A method for predicting marine shale sweet spot layers

Abstract

This application discloses a method for predicting sweet spots in marine shale, relating to the field of oil and gas geological exploration and shale sweet spot prediction technology. The method includes: constructing a quantitative relationship model between Frod number, Reynolds number, and total organic carbon content, and determining the optimal hydrodynamic window for organic matter enrichment; repositioning and standardizing core data and well logging curves to obtain repositioned core data and standardized well logging curves; using a nonlinear mapping model based on the repositioned core data and standardized well logging curves to obtain grain size parameter prediction curves and geochemical index prediction curves; calculating ancient Frod number curves and ancient Reynolds number curves based on the well logging curves, grain size parameter prediction curves, and geochemical index prediction curves; and using the optimal hydrodynamic window for organic matter enrichment to screen the ancient Frod number curves and ancient Reynolds number curves to obtain the sweet spot of the target marine shale, thus completing the sweet spot prediction. This application can improve the accuracy of sweet spot prediction.

Description

Technical Field

[0001] This application relates to the field of oil and gas geological exploration and shale sweet spot prediction technology, and in particular to a method for predicting sweet spots in marine shale. Background Technology

[0002] Shale gas has become a very important global oil and gas resource. The enrichment and high production of shale gas are mainly controlled by the "sweet spot," and one of the core characteristics of the "sweet spot" is its high organic matter content.

[0003] In some cases, research on the enrichment patterns of organic matter in shale has focused primarily on identifying geochemical indicators such as paleoproductivity and redox conditions, while insufficient attention has been paid to the sedimentary hydrodynamic processes controlling organic matter enrichment and preservation, and quantitative research methods are lacking. Major sweet spot prediction techniques rely heavily on well logging curves and geophysical inversion, but these methods are largely "effect-based" rather than "causal," limiting their ability to explain the intrinsic mechanisms of organic matter enrichment and predict its spatial distribution. Particularly for shale formations with strong heterogeneity, such as the southern marine shale, using these methods will lead to inaccurate predictions of sweet spots, increasing the risks of subsequent exploration. Summary of the Invention

[0004] The purpose of this application is to provide a method for predicting sweet spots in marine shale, which can improve the accuracy of sweet spot prediction.

[0005] To achieve the above objectives, this application provides the following solution.

[0006] This application provides a method for predicting sweet spots in marine shale, including:

[0007] Based on sample experimental data, a quantitative relationship model of Frod number-Reynolds number-total organic carbon content was constructed, and the optimal hydrodynamic window for organic matter enrichment was determined; the optimal hydrodynamic window for organic matter enrichment includes: the optimal Frod number window and the optimal Reynolds number window;

[0008] Obtain core data and well logging curves of the target marine shale; the core data includes: grain size parameters and geochemical indices;

[0009] The core data and the logging curves are respectively repositioned and standardized to obtain repositioned core data and standardized logging curves;

[0010] Based on the repositioned core data and the standardized logging curves, a nonlinear mapping model is used to obtain the grain size parameter prediction curve and the geochemical index prediction curve; the nonlinear mapping model is obtained by training a random forest model.

[0011] Based on the well logging curve, the grain size parameter prediction curve, and the geochemical index prediction curve, calculate the Gufrod number curve and the Gureynolds number curve;

[0012] The optimal organic matter enrichment hydrodynamic window is used to screen the ancient Frode number curve and the ancient Reynolds number curve to obtain the sweet spot layer of the target marine shale, thus completing the sweet spot layer prediction.

[0013] According to the specific embodiments provided in this application, this application has the following technical effects:

[0014] This application constructs a quantitative relationship model of Frod number, Reynolds number, and total organic carbon content based on sample experimental data. It establishes a causal relationship between Frod number, Reynolds number, and total organic carbon content, providing a comparable and unified benchmark for shale formations in different basins and strata within marine facies. Furthermore, by establishing the causal relationship between Frod number, Reynolds number, and total organic carbon content, it determines the optimal hydrodynamic window for organic matter enrichment. This elevates the prediction of sweet spots from a vague qualitative assessment of "deep-water areas rich in organic matter" to a quantitative prediction with a specific coupling relationship between Frod number, Reynolds number, and total organic carbon content, thereby improving the prediction accuracy of sweet spots. In practical applications, after repositioning and standardizing the core data and well logging curves of the target marine shale, a nonlinear mapping model is used to generate prediction curves from data related to grain size parameters and geochemical indices. This enables the calculation of the Gufrod number curve and the Gureynolds number curve. Then, by filtering the Gufrod number curve and the Gureynolds number curve through the optimal organic matter enrichment hydrodynamic window, the sweet spot layer of the target marine shale is obtained. This achieves the determination of the optimal organic matter enrichment layer through the Gufrod number curve and the Gureynolds number curve, thereby improving the accuracy of sweet spot layer prediction. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a method for predicting sweet spots in marine shale, provided as an embodiment of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] In one exemplary embodiment, such as Figure 1 As shown, a method for predicting sweet spots in marine shale is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is described using a server as an example, and includes the following steps S1 to S8.

[0020] Step S1: Based on the sample experimental data, construct a quantitative relationship model of Frod number-Reynolds number-total organic carbon content, and determine the optimal hydrodynamic window for organic matter enrichment; the optimal hydrodynamic window for organic matter enrichment includes: the optimal Frod number window and the optimal Reynolds number window.

[0021] As one possible implementation, step S1 includes steps S11 to S14:

[0022] Step S11: Obtain sample experimental data; the sample experimental data includes: experimental parameters and total organic carbon content of the experimental sediments; the experimental parameters include: flow rate, water depth and temperature.

[0023] Specifically, the sample experimental data were obtained through physical simulation experiments.

[0024] First, the physical simulation experimental system is constructed: a variable slope circulating water tank experimental system that can accurately control and measure hydrodynamic parameters is provided. The system includes: (1) the main body of the water tank: a transparent water tank with a length of not less than 4m and a width of not less than 0.3m, whose bottom slope can be accurately adjusted within the range of 0-5°; (2) the fluid drive and control system: including a variable frequency water pump, a high-precision flow meter and a flow stabilization device, which can achieve the stabilization and control of the flow velocity within the range of 0.1~1.5m / s; (3) the parameter measurement system: integrating a laser Doppler current meter (for measuring the vertical flow velocity profile), an ultrasonic water depth sensor (for accurately measuring water depth), and online monitoring sensors for temperature, pH value and dissolved oxygen; (4) the data acquisition system: used to synchronously record flow velocity, water depth, temperature and other related parameters.

[0025] Secondly, the simulated sediment was prepared: based on the petrological and geochemical characteristics of the target geological strata, a representative ternary component simulated sediment suspension was prepared: (1) Clastic component: accounting for about 70% by weight, composed of clay minerals (such as kaolinite and illite) and silt-grade quartz and feldspar particles, with a particle size controlled at <62.5μm; (2) Organic component: accounting for about 15% by weight, prepared by drying and grinding algae (such as Chlorella) cultured in the laboratory, simulating marine organic matter; (3) Carbonate component: accounting for about 15% by weight, using fine-grained calcite powder (<20μm); in order to facilitate subsequent source tracking and differentiation, a trace amount (about 0.1%) of fluorescent microspheres was added to the organic component, and a trace amount (about 0.5%) of magnetite powder was added to the siliceous clastic component as a tracer.

[0026] Finally, experiments and data acquisition: a multi-factor, multi-level experimental design was adopted to systematically change key variables such as flow velocity and water depth to cover the target Frod number and Reynolds number range: (1) Experimental matrix design: set flow velocity gradient (e.g., 0.1, 0.3, 0.5 m / s), water depth gradient (e.g., 0.05, 0.10, 0.15 m) and suspended matter concentration gradient (e.g., 5%, 10%, 15%) to form a comprehensive experimental parameter combination; (2) Standardized experimental procedure execution: for each parameter combination, first, adjust the water tank to the predetermined slope, start the fluid drive and control system, and after the flow field stabilizes, precisely adjust to the target flow velocity and water depth; then, inject the pre-prepared simulated sedimentation suspension evenly into the upstream of the water tank through a constant flow delivery device; maintain hydraulic Under stable conditions, a sedimentation process lasting no less than 1 hour is carried out; finally, a slow drainage method (drainage speed no greater than 0.5 cm / s) is adopted to prevent the sediment from being resuspended; each parameter combination experiment is repeated at least three times to ensure the reproducibility and statistical significance of the data; (3) Real-time measurement and calculation of hydrodynamic parameters: During the sedimentation stage, the following is achieved through an integrated measurement system: the cross-sectional velocity distribution is measured in real time using a laser Doppler current meter, and the average velocity per unit time is calculated; the water depth is continuously monitored by a high-precision water depth sensor; the water temperature is recorded synchronously, and the corresponding kinematic viscosity coefficient is determined according to the international standard water property parameter table; based on the real-time measured average velocity, water depth parameters and determined kinematic viscosity coefficient, the Frod number and Reynolds number of each experiment are calculated according to the following formula.

[0027] Froude number ( Calculation formula: .

[0028] Reynolds number ( Calculation formula: .

[0029] in, For flow rate, Because of the water depth, Let be the coefficient of kinematic viscosity. The acceleration due to gravity is taken as 9.8 m / s². 2 .

[0030] Step S12: Calculate the Frode number and Reynolds number based on the experimental parameters.

[0031] Step S13: Based on the Floyd number, Reynolds number, and total organic carbon content, a quantitative relationship model of Floyd number-Reynolds number-total organic carbon content is constructed using response surface methodology.

[0032] As one possible implementation, step S13 includes steps S131 to S132:

[0033] Step S131: Using the Frod number and the Reynolds number as independent variables and the total organic carbon content as the dependent variable, fit the response surface model to obtain the model coefficients of the response surface model.

[0034] Step S132: Based on the model coefficients of the response surface model, construct a quantitative relationship model of Frode number-Reynolds number-total organic carbon content.

[0035] As an implementable approach, the quantitative relationship model can be expressed as follows:

[0036] .

[0037] in, This refers to the total organic carbon content. , , , , and All are model coefficients. For Freud number, Let Reynolds number be 1. The parameters are all determined by least squares fitting, which represents random error terms. For about and Continuous functions, .

[0038] Specifically, data analysis and model building: Through systematic testing and mathematical modeling of standardized experimental products, a quantitative predictive relationship between hydrodynamic parameters and rock properties is established. The specific implementation steps are as follows: (1) Multi-parameter test and analysis of sediment samples: A series of tests are conducted on the collected sediment samples, including: geochemical analysis. According to national standards, the total organic carbon content is determined after pretreatment of the samples using an elemental analyzer, with a measurement accuracy better than 0.1%; the particle size distribution of the samples is measured using a laser diffraction particle size analyzer to obtain the median particle size parameter; the microstructure and bedding characteristics of the samples are observed using a scanning electron microscope; (2) Quantitative relationship model building: A mathematical correlation is established between the experimentally measured hydrodynamic parameters and rock property parameters: , As the independent variable, As the dependent variable, at least one mathematical method from response surface methodology, multiple regression analysis, binning statistics, or machine learning algorithms is used to establish the " - - "Quantitative relationship model; taking response surface methodology as an example, it is applied to the nonlinear influence of multiple independent variables on a single dependent variable. The specific implementation process of constructing and predicting the analytical method is as follows: the input data are the two hydrodynamic parameters, Froude number and Reynolds number, and the dependent variable is the total organic carbon content (..." This forms n sets of one-to-one data pairs. , , ),( , , ),...,( , , This constitutes the original dataset for modeling. To avoid the influence of differences in units and numerical ranges on the model, the independent variables... , and dependent variable Each distribution is standardized to a mean of 0 and a standard deviation of 1. The transformation formula is as follows: ,in represent , or , The mean, is the standard deviation. For the fitted response surface model, is the standard deviation for each sample. An eigenvector containing constant, linear, interaction, and quadratic terms is constructed, and finally an N×6 dimensional design matrix is ​​built, with each row corresponding to a sample, i.e., a quantitative relationship model of Frode number-Reynolds number-total organic carbon content.

[0039] Step S14: Based on the Frod number, Reynolds number, and total organic carbon content, a clustering algorithm is used to determine the optimal hydrodynamic window for organic matter enrichment.

[0040] As one possible implementation, step S14 includes steps S141 to S148:

[0041] Step S141: Construct a three-dimensional feature vector based on the Frod number, the Reynolds number, and the total organic carbon content.

[0042] Step S142: Normalize the three-dimensional feature vector to obtain a standardized three-dimensional feature vector.

[0043] Step S143: Determine the optimal cluster by combining the profile coefficient analysis method with the elbow rule.

[0044] Step S144: Based on the optimal cluster, the K-means algorithm is used to cluster the standardized three-dimensional feature vectors to obtain multiple clusters.

[0045] Step S145: Calculate the total organic carbon content of each cluster.

[0046] Step S146: Select clusters with total organic carbon content greater than the average threshold and greater than the fluctuation coefficient threshold to obtain clusters with high total organic carbon content.

[0047] Step S147: Based on the high total organic carbon content cluster, use t-distribution statistics to select the Frod number window within the first preset confidence interval and the Reynolds number window within the second preset confidence interval.

[0048] Step S148: Determine the Frod number in the first preset confidence interval and the Reynolds number in the second preset confidence interval as the optimal hydrodynamic window for organic matter enrichment.

[0049] Specifically, based on a quantitative relationship model, a data-driven intelligent analysis method is adopted. By using at least one of binning statistical optimization, clustering algorithm, multivariate regression extreme value analysis and machine learning prediction method, the numerical range of Frod number and Reynolds number with the highest organic matter enrichment efficiency is accurately defined, namely the "optimal organic matter enrichment hydrodynamic window".

[0050] Taking clustering algorithms as an example, the specific implementation steps are as follows: ① Data standardization and feature space construction: Each experimental sample is represented as a three-dimensional feature vector. ① The Z-score standardization method is used to normalize the data of each dimension to eliminate differences in dimensions and ensure that each feature has equal weight in the cluster analysis; ② Determination and execution of optimal clusters: The optimal clusters are determined by combining the silhouette coefficient analysis method with the elbow rule. First, the silhouette coefficient from 2 to a preset upper limit (e.g., 10) is calculated to find the K value with the best inter-cluster separation and the strongest intra-cluster cohesion. At the same time, the sum of squared distances (TSS) from all samples to the center of their respective clusters under different K values ​​is calculated, and the K-TSS curve is plotted to determine the inflection point of the curve, which represents the denser the sample clusters. After determining the optimal number of clusters based on double verification, the K-means algorithm is used to calculate the Euclidean distance from the standardized 3D dataset (feature vectors) to the K cluster centers and assign them to the nearest cluster centers. The mean of all points in each cluster is recalculated as the new cluster center, and the above steps are repeated until the change to the cluster center is less than a preset threshold or the maximum number of iterations is reached, and the final optimal cluster is output. ③ Identification and statistical verification of high total organic carbon content clusters: After clustering, the total organic carbon content of each cluster is calculated. A "high total organic carbon content cluster" is defined to meet the following conditions simultaneously: the average total organic carbon content within the cluster is higher than 1.5 standard deviations of the average total organic carbon content of all samples; the coefficient of variation of the total organic carbon content within the cluster is less than 0.3. ④ Precise quantification and boundary determination of the hydrodynamic window: For the identified high total organic carbon content clusters, the optimal organic matter enrichment hydrodynamic window is determined in the following way: a. Parameter distribution statistics: The minimum, maximum, average, and standard deviation of the cluster in the Floyd number and Reynolds number dimensions are calculated respectively. b. Window Boundary Determination: The optimal window boundary is preferentially determined by the 95% confidence interval of the cluster in both the Floyd number and Reynolds number dimensions, while ensuring that the window largely lies within the minimum and maximum ranges. c. Window Definition: The two-dimensional parameter space region jointly formed by the Floyd number range and the Reynolds number range determined by the above methods is defined as the optimal hydrodynamic window for organic matter enrichment. This window characterizes the paleocurrent conditions most favorable for organic matter enrichment and preservation in both sedimentological and hydrodynamic terms.

[0051] The calculation steps for the 95% confidence interval are as follows: (1) Calculate the sample mean (x̄) and sample standard deviation (s) of the high total organic carbon content cluster in the Frod number and Reynolds number dimensions, respectively; (2) Determine the degrees of freedom df=n-1 based on the sample size n, and obtain the two-sided critical value t(0.025, df) at a confidence level of 95% based on the t-distribution; (3) Calculate the standard errors of Fr and Re, respectively. (4) Calculate the 95% confidence interval boundaries for Fr and Re respectively: the lower limit of the 95% confidence interval = x̄ - t(0.025, The upper limit of the 95% confidence interval = x̄ + t(0.025, .

[0052] Step S2: Obtain core data and logging curves of the target marine shale; core data includes: grain size parameters and geochemical indices.

[0053] Step S3: Perform repositioning and standardization processing on the core data and logging curves respectively to obtain the repositioned core data and standardized logging curves.

[0054] As an implementable method, step S3 includes: using logging curves to reposition the core data to obtain repositioned core data; and standardizing the logging curves to obtain standardized logging curves.

[0055] Specifically, the core data and logging curves of the target marine shale are multi-source data, so preprocessing is required: a systematic multi-source data integration and standardization process is constructed to achieve unified processing of geological data of the target strata. (1) Core data repositioning: based on the correspondence between core gamma scan data and logging natural gamma curves, depth correction and repositioning are performed to eliminate the systematic deviation between core depth and logging depth, ensure the accurate correspondence between core sample points and logging measurement points, and control the core data and logging depth matching error within ±0.1m; (2) using sedimentary structural marker layers such as flooding surfaces, establish the correspondence between key control points of core and logging, and achieve depth alignment under geological constraints; (3) logging data standardization: logging data acquisition is a continuous or discrete measurement process with depth as the independent variable, forming a correspondence between depth and logging value, with each 0.125m as a sampling point. Acquisition includes natural gamma ( GR ), deep lateral resistivity ( RT ), sound wave time difference ( AC ),density( DEN ) and neutron ( CNL The logging curves include those from the data collected. The acquired logging data underwent environmental correction, depth matching, and standardization processing sequentially. The standardization process used the following formula: [Formula omitted for brevity]. GR Taking the curve as an example, ,in, This is the average value of the corrected natural gamma curve. The average natural gamma ray value for the reference well section; The standard deviation is used to eliminate the influence of instrument differences and wellbore environment.

[0056] Step S4: Based on the repositioned core data and standardized logging curves, a nonlinear mapping model is used to obtain the grain size parameter prediction curve and the geochemical index prediction curve; the nonlinear mapping model is obtained by training a random forest model.

[0057] As an implementable approach, nonlinear mapping models include granular parameter linear models and geochemical index mapping models. Step S4 includes steps S41 to S45:

[0058] Step S41: Extract features from the grain size parameters in the repositioned core data and the standardized logging curves to obtain grain size parameter features.

[0059] Step S42: Input the granularity parameter features into the granularity parameter linear model to obtain the granularity parameter prediction curve.

[0060] Step S43: Extract features from the geochemical indices in the returned core data and the standardized logging curves to obtain geochemical index features.

[0061] Step S44: Input the geochemical index features into the geochemical index mapping model to obtain the geochemical index prediction curve.

[0062] Specifically, a nonlinear mapping model is constructed to calibrate high-precision geological information at the core scale to the well logging scale through feature mapping relationships. This involves combining the repositioned core data (such as grain size parameters, geochemical element indices, etc.) with standardized well logging curves. GR , RT , AC , DEN , CNL Establish relevant relationships, extend the repositioned core data to continuous well sections, and generate continuous grain size parameter prediction curves and geochemical index prediction curves with depths consistent with standardized logging curves.

[0063] Nonlinear mapping model: First, under the constraint of a depth matching error of less than ±0.1m, a model is established by comparing core measured data (grain size parameters, geochemical indices) with well logging response (…). GR , RT , AC , DEN The nonlinear mapping model (etc.) is used, and machine learning methods such as random forest algorithm (but not limited to this method) are employed to obtain the granularity parameter prediction curve and the geochemical index prediction curve, ensuring that the coefficient of determination R between the predicted and measured values ​​is... 2 Greater than 0.8.

[0064] Taking the random forest model as an example, the specific implementation process of construction and prediction is as follows: First, on core data and well logging curve data pairs with precise depth matching (error < ±0.1 meters), feature engineering is performed to form a standardized well logging curve (e.g., GR , RT , AC , DEN Features and derived parameters A training set labeled y with measured core grain size parameters or geochemical indices. During model training, the algorithm constructs a large number of (e.g., ...) sets in parallel. T =500 trees) differentiated decision trees: each tree is generated by sampling a randomly selected subset of samples (Bootstrap sampling) and a randomly selected subset of features (e.g. Among the features, the optimal splitting feature and threshold (e.g., maximizing the mean square error) are recursively searched. MSE The reduction Δ MSE The tree grows until a stopping condition is met (e.g., the number of node samples is less than 5). The predicted value of a single tree for sample i is... Ultimately, the model's ensemble predictions are achieved through a simple averaging of all decision tree predictions: granularity parameters. After training, the model can output continuous grain size parameters and geochemical index curves for the entire well section. It can also calculate the total reduction in impurity resulting from splitting nodes of each feature across all trees, yielding results such as... DEN >Δ logR > GR The ranking of feature importance provides insights into geological origins. Among them, M Given the total number of eigenvalues, the mean squared error (MSE) is used as a measure of impurity. Specifically, it is assumed that the MSE of the nodes before the split is... MSE_before After splitting, the left and right child nodes MSE They are MSE_lef t and MSE_right And the weights of the number of child node samples relative to the total number are respectively w_left and w_right The reduction in impurity resulting from this split is Δ. Impurity = MSE_before -( w_left MSE_left + w_right MSE_right The model will traverse all trees and all split points that have used the feature, reducing this impurity by Δ. Impurity By summing these scores, we obtain the total importance score for that feature. Finally, by normalizing or ranking the importance scores of all features, we obtain a similar result. DEN> Δ logR >GR In simple terms, the more frequently a feature is used for splitting, and the greater the improvement in prediction accuracy brought by each split, the higher its importance score. This transforms the "black box" output of machine learning into geological insights with clear physical meaning.

[0065] Step S5: Calculate the Goufredo number curve and Goureynolds number curve based on the well logging curve, grain size parameter prediction curve, and geochemical index prediction curve.

[0066] As one possible implementation, step S5 includes steps S51 to S56:

[0067] Step S51: Calculate paleotemperature and paleosalinity based on the well logging curves.

[0068] Specifically, key elements are obtained by combining natural gamma spectroscopy logging and element capture logging. For example, Sr and Ba elements are used to invert paleotemperature of sedimentary water bodies, and Mg and Ca elements are used to invert paleosalinity of sedimentary water bodies.

[0069] Step S52: Calculate paleowater depth based on the predicted curve of the geochemical index.

[0070] Specifically, paleowater depth is determined by predicting curves using geochemical indicators. For example, based on the above, a thorium-uranium ratio (TURBO) is established. Nonlinear mapping model between logging curves The quantitative relationship of paleowater depth was obtained by inversion using an empirical model calibrated in a specific region and constrained by sedimentary facies. .

[0071] Step S53: Calculate the paleovelocity based on the predicted curve of the particle size parameter.

[0072] Specifically, by utilizing the largest suspended particle size in the sediment ( The critical velocity can be inverted using the final settling velocity formula, and the particle size can also be obtained as a continuous value through nonlinear mapping of the particle size parameter prediction curve mentioned above. .

[0073] Step S54: Calculate the paleotemperature and paleosalinity based on the predicted particle size parameters.

[0074] Step S55: Calculate the paleofrode number curve based on the paleowater depth and the paleocurrent velocity.

[0075] As an feasible approach, the formula for calculating the Gufrod number curve is:

[0076] .

[0077] .

[0078] .

[0079] .

[0080] .

[0081] in, The curve represents the ancient Freud number. For ancient flow velocity, It is the acceleration due to gravity. Because the ancient water was deep, Particle density, For the density of water, The coarsest suspended particle size, This is the drag coefficient. The particle size of the coarsest suspended particle in the particle size parameter prediction curve. Because the ancient water was deep, and All are specific coefficients. The thorium-uranium ratio in the geochemical index prediction curve. As an environmental correction item, it can be dynamically adjusted based on redox indicators such as Mo in the same layer to correct for the impact of organic matter enrichment. The abnormal adsorption effect This is a prediction curve for geochemical indicators.

[0082] Step S56: Calculate the paleo-Reynolds number curve based on the paleo-temperature, paleo-salinity, paleo-water depth, paleo-flow velocity, and paleo-water kinematic viscosity coefficient.

[0083] As an feasible approach, the formula for calculating the Gurey number curve is:

[0084] .

[0085] .

[0086] in, The curve represents the Gurey number. The viscosity coefficient of ancient water movement. The coefficient of kinematic viscosity under standard conditions. This is a temperature correction factor. For ancient temperature, Ancient temperatures under standard conditions This is the salinity correction factor. For ancient salinity, Paleosalinity under standard conditions.

[0087] Step S6: Use the optimal organic matter enrichment hydrodynamic window to screen the ancient Frod number curve and the ancient Reynolds number curve to obtain the sweet spot layer of the target marine shale and complete the sweet spot layer prediction.

[0088] Specifically, the prediction of the sweet spot layer, or the prediction of the organic matter enrichment sweet spot layer (segment): This involves comparing the inverted paleohydrodynamic parameters (paleo-Frod number curves and paleo-Reynolds number curves) with a pre-established quantitative relationship model. Coupled analysis was performed, enabling a leap from single-point prediction to continuous prediction across the entire well section. For single-point data, the total organic carbon content (i.e., the expected rock property parameters) was directly calculated using the quantitative relational model. In the vertical profile prediction, a sliding window analysis method was used to continuously calculate the entire well section, with an optimal window scale of 0.5–1 m. The Gugufrod number and Gugureynolds number curves were compared with the optimal organic matter enrichment hydrodynamic window determined by the quantitative relational model to identify depth segments that simultaneously meet the window conditions, which were marked as vertical sweet spot segments. Therefore, the method of this application can not only predict sweet spot layers but also calculate the total organic carbon content of sweet spot layers through the quantitative relational model, realizing both quantitative analysis and prediction of organic matter enrichment data.

[0089] Among them, the sliding window analysis method is a computational technique for processing sequential data. It slides a fixed-length window along the data sequence step by step, performs local analysis or model calculation on the data within the window at each stopping position, and assigns the result to the center point of the window. This method can effectively smooth data noise, maintain the continuity and trend of the sequence, and in geological applications, it can transform discrete hydrodynamic parameters into continuous and geologically reasonable organic matter content prediction curves.

[0090] The beneficial effects of the method for predicting sweet spots in marine shale proposed in this application are mainly reflected in the following aspects:

[0091] 1. Through experiments, a causal framework was constructed for Frod number, Reynolds number, and total organic carbon content, providing a comparable and unified benchmark for shale formations in different basins and strata within marine facies. By establishing the causal relationship between Frod number, Reynolds number, and total organic carbon content, the optimal hydrodynamic window for organic matter enrichment was determined. This improved the prediction accuracy of sweet spot layers from a vague qualitative assessment of "deep water areas rich in organic matter" to a quantitative prediction with a specific coupling relationship between Frod number, Reynolds number, and total organic carbon content.

[0092] 2. In practical applications, a nonlinear mapping model is used to generate prediction curves from data related to grain size parameters and geochemical indicators. This enables the calculation of the ancient Frod number curve and the ancient Reynolds number curve. Then, the ancient Frod number curve and the ancient Reynolds number curve are screened through the optimal organic matter enrichment hydrodynamic window to obtain the sweet spot layer of the target marine shale. This achieves the determination of the optimal organic matter enrichment layer through the ancient Frod number curve and the ancient Reynolds number curve, thereby improving the accuracy of sweet spot layer prediction.

[0093] 3. Clear Mechanism: Starting from the origin of sedimentary dynamics, the key physical processes controlling organic matter enrichment are revealed, making predictions reliable. It overcomes the shortcomings of existing technologies, such as insufficient explanation of the genesis of shale organic matter heterogeneity and strong ambiguity in sweet spot predictions, providing a new theoretical basis and key technical means for shale exploration.

[0094] 4. Quantification and standardization: The introduction of two standardized dimensionless parameters, Frode number and Reynolds number, provides a comparable and unified benchmark for the evaluation of shale in different basins and strata.

[0095] 5. High prediction accuracy: The prediction of sweet spots is improved from the fuzzy qualitative prediction of "deep water areas rich in organic matter" to the quantitative prediction of "phase bands with specific Fr-Re coupling relationship", which significantly reduces the ambiguity.

[0096] 6. High applicability: This method can be directly integrated with conventional geological and logging workflows, is highly operable, and is easy to promote and apply in exploration and production, effectively reducing exploration risks and guiding well location deployment design.

[0097] 7. To address the shortcomings of existing technologies that rely on static cognition (emphasizing current geochemical indicators while neglecting dynamic sedimentary processes).

[0098] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0099] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0100] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0101] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of predicting sweet spots in marine shale, characterized by, The method for predicting sweet spots in marine shale includes: Based on sample experimental data, a quantitative relationship model of Frod number-Reynolds number-total organic carbon content was constructed, and the optimal hydrodynamic window for organic matter enrichment was determined; the optimal hydrodynamic window for organic matter enrichment includes: the optimal Frod number window and the optimal Reynolds number window; Obtain core data and well logging curves of the target marine shale; the core data includes: grain size parameters and geochemical indices; The core data and the logging curves are respectively repositioned and standardized to obtain repositioned core data and standardized logging curves; Based on the repositioned core data and the standardized logging curves, a nonlinear mapping model is used to obtain the grain size parameter prediction curve and the geochemical index prediction curve; the nonlinear mapping model is obtained by training a random forest model. Based on the well logging curve, the grain size parameter prediction curve, and the geochemical index prediction curve, calculate the Gufrod number curve and the Gureynolds number curve; The optimal organic matter enrichment hydrodynamic window is used to screen the ancient Frode number curve and the ancient Reynolds number curve to obtain the sweet spot layer of the target marine shale, thus completing the sweet spot layer prediction.

2. The method for predicting sweet spots in marine shale according to claim 1, characterized in that, Based on sample experimental data, a quantitative relationship model of Frode number, Reynolds number, and total organic carbon content was constructed, and the optimal hydrodynamic window for organic matter enrichment was determined, specifically including: Acquire sample experimental data; the sample experimental data includes: experimental parameters and total organic carbon content of experimental sediments; the experimental parameters include: flow rate, water depth, and temperature; Based on the experimental parameters, calculate the Frode number and Reynolds number; Based on the Frod number, the Reynolds number, and the total organic carbon content, a quantitative relationship model of Frod number-Reynolds number-total organic carbon content is constructed using response surface methodology. Based on the Frod number, the Reynolds number, and the total organic carbon content, a clustering algorithm is used to determine the optimal hydrodynamic window for organic matter enrichment.

3. The method for predicting sweet spots in marine shale according to claim 2, characterized in that, Based on the Frod number, the Reynolds number, and the total organic carbon content, a quantitative relationship model of Frod number-Reynolds number-total organic carbon content is constructed using response surface methodology, specifically including: Using the Frod number and the Reynolds number as independent variables and the total organic carbon content as the dependent variable, the response surface model is fitted to obtain the model coefficients of the response surface model; Based on the model coefficients of the response surface model, a quantitative relationship model of Frode number, Reynolds number, and total organic carbon content is constructed.

4. The method for predicting sweet spots in marine shale according to claim 3, characterized in that, The expression for the quantitative relationship model is: ； in, This refers to the total organic carbon content. , , , , and All are model coefficients. For Freud number, Let Reynolds number be 1. This is the random error term.

5. The method for predicting sweet spots in marine shale according to claim 2, characterized in that, Based on the Frod number, the Reynolds number, and the total organic carbon content, a clustering algorithm is used to determine the optimal hydrodynamic window for organic matter enrichment, specifically including: A three-dimensional feature vector is constructed based on the Frod number, the Reynolds number, and the total organic carbon content. The three-dimensional feature vectors are normalized to obtain standardized three-dimensional feature vectors. The optimal cluster was determined by combining the silhouette coefficient analysis method with the elbow rule. Based on the optimal clustering cluster, the K-means algorithm is used to cluster the standardized three-dimensional feature vectors to obtain multiple clusters; Calculate the total organic carbon content of each cluster; Clusters with total organic carbon content greater than the average threshold and greater than the fluctuation coefficient threshold are selected to obtain high total organic carbon content clusters; Based on the clusters with high total organic carbon content, the Frod number window within the first preset confidence interval and the Reynolds number window within the second preset confidence interval are selected using t-distribution statistics. The Frod number within the first preset confidence interval and the Reynolds number within the second preset confidence interval are determined as the optimal hydrodynamic window for organic matter enrichment.

6. The method for predicting sweet spots in marine shale according to claim 1, characterized in that, The core data and logging curves are respectively subjected to repositioning and standardization processes to obtain repositioned core data and standardized logging curves, specifically including: The core data was repositioned using well logging curves to obtain the repositioned core data. The logging curves are standardized to obtain standardized logging curves.

7. The method for predicting sweet spots in marine shale according to claim 1, characterized in that, The nonlinear mapping model includes a granular parameter linear model and a geochemical index mapping model. Based on the repositioned core data and the standardized logging curves, a nonlinear mapping model is used to obtain grain size parameter prediction curves and geochemical index prediction curves, specifically including: The grain size parameters in the repositioned core data and the standardized logging curves are extracted to obtain grain size parameter features; The granularity parameter features are input into the granularity parameter linear model to obtain the granularity parameter prediction curve; Geochemical features are extracted from the repositioned core data and the standardized logging curves to obtain geochemical feature characteristics. The geochemical index features are input into the geochemical index mapping model to obtain the geochemical index prediction curve.

8. The method for predicting sweet spots in marine shale according to claim 1, characterized in that, Based on the well logging curves, the grain size parameter prediction curves, and the geochemical index prediction curves, the Gulfrod number curves and the Gurey number curves are calculated, specifically including: Based on the well logging curves, paleotemperature and paleosalinity were calculated; Based on the predicted curves of the aforementioned geochemical indicators, paleowater depth was calculated. Based on the predicted curve of the particle size parameter, the paleovelocity is calculated; Based on the paleotemperature and paleosalinity, the paleowater motion viscosity coefficient was calculated; Based on the paleowater depth and paleocurrent velocity, calculate the paleofrod number curve; Based on the paleotemperature, paleosalinity, paleowater depth, paleoflow velocity, and paleowater kinetic viscosity, the paleoReynolds number curve is calculated.

9. The method for predicting sweet spots in marine shale according to claim 8, characterized in that, The formula for calculating the Goufrod number curve is as follows: ； ； ； ； ； in, The curve represents the ancient Freud number. For ancient flow velocity, It is the acceleration due to gravity. Because the ancient water was deep, Particle density, For the density of water, The coarsest suspended particle size, This is the drag coefficient. The particle size of the coarsest suspended particle in the particle size parameter prediction curve. Because of the water depth, and All are specific coefficients. The thorium-uranium ratio in the geochemical index prediction curve. For environmental correction items, The curve represents the geochemical index prediction curve; GR represents natural gamma; RT represents deep lateral resistivity; AC represents acoustic transit time; and DEN represents density.

10. The method for predicting sweet spots in marine shale according to claim 9, characterized in that, The formula for calculating the Gurey number curve is as follows: ； ； in, The curve represents the Gurey number. The viscosity coefficient of ancient water movement. The coefficient of kinematic viscosity under standard conditions. This is a temperature correction factor. For ancient temperature, Ancient temperatures under standard conditions This is the salinity correction factor. For ancient salinity, Paleosalinity under standard conditions.

Images