it uses more characteristic parameters. Based on the idea of integrating geology and geophysics, it greatly utilizes the advantages of artificial intelligence means to classify lithofacies. The classification method is developing towards being more quantitative, non-linear, and from low-dimensional space to high-dimensional space, with higher accuracy. However, this type of method also has disadvantages. For reservoirs with complex fracture, pore and cavity configurations and strong heterogeneity, the lithofacies types are extremely diverse, and the identification of lithofacies types far exceeds the identification accuracy of logging data. How to reasonably merge lithofacies within the scope of logging identification ability without being interfered by human factors while ensuring that the classification results have geological significance is an urgent problem to be solved currently.

[0007] Therefore, there is an urgent need to propose a method for fine classification of lithofacies of fracture-cavity type reservoirs that is simple in logic, accurate and reliable. Summary of the Invention

[0008] In view of the above problems, the purpose of the present invention is to provide a method for fine classification of lithofacies of fracture-cavity type reservoirs. The technical solution adopted by the present invention is as follows:

[0009] A method for fine classification of lithofacies of fracture-cavity type reservoirs, which includes the following steps:

[0010] Obtain single-well data of the area to be measured and preprocess the single-well data;

[0011] Combine the preprocessed single-well data to perform depth alignment on the required core physical property data;

[0012] Perform principal component analysis on the preprocessed single-well data to obtain the dimension-reduced and noise-reduced fusion characterization data for lithofacies identification;

[0013] Use the dimension-reduced and noise-reduced fusion characterization data to establish the clustering dataset and the test dataset for the lithofacies identification model;

[0014] Adopt the elbow method to select the optimal classification clusters for the clustering dataset to obtain the number of optimal classification clusters;

[0015] Use the clustering algorithm improved by fuzzy C-means to classify the clustering dataset of the lithofacies identification model, and thus establish the identification model of the fracture-vuggy reservoir lithofacies;

[0016] Adopt the silhouette coefficient to evaluate the identification model of the fracture-vuggy reservoir lithofacies;

[0017] For the identification model of the fracture-vuggy reservoir lithofacies, combine the test dataset to test it;

[0018] Use the identification model of the fracture-vuggy reservoir lithofacies to output the test clustering results, and perform sensitivity analysis using the rank correlation matrix to obtain the sensitivity analysis results;

[0019] Adopt the sensitivity curve to perform intersection to obtain the identification chart of the fracture-vuggy reservoir lithofacies.

[0020] Furthermore, the single-well data includes natural gamma curve, compensated neutron curve, compensated acoustic curve, density curve, deep lateral resistivity and shallow lateral resistivity curves, mass photoelectric absorption cross-section index curve, potassium energy spectrum curve, uranium energy spectrum curve, and thorium energy spectrum curve.

[0021] Furthermore, the expression of the dimension-reduced and noise-reduced fusion characterization data var(X) is:

[0022]

[0023] where a i represents the value of the feature X of the preprocessed single-well data; μ x represents the mean of the preprocessed single-well data.

[0024] Furthermore, the expression of the identification model of the fracture-vuggy reservoir lithofacies is:

[0025]

[0026] s.t.p = log e 4n(e - 1)-1

[0027] where e represents the number of classifications; p represents the parameter to make up for the defects of the partition coefficient and the clustering function; b ij represents the sample point x iMembership degree to the clustering center k j ; Indicates that the sample point x i and the clustering center k j The density-sensitive distance; ρ represents the scaling factor; ω represents the adjustment factor of the fuzzy entropy; G represents the identification model of the fracture-vuggy reservoir rock facies; s.t. p represents the calculation formula for the parameter that makes up for the defects of the partition coefficient and the clustering function.

[0028] Furthermore, the initial clustering center is no longer randomly selected, but is calculated by the average density-sensitive distance, which can effectively solve the problem of noise interference and reduce the number of iterations. Its function is:

[0029]

[0030] In the formula, is the average density-sensitive distance; is the combination number of selecting any two sample points from n samples.

[0031] Furthermore, the expression of the rank correlation matrix is:

[0032]

[0033] Among them, ρ represents the correlation, x i , y i represent any two groups of clustered data, represents the average value of the clustered data x i ; represents the average value of the clustered data y i .

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] (1) The present invention uses natural gamma curve, compensated neutron curve, compensated acoustic curve, density curve, deep lateral resistivity and shallow lateral resistivity curves, mass photoelectric absorption cross section index curve, potassium energy spectrum curve, uranium energy spectrum curve, thorium energy spectrum curve, etc. as the logging data of the work area of the standard well, and can extract useful information through multi-source complex data as much as possible.

[0036] (2) The present invention performs principal component analysis on the logging curves to obtain the dimensionality reduction and noise reduction fusion characterization data for rock facies identification. The advantage is that it reduces the noise in the multi-source data, increases the proportion of useful information, and at the same time reduces the dimension, greatly shortening the calculation amount.

[0037] (3) The present invention uses the elbow method to select the optimal classification cluster for the data. The advantage is that it judges the similarity of the data through the characteristics of the data itself, can greatly reduce the influence of human factors, and the clustering characteristics are more objective and reasonable.

[0038] (4) The present invention classifies data using a clustering algorithm improved by fuzzy C-means and establishes an identification model for the rock facies of fractured-vuggy reservoirs. In addition, an identification model for the rock facies of fractured-vuggy reservoirs is established based on the clustering algorithm improved by fuzzy C-means, and the effect is tested using a test data set, which makes the clustering result more reasonable and credible.

[0039] (5) The present invention conducts sensitivity analysis using a rank correlation matrix. Then, according to the sensitivity analysis results, sensitive curves are selected for crossplotting to obtain an identification chart for the rock facies of fractured-vuggy reservoirs. The advantage is that sensitive parameters are screened, and the obtained identification chart can be quickly applied to oilfields of the same type, greatly improving the applicability and interpretability of the method.

[0040] In summary, the present invention has the advantages of simple logic, accuracy, reliability, etc., and has high practical value and popularization value in the technical field of oil and gas reservoir exploration and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope of protection. For those skilled in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0042] Figure 1 is the logic flow chart of the present invention.

[0043] Figure 2 is the crossplot identification chart for the AC-DEN rock facies of the Deng 4th Member in the present invention.

[0044] Figure 3 is the crossplot identification chart for the AC-DEN rock facies of the Deng 2nd Member in the present invention.

[0045] Figure 4 is the classification prediction and evaluation result chart of the rock facies of a certain well in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] To make the objectives, technical solutions, and advantages of the present application clearer, the present invention will be further described below with reference to the drawings and embodiments. The embodiments of the present invention include, but are not limited to, the following embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present application.

[0047] In this embodiment, the term "and / or" is merely a relational term describing associated objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0048] In the description of the specification and claims of this embodiment, terms such as "first" and "second" are used to distinguish different objects, rather than to describe a specific order of the objects. For example, the first target object and the second target object are used to distinguish different target objects, rather than to describe a specific order of the target objects.

[0049] In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0050] In the description of the embodiments of this application, unless otherwise specified, the meaning of "a plurality of" refers to two or more. For example, a plurality of processing units refers to two or more processing units; a plurality of systems refers to two or more systems.

[0051] As Figures 1 to 4 shown, this embodiment provides a method for fine division of fracture-cavity reservoir rock facies. Taking the fracture-cavity reservoir of the Dengying Formation (Deng 2 Member, Deng 4 Member) in the Sichuan Basin as an example, this embodiment includes the following steps:

[0052] First step, select the single well with the most complete data in the work area of the area to be measured as the standard well.

[0053] Second step, perform logging preprocessing on the logging data of the work area according to the standard well. Among them, the logging data of the work area includes natural gamma curve, compensated neutron curve, compensated acoustic curve, density curve, deep lateral resistivity and shallow lateral resistivity curves, mass photoelectric absorption cross-section index curve, potassium energy spectrum curve, uranium energy spectrum curve, thorium energy spectrum curve, etc.

[0054] Third step, perform depth alignment on the core physical property data to be used subsequently, that is, align it to the logging depth. Here, there is generally a certain depth error between the core data and the wellbore logging test data, and the error is eliminated through depth alignment.

[0055] Fourth step, perform principal component analysis on the preprocessed single well data to obtain the dimension-reduced and noise-reduced fusion characterization data for rock facies identification, and its expression is:

[0056]

[0057] Among them, a i represents the value of the feature X of the single-well data after preprocessing; μ x represents the mean value of the single-well data after preprocessing.

[0058] Here, the statistical representation of the data by the fusion method based on PCA dimensionality reduction and noise reduction is shown in Table 1:

[0059] Table 1 Statistical Representation of Data by the Fusion Method Based on PCA Dimensionality Reduction and Noise Reduction

[0060]

[0061]

[0062] In the fifth step, a clustering data set and a test data set for the rock facies identification model are established by using the dimensionality reduction and noise reduction fusion-represented data. Among them, 80% of the data is used as the model clustering data, and the remaining 20% of the data is used as the model test data.

[0063] In the sixth step, for the clustering data set, the elbow method is used to select the optimal classification clusters for the data to obtain the number of the most optimal classification clusters.

[0064] In the seventh step, for the number of the most optimal classification clusters, an intelligent clustering method (FCM-DSDFP) improved by fuzzy C-means, which has more advantages in complex mapping relationships, is selected to perform intelligent classification on the data, and a fine identification model for the rock facies of fractured-vuggy reservoirs is established. Its expression is:

[0065]

[0066] s.t.p = log e 4n(e - 1)-1

[0067] Among them, e represents the number of classifications; p represents the parameter to make up for the defects of the partition coefficient and the clustering function; b ij represents the membership degree of the sample point x i to the clustering center k j ; represents the sample point x i and the clustering center k jThe density-sensitive distance; ρ represents the scaling factor; ω represents the adjustment factor of fuzzy entropy; G represents the identification model of the fracture-cavity reservoir rock facies; s.t.p represents the calculation formula for compensating the defects of the partitioning coefficient and the clustering function. Here, FCM-DSDFP (i.e., FCM Algorithm Based on Density Sensitive Distance and Fuzzy Partrition) is an improvement of the FCM unsupervised clustering method. Among them, FCM adopts mechanisms such as membership degree and clustering center, uses the alternating optimization algorithm for intelligent solution and clustering, with fewer parameters set, and has more advantages in complex mapping relationships. However, due to the random selection characteristic of the initial clustering center of this method and the local optimality of the Euclidean distance algorithm, FCM is easily interfered by noise points, the clustering effect is unstable and the global consistency is ignored, resulting in a decrease in clustering accuracy.

[0068] Compared with the traditional fuzzy C-means (FCM) algorithm, in the improved intelligent clustering method (FCM-DSDFP), the initial clustering center is no longer randomly selected, but calculated through the average density-sensitive distance, which can effectively solve the problem of noise point interference and reduce the number of iterations. Its function is:

[0069]

[0070] In the formula, is the average density-sensitive distance; is the combination number of selecting any two sample points from n samples.

[0071] FCM-DSDFP overcomes the defects of FCM, realizes global optimal clustering by introducing density-sensitive distance and fuzzy entropy, and can be used to solve the difficult problem of predicting the fracture-cavity reservoir rock facies.

[0072] In the eighth step, the silhouette coefficient is used to evaluate the identification model of the fracture-cavity reservoir rock facies, and the evaluation threshold is preset. When the evaluation result is less than the evaluation threshold, return to the seventh step for classification.

[0073] In the ninth step, for the identification model of the fracture-cavity reservoir rock facies, it is tested in combination with the test data set. Similarly, the identification model is evaluated and analyzed until it is greater than the evaluation threshold.

[0074] In the tenth step, the test clustering result is output by using the identification model of the fracture-cavity reservoir rock facies, and sensitivity analysis is carried out by using the rank correlation matrix to obtain the sensitivity analysis result. See Table 2 for details:

[0075] Table 2 Sensitivity Analysis Based on the Rank Matrix

[0076]

[0077] Step 11: Select sensitive curves according to the sensitivity analysis results and perform crossplotting to obtain the identification chart of fracture-vuggy reservoir rock facies, as Figure 2 and Figure 3 shown. Figure 2 and Figure 3 show that the improved intelligent clustering method (FCM-DSDFP) has obvious classification effects on the rock facies of the 4th and 2nd members of the Dengying Formation. Each type of rock facies has a high coincidence with the core. The rock facies of the 4th and 2nd members of the Dengying Formation can be divided into 7 categories respectively by the improved intelligent clustering method (FCM-DSDFP).

[0078] Among them, Figure 4 (a) in Figure 4 and Figure 4 respectively represent the inspection results of the rock facies classification of the 4th and 2nd members of the Dengying Formation. It can be seen from Figure 4 (a) that it represents a core well in the 4th member of the Dengying Formation with a core length of 16 m. The coincidence rate between the rock facies classification result obtained by the improved intelligent clustering method (FCM-DSDFP) model and the rock facies divided by the core reaches 88.26%, and the coincidence length is 14.12 m; Figure 4 (b) represents a core well in the 4th member of the Dengying Formation with a core length of 13 m. The coincidence rate between the rock facies classification result obtained by the improved intelligent clustering method (FCM-DSDFP) model and the rock facies divided by the core reaches 87.12%, and the coincidence length is 11.32 m. The rock facies classification prediction results evaluated by the method of this embodiment have all achieved good evaluation effects.

[0079] The above embodiments are only the preferred embodiments of the present invention, and do not limit the protection scope of the present invention. Any changes made by using the design principle of the present invention and non-creative labor on this basis shall fall within the protection scope of the present invention.

Claims

1. A method for fine division of rock phases of fracture-cavity reservoirs, characterized in that: The following steps are involved: Obtain single well data of the area to be tested and preprocess the single well data; Combined with the pre-processed single well data, the required core physical property data is depth-relocated; The pre-processed single well data is subjected to principal component analysis to obtain dimension reduction, noise reduction and fusion characterization data for rock phase identification; The clustering data set and test data set of the rock phase identification model are established by using dimensionality reduction, noise reduction and fusion characterization data; The elbow rule is used to select the optimal classification cluster for the clustering data set to obtain the most classification cluster number; The clustering algorithm improved by fuzzy C-means is used to classify the clustering data set of the rock phase identification model, so as to establish the identification model of the rock phase of the fracture-cavity reservoir. The silhouette coefficient is used to evaluate the identification model of fracture-vuggy reservoir rock phases; The rock phase identification model of fracture-cavity reservoir is tested in combination with the test data set; The clustering results were tested using the output of the identification model of fracture-cavity reservoir rock phases, and the rank correlation matrix was used to perform sensitivity analysis to obtain the sensitivity analysis results. The identification plate of fracture-vuggy reservoir rock phase is obtained by using sensitive curve intersection.

2. A method for fine division of rock phases of fracture-cavity reservoirs according to claim 1, characterized in that: The single well data include natural gamma curve, compensated neutron curve, compensated acoustic wave curve, density curve, deep lateral resistivity and shallow lateral resistivity curve, mass photoelectric absorption cross-section index curve, potassium spectrum curve, uranium spectrum curve, and thorium spectrum curve.

3. The method for fine division of rock phases of fracture-cavity reservoir according to claim 1, characterized in that: The expression of the dimension reduction, noise reduction and fusion characterization data var(X) is: Among them, a i represents the value of feature X of the preprocessed single well data; μ x Represents the mean of the preprocessed single well data.

4. A method for fine division of rock phases of fracture-cavity reservoirs according to claim 1, 2 or 3, characterized in that: The expression of the identification model of the fracture-cavity reservoir rock phase is: s.t.p=log e 4n(e-1)-1 Among them, e represents the number of classifications; p represents the parameter that makes up for the defects of the partition coefficient and clustering function; b ij Represents the sample point x i For cluster center k j The degree of membership; Represents the sample point x i With cluster center k j The density sensitive distance; ρ represents the scaling factor; ω represents the adjustment factor of fuzzy entropy; G represents the identification model of fracture-cavity reservoir rock phase; s.t. p represents the calculation formula for making up for the defective parameters of the partition coefficient and clustering function.

5. A method for fine division of rock phases of fracture-cavity reservoirs according to claim 4, characterized in that: The expression of the rank correlation matrix is: Among them, ρ represents the correlation, x i ,y i represents any two sets of clustered data, Represents clustered data x i The average value of Represents clustered data y i The average value of .

6. A method for fine division of rock phases of fracture-cavity reservoirs according to claim 4, characterized in that: The sample point x i With cluster center k j Density sensitive distance Average density sensitive distance The expression is obtained as follows: in, is the average density sensitive distance; is the number of combinations of selecting any two sample points from n samples.

Images