A method for shale gas reservoir sweet spot evaluation and analysis

Abstract

The application discloses a shale gas reservoir dessert evaluation and analysis method, including: collecting seismic data, well logging curve data and geological data of strata of a historical shale gas reservoir to obtain standard characteristic variables; calculating a comprehensive index, a coupling index and a complexity index of the strata; calculating a reservoir dessert index Y; constructing a training data set; constructing a physical constraint neural network PCNN and adding a physical regularization term to output the trained physical constraint neural network PCNN; collecting seismic data, well logging curve data and geological data of strata of a target shale gas reservoir and inputting the trained physical constraint neural network PCNN to output a reservoir dessert index of the target shale gas reservoir, and then evaluating the target shale gas reservoir dessert. The application constructs an interpretable shale gas reservoir dessert intelligent evaluation method, and realizes accurate evaluation of advantages of a shale gas reservoir dessert through the predicted reservoir dessert index.

Description

Technical Field

[0001] This invention relates to the field of shale gas exploration, and more specifically to a method for evaluating and analyzing sweet spots in shale gas reservoirs. Background Technology

[0002] Shale gas sweet spots refer to areas rich in shale gas and easily exploitable. Current shale gas sweet spot evaluation faces challenges such as high data heterogeneity, complex mechanistic coupling, and significant prediction uncertainty. Traditional methods, such as multiple regression, fuzzy comprehensive evaluation, or single machine learning models, have the following specific drawbacks:

[0003] The feature construction lacks geological mechanism guidance: it only uses the original geophysical or well logging parameters and fails to construct composite features that can reflect the complex process of "generation-reservoir-preservation-permeability-breakage" of shale reservoirs.

[0004] Uncertainty is not systematically quantified: the prediction results are only point estimates, without providing confidence intervals or probability distributions, and cannot support risk decision-making. Summary of the Invention

[0005] To address the aforementioned shortcomings of existing technologies, this invention provides a method for evaluating and analyzing sweet spots in shale gas reservoirs. This method quantitatively predicts the quality of sweet spots in shale gas reservoirs and enables comprehensive analysis of sweet spots in shale gas reservoirs based on composite characteristics.

[0006] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0007] A method for evaluating and analyzing sweet spots in shale gas reservoirs is provided, comprising:

[0008] Step S1: Collect historical seismic data, well logging data, and geological data of shale gas reservoirs as feature variables, and perform standardization processing to obtain standard feature variables;

[0009] Step S2: Calculate the comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of the rock formation. GPI A coupling index characterizing the reservoir properties and compressibility of rock formations RFI and the complexity index characterizing the complexity of the microscopic pore structure of rock strata. ;

[0010] Step S3: Utilize the composite index GPI Coupling index RFI and complexity index Calculate the reservoir sweetness index Y ;

[0011] Step S4: Use the standard characteristic variables of historical shale gas reservoirs as input data, including the reservoir sweet spot index. YAs output data, training samples are formed for each historical shale gas reservoir, and a training dataset is constructed.

[0012] Step S5: Construct a Physically Constrained Neural Network (PCNN) and apply the loss function to the PCNN. Add a physical regularization term; train the Physical Constraint Neural Network (PCNN) using the training dataset, and output the trained PCNN.

[0013] Step S6: Collect seismic data, well logging data, and geological data of the target shale gas reservoir, and perform standardization processing to obtain the input vector of the target shale gas reservoir. Input this vector into a trained Physically Constrained Neural Network (PCNN) to output the reservoir sweet spot index of the target shale gas reservoir. This allows for the evaluation of the sweet spot in the target shale gas reservoir.

[0014] Furthermore, step S1 specifically includes:

[0015] Historical seismic data, well logging data, and geological data of shale gas reservoirs were collected as characteristic variables. Furthermore, a combination of interval scaling and distribution alignment was used to standardize the feature variables, resulting in standard feature variables. ;

[0016] ;

[0017] Where i is the number of the feature variable, For characteristic variables The 0.01 quantile, For characteristic variables The 0.99 quantile, It is the inverse function of the standard normal distribution. For characteristic variables The empirical cumulative distribution function, These are the weighting coefficients.

[0018] Furthermore, a comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of rock formations. GPI The calculation method is as follows:

[0019] The total organic carbon (TOC) content of rock strata is used to calculate a comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of rock strata. GPI ;

[0020] ;

[0021] in, The content of free hydrocarbons in the rock strata. These are correction factors for shale kerogen type and maturity, respectively. These are the carbon isotope values ​​of methane. The threshold for carbon isotope loss. This represents the measured reflectance of the vitrinite. As a hydrocarbon generation threshold, This is the scale parameter.

[0022] Furthermore, the complexity index characterizes the complexity of the microscopic pore structure of rock strata. The calculation method is as follows:

[0023] Based on the specific surface area of ​​the rock strata Calculate the complexity index that characterizes the complexity of the microscopic pore structure of rock strata. ;

[0024] ;

[0025] in, The pore size corresponding to the peak value of the pore size distribution. The fractal dimension of the pores. For mesopore volume, For micropore volume, This refers to the volume of the macropore.

[0026] Furthermore, the coupling index characterizing the reservoir properties and compressibility of rock formations... RFI The calculation method is as follows:

[0027] Based on the experimentally measured rock porosity Data calculations characterize the coupling index between reservoir properties and compressibility of rock formations. RFI Porosity Including effective porosity and total porosity ;

[0028] ;

[0029] in, These are the effective porosity and total porosity of the rock strata, respectively. This is the critical value for porosity. To restrict water saturation, These are the brittleness index and the critical value of the brittleness index of the rock strata, respectively. These represent the maximum and minimum horizontal stresses of the rock strata, respectively. For the optimal stress ratio, For stress ratio tolerance, These are the weighting indices.

[0030] Furthermore, the reservoir sweetness index Y The calculation method is as follows:

[0031] ;

[0032] in, These are the comprehensive indices corresponding to historically discovered reservoir sweet spots. GPI Complexity index and coupling index RFI The average value, These are the weighting coefficients for hydrocarbon generation potential and storage conditions, micropore structure complexity, and storage capacity and compressibility, respectively.

[0033] Further, step S5 includes:

[0034] Step S51: Construct a Physically Constrained Neural Network (PCNN) and apply the loss function to the PCNN. Add a physical regularization term to it;

[0035] ;

[0036] in, n The training sample number, For the first n The reservoir sweetness index predicted from each training sample. For the first n The actual reservoir sweetness index of each training sample. As a monotonicity constraint, Boundary condition constraints;

[0037] The output layer of the Physically Constrained Neural Network (PCNN) is: , Weight vector , To form an input vector from the input data in the training samples, For activation functions;

[0038] Step S52: Train the Physically Constrained Neural Network (PCNN) using the training dataset, and correct the weight coefficients in the weight vector until the PCNN converges, thus obtaining the trained PCNN.

[0039] The beneficial effects of this invention are as follows: This invention constructs an interpretable intelligent evaluation method for shale gas reservoir sweet spots, achieving accurate evaluation of the sweet spot advantage of shale gas reservoirs through the predicted reservoir sweet spot index. This is achieved by integrating multiple types of data into a comprehensive index of hydrocarbon generation potential and preservation conditions. GPI Storage and compressibility coupling index RFI and the complexity index of micropore structure It achieves the fusion of composite characteristics that can reflect the complex process of "generation-reservoir-preservation-permeability-breakage" in shale reservoirs, improves the physical interpretability and predictive reliability of multiple types of characteristics, and can be adapted to the sweet spot evaluation of shale gas, tight gas and even shale oil reservoirs in different geological backgrounds, with wide applicability. Attached Figure Description

[0040] Figure 1 This is a flowchart of the evaluation and analysis method for sweet spots in shale gas reservoirs. Detailed Implementation

[0041] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0042] like Figure 1 As shown, a method for evaluating and analyzing sweet spots in shale gas reservoirs includes:

[0043] Step S1: Collect historical seismic data, well logging data, and geological data of shale gas reservoirs as characteristic variables. Furthermore, a combination of interval scaling and distribution alignment was used to standardize the feature variables, resulting in standard feature variables. ;

[0044] ;

[0045] in, i The identifier is the number of the feature variable. For characteristic variables The 0.01 quantile, For characteristic variables The 0.99 quantile, It is the inverse function of the standard normal distribution. For characteristic variables The empirical cumulative distribution function, These are the weighting coefficients;

[0046] This embodiment utilizes the thousandths of the feature variables to eliminate the influence of extreme values, and weighting coefficients are used to balance interval scaling and normal transformation. Typically, the weighting coefficients... Take 0.7.

[0047] The seismic data acquired in this embodiment includes amplitude, frequency, and impedance, while the well logging data includes natural gamma data (GR), density data (DEN), acoustic data (AC), and formation resistivity data (RT). The geological data of the rock formations includes the experimentally measured total organic carbon (TOC) content and mineral composition ratio. V Porosity Penetration rate k gas content G and maturity R wait;

[0048] Step S2: Calculate the comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of the rock formation. GPI A coupling index characterizing the reservoir properties and compressibility of rock formations RFI and the complexity index characterizing the complexity of the microscopic pore structure of rock strata. ;

[0049] The total organic carbon (TOC) content of rock strata is used to calculate a comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of rock strata. GPI ;

[0050] ;

[0051] in, The content of free hydrocarbons in the rock strata. These are correction factors for shale kerogen type and maturity, respectively. These are the carbon isotope values ​​of methane. The threshold for carbon isotope loss. This represents the measured reflectance of the vitrinite. As a hydrocarbon generation threshold, For scale parameters;

[0052] The free hydrocarbon content in the rock strata of this embodiment Shale kerogen type correction factor used to characterize the current hydrocarbon content of rock strata. Values ​​are assigned based on kerogen type: 1.2 for type I, 1.0 for type II, and 0.8 for type III; methane carbon isotope values. Used to indicate the degree of gas loss; the smaller the negative value, the stronger the loss. Carbon isotope loss threshold. Usually taken .

[0053] Complexity index, which characterizes the complexity of the microscopic pore structure of rock strata This approach unifies hydrocarbon generation processes and preservation conditions into a single mathematical framework, reflecting the entire process of hydrocarbon generation and preservation. It directly quantifies the impact of hydrocarbon loss risk on dessert quality, reflects the differences in thermal evolution stages through maturity, and considers the differences in hydrocarbon generation potential among different kerogen types.

[0054] Total organic carbon (TOC) content in rock formations represents the percentage of organic carbon per unit mass of rock. It is the material basis for hydrocarbon generation and determines the upper limit of maximum hydrocarbon generation potential. Effective TOC in shale gas is typically >1.5%, and in high-quality sweet spots, it is >3%. Free hydrocarbon content in rock formations Characterizing the amount of free hydrocarbons already generated and pyrolyzable in shale, reflecting the current hydrocarbon content, and representing the actual result of TOC hydrocarbon generation and conversion. Hydrocarbon generation threshold. Typically, it is taken as 0.5% or 0.7%, which is below the hydrocarbon generation threshold. The organic matter has not entered the effective hydrocarbon generation stage. Scale parameters The exp function is used to control the growth rate, which can effectively simulate the gradual process of organic matter converting into hydrocarbons.

[0055] The complexity index of the micropore structure of the rock strata constructed in this invention. The calculation method reflects the cascade of hydrocarbon generation and storage processes, forming multiple advantageous factors that work together to produce superlinear gains.

[0056] Based on the experimentally measured rock porosity Data calculations characterize the coupling index between reservoir properties and compressibility of rock formations. RFI Porosity Including effective porosity and total porosity ;

[0057] ;

[0058] in, These are the effective porosity and total porosity of the rock strata, respectively. This is the critical value for porosity. To restrict water saturation, These are the brittleness index and the critical value of the brittleness index of the rock strata, respectively. These represent the maximum and minimum horizontal stresses of the rock strata, respectively. For optimal stress ratio, a ratio close to 1 is theoretically beneficial for forming complex mesh structures. For stress ratio tolerance, These are the weighting indices, and in this embodiment, the weighting indices are... The value range is 0.5-1;

[0059] As a standardized term for storage capacity, it is used to characterize the pore space available for gas storage and flow, eliminating ineffective pores occupied by bound water. As a compressibility normalization term, it is used to characterize the relative importance of storage and compressibility. As a geostress compatibility term, it is a coupling index characterizing the reservoir properties and compressibility of rock formations. RFIThe actual exploitability of shale gas is characterized by three aspects: storage capacity, compressibility, and geostress.

[0060] Based on the specific surface area of ​​the rock strata Calculate the complexity index that characterizes the complexity of the microscopic pore structure of rock strata. ;

[0061] ;

[0062] in, The pore size corresponding to the peak value of the pore size distribution. The fractal dimension of the pores. For mesopore volume, For micropore volume, For large pore volume;

[0063] Complexity index of the invention Characterizing the gas storage properties of shale from pore volume to pore structure. Complexity index. The larger the shale, the stronger its gas storage capacity, and vice versa.

[0064] As a synergistic term of specific surface area and pore size, it represents the specific surface area provided by unit logarithmic pore size. A larger value indicates both a high specific surface area (favorable for adsorption) and a relatively large, advantageous pore size (favorable for seepage). A smaller value indicates either a low specific surface area or excessively small pore size that obstructs seepage. Specific surface area The number of adsorption sites available per gram of rock is positively correlated with organic matter abundance and clay mineral content; the larger the specific surface area, the more tortuous the gas flow path; pore size... The maximum point of the pore size distribution curve represents the most developed pore size. This is achieved by analyzing the pore size... Taking the logarithm to compress the order-of-magnitude difference with the specific surface area, the effect of pore size on seepage capacity follows a power-law relationship, and the logarithmic transformation is more in line with physical reality.

[0065] As a fractal characteristic term of pore surfaces, it is used to characterize the complexity and self-similarity of pore structures. The larger the value, the more favorable it is for gas flow. A fractal dimension of 2 represents an ideal smooth surface.

[0066] This is the pore volume synergistic term, which is the volume ratio of mesopores to the sum of micropores and macropores, emphasizing the bridging role of mesopores in connecting pores of different scales.

[0067] Step S3: Utilize the composite index GPI Coupling index RFI and complexity index Calculate the reservoir sweetness index Y ;

[0068] ;

[0069] in, These are the comprehensive indices corresponding to historically discovered reservoir sweet spots. GPI Complexity index and coupling index RFI The average value, These are weighting coefficients for hydrocarbon generation potential and storage conditions, micropore structure complexity, and storage capacity and compressibility, respectively, and satisfying the following conditions: In this embodiment, we take... In reservoir sweetness assessment, hydrocarbon generation potential and preservation conditions are more important, as they determine the amount of shale gas reserves. Reservoir sweetness index. Y The sweet spot index indicates the advantage of shale gas reservoirs. Y The larger the shale gas reservoir, the higher its quality.

[0070] Step S4: Convert the standard characteristic variables corresponding to historical shale gas reservoirs. As input data, reservoir sweetness index Y As output data, training samples are generated for each historical shale gas reservoir, and a training dataset is constructed. The number of training samples in the training dataset is [number missing]. N ;

[0071] Step S5: Construct a Physically Constrained Neural Network (PCNN) and apply the loss function to the PCNN. Add a physical regularization term; train the Physical Constraint Neural Network (PCNN) using the training dataset, and output the trained PCNN. Step S5 specifically includes:

[0072] Step S51: Construct a Physically Constrained Neural Network (PCNN) and apply the loss function to the PCNN. Add a physical regularization term to it;

[0073] ;

[0074] in, n The training sample number, For the first n The reservoir sweetness index predicted from each training sample. For the first n The actual reservoir sweetness index of each training sample. As a monotonicity constraint, Boundary condition constraints;

[0075] Monotonicity constraints This is used to constrain the correlation between characteristic variables and gas content. For example, if it is known that total organic carbon (TOC) is positively correlated with the gas content of a rock stratum, then this constraint... , Minimum tolerance; boundary condition constraints Used to constrain the limiting effects of characteristic variables, for example, when porosity When the value is less than the critical value, the Physically Constrained Neural Network (PCNN) is forced to output a predicted value close to 0.

[0076] The output layer of the Physically Constrained Neural Network (PCNN) is: , Weight vector , To form an input vector from the input data in the training samples, For activation functions;

[0077] Step S52: Train the Physically Constrained Neural Network (PCNN) using the training dataset, and correct the weight coefficients in the weight vector until the PCNN converges, thus obtaining the trained PCNN.

[0078] Step S6: Collect seismic data, well logging data, and geological data of the target shale gas reservoir, perform standardization processing to obtain the input vector of the target shale gas reservoir, input it into the trained Physically Constrained Neural Network (PCNN), and output the reservoir sweet spot index of the target shale gas reservoir. ;

[0079] According to the reservoir sweetness index The size of the sweet spot in the target shale gas reservoir is used to evaluate its advantages, and the sweet spot index is used to determine its strength. The larger the value, the higher the quality of the target shale gas reservoir sweet spot; conversely, the lower the quality of the target shale gas reservoir sweet spot.

[0080] This invention constructs an interpretable intelligent evaluation method for shale gas reservoir sweet spots, achieving accurate evaluation of the sweet spot advantage of shale gas reservoirs through a predicted reservoir sweet spot index. This is achieved by integrating multiple types of data into a comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of the formation. GPI A coupling index characterizing the reservoir properties and compressibility of rock formations RFI and the complexity index characterizing the complexity of the microscopic pore structure of rock strata. It achieves the fusion of composite characteristics that can reflect the complex process of "generation-reservoir-preservation-permeability-breakage" in shale reservoirs, improves the physical interpretability and predictive reliability of multiple types of characteristics, and can be adapted to the sweet spot evaluation of shale gas, tight gas and even shale oil reservoirs in different geological backgrounds, with wide applicability.

Claims

1. A method for evaluating and analyzing sweet spots in shale gas reservoirs, characterized in that, include: Step S1: Collect historical seismic data, well logging data, and geological data of shale gas reservoirs as feature variables, and perform standardization processing to obtain standard feature variables; Step S2: Calculate the comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of the rock formation. GPI A coupling index characterizing the reservoir properties and compressibility of rock formations RFI and the complexity index characterizing the complexity of the microscopic pore structure of rock strata. ; Step S3: Utilize the composite index GPI Coupling index RFI and complexity index Calculate the reservoir sweetness index Y ; Step S4: Use the standard characteristic variables of historical shale gas reservoirs as input data, including the reservoir sweet spot index. Y As output data, training samples are formed for each historical shale gas reservoir, and a training dataset is constructed. Step S5: Construct a Physically Constrained Neural Network (PCNN) and apply the loss function to the PCNN. Add a physical regularization term; train the Physical Constraint Neural Network (PCNN) using the training dataset, and output the trained PCNN. Step S6: Collect seismic data, well logging data, and geological data of the target shale gas reservoir, and perform standardization processing to obtain the input vector of the target shale gas reservoir. Input this vector into a trained Physically Constrained Neural Network (PCNN) to output the reservoir sweet spot index of the target shale gas reservoir. This allows for the evaluation of the sweet spot of the target shale gas reservoir. The comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of rock formations GPI The calculation method is as follows: The total organic carbon (TOC) content of rock strata is used to calculate a comprehensive index characterizing the hydrocarbon generation potential and preservation conditions of rock strata. GPI ; ； in, The content of free hydrocarbons in the rock strata. These are correction factors for shale kerogen type and maturity, respectively. These are the carbon isotope values ​​of methane. The threshold for carbon isotope loss. This represents the measured reflectance of the vitrinite. As a hydrocarbon generation threshold, Scale parameter; Shale kerogen type correction factor The values ​​are determined based on the type of kerogen: 1.2 for type I, 1.0 for type II, and 0.8 for type III. The complexity index characterizing the complexity of the microscopic pore structure of rock strata The calculation method is as follows: Based on the specific surface area of ​​the rock strata Calculate the complexity index that characterizes the complexity of the microscopic pore structure of rock strata. ; ； in, The pore size corresponding to the peak value of the pore size distribution. The fractal dimension of the pores. For mesopore volume, For micropore volume, For large pore volume; The coupling index characterizing the reservoir properties and compressibility of rock formations RFI The calculation method is as follows: Based on the experimentally measured rock porosity Data calculations characterize the coupling index between reservoir properties and compressibility of rock formations. RFI Porosity Including effective porosity and total porosity ; ； in, These are the effective porosity and total porosity of the rock strata, respectively. This is the critical value for porosity. To restrict water saturation, These are the brittleness index and the critical value of the brittleness index of the rock strata, respectively. These represent the maximum and minimum horizontal stresses of the rock strata, respectively. For the optimal stress ratio, For stress ratio tolerance, These are the weighting indices; Step S5 includes: Step S51: Construct a Physically Constrained Neural Network (PCNN) and apply the loss function to the PCNN. Add a physical regularization term to it; ； in, n The training sample number, For the first n The reservoir sweetness index predicted from each training sample. For the first n The actual reservoir sweetness index of each training sample. As a monotonicity constraint, Boundary condition constraints; Monotonicity constraints Boundary condition constraints are used to constrain the correlation between characteristic variables and gas content. Used to constrain the limiting effects of characteristic variables; The output layer of the Physically Constrained Neural Network (PCNN) is: , For the weight vector, To form an input vector from the input data in the training samples, For activation functions; Step S52: Train the Physically Constrained Neural Network (PCNN) using the training dataset, and correct the weight coefficients in the weight vector until the PCNN converges, thus obtaining the trained PCNN.

2. The method for evaluating and analyzing sweet spots in shale gas reservoirs according to claim 1, characterized in that, Step S1 specifically involves: Historical seismic data, well logging data, and geological data of shale gas reservoirs were collected as characteristic variables. Furthermore, a combination of interval scaling and distribution alignment was used to standardize the feature variables, resulting in standard feature variables. ; ； in, i The identifier is the number of the feature variable. For characteristic variables The 0.01 quantile, For characteristic variables The 0.99 quantile, It is the inverse function of the standard normal distribution. For characteristic variables The empirical cumulative distribution function, These are the weighting coefficients.

3. The method for evaluating and analyzing sweet spots in shale gas reservoirs according to claim 1, characterized in that, The reservoir sweetness index Y The calculation method is as follows: ； in, These are the comprehensive indices corresponding to historically discovered reservoir sweet spots. GPI Complexity index and coupling index RFI The average value, These are the weighting coefficients for hydrocarbon generation potential and storage conditions, micropore structure complexity, and storage capacity and compressibility, respectively.

Images