A cross-scale numerical simulation method based on micro-macro mixing

Abstract

The application discloses a cross-scale numerical simulation method based on micro-macro mixing. The method comprises the following steps: S1, realizing downscaling generation from a macroscopic reservoir scale grid to a micro-pore scale grid; S2, establishing a dual grid system consistent with the size of the macroscopic reservoir scale grid; S3, considering the boundary condition, decomposing pressure and flow rate into a plurality of base functions to construct a local micro-flow equation; S4, establishing a quantitative conversion relationship between micro-pore flow characteristics and macro-flow parameters; and S5, realizing cross-scale conversion of the macroscopic scale and the micro-pore scale. The method has the beneficial effects that the simulation precision is improved while the calculation efficiency is ensured, local heterogeneity is accurately captured, and remaining oil is accurately identified.

Description

Technical Field

[0001] This invention relates to the field of reservoir numerical simulation technology, and in particular to a cross-scale numerical simulation method based on micro-macro-scale mixing. Background Technology

[0002] In the field of oil and gas field development, reservoir numerical simulation is a core technical means to predict reservoir development dynamics and optimize development plans. Oil reservoirs are typical multi-scale complex systems, involving reservoir areas at the kilometer level on a macroscopic level and pore structures at the micrometer level on a microscopic level. The flow behavior of fluids in oil reservoirs is simultaneously influenced by both macroscopic reservoir properties and microscopic pore topology.

[0003] Traditional reservoir simulation methods are mainly divided into two categories: one is macroscale simulation, which mainly uses macroscale grids of 10 to 50 meters to perform numerical simulation calculations. This type of method simplifies and averages the motion equations in porous media to quickly calculate the macroscopic production trend of the reservoir (such as daily oil production and water cut changes). However, because it uses parameters such as permeability, porosity, relative permeability, and capillary force to equivalently characterize complex physical phenomena, it ignores the influence of microscopic pore structure on seepage to some extent, and masks the non-uniformity of flow inside porous media. This results in insufficient prediction accuracy for microscopic sensitive issues such as remaining oil distribution and water drive front advancement, making it difficult to meet the needs of precise development. Another type is microscale simulation, which mainly uses core CT scans to reconstruct or randomly generate micrometer-scale pore network models, and employs direct numerical simulation (such as the Lattice Boltzmann Method, LBM) to describe microscopic seepage. Although this type of method can accurately depict the details of fluid flow within pores, the computational load is extremely large—for a 10-meter-scale macroscopic reservoir grid, a full microscopic simulation would require constructing millions to hundreds of millions of pore grids, and a single simulation could take weeks or even months, resulting in extremely low computational efficiency and making it unsuitable for large-scale reservoir simulation in actual oilfields.

[0004] Given that existing technologies cannot balance the accuracy and efficiency of cross-scale simulation, there is an urgent need for a hybrid micro-macro-scale numerical simulation method that can achieve precise matching of macro and micro grids, reduce coupling errors, and balance computational efficiency, in order to meet the production needs of precise development of water-driven sandstone reservoirs. Summary of the Invention

[0005] The purpose of this invention is to provide a cross-scale numerical simulation method based on micro-macro hybridization. By constructing a micro-pore scale grid and its dual grid, and using basis functions to equate the physical effects at the micro-pore scale to the macro-scale, this method improves simulation accuracy while ensuring computational efficiency, accurately captures local heterogeneity, and thus accurately identifies remaining oil, providing a reference for optimizing reservoir development schemes.

[0006] To achieve the above objectives, the present invention adopts the following technical solution, comprising the following steps: S1. Based on the data volume of the numerical simulation software Eclipse, extract the attribute data at the reservoir scale of the target oilfield; and use a pore topology generation algorithm based on conditional generative adversarial networks, combined with core CT scan data, to achieve downscaling from macroscopic reservoir-scale grids to microscopic pore-scale grids. S2. Based on the micro-pore-scale grid generated by downscaling, a dual grid system with the same size as the macro-reservoir-scale grid is established, with the center point of the macro-reservoir-scale grid as the corner point of the dual grid and the boundary of the dual grid being orthogonal to the boundary of the macro-reservoir-scale grid. S3. Combining the established dual grid system, considering the boundary conditions, the local micro-flow equations are constructed by decomposing pressure and velocity into multiple basis functions. S4. Based on the basis functions, the micro-flow at the pore scale is equivalent to the macro-scale using conductivity, and a quantitative conversion relationship between micro-flow characteristics and macro-flow parameters is established. S5. Taking the error between the flow calculated at the macroscale boundary and the flow inverted at the microscale boundary as less than 5% as the convergence condition for the iteration, the boundary parameters obtained by iteratively solving the correction function based on the macroscale finite volume method are used to reverse downscale and correct the boundary conditions of the basis function at the microscale pore scale. After multiple rounds of iterative optimization, the cross-scale conversion between the macroscale and microscale pore scales is realized.

[0007] Preferably, in step S1, the attribute data includes porosity, permeability, oil saturation, crude oil viscosity, water permeability ratio, grid water drive velocity, and grid formation pressure; the core CT scan data includes the porosity, permeability, pore radius, sphericity, and pore throat radius ratio of the formation rocks.

[0008] Preferably, in step S1, the pore topology generation algorithm based on conditional generative adversarial networks is to establish a micro-pore scale grid model that satisfies the average porosity error of the corresponding macro-unit ≤5% by inputting core CT scan data and attribute data of the target oilfield, setting the network structure and loss function, and the micro-pore scale grid model is composed of multiple micro-pore scale grids, which is a downscaling generation from macro-reservoir scale grids to micro-pore scale grids.

[0009] Preferably, in step S1, the size of the macroscopic reservoir-scale grid is on the order of 10 to 50 meters, and the size of the microscopic pore-scale grid is on the order of 1 to 100 micrometers.

[0010] Preferably, in step S2, during the process of establishing a dual grid system with the same size as the macroscopic reservoir-scale grid, constraints are also applied. The constraints are: by giving the local conditions that the pressure and velocity within the dual grid are linearly determined by interpolation of the macroscopic pressure and velocity fields, the normal velocity is 0, and the velocity and pressure gradient at the rock particle boundary are 0, and the cross-scale flow at the dual grid boundary satisfies the continuity of the normal velocity, the flow error is reduced by 12% compared to the non-orthogonal grid, and at the same time, the solid boundaries such as rock particles satisfy the no-slip condition.

[0011] Preferably, in step S3, the linear combination equation of the basis functions is expressed as follows: (1) (2) In equations (1) and (2): p is the pressure value at any position within the dual grid, Pa, reflecting the magnitude of the pressure exerted on the fluid in the pores, and is one of the key factors driving fluid flow; v is the vector velocity at any position within the dual grid, m / s, whose direction and magnitude describe the flow state of the fluid in the pores; φi is the pressure basis function, dimensionless, characterizing the contribution pattern of the pressure at the i-th node to the pressure distribution within the entire dual grid, reflecting the influence of the micropore structure on pressure propagation; Ψi is the velocity basis function, m / (s·Pa), constructed based on Darcy's law, characterizing the influence of the pressure change at the i-th node on the velocity distribution within the entire dual grid, reflecting the influence of the micropore structure on the fluid flow velocity; P i Let be the pressure value, in Pa, for the i-th macroscopic reservoir-scale grid node.

[0012] Preferably, in step S4, the inter-grid conductivity at the macroscopic reservoir scale is obtained by integrating the basis function space, and its expression is: (3) In the formula: T ij m represents the conductivity between the interfaces of grids i and j at the macroscopic reservoir scale. 3 / (Pa·s) represents the magnitude of fluid conductivity between grids at the macroscopic reservoir scale; K is the average permeability of the micropore domain, calculated from the microscopic topology generated by S1, m 2 1mD=9.87×10 -16 m 2 μ is the fluid viscosity. This represents the interface between two adjacent grid cells i and j at the macroscopic reservoir scale; denoted as the gradient of the pressure basis function, representing its spatial variation trend, dimensionless, and reflecting the influence of micropore structure on pressure propagation; denoted as n, the unit normal vector of the dual mesh boundary, dimensionless, used to determine the integration direction; denoted as ds, the infinitesimal area of ​​the dual mesh boundary element; and denoted as m. 2 .

[0013] Preferably, in step S5, the finite volume method discretizes the continuous reservoir space, dividing the entire reservoir region into multiple interconnected control volumes. For each control volume (Vx, Vy), the Gaussian formula is applied to convert the area integral into a volume integral, resulting in the discretized equation expression: (4) In the formula: N(i) is the control volume V adjacent to the control volume V. i The set that constitutes; P i P j Control volume V i V j Pressure value at the center node, Pa; ΔV i , ΔV j Control volume V i V j The volume, m 3 Q i To control the total injection and production volume, m 3 / s.

[0014] Preferably, in step S5, the finite volume method is used to iteratively solve the correction function equation to obtain the pressure correction term and velocity correction term after macroscopic reservoir-scale grid correction. The correction term is then superimposed onto the original grid, and the expressions for the corrected pressure and velocity are as follows: (5) (6) In the formula: P knew For the corrected macroscopic reservoir-scale grid pressure, Pa; v knew The corrected macroscopic reservoir-scale grid velocity is in m / s; P k The pressure (Pa) is calculated based on the discretized reservoir space, taking into account factors such as conductivity. k The velocity is based on the discretized reservoir space, taking into account factors such as conductivity, in m / s; λ is a correction coefficient, determined by core experiment fitting, and is dimensionless. This represents the flow error, expressed as a decimal.

[0015] The beneficial effects of this invention are: by constructing a micro-pore-scale grid and its dual grid, and using basis functions to equate the physical effects at the micro-pore scale to the macro-scale, the simulation accuracy is improved while ensuring computational efficiency, accurately capturing local heterogeneity, thereby accurately identifying remaining oil and providing a reference for optimizing reservoir development schemes. Attached Figure Description

[0016] Figure 1 This is a flowchart of a cross-scale numerical simulation method based on micro-macro hybridization according to the present invention.

[0017] Figure 2 This is a schematic diagram of the dual mesh in this invention.

[0018] Figure 3 This is a schematic diagram of the pressure basis function in this invention.

[0019] Figure 4 This is a schematic diagram of the velocity basis function in this invention.

[0020] Figure 5 This is a comparison chart of the fitting curves of traditional numerical simulation methods and the present invention.

[0021] Figure 6 This is a planar residual oil distribution map obtained using traditional numerical simulation methods.

[0022] Figure 7 This is a planar residual oil distribution map obtained by the present invention. Detailed Implementation

[0023] The invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0024] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not exclude the presence or addition of one or more other elements or combinations thereof.

[0025] Example like Figure 1-7 As shown, the present invention provides a cross-scale numerical simulation method based on micro-macro hybridization, comprising: S1. Based on the data volume of the numerical simulation software Eclipse, extract the attribute data at the reservoir scale of the target oilfield; and use a pore topology generation algorithm based on conditional generative adversarial networks, combined with core CT scan data, to achieve downscaling from macroscopic reservoir-scale grids to microscopic pore-scale grids. The attribute data includes porosity, permeability, oil saturation, crude oil viscosity, water permeability ratio, grid water drive velocity, and grid formation pressure. The core CT scan data includes the porosity, permeability, pore radius, sphericity, and pore throat radius ratio of the formation rocks; The pore topology generation algorithm based on conditional generative adversarial networks is to input core CT scan data and target oilfield attribute data, set the network structure and loss function, and establish a micro-pore scale grid model that satisfies the average porosity error of the corresponding macro-unit ≤5%. The micro-pore scale grid model is composed of multiple micro-pore scale grids, which is a downscaling generation from macro-reservoir scale grids to micro-pore scale grids. The size of the macroscopic reservoir-scale grid is 10 to 50 meters, and the size of the microscopic pore-scale grid is 1 to 100 micrometers.

[0026] S2. Based on the micro-pore-scale grid generated by downscaling, a dual grid system with the same size as the macro-reservoir-scale grid is established, with the center point of the macro-reservoir-scale grid as the corner point of the dual grid and the boundary of the dual grid being orthogonal to the boundary of the macro-reservoir-scale grid. In the process of establishing a dual grid system with the same size as the macroscopic reservoir-scale grid, constraints were also applied. The constraints are as follows: the pressure and velocity within the dual grid are linearly varied by interpolation of the macroscopic pressure and velocity fields, the normal velocity is 0, and the velocity and pressure gradient at the rock particle boundary are 0. Furthermore, the cross-scale flow at the dual grid boundary satisfies the continuity of the normal velocity, reducing the flow error by 12% compared to the non-orthogonal grid. At the same time, the solid boundaries such as rock particles satisfy the no-slip condition.

[0027] S3. Combining the established dual grid system, considering the boundary conditions, the local micro-flow equations are constructed by decomposing pressure and velocity into multiple basis functions. In step S3, the linear combination equation of the basis functions is expressed as follows: (1) (2) In equations (1) and (2): p is the pressure value at any position within the dual grid, Pa, reflecting the magnitude of the pressure exerted on the fluid in the pores, and is one of the key factors driving fluid flow; v is the vector velocity at any position within the dual grid, m / s, whose direction and magnitude describe the flow state of the fluid in the pores; φi is the pressure basis function, dimensionless, characterizing the contribution pattern of the pressure at the i-th node to the pressure distribution within the entire dual grid, reflecting the influence of the micropore structure on pressure propagation; Ψi is the velocity basis function, m / (s·Pa), constructed based on Darcy's law, characterizing the influence of the pressure change at the i-th node on the velocity distribution within the entire dual grid, reflecting the influence of the micropore structure on the fluid flow velocity; P i Let be the pressure value, in Pa, for the i-th macroscopic reservoir-scale grid node.

[0028] S4. Based on the basis functions, the micro-flow at the pore scale is equivalent to the macro-scale using conductivity, and a quantitative conversion relationship between micro-flow characteristics and macro-flow parameters is established. The inter-grid conductivity at the macro-reservoir scale is obtained through integral of the basis function space, and its expression is: (3) In the formula: T ij m represents the conductivity between the interfaces of grids i and j at the macroscopic reservoir scale. 3 / (Pa·s) represents the magnitude of fluid conductivity between grids at the macroscopic reservoir scale; K is the average permeability of the micropore domain, calculated from the microscopic topology generated by S1, m 2 1mD=9.87×10 -16 m 2 μ is the fluid viscosity. This represents the interface between two adjacent grid cells i and j at the macroscopic reservoir scale; denoted as the gradient of the pressure basis function, representing its spatial variation trend, dimensionless, and reflecting the influence of micropore structure on pressure propagation; denoted as n, the unit normal vector of the dual mesh boundary, dimensionless, used to determine the integration direction; denoted as ds, the infinitesimal area of ​​the dual mesh boundary element; and denoted as m. 2 .

[0029] S5. The convergence condition for the iterative process is that the error between the flow calculated at the macroscale boundary and the flow inverted at the microscale boundary is less than 5%. The boundary parameters are obtained by iteratively solving the correction function based on the macroscale finite volume method. The boundary conditions of the microscale basis function are corrected by reverse downscaling. After multiple rounds of iterative optimization, the cross-scale conversion between the macroscale and the microscale is realized. The finite volume method discretizes the continuous reservoir space, dividing the entire reservoir region into multiple interconnected control volumes. For each control volume (Vx, Vy), the Gaussian formula is applied to convert the surface integral into a volume integral, resulting in the discretized equation expression: (4) In the formula: N(i) is the control volume V adjacent to the control volume V. i The set that constitutes; P i P j Control volume V i V j Pressure value at the center node, Pa; ΔV i , ΔV j Control volume V i V j The volume, m 3 Q i To control the total injection and production volume, m3 / s; The correction function equation is solved iteratively using the finite volume method to obtain the pressure and velocity correction terms after macroscopic reservoir-scale grid correction. These correction terms are then superimposed onto the original grid, resulting in the following expressions for the corrected pressure and velocity: (5) (6) In the formula: P knew For the corrected macroscopic reservoir-scale grid pressure, Pa; v knew The corrected macroscopic reservoir-scale grid velocity is in m / s; P k The pressure (Pa) is calculated based on the discretized reservoir space, taking into account factors such as conductivity. k The velocity is based on the discretized reservoir space, taking into account factors such as conductivity, in m / s; λ is a correction coefficient, determined by core experiment fitting, and is dimensionless. This represents the flow error, expressed as a decimal.

[0030] In summary, this invention provides a cross-scale numerical simulation method based on micro-macro hybridization. By constructing a micro-pore-scale grid and its dual grid, and using basis functions to equate the physical effects at the micro-pore scale to the macro-scale, it improves simulation accuracy while ensuring computational efficiency, accurately captures local heterogeneity, and thus precisely identifies remaining oil, providing a reference for optimizing reservoir development schemes.

[0031] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A cross-scale numerical simulation method based on micro-macro hybridization, characterized in that, Includes the following steps: S1. Based on the data volume of the numerical simulation software Eclipse, extract the attribute data at the reservoir scale of the target oilfield; and use a pore topology generation algorithm based on conditional generative adversarial networks, combined with core CT scan data, to achieve downscaling from macroscopic reservoir-scale grids to microscopic pore-scale grids. S2. Based on the micro-pore-scale grid generated by downscaling, a dual grid system with the same size as the macro-reservoir-scale grid is established, with the center point of the macro-reservoir-scale grid as the corner point of the dual grid and the boundary of the dual grid being orthogonal to the boundary of the macro-reservoir-scale grid. S3. Combining the established dual grid system, considering the boundary conditions, the local micro-flow equations are constructed by decomposing pressure and velocity into multiple basis functions. S4. Based on the basis functions, the micro-flow at the pore scale is equivalent to the macro-scale using conductivity, and a quantitative conversion relationship between micro-flow characteristics and macro-flow parameters is established. S5. Taking the error between the flow calculated at the macroscale boundary and the flow inverted at the microscale boundary as less than 5% as the convergence condition for the iteration, the boundary parameters obtained by iteratively solving the correction function based on the macroscale finite volume method are used to reverse downscale and correct the boundary conditions of the basis function at the microscale pore scale. After multiple rounds of iterative optimization, the cross-scale conversion between the macroscale and microscale pore scales is realized.

2. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 1, characterized in that: In step S1, the attribute data includes porosity, permeability, oil saturation, crude oil viscosity, water permeability ratio, grid water drive velocity, and grid formation pressure; the core CT scan data includes the porosity, permeability, pore radius, sphericity, and pore throat radius ratio of the formation rocks.

3. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 2, characterized in that: In step S1, the pore topology generation algorithm based on conditional generative adversarial networks (GANs) involves inputting core CT scan data and target oilfield attribute data, setting the network structure and loss function, and establishing a micro-pore-scale grid model that satisfies the requirement that the average porosity error of the corresponding macro-units be ≤5%. The micro-pore-scale grid model consists of multiple micro-pore-scale grids. Downscaling generation from macroscopic reservoir-scale grids to microscopic pore-scale grids.

4. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 3, characterized in that: In step S1, the size of the macroscopic reservoir-scale grid is on the order of 10 to 50 meters, and the size of the microscopic pore-scale grid is on the order of 1 to 100 micrometers.

5. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 1, characterized in that, In step S2, during the process of establishing a dual grid system with the same size as the macroscopic reservoir-scale grid, constraints are also applied. The constraints are: by giving the local conditions that the pressure and velocity within the dual grid are linearly varied by interpolation of the macroscopic pressure and velocity fields, the normal velocity is 0, and the velocity and pressure gradient at the rock particle boundary are 0, and the cross-scale flow at the dual grid boundary satisfies the continuity of the normal velocity, the flow error is reduced by 12% compared to the non-orthogonal grid, and the solid boundaries such as rock particles satisfy the no-slip condition.

6. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 1, characterized in that: In step S3, the linear combination equation of the basis functions is expressed as follows: （1） （2） In equations (1) and (2): p is the pressure value at any position within the dual grid, Pa, reflecting the magnitude of the pressure exerted on the fluid in the pores, and is one of the key factors driving fluid flow; v is the vector velocity at any position within the dual grid, m / s, whose direction and magnitude describe the flow state of the fluid in the pores; φi is the pressure basis function, dimensionless, characterizing the contribution pattern of the pressure at the i-th node to the pressure distribution within the entire dual grid, reflecting the influence of the micropore structure on pressure propagation; Ψi is the velocity basis function, m / (s·Pa), constructed based on Darcy's law, characterizing the influence of the pressure change at the i-th node on the velocity distribution within the entire dual grid, reflecting the influence of the micropore structure on the fluid flow velocity; P i Let be the pressure value, in Pa, for the i-th macroscopic reservoir-scale grid node.

7. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 1, characterized in that: In step S4, the inter-grid conductivity at the macroscopic reservoir scale is obtained through basis function space integration, and its expression is: （3） In the formula: T ij m represents the conductivity between the interfaces of grids i and j at the macroscopic reservoir scale. 3 / (Pa·s) represents the magnitude of fluid conductivity between grids at the macroscopic reservoir scale; K is the average permeability of the micropore domain, calculated from the microscopic topology generated by S1, m 2 1mD=9.87×10 -16 m 2 μ is the fluid viscosity. This represents the interface between two adjacent grid cells i and j at the macroscopic reservoir scale; denoted as the gradient of the pressure basis function, representing its spatial variation trend, dimensionless, and reflecting the influence of micropore structure on pressure propagation; denoted as n, the unit normal vector of the dual mesh boundary, dimensionless, used to determine the integration direction; denoted as ds, the infinitesimal area of ​​the dual mesh boundary element; and denoted as m. 2 .

8. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 1, characterized in that: In step S5, the finite volume method discretizes the continuous reservoir space, dividing the entire reservoir region into multiple interconnected control volumes. For each control volume (Vx, Vy), the Gaussian formula is applied to convert the area integral into a volume integral, resulting in the discretized equation expression: （4） In the formula: N(i) is the control volume V adjacent to the control volume V. i The set that constitutes; P i P j Control volume V i V j Pressure value at the center node, Pa; ΔV i , ΔV j Control volume V i V j The volume, m 3 Q i To control the total injection and production volume, m 3 / s.

9. The cross-scale numerical simulation method based on micro-macro hybridity according to claim 1, characterized in that: In step S5, the finite volume method is used to iteratively solve the correction function equation, obtaining the pressure and velocity correction terms after macroscopic reservoir-scale grid correction. These correction terms are then superimposed onto the original grid, resulting in the following expressions for the corrected pressure and velocity: （5） （6） In the formula: P knew For the corrected macroscopic reservoir-scale grid pressure, Pa; v knew The corrected macroscopic reservoir-scale grid velocity is in m / s; P k The pressure (Pa) is calculated based on the discretized reservoir space, taking into account factors such as conductivity. k The velocity, in m / s, is based on the discretized reservoir space and takes into account factors such as conductivity. λ is a correction coefficient, determined by fitting through core experiments, and is dimensionless; This represents the flow error, expressed as a decimal.

Images