Reservoir seepage characteristic evaluation method and system

Abstract

The invention discloses a reservoir seepage characteristic evaluation method and system, and relates to the technical field of oil reservoir development. The method comprises the following steps: acquiring key data of a target reservoir; the key data comprises rock core CT scanning data, historical pressure and yield data of a production well and tracer agent monitoring data; constructing a dynamic response model based on a fluid potential energy theory and the key data; the dynamic response model comprises an improved deep forest mechanism algorithm and an attention mechanism weighting algorithm; wherein the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture; and performing seepage capacity classification by utilizing the dynamic response model, and performing three-dimensional visual simulation on a seepage channel according to a classification result. The limitation of a traditional method can be broken through, and technical support is provided for efficient development of complex reservoirs.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir development, and particularly to a method and system for evaluating reservoir seepage characteristics. Background Art

[0002] In the field of traditional oil and gas reservoir research, the evaluation of seepage characteristics has long faced technical bottlenecks. Existing methods mainly rely on two major paths: core experiments and static geological modeling. The former requires physical tests on core samples in the laboratory to obtain parameters such as permeability, and the latter constructs a static reservoir model based on geostatistics. However, both of these two types of technologies have significant defects: Core experiments require a large amount of manpower and material resources for sampling, processing, and testing, and are limited by the sample quantity and representativeness, making it difficult to comprehensively reflect the dynamic characteristics of the formation; Although static geological modeling can provide a macroscopic framework, it lacks the ability to depict reservoir heterogeneity and dynamic changes, resulting in limited model prediction accuracy.

[0003] As a dynamic analysis method, traditional well test interpretation technology is limited by the data acquisition range of a single well point. When dealing with complex heterogeneous reservoirs, the interpretation results are often distorted due to the spatial variability of reservoir parameters. Other evaluation methods based on dynamic information capture do not consider multiphase flow effects (such as the differences in the seepage laws of oil, gas, and water three-phase flows), and are limited in application in complex reservoirs such as low-permeability and fractured reservoirs.

[0004] In recent years, parameter inversion technology driven by production data has become a research hotspot. By inverting reservoir seepage parameters through historical production data (such as wellhead pressure and production), dynamic evaluation can theoretically be achieved. However, this direction still faces two major challenges: First, the algorithm convergence is insufficient, and the inversion process is prone to falling into local optimal solutions, resulting in parameter estimation deviations; Second, the lack of a multi-scale data fusion mechanism makes it difficult to collaboratively process wellbore-scale production data and macro-scale information such as geology / seismic, restricting the practicality of the method. Summary of the Invention

[0005] The object of the present invention is to provide a method and system for evaluating reservoir seepage characteristics, which can break through the limitations of traditional methods and provide technical support for the efficient development of complex reservoirs.

[0006] To achieve the above object, the present invention provides the following solution:

[0007] A method for evaluating reservoir seepage characteristics, comprising:

[0008] Obtaining key data of a target reservoir; the key data includes core CT scan data, historical pressure and production data of production wells, and tracer monitoring data;

[0009] Construct a dynamic response model based on the fluid potential energy theory and the key data; the dynamic response model includes an improved deep forest mechanism algorithm and an attention mechanism weighting algorithm; among them, the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture;

[0010] Use the dynamic response model to classify the seepage capacity, and perform three-dimensional visualization simulation on the seepage channels according to the classification results.

[0011] Optionally, the dynamic response model is specifically expressed as:

[0012]

[0013] Among them, λ eff represents the equivalent flow coefficient, β represents the inertia factor, K ro (S w ) represents the relative permeability of the oil phase, S w represents the current water saturation, μ o represents the viscosity of the crude oil, ρ o represents the density of the crude oil, and v represents the seepage velocity.

[0014] Optionally, the inertia factor in the dynamic response model is obtained by fitting the pressure difference-flow velocity curve through a core high-speed displacement experiment.

[0015] Optionally, before constructing the dynamic response model, it further includes: preprocessing the key data; the preprocessing process is: using wavelet transform to eliminate noise, and forming a spatio-temporal aligned data set through normalization processing.

[0016] Optionally, when constructing the dynamic response model, it further includes: training the dynamic response model;

[0017] The process of training the dynamic response model specifically includes:

[0018] Construct a training data set based on the key data; the training data set includes key factors and corresponding label data;

[0019] Construct a pre-training network;

[0020] Input the training data set into the pre-training network, with the goal of minimizing the loss between the network output and the label data, and perform training according to the gradient descent strategy, and determine the trained network as the final dynamic response model.

[0021] Optionally, constructing a training data set based on the key data specifically includes:

[0022] Based on the key data, extract core experiment data, well logging interpretation data and production dynamic data;

[0023] Normalize the core experiment data, the logging interpretation data, and the production performance data through a feature matrix to construct a preliminary data set with spatio-temporal correlation information;

[0024] Perform annotation based on the preliminary data set to obtain a training data set composed of key factors and corresponding label data.

[0025] Optionally, the core experiment data is micron-level data, the logging interpretation data is decimeter-level data, and the production performance data is kilometer-level data.

[0026] The present invention also provides a reservoir seepage characteristic evaluation system, including:

[0027] A data acquisition unit for acquiring key data of a target reservoir; the key data includes core CT scan data, historical pressure and production data of production wells, and tracer monitoring data;

[0028] A model construction unit for constructing a dynamic response model based on the fluid potential energy theory and the key data; the dynamic response model includes an improved deep forest mechanism algorithm and an attention mechanism weighted algorithm; wherein, the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture;

[0029] A seepage evaluation unit for classifying seepage capacity by using the dynamic response model and performing three-dimensional visualization simulation of seepage channels according to the classification results.

[0030] According to the specific embodiments provided by the present invention, the following technical effects are disclosed:

[0031] The present invention discloses a reservoir seepage characteristic evaluation method and system. The method includes acquiring key data of a target reservoir; the key data includes core CT scan data, historical pressure and production data of production wells, and tracer monitoring data; constructing a dynamic response model based on the fluid potential energy theory and the key data; the dynamic response model includes an improved deep forest mechanism algorithm and an attention mechanism weighted algorithm; wherein, the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture; classifying seepage capacity by using the dynamic response model and performing three-dimensional visualization simulation of seepage channels according to the classification results. The present invention can break through the limitations of traditional methods and provide technical support for the efficient development of complex reservoirs. Description of the Drawings

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0033] Figure 1 It is a schematic flow diagram of the method for evaluating the reservoir seepage characteristics of the present invention. Detailed implementation manners

[0034] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0035] The purpose of the present invention is to provide a method and system for evaluating reservoir seepage characteristics, which can break through the limitations of traditional methods and provide technical support for the efficient development of complex reservoirs.

[0036] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the drawings and specific implementation manners.

[0037] As Figure 1 shown, the present invention provides a method for evaluating reservoir seepage characteristics. By multi-scale fusion of dynamic production data and static geological data, and three-dimensional visualization of seepage capacity classification and dominant seepage channels based on an improved deep forest algorithm, the evaluation of reservoir seepage characteristics is realized. It mainly includes:

[0038] Step 100: Obtain key data of the target reservoir; the key data includes core CT scan data, historical pressure and production data of production wells, and tracer monitoring data.

[0039] Step 200: Construct a dynamic response model based on the fluid potential energy theory and the key data; the dynamic response model includes an improved deep forest mechanism algorithm and an attention mechanism weighted algorithm; among them, the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture.

[0040] Step 300: Use the dynamic response model to classify the seepage capacity and perform three-dimensional visualization simulation of the seepage channels according to the classification results.

[0041] As a specific implementation manner, the dynamic response model is specifically expressed as:

[0042]

[0043] Among them, λ eff represents the equivalent flow coefficient, β represents the inertia factor, and K ro (S w ) represents the relative permeability of the oil phase, and S w represents the current water saturation, μ o represents the viscosity of the crude oil, ρ o represents the density of the crude oil, and v represents the seepage velocity. Among them, the inertia factor is obtained by fitting the pressure difference - flow velocity curve through the high - speed core displacement experiment.

[0044] As a specific implementation manner, before constructing the dynamic response model, it further includes: pre - processing the key data; the pre - processing process is: using wavelet transform to eliminate noise and forming a spatio - temporal aligned data set through normalization processing.

[0045] As a specific implementation manner, when constructing the dynamic response model, it further includes: training the dynamic response model, and the process specifically includes:

[0046] Constructing a training data set based on the key data; the training data set includes key factors and corresponding label data; constructing a pre - trained network; inputting the training data set into the pre - trained network, aiming at minimizing the loss between the network output and the label data, and training according to the gradient descent strategy, and determining the trained network as the final dynamic response model.

[0047] As a specific implementation manner, constructing a training data set based on the key data specifically includes:

[0048] Based on the key data, extracting core experiment data, well logging interpretation data, and production dynamic data; normalizing the core experiment data, the well logging interpretation data, and the production dynamic data through a feature matrix to construct a preliminary data set with spatio - temporal correlation information; performing annotation based on the preliminary data set to obtain a training data set composed of key factors and corresponding label data. Among them, the core experiment data is in the micrometer - level, the well logging interpretation data is in the decimeter - level, and the production dynamic data is in the kilometer - level.

[0049] Based on the above technical solutions, the following specific embodiments are provided.

[0050] In the process of oil and gas field development, accurately evaluating the reservoir seepage capacity and visualizing the seepage channels are of great significance for optimizing the exploitation strategy and enhancing the recovery rate. However, traditional methods often rely on a single data source and are difficult to comprehensively reflect the dynamic characteristics of the reservoir. This embodiment proposes a method for constructing a dynamic response model based on multi-source data fusion, aiming to achieve accurate classification and three-dimensional visualization simulation of the reservoir seepage capacity.

[0051] The specific implementation steps include:

[0052] Step 100: Obtain the key data of the target reservoir

[0053] Core CT scan data: Obtain the three-dimensional structure information of the target reservoir core through high-precision CT scanning technology, and analyze parameters such as porosity and permeability.

[0054] Historical pressure and production data of production wells: Collect the historical pressure and production data of the production wells in the target reservoir, and analyze the dynamic change characteristics of the reservoir.

[0055] Tracer monitoring data: Inject tracers into the reservoir, monitor the diffusion of tracers in the reservoir, and evaluate the reservoir connectivity.

[0056] Step 200: Data preprocessing

[0057] Noise elimination: Use wavelet transform technology to eliminate noise from the core CT scan data, historical pressure and production data of production wells, and tracer monitoring data to improve the data quality.

[0058] Standardization processing: Perform standardization processing on the data after noise elimination to form a spatio-temporal aligned data set, providing a unified data basis for subsequent model construction.

[0059] Step 300: Construct a dynamic response model

[0060] Model architecture design: Based on the fluid potential energy theory, design a dynamic response model that includes an improved deep forest mechanism algorithm (using the DF-ResNet network architecture) and an attention mechanism weighted algorithm.

[0061] Determination of model parameters: Equivalent flow coefficient (λeff): Determined according to the reservoir characteristic parameters, Inertial factor (β): Obtained by fitting the pressure difference - flow velocity curve through high-speed core displacement experiments. Oil relative permeability (Kro(Sw)), current water saturation (Sw), crude oil viscosity (μo), crude oil density (ρo), seepage velocity (v): Calculated or obtained based on well logging interpretation data and production dynamic data.

[0062] Step 400: Model training

[0063] Construct a training data set:

[0064] Extract core experimental data (micrometer level), logging interpretation data (decimeter level), and production dynamic data (kilometer level). Normalize the above data through a feature matrix to construct a preliminary data set with spatio-temporal correlation information. Based on the preliminary data set, perform annotation to obtain a training data set composed of key factors and corresponding label data.

[0065] Construct a pre-trained network: Use a deep learning framework to construct a pre-trained network, and the network structure includes an improved deep forest mechanism algorithm and an attention mechanism weighted algorithm.

[0066] Model training: Input the training data set into the pre-trained network, aim to minimize the loss between the network output and the label data, and perform training according to the gradient descent strategy until the model converges.

[0067] Step 500: Classification of seepage capacity and three-dimensional visualization simulation

[0068] Classification of seepage capacity: Use the trained dynamic response model to classify the seepage capacity of the target reservoir and evaluate the seepage performance of different regions.

[0069] Three-dimensional visualization simulation: Based on the classification results of seepage capacity, use three-dimensional visualization technology to simulate the spatial distribution of reservoir seepage channels and intuitively display the seepage characteristics of the reservoir.

[0070] Therefore, through the method of this embodiment, a dynamic response model based on multi-source data fusion is successfully constructed, realizing the accurate classification of the seepage capacity of the target reservoir and three-dimensional visualization simulation. This method not only improves the accuracy of reservoir seepage capacity evaluation but also provides strong support for the optimization of oil and gas field development strategies. At the same time, through three-dimensional visualization simulation, the spatial distribution characteristics of reservoir seepage channels are intuitively displayed, providing an important basis for subsequent injection-production adjustment, well pattern optimization, etc.

[0071] As can be seen from this embodiment, the following beneficial effects exist:

[0072] This solution integrates various data sources such as core CT scan data, historical pressure and production data of production wells, and tracer monitoring data to construct a dynamic response model, realizing the accurate evaluation and visualization display of reservoir seepage capacity. In the future, more data source fusion methods can be further explored, the model architecture and algorithms can be optimized, and the model prediction accuracy and generalization ability can be improved to provide more scientific and effective technical support for oil and gas field development.

[0073] In addition, the present invention also provides a reservoir seepage characteristic evaluation system, including:

[0074] A data acquisition unit for obtaining key data of a target reservoir; the key data includes core CT scan data, historical pressure and production data of production wells, and tracer monitoring data;

[0075] A model construction unit for constructing a dynamic response model based on the fluid potential energy theory and the key data; the dynamic response model includes an improved deep forest mechanism algorithm and an attention mechanism weighted algorithm; wherein, the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture;

[0076] A seepage evaluation unit for classifying seepage capacity by using the dynamic response model and performing three-dimensional visualization simulation on seepage channels according to the classification results.

[0077] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same and similar parts among the various embodiments, reference can be made to each other.

[0078] Specific examples are used in this article to elaborate on the principles and implementation manners of the present invention. The descriptions of the above embodiments are only used to help understand the core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for evaluating reservoir seepage characteristics, characterized in that, Including: Obtaining key data of the target reservoir; the key data includes core CT scan data, historical pressure and production data of production wells, and tracer monitoring data; Constructing a dynamic response model based on the fluid potential energy theory and the key data; the dynamic response model includes an improved deep forest mechanism algorithm and an attention mechanism weighted algorithm; wherein, the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture; Using the dynamic response model to classify the seepage capacity and performing three-dimensional visualization simulation of the seepage channel according to the classification result.

2. The reservoir seepage characteristic evaluation method according to claim 1, wherein The dynamic response model is specifically expressed as: Among them, λ eff represents the equivalent flow coefficient, β represents the inertia factor, K ro (S w ) represents the relative permeability of the oil phase, S w represents the current water saturation, μ o represents the viscosity of the crude oil, ρ o represents the density of the crude oil, and v represents the seepage velocity.

3. The reservoir seepage characteristic evaluation method according to claim 2, wherein The inertia factor in the dynamic response model is obtained by fitting the pressure difference-flow velocity curve through a core high-speed displacement experiment.

4. The reservoir seepage characteristic evaluation method according to claim 1, wherein Before constructing the dynamic response model, it also includes: preprocessing the key data; the preprocessing process is: using wavelet transform to eliminate noise and forming a spatio-temporally aligned data set through normalization processing.

5. The reservoir seepage characteristics evaluation method according to claim 1, characterized in that When constructing the dynamic response model, it also includes: training the dynamic response model; The process of training the dynamic response model specifically includes: Constructing a training data set based on the key data; the training data set includes key factors and corresponding label data; Constructing a pre-trained network; Inputting the training data set into the pre-trained network, aiming at minimizing the loss between the network output and the label data, and training according to the gradient descent strategy, and determining the trained network as the final dynamic response model.

6. The reservoir seepage characteristic evaluation method according to claim 5, characterized in that Constructing a training data set based on the key data specifically includes: Based on the key data, extracting core experiment data, well logging interpretation data, and production dynamic data; Normalizing the core experiment data, the well logging interpretation data, and the production dynamic data through a feature matrix to construct a preliminary data set with spatio-temporal correlation information; Performing annotation based on the preliminary data set to obtain a training data set composed of key factors and corresponding label data.

7. The reservoir seepage characteristic evaluation method according to claim 6, wherein The core experiment data is micron-level data, the well logging interpretation data is decimeter-level data, and the production dynamic data is kilometer-level data.

8. A reservoir seepage characteristic evaluation system, characterized in that Including: A data acquisition unit for obtaining key data of the target reservoir; the key data includes core CT scan data, historical pressure and production data of production wells, and tracer monitoring data; A model construction unit for constructing a dynamic response model based on the fluid potential energy theory and the key data; the dynamic response model includes an improved deep forest mechanism algorithm and an attention mechanism weighted algorithm; wherein, the improved deep forest mechanism algorithm adopts a DF-ResNet network architecture; A seepage evaluation unit for classifying the seepage capacity using the dynamic response model and performing three-dimensional visualization simulation of the seepage channel according to the classification result.

Images