A multi-parameter modeling and predicting method for shale oil mobility

Abstract

The present application belongs to the technical field of shale oil exploration, and relates to a multi-parameter modeling and predicting method for shale oil mobility. A plurality of rock facies are divided, and a representative sample of each type of rock facies is subjected to a gas drive-nuclear magnetic resonance combined experiment one by one to obtain the movable oil saturation of each representative sample; multi-dimensional parameters of the representative sample are obtained; factor analysis is performed on the multi-dimensional parameters to extract three mutually independent first main control factors E 1 , second main control factors E 2 and third main control factors; factor score coefficients determined by the factor analysis are used to calculate the score of each representative sample on the corresponding main control factor E 1k , E 2k and E 1k , E 2k and E 1 , E 2 and a multiple linear regression model between the movable oil saturation S and the three main control factors is established. Thus, the problems of simulation distortion, difficulty in multi-factor coupling influence analysis and high evaluation cost of the traditional mobility evaluation method are solved.

Description

Technical Field

[0001] This invention belongs to the field of shale oil exploration and development technology, and relates to a multi-parameter modeling and prediction method for shale oil mobility. Background Technology

[0002] In shale oil exploration and development, accurately assessing the mobility of crude oil is crucial for determining the economic value of reservoir extraction and optimizing development strategies. Shale reservoirs typically possess low porosity, low permeability, strong heterogeneity, and complex pore-fracture systems, resulting in crude oil existing in multiple forms, including adsorbed and free states, with complex flow mechanisms. Therefore, scientifically evaluating the proportion of mobile oil has become one of the core challenges in the efficient development of shale oil.

[0003] Currently, common methods for assessing mobility largely rely on experimental techniques such as core permeation and centrifugation. However, existing methods have significant limitations in terms of simulation realism, analytical systematicity, and evaluation efficiency, mainly in the following three aspects:

[0004] I. Existing methods are mismatched with the driving mechanism and cannot accurately reflect the physical process of mining. Existing technologies employ displacement mechanisms primarily driven by capillary forces. However, in actual shale oil reservoir development, fluid mobility mainly relies on formation energy (pressure) or artificial water / gas injection pressure. The displacement mechanism of these existing technologies differs fundamentally from the physical mechanisms of real-world extraction processes, resulting in experimental results that cannot accurately simulate pressure-driven fluid mobility characteristics under formation conditions. Consequently, the obtained shale oil mobility assessment results lack geological representativeness and have limited predictive reliability.

[0005] Second, existing methods lack the ability to perform multi-factor coupling analysis and cannot reveal the main control mechanism of mobility. Existing technologies primarily focus on the dynamic characterization of fluid distribution and migration during the seepage process, failing to systematically integrate the multi-dimensional key parameters affecting shale oil mobility, including macroscopic rock properties, microscopic pore structure parameters, and oil-rock interaction parameters. Due to the lack of systematic integration and coupling analysis of these multi-dimensional parameters, existing methods struggle to quantitatively reveal the interactions between these factors and their synergistic control mechanisms on the proportion of mobile oil. Consequently, they cannot provide key input variables, after dimensionality reduction and feature extraction, for establishing a clearly mechanistic and highly interpretable mobility prediction model.

[0006] Third, existing methods are inefficient and costly, making it difficult to achieve large-scale and universally applicable predictions. Current technologies are essentially still limited to refined, process-descriptive experiments on single core samples. To assess the mobility of a large number of samples from different lithofacies and with high heterogeneity in a reservoir, time-consuming and costly joint experiments must be repeated for each sample. This operational model results in long evaluation cycles and high economic costs, making it difficult to meet the needs of rapid evaluation of large-scale samples in exploration and development. More importantly, existing methods cannot extract universally applicable and generalizable predictive models from experimental data of limited samples, thus lacking the ability to rapidly, batch-wise, and cost-effectively predict the mobility of unexperimented samples.

[0007] Therefore, existing technologies still cannot simultaneously meet the urgent needs of shale oil mobility assessment for "simulation realism", "multi-factor coupling analysis" and "efficient, low-cost and universally applicable prediction", that is, they cannot achieve an accurate, efficient, economical and universally applicable assessment of shale oil mobility. Summary of the Invention

[0008] The purpose of this invention is to provide a multi-parameter modeling and prediction method for shale oil mobility, so as to solve the problems of simulation distortion, difficulty in analyzing the effects of multiple coupled factors, high evaluation cost and poor universality of traditional mobility evaluation methods.

[0009] To achieve the above objectives, the present invention provides a multi-parameter modeling and prediction method for shale oil mobility, the specific technical solution of which is as follows: Specifically, it includes: S1. Obtain a set of rock samples, and identify and classify various different rock facies by observing the cores of the set of rock samples; S2. For various rock facies, representative samples are selected from each facies, and gas-driven nuclear magnetic resonance combined experiments are performed on each representative sample to obtain the mobile oil saturation of each representative sample. Where k is the number of the representative sample; the gas-driven nuclear magnetic resonance combined experiment includes: Under simulated formation temperature and confining pressure conditions, the total oil content of representative samples saturated with simulated formation oil was determined using nuclear magnetic resonance (NMR) technology. C O1 ; Gas was injected into a representative sample that had been simulated to be saturated with formation oil for displacement. After displacement, the residual oil content was determined again using nuclear magnetic resonance (NMR) technology. C O2 ; in accordance with and C O1 and C O2 The correspondence is used to determine the movable oil saturation of the representative sample; S3. Collect the multi-dimensional parameters of all representative samples to form a multi-dimensional parameter dataset, and perform factor analysis on the multi-dimensional parameter dataset to extract three independent first principal control factors. E 1 Second controlling factor E 2 and the third controlling factor Using the factor score coefficients determined by the factor analysis, the score of each representative sample on the first principal control factor is calculated. E 1k Score on the second major control factor E 2k And the score on the third controlling factor ; S4. Determine the movable oil saturation of each representative sample. , and the corresponding E 1k , E 2k as well as The samples are paired to form multiple sets of paired sample data; based on the multiple sets of paired sample data, the movable oil saturation is established. and E 1 , E 2 and Multiple linear regression model between them; S5. Using the aforementioned multiple linear regression model, predict the mobility of shale oil.

[0010] Preferably, the simulated formation temperature is 80°C, the confining pressure is 50 MPa, the injected gas is CO2, and the injection process is carried out under a constant pressure of 10 MPa.

[0011] Preferably, among them, and C O1 and C O2 The specific correspondence is as follows: in, C O2 The displacement was maintained for 3 days during constant-pressure gas injection, with the mass of the representative sample measured every 6 hours. Displacement equilibrium was determined when the mass of the representative sample remained constant for three consecutive measurements. Subsequently, nuclear magnetic resonance (NMR) experiments were performed on the representative sample to obtain the residual oil content. C O2 .

[0012] Preferably, the multidimensional parameters include rock physical properties, pore structure parameters, and oil-rock interaction parameters; the rock physical properties include porosity. φ and penetration rate K The pore structure parameters include the average pore radius. r and exhaust pressure P c The oil-rock interaction parameters include the content of brittle minerals. M P clay mineral content M C and specific surface area A。

[0013] Preferably, the factor score coefficients make E 1 Characterizing rock physical properties, E 2 Characterizing pore structure, E 3 Characterize oil-rock interaction.

[0014] Preferably, the multiple linear regression model is = * E 1 + * E 2 + * E 3 +ε in, It is the independent variable E 1 The partial regression coefficient, Independent variable E 2 The partial regression coefficient, It is the independent variable E 3 The partial regression coefficients, where ε is the random error term.

[0015] Preferably, the prediction includes: using the multiple linear regression model, for newly acquired rock samples that have not undergone the gas-driven-nuclear magnetic resonance combined experiment described in S2, calculating the corresponding master control factor value based on their multidimensional parameter data and the factor score coefficient, and then calculating the mobile oil saturation through the multiple linear regression model to achieve rapid prediction of the mobility of shale oil in the new sample.

[0016] The shale oil mobility modeling and prediction method of the present invention has the following advantages: (1) The accuracy of evaluation is fundamentally improved: By adopting the "gas-driven-nuclear magnetic resonance combined experiment" that simulates the formation conditions, the present invention is closer to the actual mining process in terms of driving mechanism and temperature and pressure environment, fundamentally solving the problem of simulation distortion in traditional methods, and providing a highly reliable data benchmark for mobility prediction.

[0017] (2) Scientific Quantification Through Causal Analysis: By introducing "factor analysis", this invention successfully reduces and integrates multiple complex and coupled geological and engineering parameters affecting mobility into a few key factors with clear physical meaning. This not only clearly reveals the three key dimensions of "physical properties, pore structure, and interaction", but also lays a solid foundation for establishing a mathematical model with a clear mechanism.

[0018] (3) Evaluation efficiency achieves a paradigm shift: This invention pioneers an application paradigm of "modeling-prediction based on representative samples". This paradigm transforms the evaluation work from the "exhaustive mode" that requires expensive and time-consuming core experiments on every sample to the dynamic oil saturation S obtained from experiments on only a small number of representative samples. k The three main control factors derived from the basic parameters that can be obtained in large quantities through conventional, low-cost experiments ( E 1 , E 2 and E 3 Establish a multiple linear regression model. Once the multiple linear regression model is established, for new samples from the same region and the same strata, only routine experiments are needed to calculate... E 1 , E 2 and E 3 This allows for the rapid and low-cost prediction of the movable oil saturation S using a multiple linear regression model, eliminating the need for time-consuming and costly gas-driven nuclear magnetic resonance experiments on each new sample. This enables rapid and batch-based mobility prediction, significantly reducing evaluation costs and timelines, and facilitating efficient and economical large-scale applications.

[0019] Therefore, by providing a reliable, efficient, and economical prediction method, this invention can help oilfield enterprises find oil more accurately, extract oil more efficiently, and calculate costs more scientifically, thereby comprehensively improving the overall development benefits of shale oil reservoirs and having significant practical value for promoting the economical and effective development of shale oil, an important resource. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating a multi-parameter modeling and prediction method for shale oil mobility according to the present invention. Figure 2This is a correlation diagram between the predicted movable oil saturation and the measured movable oil saturation of the present invention. Detailed Implementation

[0021] The technical solutions of this application will now be described clearly and in detail with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " indicates "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more. The terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature.

[0022] like Figure 1 As shown, this invention provides a flowchart of a multi-parameter modeling and prediction method for shale oil mobility. This multi-parameter modeling and prediction method for shale oil mobility specifically includes: S1. Obtain a set of rock samples, and identify and classify various different rock facies by observing the core of the rock samples.

[0023] Among them, representative samples refer to samples that can fully represent the overall average condition or the most common state of a specific "rock facies" (such as "massive medium sandstone" or "laminated shale") in terms of macroscopic petrological characteristics and microscopic physical properties (such as porosity, permeability, mineral composition, etc.).

[0024] Its core meaning is as follows, with the aim of ensuring the validity of the experiment and the universality of the model: Representativeness and avoidance of bias: When sampling from a particular rock facies, samples that are extremely unusual or deviate from the overall characteristics of that facies should not be selected (for example, a medium sandstone sample with large fractures may not represent the average pore structure of "massive medium sandstone"). The selected samples should represent the mainstream characteristics of that rock facies.

[0025] Ensuring model reliability: The ultimate goal of this invention is to establish a model that can predict the mobility of a large number of samples based on a small number of samples. If the samples used to build the model (i.e., to conduct core gas-driven NMR experiments) are "typical," then the predictive model built based on their data is more likely to accurately predict the mobile oil saturation of other "typical" or even most samples in the same lithofacies. Conversely, if atypical (outlier) samples are used for modeling, the predictive ability of the model will be greatly reduced.

[0026] Operational judgment criteria: In actual operation, the judgment of "representativeness" is usually based on the first step of core observation and description (such as macroscopic characteristics such as color, structure, texture, and grain size that are consistent with similar rock facies), combined with preliminary routine physical property screening (such as whether the measured porosity and permeability are within the common range of this type of rock facies) to make a comprehensive determination.

[0027] In summary, within the framework of this invention, "representative samples" are those determined by geological and engineering experts to most realistically and universally reflect the inherent properties of a particular type of rock facies. They are selected to ensure the maximum representativeness and accuracy of extrapolated predictions from subsequent costly joint experimental data and the mathematical models built upon them. While not explicitly defined in the document, this is a crucial prerequisite for achieving its goal of "predicting the whole from a small number of samples."

[0028] In one implementation case, the identified lithofacies included: massive medium-grained sandstone, massive fine-grained sandstone, massive siltstone, massive argillaceous fine-grained sandstone, massive argillaceous siltstone, layered sandy shale, thickly layered sandy shale, and thinly layered shale. This step laid the geological foundation for subsequent differentiated experiments and analyses targeting different lithofacies types.

[0029] S2. For various rock facies, representative samples are selected from each facies. Then, gas-driven nuclear magnetic resonance combined experiments are performed on each of the representative samples to obtain the mobile oil saturation of each representative sample. .

[0030] It should be noted that k is the number of the representative sample, with values ​​of 1, 2, ..., m.

[0031] This step aims to obtain highly reliable "mobile oil saturation" for representative samples of each lithofacies through physical experiments simulating formation conditions. The value is as follows: The specific steps are as follows:

[0032] (1) Preparation of experimental samples: The representative sample selected was a plunger sample with a standard size of 2.5 cm in diameter and 2 cm in length. First, the representative sample was subjected to Soxhlet extraction with chloroform at 70°C for 72 hours to thoroughly wash away the primary hydrocarbons. Subsequently, the representative sample was dried in an 80°C drying oven for 24 hours to remove residual solvent and moisture. After the sample cooled to room temperature, it was sealed with plastic wrap to prevent contamination and moisture absorption.

[0033] (2) Simulated formation oil configuration and rock sample saturation: Degassed crude oil and associated gas produced from the wellhead were recombined in a specific ratio to simulate the fluid properties of the target reservoir under geological conditions of 80℃ underground temperature and 50MPa confining pressure, thus serving as the simulated formation oil. In the experimental setup, prepared dry rock samples were placed under a high pressure of 35MPa and fully saturated with the simulated formation oil, maintaining this pressure for 36 hours. Afterward, the pressure was released, representative samples were removed and weighed. When the weight results remained unchanged after three consecutive weighings, the rock sample was considered to have been fully saturated with crude oil, reaching its original underground oil-bearing state.

[0034] (3) Gas-driven nuclear magnetic resonance combined experimental procedure: Initial oil content measurement: Nuclear magnetic resonance experiments were performed on the representative samples fully saturated with crude oil to non-destructively determine their total saturated oil content. C O1 This value represents the total oil content baseline of the rock sample prior to "mining".

[0035] Simulated gas displacement extraction: A representative sample, fully saturated with crude oil, is placed in an experimental chamber simulating a formation environment. The temperature is controlled at 80°C, and a confining pressure of 50 MPa is applied to simulate the pressure of the overlying strata. Then, Gas was injected into a representative saturated oil sample at a constant pressure of 10 MPa to simulate the process of gas injection to enhance oil recovery, and the displacement process lasted for 3 days. During this period, the mass of the representative sample was measured every 6 hours to monitor the displacement dynamics.

[0036] Residual oil content measurement and experimental endpoint determination: When the mass of the representative sample remains constant for three consecutive measurements, it indicates that under the experimental conditions, the mobile crude oil has been sufficiently displaced, and the system has reached equilibrium. At this point, the sample is immediately subjected to another nuclear magnetic resonance experiment to determine its residual oil content. C O2 This refers to the amount of crude oil that remains in the rock sample after gas displacement.

[0037] (4) Calculation of movable oil saturation: The total oil content measured based on the above experiments C O1 and residual oil content C O2 The movable oil saturation of the representative sample was calculated according to formula (1). .this The value is the "gold standard" data for evaluating the mobility of shale oil, obtained directly through high-fidelity simulation experiments.

[0038] (1) (5) Lithofacies analysis: For each rock facies identified in S1, the joint experiment was carried out one by one on representative samples selected from its category according to the above procedures (1) to (4), so as to obtain a complete benchmark dataset covering all major rock facies types.

[0039] S3. Collect the multi-dimensional parameters of all representative samples to form a multi-dimensional parameter dataset, and perform factor analysis on the multi-dimensional parameter dataset to extract three independent principal control factors. The first principal control factor is... E 1 Second controlling factor E 2 and the third controlling factor Using the factor score coefficients determined by this factor analysis, the score of each representative sample on the first principal control factor is calculated. E 1k Score on the second major control factor E 2k And the score on the third controlling factor .

[0040] Preferably, the factor score coefficient makes E 1 Characterizing rock physical properties, E 2 Characterizing pore structure, E 3 Characterize oil-rock interaction.

[0041] This step aims to improve upon the measurements already obtained in step S2. A systematic analysis of the multidimensional fundamental parameters of all representative samples was conducted to reveal the main controlling factors affecting mobility, and the high-dimensional parameters were reduced to a few comprehensive indicators (main controlling factors) with clear physical meaning, specifically including: S31. Obtain a multi-dimensional parameter dataset.

[0042] For each representative sample in S2 where the experiment has been completed, the following three types of parameters were measured to construct its multi-dimensional feature vector: In a specific, preferred implementation case, to fully cover the above three dimensions, the following seven parameters with clear physical meaning and easy to obtain through conventional experiments were selected.

[0043] Porosity was obtained by measuring porosity and permeability using a helium gas porosity-permeability analyzer. φ Units: %) and penetration rate ( K (Unit: mD)

[0044] The average pore radius was determined by high-pressure mercury intrusion porosimetry. r (unit: nm) and exhaust pressure (P c (Unit: MPa)

[0045] The content of brittle minerals was obtained by X-ray diffraction analysis. M P , Unit: %) and clay mineral content ( M C , unit: %); Specific surface area (A, unit: m² / g) was calculated from nitrogen adsorption-desorption experimental data.

[0046] In one embodiment of the present invention, N representative samples are selected for multi-dimensional parameter testing and principal factor extraction, resulting in factor score formulas as shown in equations (1)-(3) and a regression model as shown in equation (4). Therefore, the measured values ​​of the above 7 parameters of all representative samples are organized to form an N-row × 7-column multi-dimensional parameter dataset, where N is the total number of representative samples. In a specific embodiment, for example, N=8, the dataset is shown in Table 1.

[0047] Table 1. Original basic parameters of representative samples S32, Perform factor analysis.

[0048] Import the aforementioned multidimensional parameter dataset into professional statistical analysis software (such as SPSS, R, or Python's factor_analyzer library) and perform factor analysis. Specific steps include: Data standardization: Standardize the original data for the seven variables to eliminate the influence of dimensions. Suitability testing: Perform the KMO test and Bartlett's test of sphericity to confirm that the data is suitable for factor analysis (in this example, the data passes the tests).

[0049] Principal component extraction: The number of principal control factors to be extracted was scientifically determined and screened based on the criterion of "eigenvalue greater than 1". The "explained total variance" results of the factor analysis output are shown in Table 2 below:

[0050] Table 2 As shown in Table 2 above, the eigenvalues ​​of the first three components are all greater than 1, and their cumulative variance contribution rate reaches 96.451%. This indicates that extracting three principal control factors can explain 96.451% of the information from the original seven parameters, demonstrating a significant dimensionality reduction effect with minimal information loss. Therefore, three principal control factors are selected and designated as the first principal control factor. E 1 Second controlling factor E 2 and the third controlling factor E 3 .

[0051] S33. Explain the controlling factor.

[0052] The physical meaning of the controlling factor is determined by analyzing the rotated factor loading matrix (as shown in Table 3 below). The factor loadings represent the correlation between the original parameters and the controlling factor; the closer the absolute value is to 1, the stronger the relationship.

[0053] Table 3. Rotated factor loading matrix After obtaining the rotated factor loading matrix, each extracted controlling factor is assigned a specific physical meaning. The determination is based on the absolute value of the factor loadings of each original parameter on the corresponding controlling factor. The closer the absolute value of the loading is to 1, the stronger the linear correlation between the original parameter and the controlling factor; that is, the more significant the parameter's "contribution" and "definition" of the factor. By identifying the set of original parameters with significantly high absolute loadings on a specific factor (e.g., absolute values ​​greater than 0.7), the latent physical dimension represented by that controlling factor can be determined.

[0054] Specifically, in this embodiment: First controlling factor E 1 Porosity φ and penetration rate K The absolute values ​​of the loads on them are as high as 0.961 and 0.947, respectively, significantly higher than other parameters. This indicates that... E 1 It strongly and comprehensively reflects the covariant information of rock porosity and permeability, therefore it is the first controlling factor. E 1 Defined as "rock physical property factors"; porosity ( φ ) and penetration rate ( K Defined as rock physical property parameters:

[0055] Second controlling factor E 2 Average pore radius r and exhaust pressure P c The absolute values ​​of the loads on them are 0.893 and 0.973, respectively. The exhaust pressure exhibits a high absolute value load. This clearly indicates... E 2 The main focus was on capturing and characterizing pore structure features (pore size and fluid entry difficulty), therefore... E 2 Defined as "pore structure factor"; the average pore radius r and exhaust pressure P cDefined as pore structure parameters;

[0056] Third controlling factor E 3 : Brittle mineral content M P clay mineral content M C The absolute values ​​of the loads on the specific surface area A are 0.980, 0.980, and 0.762, respectively. This indicates that... E 3 This reflects the mineral composition and surface properties of the rock, therefore... E 3 Defined as "oil-rock interaction factor"; brittle mineral content M P clay mineral content M C The specific surface area A is defined as the oil-rock interaction parameter.

[0057] Thus, through analysis based on the absolute value of the load, three abstract mathematical factors ( E 1 , E 2 and E 3 The analysis is explained as focusing on three main controlling dimensions with clear geological significance: "rock properties," "pore structure," and "oil-rock interaction." This completes the scientific decoupling and quantitative characterization of the multi-factor coupling effect. S34: Calculate the factor score coefficients.

[0058] Factor analysis also outputs a factor score coefficient matrix (as shown in Table 4 below). This coefficient is used to convert the original seven parameter values ​​of any sample into its scores on the three main control factors (i.e., E 1 , E 2 and E 3 (Specific numerical value).

[0059] Table 4 Factor Score Coefficient Matrix Therefore, for any new sample (or a sample that has already been measured), its standardized parameter values ​​[ φ , K , r , P c , M P , M CThe main control factor score of the sample can be obtained by multiplying A and B by the coefficients in the corresponding columns of Table 3 and summing the results. (2) At this point, step S3 is complete, and a comprehensive index for quantifying the sample under the coupled influence of multiple factors is obtained. E 1 , E 2 and E 3 ) and its calculation method.

[0060] It should be noted that the above parameter combinations are preferred embodiments of the present invention, and not the sole limitation on the range of "multi-dimensional parameters". The core of the present invention lies in the fact that the "multi-dimensional parameters" should systematically cover the three core dimensions of "physical properties, structure, and interaction". In practical applications, those skilled in the art can flexibly select or supplement other relevant parameters that can characterize the same dimension within the above framework, based on the specific geological characteristics of the target reservoir, experimental conditions, and data availability.

[0061] For example, in the dimension of "oil-rock interaction", in addition to mineral content and specific surface area, parameters such as wettability contact angle and surface zeta potential can also be considered. In the dimension of "pore structure", parameters such as pore shape factor and tortuosity can be added; In the dimension of "rock physical properties", derivative physical parameters such as sonic transit time and density logging can be combined.

[0062] As long as the selected set of parameters can effectively reflect the above-mentioned multi-dimensional characteristics affecting the mobility of shale oil, and is sufficient to extract the main controlling factors with geological significance through factor analysis, it falls within the protection scope of this invention.

[0063] S4. Determine the movable oil saturation of each representative sample. , and the corresponding E 1k , E 2k as well as Pairing is performed to form multiple sets of paired sample data; based on the multiple sets of paired sample data, a relationship is established between the movable oil saturation S and... E 1 , E 2 and The multiple linear regression model between them.

[0064] First, this step pairs the experimentally measured "results" with the "features" extracted by mathematical analysis.

[0065] Target variable (Y): Movable oil saturation directly measured through a high-fidelity simulated gas-driven nuclear magnetic resonance combined experiment. .

[0066] Characteristic variable (X): Through factor analysis in step S3, the scores of the three principal control factors calculated for this representative sample using factor score coefficients are... E 1 , E 2 and E 3 .

[0067] Specifically, all representative samples ( , E 1 , E 2 , E 3 The data is organized into tables to form a "paired sample dataset" for modeling.

[0068] In this embodiment, the paired sample dataset contains 8 samples (corresponding to 8 representative samples), and its structure is shown in Table 5 below: Table 5 Paired Sample Dataset Import the above sample dataset into statistical analysis software (such as SPSS, R, Python's statsmodels, or the scikit-learn library) to obtain the movable oil saturation. As the dependent variable, with three main control factors E 1 , E 2 , E 3 Using as the independent variable, perform multiple linear regression analysis.

[0069] The general form of the regression model used is formula (3): =β1* E 1 +β2* E 2 +β3* E 3 +ε (3) Where β1, β2, and β3 are the independent variables. E 1 , E 2, E 3 The partial regression coefficient represents the effect caused by a one-unit change in this factor, assuming other factors remain constant. The average change; ε is the random error term.

[0070] In one specific embodiment, the specific values ​​in Table 5 are shown in Table 6 below. The following prediction model (4) is obtained through regression analysis:

[0071] Table 6 Paired Sample Data Values (4) Specifically, to visually demonstrate the model's predictive performance, Figure 2 Scatter plots and fitted lines were plotted for the "predicted movable oil saturation" and "measured movable oil saturation" of each representative sample. Figure 2 The data shows that the data points are closely distributed around the ideal fitted line (y = x), and the labeled coefficient of determination R² = 0.803, further verifying the high consistency between the model's predicted values ​​and the measured values. Figure 2 The results corroborate the aforementioned statistical test results, jointly demonstrating that the established multiple linear regression model has good predictive and generalization capabilities.

[0072] At this point, step S4 is complete, and a system based on the main control factor, which can be used to quickly predict the movable oil saturation of shale oil, has been successfully established. The multiple linear regression model is used. This model quantitatively characterizes the mobility law under the coupling effect of multiple factors and is the core mathematical tool for achieving "efficient, low-cost, and universal prediction" in this invention.

[0073] S5. Using the established multiple linear regression model, predict the mobility of shale oil.

[0074] This step is the core of applying the complete methodology established in the previous steps to practical engineering evaluation. Its core lies in using the established predictive model to quickly and cost-effectively estimate the movable oil saturation of any new target shale sample that has not undergone high-fidelity simulation experiments. This allows us to evaluate its mobility.

[0075] The specific process of the prediction operation is as follows: (1) Obtaining geological and basic parameters of new samples: For the target shale sample to be evaluated (hereinafter referred to as the "new sample"): Lithofacies identification: First, core observation and description of the new sample are performed, and the lithofacies are named according to the same standards as in step S1. This step is used to confirm whether the lithofacies type of the new sample is included within the lithofacies range of the representative sample used in modeling, in order to assess the applicability of the model application.

[0076] Multidimensional parameter measurement: Perform the same routine, rapid, and low-cost experiments as in step S3 on the new sample to obtain its multidimensional fundamental parameters.

[0077] (2) Calculate the main control factor score of the new sample: The seven basic parameter values ​​of the new sample obtained in step (1) are [ φ, K, r, P c ，M P ，M C ， A], Substitute into the fixed factor score coefficient formula determined in the embodiment of step S3 (i.e., the given calculation formula (2): The main control factor scores (E1_new, E2_new, E3_new) corresponding to the new sample are calculated.

[0078] (3) Substitute into the model for prediction: The master control factor scores (E1_new, E2_new, E3_new) of the new sample calculated in step (2) are directly substituted into the fixed multiple linear regression prediction model established in the embodiment of step S4: Sk_pred = 3.071 * E1_new - 0.082 * E2_new + 0.094 * E3_new (5) The calculated Sk_pred is the predicted movable oil saturation of the new sample.

[0079] (4) Results output and mobility evaluation: The obtained Sk_pred value is the quantitative evaluation result of the mobility of the new shale oil sample. Based on this value:

[0080] It can compare and rank the mobility of different samples to identify highly mobile "sweet spots".

[0081] Based on geological understanding, reservoir quality can be graded and evaluated.

[0082] The prediction results can be directly used as key input parameters for development scheme optimization and economic evaluation.

[0083] Example demonstration: Suppose a new sample is identified as "massive fine sandstone", and its measurement parameters are: φ =7.72%, K =0.33 mD, r =94.28 nm, P c =0.78 MPa M P =94.40%, M C =5.6%, A=0.34 m² / g.

[0084] Calculate the main control factor scores: E1_new = 0.483 * 3.0 + 0.472 * 0.08 + ... ≈ 11.29 (example value, the same below), E2_new ≈ -38.98, E3_new ≈ 43.13.

[0085] Substituting into the model prediction: Sk_pred = 3.071 * 11.29 - 0.082 * (-38.98) + 0.094 * 43.13 ≈ 62.09%.

[0086] Conclusion: The predicted movable oil saturation of this new sample is approximately 62.09%.

[0087] Thus, this invention completes the entire process from "preliminary modeling" to "post-application". Through the model and rules established in S1-S4, rapid, low-cost, and quantitative prediction of the mobility of a large number of new samples is achieved in S5, without having to repeat expensive and time-consuming gas-driven nuclear magnetic resonance experiments for each new sample, which significantly improves the evaluation efficiency and economy.

[0088] This invention provides a novel, accurate, efficient, and economical solution for shale oil mobility evaluation through a systematic method integrating highly realistic physical experiments, multi-factor data analysis, and intelligent modeling and prediction. Its beneficial effects are mainly reflected in the following three progressive levels: 1) It fundamentally solves the problem of "simulating authenticity" and significantly improves the accuracy and reliability of the evaluation.

[0089] Technical basis: "Constant pressure gas drive-nuclear magnetic resonance combined experiment under simulated formation temperature and confining pressure conditions".

[0090] Results: The driving mechanism (pressure-driven) and temperature and pressure conditions of this experiment closely match the actual mining process, completely overcoming the inherent defects of traditional simulation methods such as percolation and centrifugation, thus providing highly reliable "gold standard" data (mobile oil saturation) for the entire evaluation system. This laid the foundation for accurate and reliable evaluation results.

[0091] 2) It scientifically analyzed the complex effects of "multi-factor coupling" and achieved a leap from empirical description to mechanism quantification in evaluation.

[0092] Core technology: Introducing "factor analysis" to reduce and integrate multi-dimensional parameters such as rock properties, pore structure, and oil-rock interaction.

[0093] Results: Rock physical property factors were successfully extracted. E 1 ), Pore structure factor ( E 2 ) and oil-rock interaction factor ( E 3 Three key factors with clearly defined physical meanings were identified. This not only clearly reveals the intrinsic controlling dimensions affecting mobility and their coupling relationships, but also transforms complex high-dimensional data into concise, stable, and high-quality features suitable for mathematical modeling, providing a scientific basis for establishing predictive models with clear mechanisms and strong interpretability.

[0094] 3) It has revolutionized the evaluation model to achieve "high efficiency, low cost and universality", meeting the urgent needs of large-scale engineering applications.

[0095] Application Paradigm: Construct a two-stage "modeling-prediction" paradigm that "models based on a small number of representative samples and predicts a large number of new samples".

[0096] Effect: High efficiency and economy: In the model application phase, there is no need to repeat time-consuming and expensive high-fidelity simulation experiments for each new sample. Only rapid and low-cost conventional experiments are required, and the mobility can be predicted instantly by the model. This reduces the evaluation cost and cycle time of a single sample by several orders of magnitude.

[0097] Universality and Intelligence: The Established Multiple Linear Regression Model (S = f( E 1 , E 2 , E 3 It possesses excellent extrapolation and prediction capabilities. Once the model is established, it can rapidly, batch-wise, and accurately evaluate a large number of unknown samples covering multiple lithofacies under the same geological background, realizing the transformation from "one sample experiment" to "one model, universal prediction" intelligent evaluation mode.

[0098] In summary, this invention not only provides a single technological advancement in shale oil mobility evaluation, but also constructs a complete methodological system. It systematically solves the three core bottlenecks that have long existed in this field: "simulation distortion, biased analysis, and low efficiency." Ultimately, it forms a key technology that is realistic in simulation, clear in mechanism, and highly efficient and practical. This technology can provide reliable, rapid, and economical quantitative support for sweet spot identification, development scheme optimization, reserve assessment, and economic decision-making in shale oil reservoirs, and has significant theoretical value and broad engineering application prospects for promoting the efficient development of shale oil resources.

[0099] It is understood that the present invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A multi-parameter modeling and prediction method for shale oil mobility, characterized in that, include: S1. Obtain a set of rock samples, and identify and classify various rock facies by observing the cores of the set of rock samples; S2. For various rock facies, representative samples are selected from each facies, and gas-driven nuclear magnetic resonance combined experiments are performed on each representative sample to obtain the mobile oil saturation of each representative sample. Where k is the number of the representative sample; the gas-driven nuclear magnetic resonance combined experiment includes: Under simulated formation temperature and confining pressure conditions, the total oil content of representative samples saturated with simulated formation oil was determined using nuclear magnetic resonance (NMR) technology. C O1 ; Gas was injected into a representative sample that had been simulated to be saturated with formation oil for displacement. After displacement, the residual oil content was determined again using nuclear magnetic resonance (NMR) technology. C O2 ; in accordance with and C O1 and C O2 The correspondence determines the movable oil saturation of the representative sample. ; S3. Collect the multi-dimensional parameters of all representative samples to form a multi-dimensional parameter dataset, and perform factor analysis on the multi-dimensional parameter dataset to extract three independent first principal control factors. E 1 Second controlling factor E 2 and the third controlling factor Using the factor score coefficients determined by the factor analysis, the score of each representative sample on the first principal control factor is calculated. E 1k Score on the second major control factor E 2k And the score on the third controlling factor ; S4. Determine the movable oil saturation of each representative sample. , and the corresponding E 1k , E 2k as well as Pairing is performed to form multiple sets of paired sample data; based on the multiple sets of paired sample data, a relationship is established between the movable oil saturation S and... E 1 , E 2 and Multiple linear regression model between them; S5. Using the aforementioned multiple linear regression model, predict the mobility of shale oil.

2. The multi-parameter modeling and prediction method for shale oil mobility according to claim 1, characterized in that, The simulated formation temperature is 80℃, the confining pressure is 50MPa, the injected gas is CO2, and the injection process is carried out under a constant pressure of 10MPa.

3. The multi-parameter modeling and prediction method for shale oil mobility according to claim 1, characterized in that, in, and C O1 and C O2 The specific correspondence is as follows: in, C O2 The displacement was maintained for 3 days during constant-pressure gas injection, with the mass of the representative sample measured every 6 hours. Displacement equilibrium was determined when the mass of the representative sample remained constant for three consecutive measurements. Subsequently, nuclear magnetic resonance (NMR) experiments were performed on the representative sample to obtain the residual oil content. C O2 .

4. The multi-parameter modeling and prediction method for shale oil mobility according to claim 1, characterized in that, The multidimensional parameters include rock physical properties, pore structure parameters, and oil-rock interaction parameters; The rock physical properties include porosity. φ and penetration rate K The pore structure parameters include the average pore radius. r and exhaust pressure P c The oil-rock interaction parameters include the content of brittle minerals. M P clay mineral content M C and specific surface area A。 5. The multi-parameter modeling and prediction method for shale oil mobility according to claim 1, characterized in that, The factor score coefficient makes E 1 Characterizing rock physical properties, E 2 Characterizing pore structure, E 3 Characterize oil-rock interaction.

6. The multi-parameter modeling and prediction method for shale oil mobility according to claim 1, characterized in that, The multiple linear regression model is as follows: = * E 1 + * E 2 + * E 3 +e in, It is the independent variable E 1 The partial regression coefficient, Independent variable E 2 The partial regression coefficient, It is the independent variable E 3 The partial regression coefficients, where ε is the random error term.

7. The multi-parameter modeling and prediction method for shale oil mobility according to claim 1, characterized in that, The prediction includes: using the multiple linear regression model, for newly acquired rock samples that have not undergone the gas-driven nuclear magnetic resonance combined experiment described in S2, calculating the corresponding master control factor value based on their multidimensional parameter data and the factor score coefficient, and then calculating the mobile oil saturation through the multiple linear regression model to achieve rapid prediction of the mobility of shale oil in the new samples.

Images