CA2518240A1 - Multi-scale finite-volume method for use in subsurface flow simulation - Google Patents
Multi-scale finite-volume method for use in subsurface flow simulation Download PDFInfo
- Publication number
- CA2518240A1 CA2518240A1 CA002518240A CA2518240A CA2518240A1 CA 2518240 A1 CA2518240 A1 CA 2518240A1 CA 002518240 A CA002518240 A CA 002518240A CA 2518240 A CA2518240 A CA 2518240A CA 2518240 A1 CA2518240 A1 CA 2518240A1
- Authority
- CA
- Canada
- Prior art keywords
- fine
- coarse
- cells
- scale
- dual
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V11/00—Prospecting or detecting by methods combining techniques covered by two or more of main groups G01V1/00 - G01V9/00
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
Abstract
A multi-scale finite-volume (MSFV) method to solve elliptic problems with a plurality of spatial scales arising from single or multi-phase flows in porous media is provided. Two sets of locally computed basis functions are employed.
A first set of basis functions captures the small-scale heterogeneity of the underlying permeability field, and it is computed to construct the effective coarse-scale transmissibilities. A second set of bases functions is required to construct a conservative fine-scale velocity field. The method efficiently captures the effects of small scales on a coarse grid, is conservative, and treats tensor permeabilities correctly. The underlying idea is to construct transmissibilities that capture the local properties of a differential operator. This leads to a multi-point discretization scheme for a finite-volume solution algorithm. Transmissibilities for the MSFV method are preferably constructed only once as a preprocessing step and can be computed locally. Therefore, this step is well suited for massively parallel computers.
Furthermore, a conservative fine-scale velocity field can be constructed from a coarse-scale pressure solution which also satisfies the proper mass balance on the fine scale. A transport problem is ideally solved iteractively in two stages. In the first stage, a fine scale velocity field is obtained from solving a pressure equation. In the second stage, the transport problem is solved on the fine cells using the fine-scale velocity field. A solution may be computed on the coarse cells at an incremental time and properties, such as a mobility coefficient, may be generated for the fine cells at the incremental time. If a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
A first set of basis functions captures the small-scale heterogeneity of the underlying permeability field, and it is computed to construct the effective coarse-scale transmissibilities. A second set of bases functions is required to construct a conservative fine-scale velocity field. The method efficiently captures the effects of small scales on a coarse grid, is conservative, and treats tensor permeabilities correctly. The underlying idea is to construct transmissibilities that capture the local properties of a differential operator. This leads to a multi-point discretization scheme for a finite-volume solution algorithm. Transmissibilities for the MSFV method are preferably constructed only once as a preprocessing step and can be computed locally. Therefore, this step is well suited for massively parallel computers.
Furthermore, a conservative fine-scale velocity field can be constructed from a coarse-scale pressure solution which also satisfies the proper mass balance on the fine scale. A transport problem is ideally solved iteractively in two stages. In the first stage, a fine scale velocity field is obtained from solving a pressure equation. In the second stage, the transport problem is solved on the fine cells using the fine-scale velocity field. A solution may be computed on the coarse cells at an incremental time and properties, such as a mobility coefficient, may be generated for the fine cells at the incremental time. If a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
Description
2 1(!i SUESURFACE FLOI~ SIMULATIOtI~I
4 TECHi~ICAL FIEL~
6 The present invention relates generally to subsurface reservoir simulators, 7 and more particularly, to those simulators which use multi-scale physics to 8 simulate flow in an underground reservoir.
I3AC~ed~ROUND OF TFiE INVEf~TION
12 The level of detail available in reservoir description often exceeds the 13 computational capability of existing reservoir simulators. This resolution gap 14 is usually tackled by upscaling the fine-scale description to sizes that can be treated by a full-featured simulator. in upscaling, the original model is 16 coarsened using a computationally inexpensive process. In flow-based 17 methods, the process is based on single-phase flow. A simulation study is 18 then performed using the coarsened model. Upscaling methods such as 19 these have proven to be quite successful. However, it is not possible to have a priori estimates of the errors that are present when complex flow processes 21 are investigated using coarse models constructed via these simplified 22 settings.
24 Various fundamentally different multi-scale approaches for flow in porous media have been proposed to accommodate the fine-scale description 26 directly. As opposed to upscaling, the multi-scale approach targets the full 27 problem with the original resolution. The upscaling methodology is typically 28 based on resolving the length and time-scales of interest by maximizing local 29 operations. Arbogast et al. (T. Arbogast, Numerical subgrid upscaling of tw~
phase flow in porous media, Technical report, Texas Institute for 31 Computational and Applied Mathematics, The University of Texas at Austin, 32 1999, and T. Arbogast and S. L. Bryant, Numerical subgrid upscaling for _1_ 1 waterflood simulations, SPE 66375, 2001) presented a mixed finite-element 2 method where fine-scale effects are localized by a boundary condition 3 assumption at the coarse element boundaries. Then the small-scale influence 4~ is coupled with the coarse-scale effiects by numerical Greens functions.
Hou and Wu (T. Hou and X. H. Wu, A multiscale finite elen7ent method for elliptic 6 problems in composite materials and porous media, J. Comp. Phys., 7 134:169-189, 1997) employed a finite-element approach and constructed 8 specific basis functions which capture the small scales. Again, localization is 9 achieved by boundary condition assumptions for the coarse elements. To reduce the efFects of these boundary conditions, an oversampling technique 11 can be applied. Chen and Hou (Z. Chen and T. Y. Hou, A mixed finite 12 element method for elliptic problems with rapidly oscillating coefficients, Math.
13 Comput., June 2002) utilized these ideas in combination with a mixed 14 finite-element approach. Another approach by Beckie et al. (R. Beckie,.
A. A. Aldama, and E. F. Wood, Modeling the large-scale dynamics of 16 saturated groundwater flow using spatial filtering, Water Resources Research, 17 32:1269-1280, 1996) is based on large eddy simulation (LES) techniques 18 which are commonly used for turbulence modeling.
Lee et al. (S. H. Lee, L. J. Durlofsky, M. F. Lough, and W. H. Chen, Finite 21 difference simulation of geologically complex reservoirs with tensor 22 permeabilities, SPERE&E, pages 567-574, 1998) developed a flux-continuous 23 finite-difference (FCFD) scheme for 2D models. Lee et al. further developed a 24 method to address 3D models (S. H. Lee, H. Tchelepi, P. Jenny and L. Dechant, Implementation of a flux continuous finite-difference method for 26 stratigraphic, hexahedron grids, SPE Journal, September, pages 269-277, 27 2002). Jenny et al. (P. Jenny, C. Wolfsteiner, S. H. Lee and L. J.
Durlofsky, 28 Modeling flow in geometrically complex reservoirs using hexahedral 29 multi-block grids, SPE Journal, June, pages 149-157, 2002) later implemented this scheme in a multi-block simulator.
1 In light of the above modeling efforts, there is a need for a simulation method 2 which more efficiently captures the effects of small scales on a coarse grid.
3 Ideally, the method would be conservative and also treat tensor permeabilities 4 correctly. Further, preferably the reconstructed fine-scale solution would satisfy the proper mass balance on the fine-scale. The present invention 6 provides such a simulation method.
~~I~~GQa~~I'n ~~ ~~~ ~~~~~l~~l~~
A mufti-scale finite-volume (MSFV) approach is taught for solving elliptic or 11 parabolic problems such as those found in subsurface flow simulators.
12 Advanfiages of the present MSFV method are that it fits nicely into a 13 finite-volume framework, it allows for computing effective coarse-scale 14 transmissibiiities, treats tensor permeabilities properly, and is conservative at both the coarse and fine scales. The present method is computationally 16 efficient relative to reservoir simulation now in use and is well suited for 17 massive parallel computation. The present invention can be applied to 3D
18 unstructured grids and also to mufti-phase flow. Further, the reconstructed 19 fine-scale solution satisfies the proper mass balance on the fine-scale.
21 A mufti-scale approach is described which results in effective transmissibilities 22 for the coarse-scale problem. ~nce the transmissibilities are constructed, the 23 MSFV method uses a finite-volume scheme employing mufti-point stencils for 24 flux discretization. The approach is conservative and treats tensor permeabilities correctly. This method is easily applied using existing 26 finite-volume codes, and once the transmissibilities are computed, the method 27 is computationally very efficient. In computing the effective transmissibilities, 28 closure assumptions are employed.
A significant characteristic of the present mufti-scale method is that two sets 31 of basis functions are employed. A first set of dual basis functions is 32 computed to construct transmissibilities between coarse cells. A second set 1 of locally computed fine scale basis functions is utilized to reconstruct a 2 fine-scale velocity field from a coarse scale solution. This second set of 3 fine-scale basis functions is designed such that the reconstructed fine-scale 4. velocity solution is fully consistent with fibs transmissibilities.
Further, the solution satisfies the proper mass balance on the small scale.
7 The fVISF~I method may be used in modeling a subsurface reservoir. A fine 8 grid is first created defining a plurality of fns veils. A permeability field and 9 other fine scale properties are associated with the fine cells. ~dexfi, a coarse grid is created which defines a plurality of coarse cells having interfaces 11 between the coarse cells. The coarse cells are ideally aggregates of the fine 12 cells. A dual coarse grid is constructed defining a plurality of dual coarse 13 control volumes. The dual coarse control volumes are ideally also aggregates 14 of the fine cells. Boundaries surround the dual coarse control volumes.
16 Dual basis functions are then calculated on the dual coarse control volumes 17 by solving local elliptic or parabolic problems, preferably using boundary 13 conditions obtained from solving reduced problems along the interfaces of the 19 course cells. Fluxes, preferably integral fluxes, are then extracted across the interfaces of the coarse cells from the dual basis functions. These fluxes are 21 assembled to obtain effective transmissibilities between coarse cells of the 22 coarse cell grid. The transmissibilities can be used for coarse scale finite 23 volume calculations.
A fine scale velocity field may be established. A finite volume method is used 26 to calculate pressures in the coarse cells utilizing the transmissibilities 27 between cells. Fine scale basis functions are computed by solving local 23 elliptic or parabolic flow problems on the coarse cells and by utilizing fine 29 scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions. Finally, the fine-scale basis functions and the 31 corresponding coarse calf pressures are combined to extract the small scale 32 velocity field.
_q._ 1 A transport problem may be solved on the fine grid by using the small scale 2 velocity field. Ideally, the transport problem is solved iteratively in two stages.
3 In the first stage, a fine scale velocity field is obtained from solving a pressure 4 equation. In the second stage, the transport problem is solved on the fine cells using the fine-scale velocity field. A Schwart~ overlap technique can be 6 applied to solve the transport problem locally on each coarse cell with an 7 implicit: upwind scheme.
9 A solution may be computed on the coarse cells at an incremental time and properties, such as a mobility coefficient, may be generated for the fine cells 11 at the incremental time. If a predetermined condition is not met for all fine 12 cells inside a dual coarse control volume, then the dual and fine scale basis 13 functions in that dual coarse control volume are reconstructed, BRIEF DESCRIPTION OF THE DRAWINGS
17 These and other objects, features and advantages of the present invention 18 will become befiter understood with regard to the following description, 19 pending claims and accompanying drawings where:
21 F1G. 1 illustrates a coarse 2D grid of coarse cells with an overlying dual 22 coarse grid including a dual coarse control volume and an underlying fine grid 23 of fine cells;
FIG. 2 illustrates a coarse grid including nine adjacent coarse cells (bold solid 26 tines) with a corresponding overlying dual coarse grid (bold dashed lines) 27 including dual coarse control volumes and an underlying fine grid (thin dotted 28 lines) of fine cells;
FIG. 3 shows flux contributi~n e~~'-~ and y;f ~ due to the pressure in a particular 31 coarse cell 2;
1 FIG. 4 is a flowchart describing the overall steps used in a preferred 2 embodiment of a reservoir simulation which employs a multi-scale 3 finite-volume (MSFV) method made in accordance with this invention;
FIG. 5 is a flowchart further detailing steps used to determine transmissibilities 6 T between coarse cells;
8 FIG. 6 is a flow chart further describing steps used to construct a set of 9 fine-scale basis functions and to exiract a small scale velocity field;
11 FIG. 7 is a flowchart depicting coupling between pressure and the saturation 12 equations which utilize an implicit solution scheme and wherein IZ and ~
are 13 operators used to update total velocity and saturation, respectively, during a 14 single time step;
16 FIG. 8 is an illustration of the use of an adaptive scheme to selectively update 17 basis functions;
19 FIG. 9 is an illustration of a permeability field associated with a SPE 10 problem;
22 FIGS. 10A-B are illustrations of permeability fields of a top layer and a bottom 23 layer of cells from the SPE 10 problem;
24 . .
FIGS. 11A-B are illustrations of saturation fields of top layers of cells created 26 using the MSFV method and F1G. 11 C is an illustration of a saturation field 27 computed by a conventional fine-scale reservoir simulator;
29 FIGS. 12A-B are illustrations of saturation fields of bottom layers of cells created using the MSFV method and FIG. 12C is an illustration of a saturation 31 field computed by a conventional fine-scale reservoir computer;
1 FIGS. 13A-B are graphs of oil cut and oil recovery;
3 FIG. 14 is an illustration of a 3~ test case having a grid of 10 x 22 x 17 grid 4 cells and including injector and producer welts; and 6 FIG. 15 is a graph of oil cut and oil recovery.
8 BE~'T ~~~E~ F'~~ ~~~I~~I~~ OU'i' 'THE 1~~'E~1'I~i'~
!. FLOUT/ PROBLEM
12 A. One Phase Flow 13 Fluid flow in a porous media can be described by the elliptic problem:
0~(~,~op~= f on S2 (1) 17 vuhere p is the pressure, ~, is the mobility coefficient (permeability, K, divided 18 by fluid viscosity, p) and SZ is a volume or region of a subsurface which is to 19 be simulated. A source term f represents wells, and in the compressible case, time derivatives. Permeability heterogeneity is a dominant factor in 21 dictating the flow behavior in natural porous formations. The heterogeneity of 22 permeability K is usually represented as a complex multi-scale function of 23 space. Moreover, permeability K tends to be a highly discontinuous full 24 tensor. Resolving the spatial correlation structures and capturing the variability of permeability requires a highly detailed reservoir description.
27 The velocity a of fluid flow is related to the pressure field through Darcy's law:
29 a =- ~. ° 0~ . (2) 1 On the boundary of a volume, aS2 , the flux q = a ~ v is specified, where v is the 2 boundary unit normal vector pointing outward. Equations (1 ) and (2) describe 3 incompressible flow in a porous media. These equations apply for both single 4 and mufti-phase flows when appropriate interpretations of the mobility coefficient ~ and velocity ~ are made. This elliptic problem is a simple, yet 6 representative, description of the type of systems that should be handled '~ efficiently by a subsurfiace fifow simulator. Moreover, the ability to handle this 8 limiting case of incompressible flow ensures that compressible systems can 9 be treated as a subset.
11 B. Two Phase Flow 12 The flow of two incompressible phases in a heterogeneous domain may be 13 mathematically described by the following:
~ as° - ~ x ~Y° ap at axe ,~o axr 1 a (a) 1~ ~asw- a kk'~ ap at axi few a~i 19 on a volume S~, where p is the pressure, So,,~, are the saturations (the subscripts o and w stand for oil and water, respectively) with 0 <_ S'o,", <_ 1 and 21 So +S", ---1, k is the heterogeneous permeability, Ic,.,.", are the relative 22 permeabilities (which are functions of So,W ), ,uo,,~ the viscosities and qo,", are 23 source terms which represent the wells. The system assumes that capillary 24 pressure and gravity are negligible. Equivalently, system (3) can be written as:
27 - W a = ~lo + ~l w (4) _g_ 1 ~a~° +o' k +°k a -'R'° (5) ° w On ~ wlth u =-~.'op. (6) 7 and the total mobility 9 ~, = k(k~ + ky,,), (7) 11 where k~ --_kr~l,u~for ,j E ~o,v~~~.
13 Equation (4) is known as the "pressure equation" and equation (5) as the 14 "hyperbolic transport equation." Again, equations (4) and (5) are a representative description of the type of systems that should be handled 16 efficiently by a subsurface flow simulator. Such flow simulators, and 17 techniques employed to simulate flow, are well known to those skilled in the 18 art and are described in publications such as Petroleum Reservoir Simulation, 19 K. Aziz and A. Settari, Stanford Bookstore Custom Publishing, 1999.
21 II. MULTI-SCALE FINITE-VOLUME (MSFV) METHOD
23 A. MSFV Method for One Phase Flow 24 1. Finite-Votume Method A cell centered finite-volume method will now be briefly described. To solve 26 the problem of equation (1 ), the overall domain or volume S~ is partitioned into 27 smaller volumes {S~, }. A finite-volume solution then satisfies 29 ~~~~clS~.= ~ ' u~vdr=-~ fclS~, (8) z s2 _g_ 1 for each control volume SZ; , where v' is the unit normal vector of the volume 2 boundary aS~; pointing outward. The chal4enge is to find a good 3 approximation for ~~ ~ at ~S~i . In general, the flux is expressed as:
h a . v = ~ T kl~k . (9) k=1 7 Equation (9) is a linear combination of the pressure values, ~ , in the volumes 8 ~52,; ~ of the domain ~ . The total number ofi volumes is ~z and T~ denotes 9 transmissibility between volumes ~52; }. By definition, the fluxes of equation (9) are continuous across the interfaces of the volumes ~52; ~ and, as a result, the 11 finite-volume method is conservative.
13 2. Construction of the Effective Transmissibilities 14 The MSFV method results in multi-point stencils for coarse-scale fluxes.
For the following description, an orthogonal 2D grid 20 of grid cells 22 is used, as 16 shown in FIG. 1. An underlying tine grid 24 of fine grid cells 26 contains the 17 fine-scale permeability K information. To compute the transmissibilities T
18 between coarse grid cells 22, a dual coarse grid 30 of dual coarse control 19 volumes 32 is used. A control volume 32 of the dual grid 30, SZ , is constructed by connecting the mid-points of four adjacent coarse grid cells 22. To relate 21 the fluxes across the coarse grid cell interfaces 34 which lie inside a particular 22 control volume 32, or S2 , to the finite-volume pressures p~ (k =1,4) in the four 23 adjacent coarse grid cells 22, a local elliptical problem in the preferred 24 embodiment is defined as 26 V~(~,~Ph)=Oon S~. (10) 2?
6 The present invention relates generally to subsurface reservoir simulators, 7 and more particularly, to those simulators which use multi-scale physics to 8 simulate flow in an underground reservoir.
I3AC~ed~ROUND OF TFiE INVEf~TION
12 The level of detail available in reservoir description often exceeds the 13 computational capability of existing reservoir simulators. This resolution gap 14 is usually tackled by upscaling the fine-scale description to sizes that can be treated by a full-featured simulator. in upscaling, the original model is 16 coarsened using a computationally inexpensive process. In flow-based 17 methods, the process is based on single-phase flow. A simulation study is 18 then performed using the coarsened model. Upscaling methods such as 19 these have proven to be quite successful. However, it is not possible to have a priori estimates of the errors that are present when complex flow processes 21 are investigated using coarse models constructed via these simplified 22 settings.
24 Various fundamentally different multi-scale approaches for flow in porous media have been proposed to accommodate the fine-scale description 26 directly. As opposed to upscaling, the multi-scale approach targets the full 27 problem with the original resolution. The upscaling methodology is typically 28 based on resolving the length and time-scales of interest by maximizing local 29 operations. Arbogast et al. (T. Arbogast, Numerical subgrid upscaling of tw~
phase flow in porous media, Technical report, Texas Institute for 31 Computational and Applied Mathematics, The University of Texas at Austin, 32 1999, and T. Arbogast and S. L. Bryant, Numerical subgrid upscaling for _1_ 1 waterflood simulations, SPE 66375, 2001) presented a mixed finite-element 2 method where fine-scale effects are localized by a boundary condition 3 assumption at the coarse element boundaries. Then the small-scale influence 4~ is coupled with the coarse-scale effiects by numerical Greens functions.
Hou and Wu (T. Hou and X. H. Wu, A multiscale finite elen7ent method for elliptic 6 problems in composite materials and porous media, J. Comp. Phys., 7 134:169-189, 1997) employed a finite-element approach and constructed 8 specific basis functions which capture the small scales. Again, localization is 9 achieved by boundary condition assumptions for the coarse elements. To reduce the efFects of these boundary conditions, an oversampling technique 11 can be applied. Chen and Hou (Z. Chen and T. Y. Hou, A mixed finite 12 element method for elliptic problems with rapidly oscillating coefficients, Math.
13 Comput., June 2002) utilized these ideas in combination with a mixed 14 finite-element approach. Another approach by Beckie et al. (R. Beckie,.
A. A. Aldama, and E. F. Wood, Modeling the large-scale dynamics of 16 saturated groundwater flow using spatial filtering, Water Resources Research, 17 32:1269-1280, 1996) is based on large eddy simulation (LES) techniques 18 which are commonly used for turbulence modeling.
Lee et al. (S. H. Lee, L. J. Durlofsky, M. F. Lough, and W. H. Chen, Finite 21 difference simulation of geologically complex reservoirs with tensor 22 permeabilities, SPERE&E, pages 567-574, 1998) developed a flux-continuous 23 finite-difference (FCFD) scheme for 2D models. Lee et al. further developed a 24 method to address 3D models (S. H. Lee, H. Tchelepi, P. Jenny and L. Dechant, Implementation of a flux continuous finite-difference method for 26 stratigraphic, hexahedron grids, SPE Journal, September, pages 269-277, 27 2002). Jenny et al. (P. Jenny, C. Wolfsteiner, S. H. Lee and L. J.
Durlofsky, 28 Modeling flow in geometrically complex reservoirs using hexahedral 29 multi-block grids, SPE Journal, June, pages 149-157, 2002) later implemented this scheme in a multi-block simulator.
1 In light of the above modeling efforts, there is a need for a simulation method 2 which more efficiently captures the effects of small scales on a coarse grid.
3 Ideally, the method would be conservative and also treat tensor permeabilities 4 correctly. Further, preferably the reconstructed fine-scale solution would satisfy the proper mass balance on the fine-scale. The present invention 6 provides such a simulation method.
~~I~~GQa~~I'n ~~ ~~~ ~~~~~l~~l~~
A mufti-scale finite-volume (MSFV) approach is taught for solving elliptic or 11 parabolic problems such as those found in subsurface flow simulators.
12 Advanfiages of the present MSFV method are that it fits nicely into a 13 finite-volume framework, it allows for computing effective coarse-scale 14 transmissibiiities, treats tensor permeabilities properly, and is conservative at both the coarse and fine scales. The present method is computationally 16 efficient relative to reservoir simulation now in use and is well suited for 17 massive parallel computation. The present invention can be applied to 3D
18 unstructured grids and also to mufti-phase flow. Further, the reconstructed 19 fine-scale solution satisfies the proper mass balance on the fine-scale.
21 A mufti-scale approach is described which results in effective transmissibilities 22 for the coarse-scale problem. ~nce the transmissibilities are constructed, the 23 MSFV method uses a finite-volume scheme employing mufti-point stencils for 24 flux discretization. The approach is conservative and treats tensor permeabilities correctly. This method is easily applied using existing 26 finite-volume codes, and once the transmissibilities are computed, the method 27 is computationally very efficient. In computing the effective transmissibilities, 28 closure assumptions are employed.
A significant characteristic of the present mufti-scale method is that two sets 31 of basis functions are employed. A first set of dual basis functions is 32 computed to construct transmissibilities between coarse cells. A second set 1 of locally computed fine scale basis functions is utilized to reconstruct a 2 fine-scale velocity field from a coarse scale solution. This second set of 3 fine-scale basis functions is designed such that the reconstructed fine-scale 4. velocity solution is fully consistent with fibs transmissibilities.
Further, the solution satisfies the proper mass balance on the small scale.
7 The fVISF~I method may be used in modeling a subsurface reservoir. A fine 8 grid is first created defining a plurality of fns veils. A permeability field and 9 other fine scale properties are associated with the fine cells. ~dexfi, a coarse grid is created which defines a plurality of coarse cells having interfaces 11 between the coarse cells. The coarse cells are ideally aggregates of the fine 12 cells. A dual coarse grid is constructed defining a plurality of dual coarse 13 control volumes. The dual coarse control volumes are ideally also aggregates 14 of the fine cells. Boundaries surround the dual coarse control volumes.
16 Dual basis functions are then calculated on the dual coarse control volumes 17 by solving local elliptic or parabolic problems, preferably using boundary 13 conditions obtained from solving reduced problems along the interfaces of the 19 course cells. Fluxes, preferably integral fluxes, are then extracted across the interfaces of the coarse cells from the dual basis functions. These fluxes are 21 assembled to obtain effective transmissibilities between coarse cells of the 22 coarse cell grid. The transmissibilities can be used for coarse scale finite 23 volume calculations.
A fine scale velocity field may be established. A finite volume method is used 26 to calculate pressures in the coarse cells utilizing the transmissibilities 27 between cells. Fine scale basis functions are computed by solving local 23 elliptic or parabolic flow problems on the coarse cells and by utilizing fine 29 scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions. Finally, the fine-scale basis functions and the 31 corresponding coarse calf pressures are combined to extract the small scale 32 velocity field.
_q._ 1 A transport problem may be solved on the fine grid by using the small scale 2 velocity field. Ideally, the transport problem is solved iteratively in two stages.
3 In the first stage, a fine scale velocity field is obtained from solving a pressure 4 equation. In the second stage, the transport problem is solved on the fine cells using the fine-scale velocity field. A Schwart~ overlap technique can be 6 applied to solve the transport problem locally on each coarse cell with an 7 implicit: upwind scheme.
9 A solution may be computed on the coarse cells at an incremental time and properties, such as a mobility coefficient, may be generated for the fine cells 11 at the incremental time. If a predetermined condition is not met for all fine 12 cells inside a dual coarse control volume, then the dual and fine scale basis 13 functions in that dual coarse control volume are reconstructed, BRIEF DESCRIPTION OF THE DRAWINGS
17 These and other objects, features and advantages of the present invention 18 will become befiter understood with regard to the following description, 19 pending claims and accompanying drawings where:
21 F1G. 1 illustrates a coarse 2D grid of coarse cells with an overlying dual 22 coarse grid including a dual coarse control volume and an underlying fine grid 23 of fine cells;
FIG. 2 illustrates a coarse grid including nine adjacent coarse cells (bold solid 26 tines) with a corresponding overlying dual coarse grid (bold dashed lines) 27 including dual coarse control volumes and an underlying fine grid (thin dotted 28 lines) of fine cells;
FIG. 3 shows flux contributi~n e~~'-~ and y;f ~ due to the pressure in a particular 31 coarse cell 2;
1 FIG. 4 is a flowchart describing the overall steps used in a preferred 2 embodiment of a reservoir simulation which employs a multi-scale 3 finite-volume (MSFV) method made in accordance with this invention;
FIG. 5 is a flowchart further detailing steps used to determine transmissibilities 6 T between coarse cells;
8 FIG. 6 is a flow chart further describing steps used to construct a set of 9 fine-scale basis functions and to exiract a small scale velocity field;
11 FIG. 7 is a flowchart depicting coupling between pressure and the saturation 12 equations which utilize an implicit solution scheme and wherein IZ and ~
are 13 operators used to update total velocity and saturation, respectively, during a 14 single time step;
16 FIG. 8 is an illustration of the use of an adaptive scheme to selectively update 17 basis functions;
19 FIG. 9 is an illustration of a permeability field associated with a SPE 10 problem;
22 FIGS. 10A-B are illustrations of permeability fields of a top layer and a bottom 23 layer of cells from the SPE 10 problem;
24 . .
FIGS. 11A-B are illustrations of saturation fields of top layers of cells created 26 using the MSFV method and F1G. 11 C is an illustration of a saturation field 27 computed by a conventional fine-scale reservoir simulator;
29 FIGS. 12A-B are illustrations of saturation fields of bottom layers of cells created using the MSFV method and FIG. 12C is an illustration of a saturation 31 field computed by a conventional fine-scale reservoir computer;
1 FIGS. 13A-B are graphs of oil cut and oil recovery;
3 FIG. 14 is an illustration of a 3~ test case having a grid of 10 x 22 x 17 grid 4 cells and including injector and producer welts; and 6 FIG. 15 is a graph of oil cut and oil recovery.
8 BE~'T ~~~E~ F'~~ ~~~I~~I~~ OU'i' 'THE 1~~'E~1'I~i'~
!. FLOUT/ PROBLEM
12 A. One Phase Flow 13 Fluid flow in a porous media can be described by the elliptic problem:
0~(~,~op~= f on S2 (1) 17 vuhere p is the pressure, ~, is the mobility coefficient (permeability, K, divided 18 by fluid viscosity, p) and SZ is a volume or region of a subsurface which is to 19 be simulated. A source term f represents wells, and in the compressible case, time derivatives. Permeability heterogeneity is a dominant factor in 21 dictating the flow behavior in natural porous formations. The heterogeneity of 22 permeability K is usually represented as a complex multi-scale function of 23 space. Moreover, permeability K tends to be a highly discontinuous full 24 tensor. Resolving the spatial correlation structures and capturing the variability of permeability requires a highly detailed reservoir description.
27 The velocity a of fluid flow is related to the pressure field through Darcy's law:
29 a =- ~. ° 0~ . (2) 1 On the boundary of a volume, aS2 , the flux q = a ~ v is specified, where v is the 2 boundary unit normal vector pointing outward. Equations (1 ) and (2) describe 3 incompressible flow in a porous media. These equations apply for both single 4 and mufti-phase flows when appropriate interpretations of the mobility coefficient ~ and velocity ~ are made. This elliptic problem is a simple, yet 6 representative, description of the type of systems that should be handled '~ efficiently by a subsurfiace fifow simulator. Moreover, the ability to handle this 8 limiting case of incompressible flow ensures that compressible systems can 9 be treated as a subset.
11 B. Two Phase Flow 12 The flow of two incompressible phases in a heterogeneous domain may be 13 mathematically described by the following:
~ as° - ~ x ~Y° ap at axe ,~o axr 1 a (a) 1~ ~asw- a kk'~ ap at axi few a~i 19 on a volume S~, where p is the pressure, So,,~, are the saturations (the subscripts o and w stand for oil and water, respectively) with 0 <_ S'o,", <_ 1 and 21 So +S", ---1, k is the heterogeneous permeability, Ic,.,.", are the relative 22 permeabilities (which are functions of So,W ), ,uo,,~ the viscosities and qo,", are 23 source terms which represent the wells. The system assumes that capillary 24 pressure and gravity are negligible. Equivalently, system (3) can be written as:
27 - W a = ~lo + ~l w (4) _g_ 1 ~a~° +o' k +°k a -'R'° (5) ° w On ~ wlth u =-~.'op. (6) 7 and the total mobility 9 ~, = k(k~ + ky,,), (7) 11 where k~ --_kr~l,u~for ,j E ~o,v~~~.
13 Equation (4) is known as the "pressure equation" and equation (5) as the 14 "hyperbolic transport equation." Again, equations (4) and (5) are a representative description of the type of systems that should be handled 16 efficiently by a subsurface flow simulator. Such flow simulators, and 17 techniques employed to simulate flow, are well known to those skilled in the 18 art and are described in publications such as Petroleum Reservoir Simulation, 19 K. Aziz and A. Settari, Stanford Bookstore Custom Publishing, 1999.
21 II. MULTI-SCALE FINITE-VOLUME (MSFV) METHOD
23 A. MSFV Method for One Phase Flow 24 1. Finite-Votume Method A cell centered finite-volume method will now be briefly described. To solve 26 the problem of equation (1 ), the overall domain or volume S~ is partitioned into 27 smaller volumes {S~, }. A finite-volume solution then satisfies 29 ~~~~clS~.= ~ ' u~vdr=-~ fclS~, (8) z s2 _g_ 1 for each control volume SZ; , where v' is the unit normal vector of the volume 2 boundary aS~; pointing outward. The chal4enge is to find a good 3 approximation for ~~ ~ at ~S~i . In general, the flux is expressed as:
h a . v = ~ T kl~k . (9) k=1 7 Equation (9) is a linear combination of the pressure values, ~ , in the volumes 8 ~52,; ~ of the domain ~ . The total number ofi volumes is ~z and T~ denotes 9 transmissibility between volumes ~52; }. By definition, the fluxes of equation (9) are continuous across the interfaces of the volumes ~52; ~ and, as a result, the 11 finite-volume method is conservative.
13 2. Construction of the Effective Transmissibilities 14 The MSFV method results in multi-point stencils for coarse-scale fluxes.
For the following description, an orthogonal 2D grid 20 of grid cells 22 is used, as 16 shown in FIG. 1. An underlying tine grid 24 of fine grid cells 26 contains the 17 fine-scale permeability K information. To compute the transmissibilities T
18 between coarse grid cells 22, a dual coarse grid 30 of dual coarse control 19 volumes 32 is used. A control volume 32 of the dual grid 30, SZ , is constructed by connecting the mid-points of four adjacent coarse grid cells 22. To relate 21 the fluxes across the coarse grid cell interfaces 34 which lie inside a particular 22 control volume 32, or S2 , to the finite-volume pressures p~ (k =1,4) in the four 23 adjacent coarse grid cells 22, a local elliptical problem in the preferred 24 embodiment is defined as 26 V~(~,~Ph)=Oon S~. (10) 2?
1 For one skilled in the art, the method can easily be adapted to use a local 2 parabolic problem.
4 For an elliptic problem, ~irichlet or I~eui~ann boundary conditions are t~
be specified on boundaryaS~ . Ideally, the imposed boundary conditions should 6 approximate the true flow conditions experienced by the sub-domain in the full 7 system, These boundary conditions can be time and flow dependent. Since 8 the sub-domain is embedded in the whole system, Wallstrom et al.
9 (T. ~. Wallstrom, T. Y. Hou, iVi. A Ghristie, L. J. ~urlofsky, and ~. H.
Sharp, Application ofi a new fiwo pf~ase upscaling technique fo realistic r~eser~oir cross 11 sections, SPE 51939, presented at the SPE Symposium on Reservoir 12 Simulation, Houston, 1999) found that a constant pressure condition at the 13 sub-domain boundary tends to overestimate flow contributions from high 14 permeability areas. If the correlation length of permeability is not much larger than the grid size, the filow contribution from high permeability areas is not 16 proportional to the nominal permeability ratio. The transmissibility between 17 two cells is a harmonic mean that is closer to the lower permeability. As a 18 result, uniform flux conditions along the boundary often yield much better 19 numerical results for a sub-domain problem than linear or constant pressure conditions.
22 Hou and Wu (T. Hou and W. H. Wu, A multiscale finite element method for 23 elliptic problems in composite materials and porous media, J. Gomp. Phys, 24 134:169-189, 1997) also proposed solving a reduced problem 26 ~~ ~,;~ ~p = 0, (11 ) .~ , 28 to specify the boundary conditions for the local problem. The subscript t 29 denotes the component parallel to the boundary of the dual coarse control volume 32 orc7~2 . For eguation (11 ) and for the following part of this 1 specification, Einstein summation convention will be used. The elliptic 2 problem on a control volume S~ with boundary conditions of equation (11 ) on 3 as2 can be solved by any appropriate numerical method. In order to obtain a 4 pressure solution that depends linearly on the pressures ~ar'(j -1,4), this preferred embodiment solves four elliptic problems, one for each cell-center 6 pressure. For instance, to get the solution for the pressure p' in the coarse 7 grid cell having node 1 at its center, p~' _ ~lr' is set. The four solutions 8 provide the dual basis functions ~1~(7~ =1,4) in control volume S2 , and the 9 pressure solution of the local elliptic problem in a control volume S2 is the linear combination 12 p = ~ pk ~k . (12) k=1 14 Accordingly, the flux q across the grid cell interfaces can be written as a linear combination 17 q = ~ pkqk , (13) 7z=1 19 where qk (k =1,4~ are the flux contributions from the corresponding dual basis functions, given all ~k (k =1,4~ from all control volumes S~ . The effective 21 transmissibilities T are computed, which can be used for finite-volume 22 simulations, by assembling the flux contributions, in the preferred embodiment 23 integral flux contributions across the cell interfaces 34.
1 Note that the domain S2 can have any fine-scale distribution of mobility 2 coefficients ~, . Of course the boundary condition given by equation (1~1 ) is an 3 approximation that allows one to decouple the local problems. The MSFV
4 and global fine-scale solutions are identical, only if equation (11 ) happens to capture the exact fine-scale pressure solution. However, numerical 6 experiments have been performed which indicate that equation (11) is an 7 excellent approximation of the boundary condition.
9 Although the MSFV approach is a finite-volume method, it resembles the mufti-scale finite-element method of ldVu and Hou, briefly mentioned above.
11 The construction of the dual basis functions is similar, though in the present 12 MSFV method they are represented on the dual coarse grid rather than on the 13 boundary of a finite element. A significant difference is that the present MSFV
14 method is a cell-centered finite-volume method and is conservative. On the other hand, the mass matrix in the mufti-scale finite-element method is 16 constructed based on a variational principle arid does not ensure local 17 conservation. In the next section, the importance is illustrated of a fine-scale 18 velocity field that is conservative.
3. Reconstruction of a Conservative Fine-Scale Velocity 21 Field 22 Fluxes across the coarse cell interfaces 34 can be accurately computed by 23 mufti-scale transmissibilities T. In some, cases, it is interesting to accurately 24 represent the small-scale velocities a (e.g., to predict the distribution of solute transported by a fluid). A straightforward approach might appear to be to use 26 the dual basis functions ~ of equation (12). However, then the reconstructed 27 fine-scale velocity field is, in general, discontinuous at the cell intertaces of the 28 dual grid 30. Therefore, large errors can occur in the divergence field, and 29 local mass balance is violated. Note that mass conservation is always satisfied for the coarse solution using the present MSFV method.
1 The construction of a second set of local fine scale basis functions ~ will now 2 be described which is fully consistent with the fluxes q across the cell 3 interfaces given by the dual basis functions ~ . This second set of fine-scale 4~ basis functions ~ allows a conservative fine-scale velocity field ~to be reconstructed.
7 FIG. 2 shows a coarse grid 20 with nine adjacent grid cells 22 and a 8 corresponding dual grid 30 of dual coarse control volumes 32 or ~ . For 9 indexing purposes, these particular cells and corresponding dual volumes shall now be identified with numerals "1-9" and letters "A-~" at their respective 11 centers. Also shown is the underlying tine grid 24 of fine grid cells 26.
The 12 coarse grid, having the nine adjacent coarse cells 1-9, is shown in bold solid 13 lines. The corresponding dual grid 30 of dual coarse control volumes A-D
are 14 depicted with bold dashed lines. The underlying fine grid 24 of fine grid cells 26 is shown with thin dotted lines.
17 To explain the reconstruction of the fine-scale velocity, the mass balance of 18 the center grid cell 5 is examined. The coarse scale pressure solution, 19 together with the dual basis functions ~ , provides the fine-scale fluxes q across the interfaces of coarse cell 5.
22 To obtain a proper representation of the fine-scale velocity field in coarse cell 23 5, (l) the fine-scale fluxes across an interface of coarse calf 5 must match, and 24 (ii) the divergence of the fine-scale velocity field within the coarse volume satisfies (14) 27 O~u=
1 where SZ$ is the coarse grid cell 5. The fine-scale flux q across the boundary 2 of grid cell 5 depends on the coarse pressure solutions in grid cells 1-9.
3 Therefore, the fine-scale velocity field within coarse grid cell 5 can be 4 expressed as a superposition of fine scale basis functions c~z (i = I,9).
~ilith the help of FIG. 3, which depicts the needed dual coarse control volumes, the 17 construction of the fine-scale basis functions ~Z will be described. Each 13 coarse cell pressure ~(i =1,9) contributes to the fine-scale fluxq . For 19 example, let the contribution of the pressure in cell 2 to the filux q in grid cell 20 5 beq(z) . Note that q(2) is composed of contributions qA~~ and qB~) coming 21 from the dual basis functions associated with node 2 of volume A and volume 22 B, respectively. To compute the fine-scale basis function ~l associated wifh 23 the pressure in a coarse cell i , p~ _ ~~ is set, and the pressure field is 24 constructed according to the following equation.
26 p= ~ ~p~c~j ' (15) kE{A,B,C,D }j=1 28 The fine-scale fluxes q are computed from the pressure field. These fluxes 29 provide the proper boundary condition for computing the fine-scale basis function ~Z . To solve the elliptic problem 32 W(~.Wh~=f~ on s2$ (16) 34 witll the boundary conditions described above, solvability must be ensured.
This is achieved by setting 1 ' f, = f aS25 dS2 ' ( 17 ) Jas 3 which is an equally distributed source term within S~5 . Finally, the solution of 4 the elliptic problem, (16) and (17), is the one-scale basis function ~t for coarse cell 5 associated with the pressure in volume z . The small-scale 6 velocity field is extracted from the superposition T
9 _ 8 P = ~ P~~S ' (18) j=1 g For incompressible flow, this velocity field is divergence free everywhere.
11 Computing the fine-scale basis functions ~l requires solving nine small 12 elliptic problems, which are of the same size as those for the transmissibility 13 calculations. Note that this step is a preprocessing step and has to be done 14 only once. Furthermore, the construction of the fine-scale basis functions ~i is independent and therefore well suited for parallel computation. The 16 reconstruction of the fine-scale velocity field is a simple superposition and is 17 ideally performed only in regions of interest.
4 For an elliptic problem, ~irichlet or I~eui~ann boundary conditions are t~
be specified on boundaryaS~ . Ideally, the imposed boundary conditions should 6 approximate the true flow conditions experienced by the sub-domain in the full 7 system, These boundary conditions can be time and flow dependent. Since 8 the sub-domain is embedded in the whole system, Wallstrom et al.
9 (T. ~. Wallstrom, T. Y. Hou, iVi. A Ghristie, L. J. ~urlofsky, and ~. H.
Sharp, Application ofi a new fiwo pf~ase upscaling technique fo realistic r~eser~oir cross 11 sections, SPE 51939, presented at the SPE Symposium on Reservoir 12 Simulation, Houston, 1999) found that a constant pressure condition at the 13 sub-domain boundary tends to overestimate flow contributions from high 14 permeability areas. If the correlation length of permeability is not much larger than the grid size, the filow contribution from high permeability areas is not 16 proportional to the nominal permeability ratio. The transmissibility between 17 two cells is a harmonic mean that is closer to the lower permeability. As a 18 result, uniform flux conditions along the boundary often yield much better 19 numerical results for a sub-domain problem than linear or constant pressure conditions.
22 Hou and Wu (T. Hou and W. H. Wu, A multiscale finite element method for 23 elliptic problems in composite materials and porous media, J. Gomp. Phys, 24 134:169-189, 1997) also proposed solving a reduced problem 26 ~~ ~,;~ ~p = 0, (11 ) .~ , 28 to specify the boundary conditions for the local problem. The subscript t 29 denotes the component parallel to the boundary of the dual coarse control volume 32 orc7~2 . For eguation (11 ) and for the following part of this 1 specification, Einstein summation convention will be used. The elliptic 2 problem on a control volume S~ with boundary conditions of equation (11 ) on 3 as2 can be solved by any appropriate numerical method. In order to obtain a 4 pressure solution that depends linearly on the pressures ~ar'(j -1,4), this preferred embodiment solves four elliptic problems, one for each cell-center 6 pressure. For instance, to get the solution for the pressure p' in the coarse 7 grid cell having node 1 at its center, p~' _ ~lr' is set. The four solutions 8 provide the dual basis functions ~1~(7~ =1,4) in control volume S2 , and the 9 pressure solution of the local elliptic problem in a control volume S2 is the linear combination 12 p = ~ pk ~k . (12) k=1 14 Accordingly, the flux q across the grid cell interfaces can be written as a linear combination 17 q = ~ pkqk , (13) 7z=1 19 where qk (k =1,4~ are the flux contributions from the corresponding dual basis functions, given all ~k (k =1,4~ from all control volumes S~ . The effective 21 transmissibilities T are computed, which can be used for finite-volume 22 simulations, by assembling the flux contributions, in the preferred embodiment 23 integral flux contributions across the cell interfaces 34.
1 Note that the domain S2 can have any fine-scale distribution of mobility 2 coefficients ~, . Of course the boundary condition given by equation (1~1 ) is an 3 approximation that allows one to decouple the local problems. The MSFV
4 and global fine-scale solutions are identical, only if equation (11 ) happens to capture the exact fine-scale pressure solution. However, numerical 6 experiments have been performed which indicate that equation (11) is an 7 excellent approximation of the boundary condition.
9 Although the MSFV approach is a finite-volume method, it resembles the mufti-scale finite-element method of ldVu and Hou, briefly mentioned above.
11 The construction of the dual basis functions is similar, though in the present 12 MSFV method they are represented on the dual coarse grid rather than on the 13 boundary of a finite element. A significant difference is that the present MSFV
14 method is a cell-centered finite-volume method and is conservative. On the other hand, the mass matrix in the mufti-scale finite-element method is 16 constructed based on a variational principle arid does not ensure local 17 conservation. In the next section, the importance is illustrated of a fine-scale 18 velocity field that is conservative.
3. Reconstruction of a Conservative Fine-Scale Velocity 21 Field 22 Fluxes across the coarse cell interfaces 34 can be accurately computed by 23 mufti-scale transmissibilities T. In some, cases, it is interesting to accurately 24 represent the small-scale velocities a (e.g., to predict the distribution of solute transported by a fluid). A straightforward approach might appear to be to use 26 the dual basis functions ~ of equation (12). However, then the reconstructed 27 fine-scale velocity field is, in general, discontinuous at the cell intertaces of the 28 dual grid 30. Therefore, large errors can occur in the divergence field, and 29 local mass balance is violated. Note that mass conservation is always satisfied for the coarse solution using the present MSFV method.
1 The construction of a second set of local fine scale basis functions ~ will now 2 be described which is fully consistent with the fluxes q across the cell 3 interfaces given by the dual basis functions ~ . This second set of fine-scale 4~ basis functions ~ allows a conservative fine-scale velocity field ~to be reconstructed.
7 FIG. 2 shows a coarse grid 20 with nine adjacent grid cells 22 and a 8 corresponding dual grid 30 of dual coarse control volumes 32 or ~ . For 9 indexing purposes, these particular cells and corresponding dual volumes shall now be identified with numerals "1-9" and letters "A-~" at their respective 11 centers. Also shown is the underlying tine grid 24 of fine grid cells 26.
The 12 coarse grid, having the nine adjacent coarse cells 1-9, is shown in bold solid 13 lines. The corresponding dual grid 30 of dual coarse control volumes A-D
are 14 depicted with bold dashed lines. The underlying fine grid 24 of fine grid cells 26 is shown with thin dotted lines.
17 To explain the reconstruction of the fine-scale velocity, the mass balance of 18 the center grid cell 5 is examined. The coarse scale pressure solution, 19 together with the dual basis functions ~ , provides the fine-scale fluxes q across the interfaces of coarse cell 5.
22 To obtain a proper representation of the fine-scale velocity field in coarse cell 23 5, (l) the fine-scale fluxes across an interface of coarse calf 5 must match, and 24 (ii) the divergence of the fine-scale velocity field within the coarse volume satisfies (14) 27 O~u=
1 where SZ$ is the coarse grid cell 5. The fine-scale flux q across the boundary 2 of grid cell 5 depends on the coarse pressure solutions in grid cells 1-9.
3 Therefore, the fine-scale velocity field within coarse grid cell 5 can be 4 expressed as a superposition of fine scale basis functions c~z (i = I,9).
~ilith the help of FIG. 3, which depicts the needed dual coarse control volumes, the 17 construction of the fine-scale basis functions ~Z will be described. Each 13 coarse cell pressure ~(i =1,9) contributes to the fine-scale fluxq . For 19 example, let the contribution of the pressure in cell 2 to the filux q in grid cell 20 5 beq(z) . Note that q(2) is composed of contributions qA~~ and qB~) coming 21 from the dual basis functions associated with node 2 of volume A and volume 22 B, respectively. To compute the fine-scale basis function ~l associated wifh 23 the pressure in a coarse cell i , p~ _ ~~ is set, and the pressure field is 24 constructed according to the following equation.
26 p= ~ ~p~c~j ' (15) kE{A,B,C,D }j=1 28 The fine-scale fluxes q are computed from the pressure field. These fluxes 29 provide the proper boundary condition for computing the fine-scale basis function ~Z . To solve the elliptic problem 32 W(~.Wh~=f~ on s2$ (16) 34 witll the boundary conditions described above, solvability must be ensured.
This is achieved by setting 1 ' f, = f aS25 dS2 ' ( 17 ) Jas 3 which is an equally distributed source term within S~5 . Finally, the solution of 4 the elliptic problem, (16) and (17), is the one-scale basis function ~t for coarse cell 5 associated with the pressure in volume z . The small-scale 6 velocity field is extracted from the superposition T
9 _ 8 P = ~ P~~S ' (18) j=1 g For incompressible flow, this velocity field is divergence free everywhere.
11 Computing the fine-scale basis functions ~l requires solving nine small 12 elliptic problems, which are of the same size as those for the transmissibility 13 calculations. Note that this step is a preprocessing step and has to be done 14 only once. Furthermore, the construction of the fine-scale basis functions ~i is independent and therefore well suited for parallel computation. The 16 reconstruction of the fine-scale velocity field is a simple superposition and is 17 ideally performed only in regions of interest.
19 III. IMPLEMENTATION OF THE MSFV METHOD
21 FIG. 4 is a flow chart summarizing the steps employed in a preferred 22 embodiment in simulating a reservoir using the MSFV algorithm of this 23 invention. The MSFV algorithm consists of six major steps:
A. compute transmissibilities Tfor coarse-scale fluxes (step 100);
26 i3. construct fine-scale basis functions (step 200);
1 C. compute a coarse solution at a new time level; (step 300);
2 D. reconstruct the fine-scale velocity field in regions of interest (step 400);
3 E. solve transport equations (step 500); and 4~ F. recompute transmissibilities and also the fine-scale basis functions in regions where the total mobility has changed more than a predetermined 6 amount (step G00).
8 Steps A-~ describe a two-scale approach. The methodology can be applied 9 recursively with successive levels of coarsening.: fn cases of extremely fine resolution, this multi-level approach should yield scalable solutions. Parts E
11 and F account for transporfi and mobility changes due to evolving phases and 12 will be described in more detail below.
14 A. Computing Transmissibilities for Coarse-Scale Fluxes -Step 100 16 The transmissibility calculations can be done in a stand alone module 17 (T-module) and are well suited for parallel computation. The transmissibilities 18 Tcan be written to a file for use by any finite-volume simulator that can handle 19 multi-point flux discretiza'tion.
, 21 Referring now to FIG. 5, a flowchart describes the individual steps which are 22 undertaken to compute the transmissibilities Tfor a coarse scale model.
First, 23 a fine-scale grid having fine cells with an associated permeability field K
are 24 created (step 110). Next, a coarse grid, having coarse cells corresponding to the fine scale grid, is created (step 120). The fine and coarse grids are then 26 passed into a transmissibility or T-module.
28 ~ual coarse control volumes S~ are constructed (step 130), one for each node 29 ef the coarse grid. For each dual coarse control volume S~ , dual or coarse scale basis functions ~h~5 are constructed (step 140) by solving local elliptic 31 problems (equation (10)) for each v~lume ~ . This local elliptic problem, as 1 described in section II.A.2 above, and the permeability field K associated with 2 the fine grid are used and the boundary conditions corresponding to equation 3 (11 ) are utilized (step 135) in solving the elliptic problem. In cases where the 4 fine and coarse grids are nonconforming (e.g., if unstructured grids are used), oversampling may be applied. Finally, the integral coarse scale fluxes 6 q across the interfaces of the coarse cells are extracted (step 150) from the 7 dual basis functions ~ . These integral coarse seals fluxes q are then 8 assembled (step 160) to obtain fi~SFV-transmissibilities T between grid cells 9 of the coarse grid.
11 The computation oftransmissibilities Tcan be viewed as an upscaling 12 procedure. That is, the constructed coarse pressure solutions are designed to 13 account for, in some manner, the fine-scale description of the permeability K
14 in the original fine scale grid model. Thus, part A - step 100 - computing transmissibilities, is preferably a separate preprocessing step used to coarsen 16 the original fine scale mode( to a size manageable by a conventional reservoir 17 simulator.
19 These firansmissibilities T may be written to a file for later use. A
finite-volume simulator that can handle multi-point flux discretization can then use these 21 transmissibilities T.
23 B. Construction of Fine-Scale Basis Function and Fine Scale 24 Velocity Field- Step 200 FIG. 6 is a flowchart describing the steps taken to construct a set of fine scale 26 basis functions ~ which can be isolated in a separate fine scale basis 2i function ~ module. These fine scale basis functions ~ can then be used 28 to create a fine scale velocity field. This module is only necessary if there is 29 an interest in reconstructing the fine-scale velocity field from the coarse _1 g_ 1 pressure solution. As described in Section II.A.3~above, if the originaLdual 2 basis functions ~ are used in reconstructing the fine-scale velocity field, large 3 mass balance errors can occur. Here, steps are described to compute the 4 fine-scale basis functions ~ , which can be used to reconstruct a conservative fine-scale velocity field. The procedure (step 200) of FIG. 4~ follows the 6 description of Section II.A.3 and has to be performed only once at the 7 beginning of a simulation and is well suited for parallel computation.
9 The fine-scale grid (step 210), with its corresponding permeability field 1'~, the coarse grid (step 220), and the dual basis functions ~ (step 230) are passed 11 into a fine scale basis function ~ . A pressure field is constructed from the 12 coarse scale pressure solution and dual basis functions (step 250). The fine 13 scale fluxes for the coarse cells are then computed {step 260). For each 14 control volume, elliptic problems are solved, using the fine scale fluxes as boundary conditions, to determine fine scale basis functions (step 270). The 16 small scale velocity field can then be computed from the superposition of cell 17 pressures and fine scale basis functions. The results may then be output 18 from the module. In many cases, the fine-scale velocity field has to be 19 reconstructed in certain regions only, as will be described in fuller detail below. Therefore, in order to save memory and computing time, one can 21 think of a in situ computation of the fine-scale basis functions ~ , which, once 22 computed, can be reused.
24 C. Computation of the Coarse Solution at the New Time - Step 300 Step 300 can be performed by virtually any multi-point stencil finite-volume 26 code by using the MSFV-transmissibilities Tfor the flux calculation. These 27 coarse fluxes effectively capture the large-scale behavior of the solution 28 without resolving the small scales.
~. Reconstruction of the Fine-Scale Velocity Field - Step 400 31 Step 400 is straight forward. Reconstruction of the fine-scale velocity field in 1 regions of interest is achieved by superposition of the fine-scale basis 2 functions ~l as described in section Ii.A.3, step S above and as shown in 3 FIG. 6. ~f course, many variations of the MSFV method can be devised. It 4 may be advantageous; however, that construction of the transmissibilities T
and fine-scale basis functions ~ can be done in modules separate from the 6 simulator.
8 E. Solving Pressure and Transport Equations 9 1. Numerical solution algorithm - eaeplicit solution Multi-phase flow problems may be solved in two stages. First, the total 11 velocity field is obtained from solving the pressure equation (4), and then the 12 hyperbolic transport equation.(5) is solved. To solve the pressure equation, 13 the MSFV method, which has been described above is used. The difference 14 from single phase flow is that in this case the mobility term ~, reflects the total mobility of both phases, and then the obtained velocity field a is the total 16 velocity in the domain. The reconstructed fine-scale velocity field a is then 17 used to solve the transport equation on the fine grid. The values of ko~W
are 18 taken from the upwind direction; time integration may be obtained using a 19 backward Euler scheme. Note that, in general, the dual and fine scale basis functions (~, ~) must be recomputed each time step due to changes in the 21 saturation (mobility) field.
23 2. Numerical Solution Algorithm - Implicit Coupling 24 In the preferred embodiment of this invention, the MSFV method utilizes an algorithm with implicit calculations. The multi-phase flow problem is solved 26 iteratively in two stages. See FIG. 7 for a diagram of this method illustrating 27 the coupling between the pressure and saturation equations.
29 First, in each Newton step, a saturation field S is established - either initial input or through an iteration (step 510). Next, a pressure equation (see 1 equation (19) below) is solved (step 520) using the MSFV techniques 2 described above to obtain (sfiep 530) the total velocity field. Then a transport 3 equation (see equation (20) below) is solved (step 540) on the fine grid by 4~ using the reconstructed fine-scale velocity field ca. In this solution, a Schwar~.
overlap technique is applied, i.e., the transport problem is solved locally on s'a each coarse volume with an implicit upwind scheme, where the saturation 7 values from the neighboring coarse volumes at the previous iteration level are 8 used for the boundary conditions. ~nce the Schwar~ overlap scheme has 9 converged (steps 550, 560) - for hyperbolic systems this method is very efficient - the new saturation distribution determines the new total mobility 11 field for the pressure problem of the next Newton iteration. Note that, in 12 genera(, some of the basis functions have to be recomputed each iteration.
14 The superscripts h and v denote the old time and iteration levels, respectively. Saturation is represented byS, the total velocity field by u, the 16 computation of the velocity by the operatorlI, and the computation of the 17 saturation bye . The new pressure field pv+1 is obtained by solving 19 p ~~k~ko~Sv~+kwCSv lVpv+l l ~ ~o -~-Rw~ (1 21 from which the new velocity field uv+1 is computed. The new saturation field 22 sv+1 is obtained by solving sv+1 _ Sn ko ~,5'v+1 24 ~ + V ~ uv+1 -~o (20) ~t ko(Sv+11+kw(Sv+11, 26 F. Recomputing Transmissibilities and Fine-Scale Basis 1 Functions - Adaptive Scheme 2 The most expensive part of the MSFV algorithm for multi-phase flow is the 3 reconstruction of the coarse scale and fine-scale basis functions ( a~, c~
).
4~ Therefore, to obtain higher efficiency, it is desirable to recompute the basis functions only where it is absolutely necessary. An adaptive scheme can be 6 used t~ update these basis functions. In the preferred exemplary 7 embodiment, if the condition 9 1 ( '~~ ~l+ E,~ (23) 1+ ~,~ ~~-11 is not fulfilled (the superscripts ~ and fa-1 denote the previous two time steps 12 and E,~ is a defined value) for all fine cells inside a coarse dual volume, then 13 the dual basis functions of that control volume have to be reconstructed.
Note 14 that condition (23) is true if ~, changes by a factor which is larger than 1/(1+~~) and smaller than 1+s~,. An illustration of this scheme is shown in 16 FIG. 8, where the fine and the coarse grid cells are drawn with thin and bold 17 lines, respectively. The black squares represent the fine cells in which 18 condition (23) is not fulfilled. The squares with bold dashed lines are the 19 control volumes for which the dual basis functions have to be reconstructed.
The shaded regions represent the coarse cells for which the fine-scale basis 21 functions have to be updated. In the schematic 2D example of FIG. 8, only 22 of 196 total dual basis functions and 117 of 324 total fine-scale basis functions 23 have to be reconstructed. Of course, these numbers depend heavily on the 24 defined threshold E~ . In general, a smaller threshold triggers more fine volumes, and as a consequence more basis functions are recomputed each 26 time step. For a wide variety of test cases, it has been found that taking ~~, 27 to be (0.2 yields marginal changes in the obtained results.
28 ' 1 IV. NUMERICAL RESULTS
3 This MSFV method, combined the implicit coupling scheme shown in FIG. 7, 4 has been tested for two phase flow ( ~~ /,~,~, -10 ) in a stiff 3~ model with more than 140,000 fine cells. It has been demonstrated that the multi-scale 6 results are in excellent agreement with the fine-scale solution. Moreover, the 7 MSFV method has proven to be approximately 27 times more efficient than 3 the established oil reservoir simulator Chears. However, in many cases the 9 computational efficiency is compromised due to the time step size restrictions inherent for IMPES schemes. This problem may be resolved by applying the 11 fully implicit MSFV method, which was described in fibs previous section.
12 Here numerical studies show the following:
13 (1 ) The results obtained with the implicit MSFV method 14 are in exceAent agreement with the fine-scale results.
(2) The results obtained with the implicit MSFV method 16 are not very sensitive to the choice of the coarse grid.
17 (3) The implicit MSFV for two phase flow overcomes the time step 18 size restriction and therefore very large time steps can be applied.
19 (4) The results obtained with the implicit MSFV method are, to a large extent, insensitive to the time step size. and 21 (5) The implicit MSFV method is very efficient.
23 For the fine-scale comparison runs, the established reservoir simulator 24 Chears was used. The efficiency of both the implicit MSFV method and the fine scale reservoir simulator depends on the choice of various parameter 26 settings which were not fully optimized.
27 A. Test Case 23 To study the accuracy and efficiency of the fully implicit MSFV algorithm, 2~
29 and 3D test cases with uniformly spaced orthogonal 60 x 220 and 60 x 220 x 85 grids were used. The 3D grid and permeability field are the same as for 1 the SPE 10 test case, which is regarded as being extremely difficult for 2 reservoir simulators. While this 3D test case is used for computational 3 efficiency assessment, the 2D test cases, which consist of top and bottom 4 layers, serves to study the accuracy of the MSFV method. FIG_ 9 illustrates the 3D test case posed by the permeability field of the SPE 10 problem. The 6 darker areas indicate lower permeability. An injector well is placed in the 7 center of the field and four producers in the corners. These well locations are 8 used for all of the following studies. The reservoir is initially filled with oil and 9 ,uQl,c~~=l0and ~i.o~~=,~a~u,.
11 B. 2D Simulation of the Top and Bottom Layers 12 The MSFV simulator used lacked a sophisticated well model. That is, wells 13 are modeled by defining the total rates for each perforated coarse volume.
14 Therefore, in order to make accuracy comparisons between MSFV and fine-scale (Chears reservoir simulator) results, each fine-scale volume inside 16 each perforated coarse volume becomes a well In the Chears runs, For large 17 3D models, this poses a technical problem since Chears reservoir simulator is 18 not designed to handle an arbitrary large number of individual wells. For this 19 reason, it was determined to do an accuracy assessment in 2D, i.e., with the top and the bottom layers of the 3D model. These two layers, for which the 21 permeability fields are shown in FIGS. 1 OA and 10B, are representative for 22 the two characteristically different regions of the full model.
24 MSFV simulations were performed with uniformly spaced 10 x 22 and 20 x 44 coarse grids. The results were compared with the fine-scale solution on a 26 60 x 220 grid. As in the full 3D test case, there are four producers at the 27 corners which are distributed over an area of 6 x 10 fine-scale volumes.
The 28 injector is located in the center of the domain and is distributed over an area 29 of 12 x 12 fine-scale volumes. The rates are the same for all fine-scale volumes (positive for the producer volumes and negative for the injector 31 volumes). FIGS. 11A-C and 12A-C show the permeability fields of the _2~_ 1 respective top and the bottom layers. The black is indicative of low 2 permeability. These two layers are representative fior the two 3 characteristically different regions ofi the fiull 3D model. FIGS. 11 A-C
and 4 12A-C shove the computed saturation gelds after 0.0933 PVI (pcare volume injected) fior the top and the bottom layers, respectively. While FIC~S. 11 C
and 6 12C show the fine-scale reference solutions, FIGS 11A and 11 B and 12A and 7 12B show the MSFV results for 10 x 22 and 20 x 44 coarse grids, 3 respectively. For both layers, it can be observed that the agreement is 9 excellent and that the mufti-scale method is hardly sensitive to the choice of the coarse grid. A more quantitative comparison is shown in FIGS. 13A and 11 13B where the fine-scale and mufti-scale oil cut and oil recovery curves are 12 plotted. Considering the difficulty of these test problems and the fact that two 13 independently implemented simulators are used for the comparisons, this 14 agreement is quite good. In the following studies, it will be demonstrated that for a model with 1,122,000 cells, the MSFV method is significantly more 16 efficient than fine-scale simulations and the results remain accurate for very 17 large time steps.
19 C. 3D Simulations While 2D studies are appropriate to study the accuracy ofi the implicit MSFV
21 method, large and stifF 3D computations are required for a meaningful 22 efficiency assessment. A 3D test case was employed as described above. A
23 coarse 10 x 22 x 17 grid, shown in FIG.14, was used and 0.5 pore volumes 24 were injected. Opposed to the MSFV runs, the wells for the CHEARS
simulations were defined on the fine-scale. Table 1 below shows CPU time 26 and required number of times steps for the CHEARS simulation and two 27 MSFV runs.
29 TASLE 1: EFFICENCY COMPARISON BETWEEN
MSFV and FINE SCALE SIMULATIONS
Simulator ~ CPU TIME ~ Time steps ~ Recomputed ~ Coarse _25_ (minutes) Basis Pressure Functions (%) Computations (%) Chears 3325 790 I~fYSFV 123 50 26 100 2 While Chears uses a control algorithm, the time step size in the multi-scale 3 simulations was fixed. It is due to the size and stiffness of the problem that 4 much smaller time steps have to be applied for a successful Chears simulation. The table shows that the implicit MSFV method can compute the 6 solution approximately 27 times faster than CHEARS. FIG. 15 shows the oil 7 cut and recovery curves obtained with multi-scale simulations using 50 and 8 200 time steps. The close agreement between the results confirms that the 9 method is very robust in respect to time step size. Since the cost for MSFV
simulation scales almost linearly with the problem size and since the dual and 11 fine-scale basis function can be computed independently, the method is 12 ideally suited for massive parallel computations and huge problems.
14 While in the foregoing specification this invention has been described in relation to certain preferred embodiments thereof, and many details have 16 been set forth for purpose of illustration, it will be apparent to those skilled in 17 the art that the invention is susceptible to alteration and that certain other 18 details described herein can vary considerably without departing from the 19 basic principles of the invention.
21 FIG. 4 is a flow chart summarizing the steps employed in a preferred 22 embodiment in simulating a reservoir using the MSFV algorithm of this 23 invention. The MSFV algorithm consists of six major steps:
A. compute transmissibilities Tfor coarse-scale fluxes (step 100);
26 i3. construct fine-scale basis functions (step 200);
1 C. compute a coarse solution at a new time level; (step 300);
2 D. reconstruct the fine-scale velocity field in regions of interest (step 400);
3 E. solve transport equations (step 500); and 4~ F. recompute transmissibilities and also the fine-scale basis functions in regions where the total mobility has changed more than a predetermined 6 amount (step G00).
8 Steps A-~ describe a two-scale approach. The methodology can be applied 9 recursively with successive levels of coarsening.: fn cases of extremely fine resolution, this multi-level approach should yield scalable solutions. Parts E
11 and F account for transporfi and mobility changes due to evolving phases and 12 will be described in more detail below.
14 A. Computing Transmissibilities for Coarse-Scale Fluxes -Step 100 16 The transmissibility calculations can be done in a stand alone module 17 (T-module) and are well suited for parallel computation. The transmissibilities 18 Tcan be written to a file for use by any finite-volume simulator that can handle 19 multi-point flux discretiza'tion.
, 21 Referring now to FIG. 5, a flowchart describes the individual steps which are 22 undertaken to compute the transmissibilities Tfor a coarse scale model.
First, 23 a fine-scale grid having fine cells with an associated permeability field K
are 24 created (step 110). Next, a coarse grid, having coarse cells corresponding to the fine scale grid, is created (step 120). The fine and coarse grids are then 26 passed into a transmissibility or T-module.
28 ~ual coarse control volumes S~ are constructed (step 130), one for each node 29 ef the coarse grid. For each dual coarse control volume S~ , dual or coarse scale basis functions ~h~5 are constructed (step 140) by solving local elliptic 31 problems (equation (10)) for each v~lume ~ . This local elliptic problem, as 1 described in section II.A.2 above, and the permeability field K associated with 2 the fine grid are used and the boundary conditions corresponding to equation 3 (11 ) are utilized (step 135) in solving the elliptic problem. In cases where the 4 fine and coarse grids are nonconforming (e.g., if unstructured grids are used), oversampling may be applied. Finally, the integral coarse scale fluxes 6 q across the interfaces of the coarse cells are extracted (step 150) from the 7 dual basis functions ~ . These integral coarse seals fluxes q are then 8 assembled (step 160) to obtain fi~SFV-transmissibilities T between grid cells 9 of the coarse grid.
11 The computation oftransmissibilities Tcan be viewed as an upscaling 12 procedure. That is, the constructed coarse pressure solutions are designed to 13 account for, in some manner, the fine-scale description of the permeability K
14 in the original fine scale grid model. Thus, part A - step 100 - computing transmissibilities, is preferably a separate preprocessing step used to coarsen 16 the original fine scale mode( to a size manageable by a conventional reservoir 17 simulator.
19 These firansmissibilities T may be written to a file for later use. A
finite-volume simulator that can handle multi-point flux discretization can then use these 21 transmissibilities T.
23 B. Construction of Fine-Scale Basis Function and Fine Scale 24 Velocity Field- Step 200 FIG. 6 is a flowchart describing the steps taken to construct a set of fine scale 26 basis functions ~ which can be isolated in a separate fine scale basis 2i function ~ module. These fine scale basis functions ~ can then be used 28 to create a fine scale velocity field. This module is only necessary if there is 29 an interest in reconstructing the fine-scale velocity field from the coarse _1 g_ 1 pressure solution. As described in Section II.A.3~above, if the originaLdual 2 basis functions ~ are used in reconstructing the fine-scale velocity field, large 3 mass balance errors can occur. Here, steps are described to compute the 4 fine-scale basis functions ~ , which can be used to reconstruct a conservative fine-scale velocity field. The procedure (step 200) of FIG. 4~ follows the 6 description of Section II.A.3 and has to be performed only once at the 7 beginning of a simulation and is well suited for parallel computation.
9 The fine-scale grid (step 210), with its corresponding permeability field 1'~, the coarse grid (step 220), and the dual basis functions ~ (step 230) are passed 11 into a fine scale basis function ~ . A pressure field is constructed from the 12 coarse scale pressure solution and dual basis functions (step 250). The fine 13 scale fluxes for the coarse cells are then computed {step 260). For each 14 control volume, elliptic problems are solved, using the fine scale fluxes as boundary conditions, to determine fine scale basis functions (step 270). The 16 small scale velocity field can then be computed from the superposition of cell 17 pressures and fine scale basis functions. The results may then be output 18 from the module. In many cases, the fine-scale velocity field has to be 19 reconstructed in certain regions only, as will be described in fuller detail below. Therefore, in order to save memory and computing time, one can 21 think of a in situ computation of the fine-scale basis functions ~ , which, once 22 computed, can be reused.
24 C. Computation of the Coarse Solution at the New Time - Step 300 Step 300 can be performed by virtually any multi-point stencil finite-volume 26 code by using the MSFV-transmissibilities Tfor the flux calculation. These 27 coarse fluxes effectively capture the large-scale behavior of the solution 28 without resolving the small scales.
~. Reconstruction of the Fine-Scale Velocity Field - Step 400 31 Step 400 is straight forward. Reconstruction of the fine-scale velocity field in 1 regions of interest is achieved by superposition of the fine-scale basis 2 functions ~l as described in section Ii.A.3, step S above and as shown in 3 FIG. 6. ~f course, many variations of the MSFV method can be devised. It 4 may be advantageous; however, that construction of the transmissibilities T
and fine-scale basis functions ~ can be done in modules separate from the 6 simulator.
8 E. Solving Pressure and Transport Equations 9 1. Numerical solution algorithm - eaeplicit solution Multi-phase flow problems may be solved in two stages. First, the total 11 velocity field is obtained from solving the pressure equation (4), and then the 12 hyperbolic transport equation.(5) is solved. To solve the pressure equation, 13 the MSFV method, which has been described above is used. The difference 14 from single phase flow is that in this case the mobility term ~, reflects the total mobility of both phases, and then the obtained velocity field a is the total 16 velocity in the domain. The reconstructed fine-scale velocity field a is then 17 used to solve the transport equation on the fine grid. The values of ko~W
are 18 taken from the upwind direction; time integration may be obtained using a 19 backward Euler scheme. Note that, in general, the dual and fine scale basis functions (~, ~) must be recomputed each time step due to changes in the 21 saturation (mobility) field.
23 2. Numerical Solution Algorithm - Implicit Coupling 24 In the preferred embodiment of this invention, the MSFV method utilizes an algorithm with implicit calculations. The multi-phase flow problem is solved 26 iteratively in two stages. See FIG. 7 for a diagram of this method illustrating 27 the coupling between the pressure and saturation equations.
29 First, in each Newton step, a saturation field S is established - either initial input or through an iteration (step 510). Next, a pressure equation (see 1 equation (19) below) is solved (step 520) using the MSFV techniques 2 described above to obtain (sfiep 530) the total velocity field. Then a transport 3 equation (see equation (20) below) is solved (step 540) on the fine grid by 4~ using the reconstructed fine-scale velocity field ca. In this solution, a Schwar~.
overlap technique is applied, i.e., the transport problem is solved locally on s'a each coarse volume with an implicit upwind scheme, where the saturation 7 values from the neighboring coarse volumes at the previous iteration level are 8 used for the boundary conditions. ~nce the Schwar~ overlap scheme has 9 converged (steps 550, 560) - for hyperbolic systems this method is very efficient - the new saturation distribution determines the new total mobility 11 field for the pressure problem of the next Newton iteration. Note that, in 12 genera(, some of the basis functions have to be recomputed each iteration.
14 The superscripts h and v denote the old time and iteration levels, respectively. Saturation is represented byS, the total velocity field by u, the 16 computation of the velocity by the operatorlI, and the computation of the 17 saturation bye . The new pressure field pv+1 is obtained by solving 19 p ~~k~ko~Sv~+kwCSv lVpv+l l ~ ~o -~-Rw~ (1 21 from which the new velocity field uv+1 is computed. The new saturation field 22 sv+1 is obtained by solving sv+1 _ Sn ko ~,5'v+1 24 ~ + V ~ uv+1 -~o (20) ~t ko(Sv+11+kw(Sv+11, 26 F. Recomputing Transmissibilities and Fine-Scale Basis 1 Functions - Adaptive Scheme 2 The most expensive part of the MSFV algorithm for multi-phase flow is the 3 reconstruction of the coarse scale and fine-scale basis functions ( a~, c~
).
4~ Therefore, to obtain higher efficiency, it is desirable to recompute the basis functions only where it is absolutely necessary. An adaptive scheme can be 6 used t~ update these basis functions. In the preferred exemplary 7 embodiment, if the condition 9 1 ( '~~ ~l+ E,~ (23) 1+ ~,~ ~~-11 is not fulfilled (the superscripts ~ and fa-1 denote the previous two time steps 12 and E,~ is a defined value) for all fine cells inside a coarse dual volume, then 13 the dual basis functions of that control volume have to be reconstructed.
Note 14 that condition (23) is true if ~, changes by a factor which is larger than 1/(1+~~) and smaller than 1+s~,. An illustration of this scheme is shown in 16 FIG. 8, where the fine and the coarse grid cells are drawn with thin and bold 17 lines, respectively. The black squares represent the fine cells in which 18 condition (23) is not fulfilled. The squares with bold dashed lines are the 19 control volumes for which the dual basis functions have to be reconstructed.
The shaded regions represent the coarse cells for which the fine-scale basis 21 functions have to be updated. In the schematic 2D example of FIG. 8, only 22 of 196 total dual basis functions and 117 of 324 total fine-scale basis functions 23 have to be reconstructed. Of course, these numbers depend heavily on the 24 defined threshold E~ . In general, a smaller threshold triggers more fine volumes, and as a consequence more basis functions are recomputed each 26 time step. For a wide variety of test cases, it has been found that taking ~~, 27 to be (0.2 yields marginal changes in the obtained results.
28 ' 1 IV. NUMERICAL RESULTS
3 This MSFV method, combined the implicit coupling scheme shown in FIG. 7, 4 has been tested for two phase flow ( ~~ /,~,~, -10 ) in a stiff 3~ model with more than 140,000 fine cells. It has been demonstrated that the multi-scale 6 results are in excellent agreement with the fine-scale solution. Moreover, the 7 MSFV method has proven to be approximately 27 times more efficient than 3 the established oil reservoir simulator Chears. However, in many cases the 9 computational efficiency is compromised due to the time step size restrictions inherent for IMPES schemes. This problem may be resolved by applying the 11 fully implicit MSFV method, which was described in fibs previous section.
12 Here numerical studies show the following:
13 (1 ) The results obtained with the implicit MSFV method 14 are in exceAent agreement with the fine-scale results.
(2) The results obtained with the implicit MSFV method 16 are not very sensitive to the choice of the coarse grid.
17 (3) The implicit MSFV for two phase flow overcomes the time step 18 size restriction and therefore very large time steps can be applied.
19 (4) The results obtained with the implicit MSFV method are, to a large extent, insensitive to the time step size. and 21 (5) The implicit MSFV method is very efficient.
23 For the fine-scale comparison runs, the established reservoir simulator 24 Chears was used. The efficiency of both the implicit MSFV method and the fine scale reservoir simulator depends on the choice of various parameter 26 settings which were not fully optimized.
27 A. Test Case 23 To study the accuracy and efficiency of the fully implicit MSFV algorithm, 2~
29 and 3D test cases with uniformly spaced orthogonal 60 x 220 and 60 x 220 x 85 grids were used. The 3D grid and permeability field are the same as for 1 the SPE 10 test case, which is regarded as being extremely difficult for 2 reservoir simulators. While this 3D test case is used for computational 3 efficiency assessment, the 2D test cases, which consist of top and bottom 4 layers, serves to study the accuracy of the MSFV method. FIG_ 9 illustrates the 3D test case posed by the permeability field of the SPE 10 problem. The 6 darker areas indicate lower permeability. An injector well is placed in the 7 center of the field and four producers in the corners. These well locations are 8 used for all of the following studies. The reservoir is initially filled with oil and 9 ,uQl,c~~=l0and ~i.o~~=,~a~u,.
11 B. 2D Simulation of the Top and Bottom Layers 12 The MSFV simulator used lacked a sophisticated well model. That is, wells 13 are modeled by defining the total rates for each perforated coarse volume.
14 Therefore, in order to make accuracy comparisons between MSFV and fine-scale (Chears reservoir simulator) results, each fine-scale volume inside 16 each perforated coarse volume becomes a well In the Chears runs, For large 17 3D models, this poses a technical problem since Chears reservoir simulator is 18 not designed to handle an arbitrary large number of individual wells. For this 19 reason, it was determined to do an accuracy assessment in 2D, i.e., with the top and the bottom layers of the 3D model. These two layers, for which the 21 permeability fields are shown in FIGS. 1 OA and 10B, are representative for 22 the two characteristically different regions of the full model.
24 MSFV simulations were performed with uniformly spaced 10 x 22 and 20 x 44 coarse grids. The results were compared with the fine-scale solution on a 26 60 x 220 grid. As in the full 3D test case, there are four producers at the 27 corners which are distributed over an area of 6 x 10 fine-scale volumes.
The 28 injector is located in the center of the domain and is distributed over an area 29 of 12 x 12 fine-scale volumes. The rates are the same for all fine-scale volumes (positive for the producer volumes and negative for the injector 31 volumes). FIGS. 11A-C and 12A-C show the permeability fields of the _2~_ 1 respective top and the bottom layers. The black is indicative of low 2 permeability. These two layers are representative fior the two 3 characteristically different regions ofi the fiull 3D model. FIGS. 11 A-C
and 4 12A-C shove the computed saturation gelds after 0.0933 PVI (pcare volume injected) fior the top and the bottom layers, respectively. While FIC~S. 11 C
and 6 12C show the fine-scale reference solutions, FIGS 11A and 11 B and 12A and 7 12B show the MSFV results for 10 x 22 and 20 x 44 coarse grids, 3 respectively. For both layers, it can be observed that the agreement is 9 excellent and that the mufti-scale method is hardly sensitive to the choice of the coarse grid. A more quantitative comparison is shown in FIGS. 13A and 11 13B where the fine-scale and mufti-scale oil cut and oil recovery curves are 12 plotted. Considering the difficulty of these test problems and the fact that two 13 independently implemented simulators are used for the comparisons, this 14 agreement is quite good. In the following studies, it will be demonstrated that for a model with 1,122,000 cells, the MSFV method is significantly more 16 efficient than fine-scale simulations and the results remain accurate for very 17 large time steps.
19 C. 3D Simulations While 2D studies are appropriate to study the accuracy ofi the implicit MSFV
21 method, large and stifF 3D computations are required for a meaningful 22 efficiency assessment. A 3D test case was employed as described above. A
23 coarse 10 x 22 x 17 grid, shown in FIG.14, was used and 0.5 pore volumes 24 were injected. Opposed to the MSFV runs, the wells for the CHEARS
simulations were defined on the fine-scale. Table 1 below shows CPU time 26 and required number of times steps for the CHEARS simulation and two 27 MSFV runs.
29 TASLE 1: EFFICENCY COMPARISON BETWEEN
MSFV and FINE SCALE SIMULATIONS
Simulator ~ CPU TIME ~ Time steps ~ Recomputed ~ Coarse _25_ (minutes) Basis Pressure Functions (%) Computations (%) Chears 3325 790 I~fYSFV 123 50 26 100 2 While Chears uses a control algorithm, the time step size in the multi-scale 3 simulations was fixed. It is due to the size and stiffness of the problem that 4 much smaller time steps have to be applied for a successful Chears simulation. The table shows that the implicit MSFV method can compute the 6 solution approximately 27 times faster than CHEARS. FIG. 15 shows the oil 7 cut and recovery curves obtained with multi-scale simulations using 50 and 8 200 time steps. The close agreement between the results confirms that the 9 method is very robust in respect to time step size. Since the cost for MSFV
simulation scales almost linearly with the problem size and since the dual and 11 fine-scale basis function can be computed independently, the method is 12 ideally suited for massive parallel computations and huge problems.
14 While in the foregoing specification this invention has been described in relation to certain preferred embodiments thereof, and many details have 16 been set forth for purpose of illustration, it will be apparent to those skilled in 17 the art that the invention is susceptible to alteration and that certain other 18 details described herein can vary considerably without departing from the 19 basic principles of the invention.
Claims (44)
1. A multi-scale finite-volume method for use in modeling a subsurface reservoir comprising:
creating a fine grid defining a plurality of fine cells and having a permeability field associated with the fine cells;
creating a coarse grid defining a plurality of coarse cells having interfaces between the coarse cells, the coarse cells being aggregates of the fine cells;
creating a dual coarse grid defining a plurality of dual coarse control volumes, the dual coarse control volumes being aggregates of the fine cells and having boundaries bounding the dual coarse control volumes;
calculating dual basis functions on the dual coarse control volumes by solving local elliptic or parabolic problems;
extracting fluxes across the interfaces of the coarse cells from the dual basis functions; and assembling the fluxes to calculate effective transmissibilities between coarse cells.
creating a fine grid defining a plurality of fine cells and having a permeability field associated with the fine cells;
creating a coarse grid defining a plurality of coarse cells having interfaces between the coarse cells, the coarse cells being aggregates of the fine cells;
creating a dual coarse grid defining a plurality of dual coarse control volumes, the dual coarse control volumes being aggregates of the fine cells and having boundaries bounding the dual coarse control volumes;
calculating dual basis functions on the dual coarse control volumes by solving local elliptic or parabolic problems;
extracting fluxes across the interfaces of the coarse cells from the dual basis functions; and assembling the fluxes to calculate effective transmissibilities between coarse cells.
2. The method of claim 1 wherein:
the local problems which are solved use boundary conditions obtained from solving reduced problems along the interfaces of the coarse cells.
the local problems which are solved use boundary conditions obtained from solving reduced problems along the interfaces of the coarse cells.
3. The method of claim 1 wherein:
integral fluxes are assembled to calculate effective transmissibilities which are conservative.
integral fluxes are assembled to calculate effective transmissibilities which are conservative.
4. The method of claim 1 wherein:
the local problems which are solved use boundary conditions obtained from solving reduced problems along the interfaces of the coarse cells;
and integral fluxes are assembled to calculate effective transmissibilities which are conservative.
the local problems which are solved use boundary conditions obtained from solving reduced problems along the interfaces of the coarse cells;
and integral fluxes are assembled to calculate effective transmissibilities which are conservative.
5. The method of claim 1 further comprising:
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse cells;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures.
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse cells;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures.
6. The method of claim 2 further comprising:
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse cells;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures.
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse cells;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures.
7. The method of claim 3 further comprising:
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse veils;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse veils;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures
8. The method of claim 4 further comprising:
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse cells;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures.
calculating pressure in the coarse cells using a finite volume method and utilizing the effective transmissibilities between coarse cells;
computing fine scale basis functions by solving local elliptic or parabolic problems on the coarse cells and utilizing fine scale fluxes across the interfaces of the coarse cells which are extracted from the dual basis functions; and extracting a small scale velocity field from the combination of the fine scale basis functions and the corresponding coarse cell pressures.
9. The method of claim 5 further comprising:
solving a transport problem on the fine grid by using the small scale velocity field.
solving a transport problem on the fine grid by using the small scale velocity field.
10. The method of claim 6 further comprising:
solving a transport problem on the fine grid by using the small scale velocity field.
solving a transport problem on the fine grid by using the small scale velocity field.
11. The method of claim 7 further comprising:
solving a transport problem on the fine grid by using the small scale velocity field.
solving a transport problem on the fine grid by using the small scale velocity field.
12. The method of claim 8 further comprising:
solving a transport problem on the fine grid by using the small scale velocity field.
solving a transport problem on the fine grid by using the small scale velocity field.
13. The method of claim 9 wherein:
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
14. The method of claim 13 wherein:
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
15. The method of claim 10 wherein:
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
16. The method of claim 15 wherein:
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
17. The method of claim 11 wherein:
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
18. The method of claim 17 wherein:
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
19. The method of claim 12 wherein:
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
the transport problem is solved iteratively in two stages;
wherein the first stage a fine scale velocity field is obtained from solving a pressure equation; and wherein the second stage the transport problem is solved on the fine cells using the reconstructed fine-scale velocity field.
20. The method of claim 19 wherein:
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
a Schwartz overlap technique is applied to solve the transport problem locally on each coarse cell with an implicit upwind scheme.
21. The method of claim 9 further comprising:
computing a solution on the coarse cells at an incremental films and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental films and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
22. The method of claim 21 wherein:
one of the fine scale properties which is generated is a mobility coefficient A and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..gamma. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient A and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..gamma. is a predetermined tolerance value.
23. The method of claim 10 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a coarse dual control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a coarse dual control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
24. The method of claim 23 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
25. The method of claim 11 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
26. The method of claim 25 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
27. The method of claim 12 wherein:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
28. The method of claim 27 wherein:
one of the fine scale properties which is generated is a mobility coefficient A and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient A and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
29. The method of claim 73 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
30. The method of claim 29 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts h and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts h and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
31. The method of claim 15 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
32. The method of claim 31 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
33. The method of claim 17 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
34. The method of claim 33 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
35. The method of claim 19 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
36. The method of claim 35 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts h and n-denote the previous two time steps and .epsilon..gamma. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts h and n-denote the previous two time steps and .epsilon..gamma. is a predetermined tolerance value.
37. The method of claim 14 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
38. The method of claim 37 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
39. The method of claim 16 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
40. The method of claim 39 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
41. The method of claim 18 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
42. The method of claim 41 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. a mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. a mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
43. The method of claim 20 further comprising:
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
computing a solution on the coarse cells at an incremental time and generating properties for the fine cells at an incremental time; and if a predetermined condition is not met for all fine cells inside a dual coarse control volume, then the dual and fine scale basis functions in that dual coarse control volume are reconstructed.
44. The method of claim 43 wherein:
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
one of the fine scale properties which is generated is a mobility coefficient .lambda. and the predetermined condition is:
where .lambda. = mobility coefficient of a fine cell, the superscripts n and n-denote the previous two time steps and .epsilon..lambda. is a predetermined tolerance value.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US10/383,908 US6823297B2 (en) | 2003-03-06 | 2003-03-06 | Multi-scale finite-volume method for use in subsurface flow simulation |
| US10/383,908 | 2003-03-06 | ||
| PCT/US2004/006817 WO2004081845A1 (en) | 2003-03-06 | 2004-03-05 | Multi-scale finite-volume method for use in subsurface flow simulation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CA2518240A1 true CA2518240A1 (en) | 2004-09-23 |
| CA2518240C CA2518240C (en) | 2013-05-14 |
Family
ID=32927154
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA2518240A Expired - Lifetime CA2518240C (en) | 2003-03-06 | 2004-03-05 | Multi-scale finite-volume method for use in subsurface flow simulation |
Country Status (6)
| Country | Link |
|---|---|
| US (2) | US6823297B2 (en) |
| EP (1) | EP1606760B1 (en) |
| AU (1) | AU2004219229B2 (en) |
| CA (1) | CA2518240C (en) |
| NO (1) | NO334387B1 (en) |
| WO (1) | WO2004081845A1 (en) |
Families Citing this family (82)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7496488B2 (en) * | 2003-03-06 | 2009-02-24 | Schlumberger Technology Company | Multi-scale finite-volume method for use in subsurface flow simulation |
| FR2870621B1 (en) * | 2004-05-21 | 2006-10-27 | Inst Francais Du Petrole | METHOD FOR GENERATING A THREE-DIMENSIONALLY THREADED HYBRID MESH OF A HETEROGENEOUS FORMATION CROSSED BY ONE OR MORE GEOMETRIC DISCONTINUITIES FOR THE PURPOSE OF MAKING SIMULATIONS |
| WO2005121840A1 (en) * | 2004-06-07 | 2005-12-22 | Exxonmobil Upstream Research Company | Method for solving implicit reservoir simulation matrix equation |
| EP1617309B1 (en) * | 2004-07-15 | 2011-01-12 | Fujitsu Limited | Simulation technique with local grid refinement |
| FR2874696B1 (en) * | 2004-09-02 | 2007-03-23 | Inst Francais Du Petrole | METHOD FOR DETERMINING MULTIPHASIC FLOW PARAMETERS OF A POROUS MEDIUM WITH ACCOUNTING OF LOCAL HETEROGENEITY |
| CN100595773C (en) * | 2005-01-24 | 2010-03-24 | 独立行政法人海洋研究开发机构 | Simulation system |
| WO2006124878A2 (en) | 2005-05-12 | 2006-11-23 | Mark Lawson Palmer | Method for achieving virtual resolution enhancement of a diagnostic imaging modality by using coupled fea analyses |
| BRPI0611500A2 (en) | 2005-05-26 | 2011-02-22 | Exxonmobil Upstream Res Co | method for analyzing the connected quality of a hydrocarbon reservoir |
| FR2886742B1 (en) * | 2005-06-02 | 2007-07-27 | Inst Francais Du Petrole | METHOD OF CHANGING SCALE OF ABSOLUTE PERMEABILITIES TO BUILD A FLOW SIMULATION MODEL |
| CA2612093C (en) * | 2005-06-14 | 2014-03-11 | Schlumberger Canada Limited | Apparatus, method and system for improved reservoir simulation using an algebraic cascading class linear solver |
| CN101238465B (en) | 2005-07-27 | 2010-10-27 | 埃克森美孚上游研究公司 | Well modeling associated with extraction of hydrocarbons from subsurface formations |
| EP1922669A2 (en) * | 2005-07-27 | 2008-05-21 | ExxonMobil Upstream Research Company | Well modeling associated with extraction of hydrocarbons from subsurface formations |
| US8249844B2 (en) * | 2005-07-27 | 2012-08-21 | Exxonmobil Upstream Research Company | Well modeling associated with extraction of hydrocarbons from subsurface formations |
| FR2891383B1 (en) * | 2005-09-26 | 2008-07-11 | Inst Francais Du Petrole | METHOD FOR SIMULATING FLUID FLOWS WITHIN A DISCRETE MEDIA BY A HYBRID MESH |
| WO2008008121A2 (en) | 2006-07-07 | 2008-01-17 | Exxonmobil Upstream Research Company | Upscaling of reservoir models by reusing flow solutions from geologic models |
| US9031819B2 (en) * | 2006-07-14 | 2015-05-12 | Exa Corporation | Cascaded eddy simulation |
| US7983883B2 (en) * | 2006-08-14 | 2011-07-19 | Exxonmobil Upstream Research Company | Enriched multi-point flux approximation |
| US7574339B2 (en) * | 2006-09-19 | 2009-08-11 | The Protomold Company | Automated generation of lean models for injection molding simulation |
| EP2156365B1 (en) | 2007-05-24 | 2013-08-14 | ExxonMobil Upstream Research Company | Method of improved reservoir simulation of fingering systems |
| US20100191516A1 (en) * | 2007-09-07 | 2010-07-29 | Benish Timothy G | Well Performance Modeling In A Collaborative Well Planning Environment |
| BRPI0820870A2 (en) | 2007-12-13 | 2015-06-16 | Exxonmobil Upstream Res Co | Method for simulating a reservoir model. |
| CN101903805B (en) * | 2007-12-21 | 2013-09-25 | 埃克森美孚上游研究公司 | Modeling in sedimentary basins |
| US20110087471A1 (en) * | 2007-12-31 | 2011-04-14 | Exxonmobil Upstream Research Company | Methods and Systems For Determining Near-Wellbore Characteristics and Reservoir Properties |
| CA2716539C (en) * | 2008-03-28 | 2015-07-21 | Exxonmobil Upstream Research Company | Computing a consistent velocity vector field from a set of fluxes |
| CA2717373A1 (en) * | 2008-04-17 | 2009-12-03 | Exxonmobil Upstream Research Company | Robust optimization-based decision support tool for reservoir development planning |
| WO2009128972A1 (en) * | 2008-04-18 | 2009-10-22 | Exxonmobil Upstream Research Company | Markov decision process-based decision support tool for reservoir development planning |
| US8775361B2 (en) * | 2008-04-21 | 2014-07-08 | Exxonmobil Upstream Research Company | Stochastic programming-based decision support tool for reservoir development planning |
| CN102165449B (en) * | 2008-05-16 | 2014-01-29 | 雪佛龙美国公司 | Multi-scale method for multi-phase flow in porous media |
| CA2729770C (en) * | 2008-07-03 | 2016-12-13 | Chevron U.S.A. Inc. | Multi-scale finite volume method for reservoir simulation |
| CA2730149A1 (en) * | 2008-09-30 | 2010-04-08 | Exxonmobil Upstream Research Company | Method for solving reservoir simulation matrix equation using parallel multi-level incomplete factorizations |
| CA2730446A1 (en) * | 2008-09-30 | 2010-04-08 | Exxonmobil Upstream Research Company | Self-adapting iterative solver |
| US8280709B2 (en) * | 2008-10-03 | 2012-10-02 | Schlumberger Technology Corporation | Fully coupled simulation for fluid flow and geomechanical properties in oilfield simulation operations |
| BRPI0919572A2 (en) * | 2008-10-09 | 2019-09-24 | Chevron Usa Inc | multi-scale computer implemented method, computer implemented system, and method for operating a subsurface reservoir |
| CA2739590C (en) * | 2008-11-20 | 2017-01-03 | Exxonmobil Upstream Research Company | Sand and fluid production and injection modeling methods |
| CN102282562B (en) | 2009-01-13 | 2015-09-23 | 埃克森美孚上游研究公司 | Optimizing well operating plans |
| US8350851B2 (en) * | 2009-03-05 | 2013-01-08 | Schlumberger Technology Corporation | Right sizing reservoir models |
| US8339396B2 (en) * | 2009-03-05 | 2012-12-25 | Schlumberger Technology Corporation | Coarsening and splitting techniques |
| US8655632B2 (en) * | 2009-09-03 | 2014-02-18 | Schlumberger Technology Corporation | Gridless geological modeling |
| WO2011043862A1 (en) * | 2009-10-07 | 2011-04-14 | Exxonmobil Upstream Research Company | Discretized physics-based models and simulations of subterranean regions, and methods for creating and using the same |
| CA2778122C (en) * | 2009-10-28 | 2019-01-15 | Chevron U.S.A. Inc. | Multiscale finite volume method for reservoir simulation |
| BR112012009045B1 (en) | 2009-11-12 | 2020-12-08 | Exxonmobil Upstream Research Company | methods to generate a three-dimensional simulation grid for a reservoir model, simulation railing apparatus and reservoir simulator |
| CN102667804A (en) * | 2009-11-23 | 2012-09-12 | 埃克森美孚上游研究公司 | Method and system for modeling geologic properties using homogenized mixed finite elements |
| CN102640163B (en) | 2009-11-30 | 2016-01-20 | 埃克森美孚上游研究公司 | For the adaptability Newton method of reservoir simulation |
| WO2011097055A2 (en) * | 2010-02-02 | 2011-08-11 | Conocophillips Company | Multilevel percolation aggregation solver for petroleum reservoir simulations |
| WO2011136861A1 (en) * | 2010-04-30 | 2011-11-03 | Exxonmobil Upstream Research Company | Method and system for finite volume simulation of flow |
| US8463586B2 (en) | 2010-06-22 | 2013-06-11 | Saudi Arabian Oil Company | Machine, program product, and computer-implemented method to simulate reservoirs as 2.5D unstructured grids |
| WO2012003007A1 (en) | 2010-06-29 | 2012-01-05 | Exxonmobil Upstream Research Company | Method and system for parallel simulation models |
| US9187984B2 (en) | 2010-07-29 | 2015-11-17 | Exxonmobil Upstream Research Company | Methods and systems for machine-learning based simulation of flow |
| US9058445B2 (en) | 2010-07-29 | 2015-06-16 | Exxonmobil Upstream Research Company | Method and system for reservoir modeling |
| AU2011283190A1 (en) | 2010-07-29 | 2013-02-07 | Exxonmobil Upstream Research Company | Methods and systems for machine-learning based simulation of flow |
| AU2011289788B2 (en) | 2010-08-09 | 2015-03-19 | Conocophillips Company | Reservoir upscaling method with preserved transmissibility |
| US8386227B2 (en) | 2010-09-07 | 2013-02-26 | Saudi Arabian Oil Company | Machine, computer program product and method to generate unstructured grids and carry out parallel reservoir simulation |
| WO2012033650A2 (en) * | 2010-09-07 | 2012-03-15 | Saudi Arabian Oil Company | Machine, computer program product and method to generate unstructured grids and carry out parallel reservoir simulation |
| US8433551B2 (en) | 2010-11-29 | 2013-04-30 | Saudi Arabian Oil Company | Machine, computer program product and method to carry out parallel reservoir simulation |
| GB2502432B (en) | 2010-09-20 | 2018-08-01 | Exxonmobil Upstream Res Co | Flexible and adaptive formulations for complex reservoir simulations |
| WO2012071090A1 (en) | 2010-11-23 | 2012-05-31 | Exxonmobil Upstream Research Company | Variable discretization method for flow simulation on complex geological models |
| US8660824B2 (en) | 2010-12-23 | 2014-02-25 | Schlumberger Technology Corporation | Discretely gridding complex fracturing for simulation |
| US9279314B2 (en) | 2011-08-11 | 2016-03-08 | Conocophillips Company | Heat front capture in thermal recovery simulations of hydrocarbon reservoirs |
| US9489176B2 (en) | 2011-09-15 | 2016-11-08 | Exxonmobil Upstream Research Company | Optimized matrix and vector operations in instruction limited algorithms that perform EOS calculations |
| US9366121B2 (en) | 2012-02-06 | 2016-06-14 | Halliburton Energy Services, Inc. | Modeling fracturing fluid leak-off |
| US9753180B2 (en) | 2012-03-28 | 2017-09-05 | Exxonmobil Upstream Research Company | Method for multiphase flow upscaling |
| EP2901363A4 (en) | 2012-09-28 | 2016-06-01 | Exxonmobil Upstream Res Co | Fault removal in geological models |
| US9811614B2 (en) * | 2013-03-13 | 2017-11-07 | The United States Of America, As Represented By The Secretary Of The Navy | System and method for correcting a model-derived vertical structure of ocean temperature and ocean salinity based on velocity observations |
| CN103246783B (en) * | 2013-05-23 | 2016-04-06 | 合肥工业大学 | A kind of multiple dimensioned Random Coupling modeling method of water-bearing media model |
| US10914864B2 (en) | 2014-07-16 | 2021-02-09 | Schlumberger Technology Corporation | Multiscale method for reservoir models |
| CA2948667A1 (en) | 2014-07-30 | 2016-02-04 | Exxonmobil Upstream Research Company | Method for volumetric grid generation in a domain with heterogeneous material properties |
| US10803534B2 (en) | 2014-10-31 | 2020-10-13 | Exxonmobil Upstream Research Company | Handling domain discontinuity with the help of grid optimization techniques |
| AU2015339883B2 (en) | 2014-10-31 | 2018-03-29 | Exxonmobil Upstream Research Company | Methods to handle discontinuity in constructing design space for faulted subsurface model using moving least squares |
| NO346059B1 (en) | 2015-05-01 | 2022-01-24 | Geoquest Systems Bv | Multiphase flow in porous media |
| US10280722B2 (en) | 2015-06-02 | 2019-05-07 | Baker Hughes, A Ge Company, Llc | System and method for real-time monitoring and estimation of intelligent well system production performance |
| CN105760602A (en) * | 2015-12-30 | 2016-07-13 | 南京航空航天大学 | Total flow field numerical simulation method for finite volume weighted essentially non-oscillatory scheme |
| US10482202B2 (en) | 2016-06-30 | 2019-11-19 | The Procter & Gamble Company | Method for modeling a manufacturing process for a product |
| HUE064459T2 (en) | 2016-12-23 | 2024-03-28 | Exxonmobil Tech And Engineering Company | Method and system for stable and efficient reservoir simulation using stability proxies |
| CN106934185B (en) * | 2017-04-27 | 2019-08-06 | 中国石油大学(华东) | A kind of multiple dimensioned flow simulating method of the fluid structurecoupling of elastic fluid |
| US10169332B2 (en) | 2017-05-16 | 2019-01-01 | International Business Machines Corporation | Data analysis for automated coupling of simulation models |
| WO2019147689A1 (en) | 2018-01-23 | 2019-08-01 | Baker Hughes, A Ge Company, Llc | Methods of evaluating drilling performance, methods of improving drilling performance, and related systems for drilling using such methods |
| CN109095617B (en) * | 2018-09-11 | 2022-02-18 | 浙江省海洋水产养殖研究所 | Ecological system for purifying seawater culture water by using compound artificial wetland in subtropical regions and purification method |
| US10808517B2 (en) | 2018-12-17 | 2020-10-20 | Baker Hughes Holdings Llc | Earth-boring systems and methods for controlling earth-boring systems |
| CN109902412B (en) * | 2019-03-07 | 2023-02-17 | 南方科技大学 | Prediction method and device of underground water flow value, server and storage medium |
| CN110781623B (en) * | 2019-10-23 | 2021-05-04 | 北京理工大学 | Unit interface generation method for finite volume method |
| CN110880048B (en) * | 2019-11-06 | 2022-06-21 | 国网湖北省电力有限公司宜昌供电公司 | Cascade reservoir ecological random optimization scheduling model and solving method |
| CN113609723B (en) * | 2021-07-19 | 2022-08-05 | 西安交通大学 | Overlapped grid parallel data transmission method in steam generator |
Family Cites Families (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5321612A (en) * | 1991-02-26 | 1994-06-14 | Swift Energy Company | Method for exploring for hydrocarbons utilizing three dimensional modeling of thermal anomalies |
| FR2726091B1 (en) * | 1994-10-19 | 1996-12-20 | Elf Aquitaine | METHOD OF ANALYSIS AND TREATMENT OF SEISMIC DATA REFLECTION FOR THE DETERMINATION OF A FINE SPEED FIELD FOR CORRECTION OF HYPERBOLICITY |
| US5729451A (en) * | 1995-12-01 | 1998-03-17 | Coleman Research Corporation | Apparatus and method for fusing diverse data |
| US5923329A (en) * | 1996-06-24 | 1999-07-13 | National Research Council Of Canada | Method of grid generation about or within a 3 dimensional object |
| US6018497A (en) * | 1997-02-27 | 2000-01-25 | Geoquest | Method and apparatus for generating more accurate earth formation grid cell property information for use by a simulator to display more accurate simulation results of the formation near a wellbore |
| US6106561A (en) | 1997-06-23 | 2000-08-22 | Schlumberger Technology Corporation | Simulation gridding method and apparatus including a structured areal gridder adapted for use by a reservoir simulator |
| GB2336008B (en) | 1998-04-03 | 2000-11-08 | Schlumberger Holdings | Simulation system including a simulator and a case manager adapted for organizing data files |
| WO1999057418A1 (en) | 1998-05-04 | 1999-11-11 | Schlumberger Evaluation & Production (Uk) Services | Near wellbore modeling method and apparatus |
| US6106581A (en) * | 1998-06-02 | 2000-08-22 | Ciba Specialty Chemicals Corporation | Ozone stability of dyed and undyed polyamide-containing material |
| US6721694B1 (en) * | 1998-10-13 | 2004-04-13 | Raytheon Company | Method and system for representing the depths of the floors of the oceans |
| US6185512B1 (en) * | 1998-10-13 | 2001-02-06 | Raytheon Company | Method and system for enhancing the accuracy of measurements of a physical quantity |
| US6826520B1 (en) | 1999-06-24 | 2004-11-30 | Exxonmobil Upstream Research Company | Method of upscaling permeability for unstructured grids |
| US6266619B1 (en) | 1999-07-20 | 2001-07-24 | Halliburton Energy Services, Inc. | System and method for real time reservoir management |
| US7006959B1 (en) | 1999-10-12 | 2006-02-28 | Exxonmobil Upstream Research Company | Method and system for simulating a hydrocarbon-bearing formation |
| AU8025000A (en) | 1999-10-13 | 2001-04-23 | Trustees Of Columbia University In The City Of New York, The | Petroleum reservoir simulation and characterization system and method |
| WO2001073476A1 (en) | 2000-03-27 | 2001-10-04 | Ortoleva Peter J | Method for simulation of enhanced fracture detection in sedimentary basins |
| GB0017227D0 (en) | 2000-07-14 | 2000-08-30 | Schlumberger Ind Ltd | Fully coupled geomechanics in a commerical reservoir simulator |
| US6631202B2 (en) | 2000-12-08 | 2003-10-07 | Landmark Graphics Corporation | Method for aligning a lattice of points in response to features in a digital image |
| FR2823877B1 (en) | 2001-04-19 | 2004-12-24 | Inst Francais Du Petrole | METHOD FOR CONSTRAINING BY DYNAMIC PRODUCTION DATA A FINE MODEL REPRESENTATIVE OF THE DISTRIBUTION IN THE DEPOSIT OF A CHARACTERISTIC PHYSICAL SIZE OF THE BASEMENT STRUCTURE |
| US7480206B2 (en) | 2004-09-13 | 2009-01-20 | Chevron U.S.A. Inc. | Methods for earth modeling and seismic imaging using interactive and selective updating |
-
2003
- 2003-03-06 US US10/383,908 patent/US6823297B2/en not_active Expired - Lifetime
-
2004
- 2004-03-05 WO PCT/US2004/006817 patent/WO2004081845A1/en active Application Filing
- 2004-03-05 CA CA2518240A patent/CA2518240C/en not_active Expired - Lifetime
- 2004-03-05 EP EP04718060A patent/EP1606760B1/en not_active Expired - Lifetime
- 2004-03-05 AU AU2004219229A patent/AU2004219229B2/en not_active Expired
- 2004-11-22 US US10/995,764 patent/US7546229B2/en not_active Expired - Lifetime
-
2005
- 2005-10-05 NO NO20054586A patent/NO334387B1/en unknown
Also Published As
| Publication number | Publication date |
|---|---|
| WO2004081845A1 (en) | 2004-09-23 |
| CA2518240C (en) | 2013-05-14 |
| US7546229B2 (en) | 2009-06-09 |
| EP1606760A4 (en) | 2010-12-29 |
| NO20054586D0 (en) | 2005-10-05 |
| NO334387B1 (en) | 2014-02-24 |
| EP1606760B1 (en) | 2013-01-09 |
| EP1606760A1 (en) | 2005-12-21 |
| US6823297B2 (en) | 2004-11-23 |
| NO20054586L (en) | 2005-12-05 |
| US20040176937A1 (en) | 2004-09-09 |
| US20050177354A1 (en) | 2005-08-11 |
| AU2004219229B2 (en) | 2010-11-18 |
| AU2004219229A1 (en) | 2004-09-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA2518240C (en) | Multi-scale finite-volume method for use in subsurface flow simulation | |
| CA2588226C (en) | Multi-scale finite-volume method for use in subsurface flow simulation | |
| Jenny et al. | Multi-scale finite-volume method for elliptic problems in subsurface flow simulation | |
| Wen et al. | Upscaling hydraulic conductivities in heterogeneous media: An overview | |
| AU2011283192B2 (en) | Methods and systems for machine-learning based simulation of flow | |
| US20080208539A1 (en) | Method, apparatus and system for reservoir simulation using a multi-scale finite volume method including black oil modeling | |
| AU2011283191A1 (en) | Methods and systems for machine-learning based simulation of flow | |
| AU2011283193A1 (en) | Methods and systems for machine-learning based simulation of flow | |
| Tatomir et al. | Modeling two phase flow in large scale fractured porous media with an extended multiple interacting continua method | |
| Faigle et al. | Efficient multiphysics modelling with adaptive grid refinement using a MPFA method | |
| Wu et al. | A physically based approach for modeling multiphase fracture–matrix interaction in fractured porous media | |
| Farah et al. | Flow modelling of unconventional shale reservoirs using a DFM-MINC proximity function | |
| Rezapour et al. | Upscaling of geological models of oil reservoirs with unstructured grids using lifting-based graph wavelet transforms | |
| Faigle et al. | Multi-physics modeling of non-isothermal compositional flow on adaptive grids | |
| Cao et al. | An improved boundary element method for modeling fluid flow through fractured porous medium | |
| Dong et al. | Numerical modeling of contaminant transport in fractured porous media using mixed finite-element and finitevolume methods | |
| Dehotin et al. | Modeling of hydrological processes using unstructured and irregular grids: 2D groundwater application | |
| Craig et al. | Using the extended finite element method for simulation of transient well leakage in multilayer aquifers | |
| Ahmed et al. | HT2DINV: A 2D forward and inverse code for steady-state and transient hydraulic tomography problems | |
| Garrido et al. | Implicit treatment of compositional flow | |
| Efendiev et al. | Multiscale simulations of porous media flows in flow-based coordinate system | |
| Sandve | Multiscale simulation of flow and heat transport in fractured geothermal reservoirs | |
| Wu et al. | A physically based numerical approach for modeling fracture-matrix interaction in fractured reservoirs | |
| Marefat et al. | Preserving vorticity improves the accuracy of multiscale solvers | |
| Ewing | Aspects of numerical methods in multiphase flows |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| EEER | Examination request |