A shale oil lithofacies identification method based on artificial intelligence and well layering modeling

Abstract

This invention relates to the field of oilfield development technology, and in particular to a method for identifying shale oil facies based on artificial intelligence and well logging stratification modeling. The method includes the following steps: selecting core wells in the shale oil system of the study area and determining the well logging sensitivity curves; rapidly identifying fractures or organic-rich lamellar shale mudstone with stratigraphic brilliance in the shale oil reservoir using fuzzy clustering; establishing well logging models for different lithologies of shale oil using core calibration logging to determine the lithology in the shale oil reservoir; identifying the bedding structure; establishing a computational model to predict the organic carbon content of shale oil wells in the study area and distinguishing organic matter types; and performing comprehensive shale oil facies identification. This invention fills the technical gap in identifying favorable facies in shale oil reservoirs, namely organic-rich lamellar shale with stratigraphic brilliance, and proposes a method for finely classifying shale oil facies.

Description

Technical Field

[0001] This invention relates to the field of oilfield development technology, and in particular to a method for identifying shale oil facies based on artificial intelligence and well logging stratification modeling. Background Technology

[0002] In recent years, with China's rapid economic development, the search for unconventional oil and gas resources has become inevitable while ensuring conventional oil and gas supplies. Shale oil, as an important resource replacement area, has become a new focus of exploration and development.

[0003] Shale oil reservoirs are characterized by complex lithology, strong heterogeneity, extremely low porosity and permeability, and diverse occurrence patterns, making it difficult to evaluate and study the facies characteristics of shale oil using well logging information. However, different facies have different development potentials. Therefore, finding favorable shale oil facies intervals and clearly and effectively delineating shale oil facies are key to identifying the main exploration and development target layers in shale systems, and conducting analysis and research on them is of great significance.

[0004] The organic-rich lamellar shale mudstone is the best in terms of reservoir capacity, oil content, mobility, and compressibility among the shale oil facies in the study area, making it a favorable facies. Therefore, well logging identification of the organic-rich lamellar shale mudstone is very important.

[0005] Currently, well logging techniques lack sufficient precision and their logging responses to sparry calcite veins are not significant. Therefore, there is no mature technology for identifying favorable lithofacies in organic-rich, layered, sparry calcareous mudstone. Furthermore, existing methods for predicting shale oil lithofacies in the study area have limited applicability and accuracy, primarily due to the following reasons:

[0006] I. The shale oil facies in the study area exhibit complex logging responses and unclear logging curve characterization. The same type of sparry calcite vein corresponds to different logging response characteristics. Furthermore, the insufficient precision of the logging curves prevents the information about the sparry calcite veins from being adequately represented on the logging curves.

[0007] II. Complex and diverse reservoir space and bedding structure. Different bedding structures result in different reservoir spaces and capacities. Layered sparry calcareous mudstone has the best reservoir capacity, with well-developed calcite intercrystalline pores and good fluorescence display; followed by layered dark calcareous mudstone, with moderate reservoir capacity and dispersed fluorescence display; finally, dense massive limestone or mudstone has low reservoir capacity, with calcareous and argillaceous micropores and distributed quartz and organic matter.

[0008] Third, the reservoir structure is complex, and general logging lithofacies models are not applicable. Summary of the Invention

[0009] The main objective of this invention is to provide a shale oil facies identification method based on artificial intelligence and well logging stratification modeling. This method can quickly identify favorable facies in shale oil reservoirs, such as organic-rich lamellar spar argillaceous mudstone, delineate favorable facies intervals for shale oil, and effectively identify shale oil facies through well logging. This provides an important basis for facies prediction and sweet spot development of shale oil wells, and is of great significance for the exploration and development of shale oil wells in the Dongying Depression.

[0010] To achieve the above objectives, the present invention adopts the following technical solution:

[0011] This invention provides a shale oil lithological identification method based on artificial intelligence and well logging stratification modeling, comprising the following steps: selecting core wells of the shale oil system in the study area and determining the well logging sensitivity curves; rapidly identifying fractures or organic-rich layered sparry mudstone in the shale oil reservoir using fuzzy clustering; establishing lithological models for different shale oil well logging layers using core calibration logging to further determine the location of fractures or organic-rich layered sparry mudstone in the shale oil reservoir; identifying the bedding structure; establishing a computational model to predict the organic carbon content of shale oil wells in the study area and distinguishing organic matter types; and performing shale oil lithological identification.

[0012] Furthermore, core wells of the shale oil system were selected, and relevant core, thin section, logging data, and experimental analysis data were collected. Based on the logging curves corresponding to the lithology of the core section in the logging data, the logging sensitivity curves were determined.

[0013] Furthermore, the well logging sensitivity curves are standardized, and fuzzy cluster analysis is performed on the standardized well logging sensitivity curves to classify different lithofacies. The clustered lithofacies are then matched with the locations of fractures and organic-rich layered sparry mudstone observed in the cores of standard shale oil wells to determine the lithofacies types corresponding to the organic-rich layered sparry mudstone and fractures, thereby determining the layer segments and locations of fractures or organic-rich layered sparry mudstone in shale oil reservoirs.

[0014] Shale oil fractures are mainly structural microfractures and diagenetic microfractures. Fractures have a significant impact on and contribution to production capacity, and the exploration of fractured oil and gas reservoirs is a key target for new reserves.

[0015] The organic-rich, lamellar, sparry, calcareous mudstone exhibits a lamellar bedding structure with unfilled bedding fractures and semi-filled high-angle structural fractures. The pore types are mainly intergranular and intercrystalline pores. The sparry calcite lamellars develop discontinuous wavy dissolution intercrystalline fractures with a maximum fracture width of 5 μm and a lamellar thickness of mostly <1 mm. It is the lithofacies with the highest reservoir capacity and the best fluorescence scanning performance among all shale oil lithofacies.

[0016] Because the resolution of well logging curves is only 0.125m, it is insufficient to identify fractures or organic-rich lamellar spar limestone mudstone in shale oil reservoirs using conventional logging methods. Well logging curves within the same lithofacies group, identified through fuzzy clustering, show significant similarity, while data from different lithofacies exhibit considerable dissimilarity. The goal of fuzzy clustering analysis is to classify lithofacies based on the similarity of these data. Cluster analysis can quantitatively determine the affinity between lithofacies samples, thus objectively classifying them. Therefore, using fuzzy clustering machine learning methods to find fractures and the most favorable lithofacies—organic-rich lamellar spar limestone mudstone—can quickly and accurately determine their intervals and locations.

[0017] Standardizing the logging sensitivity curve ensures that there are no differences between logging curves of shale oil wells when applying it to lithofacies, thus guaranteeing the accuracy of the results.

[0018] Furthermore, using core calibration logging, layered logging models of different lithologies of shale oil were established to determine the lithology of shale oil logging in the study area. Continental shale oil reservoirs exhibit strong facies heterogeneity and rapid vertical facies changes, making it impossible to find corresponding logging response characteristics using conventional logging methods. Therefore, this study proposes establishing layered logging lithology models for different shale oil types to narrow the range of facies changes and ensure that the logging response characteristics corresponding to the same facies are consistent within the same stratigraphic level. Using this method of establishing layered logging lithology models for different shale oil types, the lithologies (calcareous mudstone, argillaceous limestone, mudstone, dolomite, limestone) are effectively identified.

[0019] Furthermore, by using the P-wave / S-wave velocity ratio, the bedding structure of the shale oil layers in the study area was identified, and the bedding structure of the shale oil layers was qualitatively classified into lamellar and layered structures. According to the propagation principle of P-waves and S-waves, the displacement direction of the S-wave particles is perpendicular to the well axis. In bedding and low-angle fractures, part of the S-wave energy propagates along the bedding and low-angle fractures, thus reducing the propagation velocity of the S-waves collected by the instrument. However, the propagation direction and particle displacement direction of the P-wave are parallel to the well axis, and the bedding and low-angle fractures have little effect on its velocity. Therefore, in areas where bedding and low-angle fractures are well developed, the P-wave / S-wave velocity ratio Vp / Vs increases. The bedding structure of shale oil is mainly composed of shale, with very few blocky structures. Based on the P-wave / S-wave propagation principle, the bedding structure is classified into lamellar and layered structures.

[0020] Furthermore, when the P-wave / S-wave velocity ratio Vp / Vs > 2 μs / ft, the layered structure is lamellar; when Vp / Vs < 2 μs / ft, the layered structure is lamellar.

[0021] Furthermore, a method for predicting the organic carbon content of shale oil wells in the study area using a computational model includes the following steps: classifying the depressions in the study area; analyzing the correlation between the organic carbon content of each depression and logging parameters; selecting the logging parameters with correlation; performing multiple regression analysis; and establishing a computational model containing multiple factors.

[0022] Furthermore, the method for distinguishing organic matter types is as follows: when the organic carbon content is greater than 2%, it is considered to be rich in organic carbon; when the organic carbon content is between 1% and 2%, it is considered to contain organic carbon; and when the organic carbon content is less than 1%, it is considered to be poor in organic carbon.

[0023] Furthermore, a method for identifying shale oil facies involves comprehensively identifying the lithology, bedding structure, rock structure, and organic matter type within the facies to achieve a refined classification of shale oil facies.

[0024] The method described in this invention first uses fuzzy clustering to quickly identify fractures and favorable lithofacies in shale oil reservoirs that cannot be identified using conventional logging;

[0025] By using core calibration logging, lithological models of different shale oil well logging are established in layers to further identify other lithologies (lime mudstone, argillaceous limestone, mudstone, dolomite, limestone) in shale oil reservoirs. The comprehensive bedding structure (laminated, layered) and organic matter type (rich in organic carbon, containing organic carbon, poor in organic carbon) are used to comprehensively name the lithofacies, thus achieving comprehensive identification of shale oil lithofacies.

[0026] Compared with the prior art, the present invention has the following advantages:

[0027] This invention fills the technological gap in identifying favorable facies in shale oil reservoirs, namely, organic-rich, layered, bright crystalline shale, and proposes a method for finely classifying shale oil facies. Using the method described in this invention, the stratigraphic intervals containing high-quality shale oil facies can be efficiently identified, and different shale oil facies types can be effectively distinguished. This provides important basis for facies prediction and sweet spot development of shale oil wells, and is of great significance for the exploration and development of shale oil wells in depressions. Attached Figure Description

[0028] Figure 1 A flowchart illustrating a specific implementation of the shale oil facies identification method based on artificial intelligence and well logging stratification modeling according to the present invention;

[0029] Figure 2 To quickly identify favorable lithofacies, organic-rich lamellar spar-grayish mudstone, and fracture maps in shale oil reservoirs using fuzzy clustering algorithms;

[0030] Figure 3 To establish logging model diagrams for different lithologies of shale oil;

[0031] Figure 4 A diagram for identifying layered structure based on the longitudinal and transverse wave ratios;

[0032] Figure 5 An identification chart for organic carbon content;

[0033] Figure 6 A diagram illustrating the lithofacies classification method of "three units and four components";

[0034] Figure 7 A comprehensive lithofacies identification map of shale oil wells in the study area. Detailed Implementation

[0035] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0036] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, and / or combinations thereof.

[0037] To enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention will be described in detail below with reference to specific embodiments. Example 1

[0038] A shale oil facies identification method based on artificial intelligence and well logging stratification modeling includes the following steps:

[0039] Step 1. Select core wells of the shale oil system in the study area and determine the logging sensitivity curves; select core wells of the shale oil system, collect relevant core, thin section, logging data and experimental analysis data, and determine the logging sensitivity curves based on the logging curves corresponding to the lithology of the core section in the logging data.

[0040] Step 2. Quickly identify fractures or organic-rich, layered, bright crystalline argillaceous mudstone in shale oil reservoirs using fuzzy clustering;

[0041] The well logging sensitivity curves are standardized, and fuzzy cluster analysis is performed on the standardized well logging sensitivity curves to classify different lithofacies. The clustered lithofacies are matched with the locations of fractures and organic-rich layered sparry mudstone observed in the cores of standard shale oil wells to determine the lithofacies types corresponding to the organic-rich layered sparry mudstone and fractures, and thus determine the intervals and locations of fractures or organic-rich layered sparry mudstone in shale oil reservoirs.

[0042] Step 3. Using core calibration logging, establish logging models for different lithologies of shale oil in layers.

[0043] Further investigation was conducted to determine the location of fractures or organic-rich, layered, sparry-like calcareous mudstone in shale oil reservoirs. Core calibration logging was used to establish stratified lithological models for different shale oil reservoirs, thus identifying the lithology of shale oil logging in the study area. Continental shale oil reservoirs exhibit strong lithofacies heterogeneity and rapid vertical facies changes, making it impossible to find corresponding logging response characteristics using conventional logging methods. Therefore, this study proposes establishing stratified lithological models for different shale oil reservoirs to narrow the range of lithofacies changes and ensure that the logging response characteristics corresponding to the same lithofacies are consistent within the same stratigraphic level. The method of establishing stratified lithological models for different shale oil reservoirs effectively identifies the lithology; the lithologies include calcareous mudstone, argillaceous limestone, mudstone, dolomite, and limestone.

[0044] Step 4. Identify the layered structure; further, identify the layered structure by the ratio of longitudinal and transverse wave velocities, and qualitatively classify the layered structure into lamellar and layered structures.

[0045] When the P-wave / S-wave velocity ratio Vp / Vs > 2 μs / ft, the layered structure is lamellar; when Vp / Vs < 2 μs / ft, the layered structure is lamellar.

[0046] Step 5. Establish a computational model to predict the organic carbon content of shale oil wells in the study area and distinguish the types of organic matter;

[0047] The depressions in the study area were classified; the correlation between organic carbon content and logging parameters in each depression was analyzed; logging parameters with correlation were selected; multiple regression analysis was performed; and a calculation model with multiple factors was established.

[0048] Methods for distinguishing organic matter types: When the organic carbon content is greater than 2, it is considered to be rich in organic carbon; when the organic carbon content is between 1 and 2, it is considered to contain organic carbon; when the organic carbon content is less than 1, it is considered to be poor in organic carbon.

[0049] Step 6. Perform shale oil facies identification.

[0050] Comprehensive and detailed identification of shale oil facies is carried out based on lithology, bedding structure, rock structure, and organic matter type. Example 2

[0051] Using shale oil wells in the Dongying Depression as an example, this paper specifically illustrates a shale oil facies identification method based on artificial intelligence and well logging stratification modeling.

[0052] The method includes the following steps:

[0053] Step 1: Select core wells in the Dongying Depression shale oil system, collect relevant core, thin section, logging data and experimental analysis data, and determine the natural gamma (GR), resistivity (RD), sonic transit time (AC), density (DEN) and neutron (CNL) as the five sensitive curves of shale oil reservoirs based on the logging curves corresponding to the lithology of the core section in the logging data.

[0054] Step 2: Use fuzzy clustering to predict fractures or organic-rich layered sparry mudstone in shale oil reservoirs based on the five selected sensitive curves.

[0055] The five logging sensitivity curves—natural gamma (GR), resistivity (RD), acoustic transit time (AC), density (DEN), and neutron (CNL)—were standardized respectively.

[0056] Fuzzy clustering analysis was performed on the five standardized logging sensitivity curves (natural gamma ray GR, resistivity RT, and sonic transit time AC) to classify them into different lithofacies.

[0057] The lithofacies were divided into 15 categories using fuzzy clustering, such as Figure 2 As shown, the 15 clustered lithofacies were matched and verified with the locations of fractures and organic-rich layered sparry mudstone observed in cores and thin sections of standard shale oil wells. Based on fuzzy clustering, the locations of fractures and organic-rich layered sparry mudstone in other shale oil wells can be quickly determined, with a matching rate of over 85%.

[0058] Step 3: Using core calibration logging, establish logging models for different lithologies of shale oil in layers.

[0059] Therefore, to improve interpretation accuracy, a method of establishing lithological models for different shale oil well logging layers is used to effectively identify the lithology (calcareous mudstone, argillaceous limestone, mudstone, dolomite, limestone), such as... Figure 3 As shown. For shale oil wells in the study area, the shale oil reservoir development sections are concentrated in three sections: Chunshang 1, Chunshang 2, and Chunshang 3, with dolomite mainly concentrated in section Chunshang 3.

[0060] The logging response characteristics of different lithologies in shale oil were determined by dividing the shale oil into three layers: Chunshang 1, Chunshang 2, and Chunshang 3. Logging models for different lithologies of shale oil were then established layer by layer to improve interpretation accuracy, as shown below:

[0061] Pure 1: Mudstone (GR > 42, RT < 3, the three porosities are all biased to the left), argillaceous limestone (GR < 42, RT > 5, the three porosities are all biased to the right), calcareous mudstone (GR > 42, 3 < RT < 5, the three porosities are all biased to the right), limestone (GR < 42, RT > 40, the three porosities are all biased to the right);

[0062] Pure 2: Mudstone (GR > 55, RT < 3, the three porosities are all biased to the left), argillaceous limestone (GR < 55; RT > 5;, the three porosities are all biased to the right), calcareous mudstone (GR > 55, 3 < RT < 5, the three porosities are all biased to the right), limestone (GR < 55, RT > 40, the three porosities are all biased to the right);

[0063] Pure 3: Mudstone (GR > 62; RT < 3; the three porosities are all biased to the left), argillaceous limestone (GR < 62; RT > 5, the three porosities are all biased to the right), calcareous mudstone (GR > 62; 3 < RT < 5; the three porosities are all biased to the right), limestone (GR < 62, RT > 40, the three porosities are all biased to the right), dolomite (GR > 70; RT > 8; low values of AC and CNL).

[0064] Step 4: Identify the bedding structure through the P-wave to S-wave velocity ratio, and qualitatively classify the bedding structure into laminar and layered.

[0065] When the P-wave to S-wave velocity ratio Vp / Vs > 2 μs / ft, the bedding structure is laminar; when Vp / Vs < 2 μs / ft, the bedding structure is layered; the obtained bedding structure diagram is as Figure 4 shown.

[0066] Step 5: Fit the organic carbon content formula of shale oil lithofacies in different sags to distinguish the types of organic matter.

[0067] In the study area, the shale oil wells in the Dongying Sag are divided into three sags: Niuzhuang Sag, Boxing Sag, and Lijin Sag. The multiple regression analysis method is used to predict the organic carbon content of the shale oil wells in the study area.

[0068] Generally, rich organic matter source rocks have well logging response characteristics such as high natural gamma, high acoustic travel time, and low density. The GR value is on the high side because there are a large number of radioactive elements U, Th, and K in the formation, especially the uranium element has a good indication effect on organic matter; the density of organic matter is low, resulting in the overall low density of the source rock; organic matter will reduce the sound velocity and mature source rocks have liquid hydrocarbons that are not easy to conduct electricity, resulting in higher acoustic travel time and resistivity.

[0069] The study area exhibits logging response characteristics of high natural gamma ray, high sonic transit time, high resistivity, and low density, which form the basis for predicting TOC content. Correlation analysis of measured TOC from shale oil wells in the study area with commonly used logging curves revealed high correlation coefficients for natural gamma ray (GR), resistivity (RD), density (DEN), and sonic transit time (AC). Therefore, these four parameters were used to perform multiple regression analysis on the organic carbon content (TOC).

[0070] Niuzhuang Depression: TOC = 37.97 + 0.02 * GR + 1.67 * log(RD) - 14.56 * DEN + 0.074 * AC

[0071] Boxing Depression: TOC = 5.24 + 0.02 * GR + 1.03 * log(RD) - 2.29 * DEN + 0.19 * AC

[0072] Lijin Depression: TOC = 24.4 + 0.03 * GR + 3.64 * log(RD) - 3.87 * DEN + 0.60 * AC

[0073] In the formula: GR represents natural gamma; RD represents resistivity; DEN represents density; AC represents sound wave transit time.

[0074] Within the study area, organic carbon content was categorized as follows: organic carbon content greater than 2 was considered organically rich; organic carbon content between 1 and 2 was considered organically contained; and organic carbon content less than 1 was considered organically poor. The resulting organic carbon content identification map is shown below. Figure 5 As shown.

[0075] In step 6, the lithofacies of the study area are divided according to the "three-unit four-component" lithofacies classification method.

[0076] The "three units and four components" refers to a comprehensive classification based on lithology, bedding structure, rock texture, and organic matter type within the overall lithofacies.

[0077] Using a three-end-member diagram of rock lithology (sandy, argillaceous, carbonate), sedimentary structure (massive, layered, lamellar), rock texture (sharp, cryptocrystalline, microcrystalline), and organic matter (rich in organic matter, containing organic matter, poor in organic matter), such as... Figure 6 As shown, it can be used to achieve fine classification of shale oil facies in the study area.

[0078] The main lithologies in the study area include mudstone, limestone, sandstone, calcareous mudstone, argillaceous limestone, sandy mudstone, and sandy limestone; the bedding structures include lamellar, layered, and massive; the rock textures include sparry, cryptocrystalline, and microcrystalline; and the organic matter types include rich in organic carbon, poor in organic carbon, and containing organic carbon. A comprehensive lithological identification of shale oil wells in the study area is conducted. The resulting comprehensive lithological identification map of shale oil wells in the study area is shown below. Figure 7 As shown.

[0079] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A shale oil facies identification method based on artificial intelligence and well logging stratification modeling, characterized in that, Includes the following steps: Core wells of the shale oil system in the study area were selected to determine the logging sensitivity curves; Fuzzy clustering was used to rapidly identify fractures or organic-rich lamellar shale mudstone in shale oil reservoirs. Well logging sensitivity curves were standardized, and fuzzy clustering analysis was performed on these standardized curves to classify different lithofacies. The clustered lithofacies were matched with the locations of fractures and organic-rich lamellar shale mudstone observed in core samples from standard shale oil wells to determine the lithofacies types corresponding to the fractures and organic-rich lamellar shale mudstone, thereby identifying the stratigraphic intervals and locations of fractures or organic-rich lamellar shale mudstone in shale oil reservoirs. Core calibration logging was used to establish lithological models for different shale oil wells, determining the lithology of shale oil well logging in the study area. The bedding structure of shale oil layers in the study area was identified. The bedding structure of shale oil layers in the study area was identified by the P-wave / S-wave velocity ratio, qualitatively classifying the bedding structure of shale oil layers into lamellar and layered types. When the P-wave / S-wave velocity ratio Vp / Vs... When Vp / Vs > 2 μs / ft, the bedding structure is lamellar; when Vp / Vs < 2 μs / ft, the bedding structure is layered. A computational model is established to predict the organic carbon content of shale oil wells in the study area. The method for establishing a computational model to predict the organic carbon content of shale oil wells in the study area includes the following steps: classifying the depressions in the study area; analyzing the correlation between the organic carbon content of each depression and logging parameters, selecting logging parameters with correlation, performing multivariate regression analysis, and establishing a multi-factor computational model; distinguishing organic matter types and identifying shale oil facies.

2. The shale oil facies identification method based on artificial intelligence and well logging stratification modeling according to claim 1, characterized in that, Select core wells in the shale oil system, collect relevant core, thin section, logging data and experimental analysis data, and determine the logging sensitivity curve based on the logging curve corresponding to the lithology of the core section in the logging data.

3. The shale oil facies identification method based on artificial intelligence and well logging stratification modeling according to claim 1, characterized in that, The lithology includes calcareous mudstone, argillaceous limestone, mudstone, dolomite, and limestone.

4. The shale oil facies identification method based on artificial intelligence and well logging stratification modeling according to claim 1, characterized in that, Methods for distinguishing organic matter types: When the organic carbon content is greater than 2, it is considered to be rich in organic carbon; when the organic carbon content is between 1 and 2, it is considered to contain organic carbon; when the organic carbon content is less than 1, it is considered to be poor in organic carbon.

5. The shale oil facies identification method based on artificial intelligence and well logging stratification modeling according to claim 1, characterized in that, Methods for identifying shale oil facies: Comprehensive identification of shale oil facies based on lithology, bedding structure, rock structure, and organic matter type.

Images