Professional Documents
Culture Documents
gravity2. The gravity-segregation concept has also been used Description of the Model
to compute the fluid levels in the matrix and fractures to The developed simulator is a parallel, 3D, fully implicit,
account for the gravity contribution3. Dual porosity models equation-of-state compositional model that uses numerical
also must account for the saturation distribution in a matrix tools for solving very large, sparse linear systems arising from
block. Because the saturation is evaluated at the center of a discretization of the governing partial differential equations. A
gridblock, it represents an average value for that gridblock. generalized dual porosity model, the multiple-interacting-
The pseudo-capillary-pressure concept has been used to continua (MINC), has been implemented in this simulator.
account for both gravity effects and nonuniform saturations The matrix blocks are discretized into subgrids in both
within a matrix block2,4. The method of subgridding horizontal and vertical directions (Fig. 2) to offer a more
discretization has also been applied to this problem5-9. In this accurate transient-flow description in matrix blocks. For
approach, the matrix is divided into a number of subgrids coupling of the fracture and matrix continua, no analytical
where pressures and saturations are calculated for each approximations are made. Instead, numerical methods are used
subgrid. The subgridding approach can also take into account to treat the transient flow of fluid between matrix and
transient effects. fractures. To overcome the computationally intensive
Pruess and Narasimhan6 introduced the so called “multiple problems raised by subgridding the matrix blocks in modeling
interacting continua” (MINC) method. All the fractures are large-scale naturally fractured reservoirs with an order of a
grouped into one continuum and all the matrix blocks into million gridblocks, the dual porosity simulator was developed
another, resulting in two interacting continua coupled through with the capability to run in a parallel processing platform
a mass-transfer function determined by the size and shape of using a cluster of computers.
the blocks, as well as by the local difference in potentials Two overlapping continua, one corresponding to the
between the two continua. Gilman10 presented two possible fracture medium and one corresponding to the matrix medium,
divisions of matrix blocks into multiple subgrids, one called are considered. Thus two values for most variables and
stacked blocks and the other nested blocks (Fig. 1). Stacked parameters are attributed to each spatial location. The
blocks represent a system where horizontal fractures are the equations of flow and component mole conservation are
primary flow paths and transient flow in the lateral direction is written independently for each medium and should hold at
not important. Nested blocks are similar to the MINC every point of the fracture and matrix medium and at all times.
assumption, representing a more randomly fractured medium. Transfer of fluids between the two media is taken into
In this formulation, the solution of the matrix equations will be consideration by a source/sink (transfer) function. Isothermal
a major portion of the total computational effort. However, multicomponent and multiphase flow in a porous medium can
phase segregation inside matrix blocks is possible, improving be described using three different types of equations.
the calculation when gravity is important. Neglecting dispersion and mutual solubility between water
Beckner et al.7 presented a dual porosity model in which and hydrocarbon phases, for a system consisting of nc
matrix blocks are discretized in such way that basically hydrocarbon components and np fluid phases (excluding the
combines Gilman’s10 stacked and nested subgrids. In the aqueous phase), the three types of equations are
lateral direction, the subgrid is the same as in the MINC mathematically expressed for a control volume in the
method, reducing a two-dimensional problem to one following sections12.
dimension. Subgrids in the vertical direction would model
fluid segregation due to gravity with much better resolution. Component Mole Conservation Equations. Using
Beckner et al.7 studied the sensitivity of matrix-block multiphase and multicomponent Darcy's law, in terms of
discretization on production forecasts and found differences as moles per unit time, the hydrocarbon component conservation
large as 35% in cumulative oil recovery between discretized equations, for fracture and matrix systems are the following:
and non-discretized matrix-block models. This difference
increases with decreasing matrix permeability. Aldejain8 Fracture system (subscript f):
investigated the effect of subgridding for a single-phase tracer-
∂
(
Vb (φ f N fi ) − Vb ∇ ⋅ ∑ λ fjξ fj x fij ∇Pfj − γ fj ∇D f )
n
p
injection case using the dual porosity option of UTCHEM11.
He concluded that using a higher number of subgrids yields a ∂t j =1
more accurate result and showed that, in a very low matrix
permeability where the transient flow is dominant, the effect − qi + τ mfi = 0
of subgridding could be significant.
Recently, Naimi-Tajdar9 and Naimi-Tajdar et al.12 (1)
developed a fully implicit parallel compositional simulator for Matrix system (subscript m):
large-scale naturally fractured reservoirs, capable of modeling
∂
( )
n
These equations also hold for water by inserting the heights (no explicit gravity effect). The effect of matrix block
properties of the aqueous phase. permeability is investigated in the next series of runs and
finally, the effect of the matrix capillary pressure is presented
Volume Constraint Equations. The volume constraint states in the last series of runs. The subsequent cases are run and
that the pore volume in each cell must be filled completely by compared with the 2x2, 2x4, 4x2, 4x4, 2x6, and 6x2 subgrids
the fluid volumes. The volume constraint equations for both in which the first number refers to the number of subgrids in
fracture and matrix media are the same and are as follows: horizontal direction and the second number refers to the
number of subgrids in vertical direction.
nc +1 np
∑ N fi ∑ L jν j = 1
i =1 j =1
(fracture system) (3)
Case 1: 1D Waterflood With Explicit Gravity Effect. A 1D
waterflood in dual porosity media is used to investigate the
effect of subgridding. The fracture network consists of eight
nc +1 np
∑ N mi ∑ L jν j = 1
gridblocks of 80x80x30 ft with matrix blocks of 10x10x10 ft
(matrix system) (4) inside every fracture gridblock. In this case, water is injected
i =1 j =1
into the first fracture gridblock at a constant rate of 498.4
STB/D (2000 ft3/day) and liquids are produced from the last
Phase Equilibrium Equations. With the assumption of local fracture gridblock at a constant pressure of 3900 psia. The
thermodynamic equilibrium for the hydrocarbon phases, the initial pressure is 4000 psia. Zero capillary pressure for both
criterion of phase equilibrium applies: fracture and matrix, straight-line relative permeabilities for
( ) ( )
fracture, and Corey model relative permeabilities for the
ln f fig − ln f fio = 0 (fracture system) (5) matrix are used. The input parameters are presented in Table
1. The matrix block heights are 10 ft while the fracture
( ) ( )
ln f mig − ln f mio = 0 (matrix system) (6)
gridblock heights are 30 ft. Hence, gravity has an explicit
effect on the fluid exchange between the matrix and the
fracture. Figures 3 through 5 show the oil recovery, oil
Independent Variables. Equations 1 through 6 describe the production rate and water cut vs. time for these cases. A very
fluid flow through porous media in naturally fractured interesting fact can be seen by studying these figures. If we
reservoirs. There are 2(2nc+2) equations. Independent start with the 2x2 subgrids and increase only the number of
unknowns are chosen as lnKi, Ni, Pw, Nw (N = moles per unit horizontal subgrids to four and six (cases with 4x2 and 6x2
pore volume) in each medium, fracture and matrix, which subgrids), not much improvement is achieved compared to the
gives 2(2nc+2) primary variables. All the fluid-related results of base case of 4x4 subgrids. The case with 6x2
properties and variables in Eqs. 1 through 6 can be expressed subgrids shows about 15% lower oil recovery, since the effect
as a function of the selected independent variables. of gravity drainage is much smaller relative to the fluid flow in
horizontal direction. However, by increasing the number of
Transfer Function and Boundary Conditions. The transfer vertical subgrids, the results of these cases converge to the
function for each component (hydrocarbon components and results of the base case with 4x4 subgrids. Since there are
more subgrids in the vertical direction in the cases with 2x4
water), τ mfi are evaluated at the boundary between the matrix and 2x6 subgrids, these cases show a higher oil recovery and
and fracture media and have the following forms: lower water cut. Therefore for applications where the gravity
drainage is dominant, the number of vertical subgrids plays an
Nb
∂
τ mfi = NM ⋅ Vb ∑ (φ m N mi )l (7) important role and simulators that do not discretize in the
l =1 ∂t vertical direction could produce erroneous results.
Where NM is the number of matrix blocks within a Case 2: 2D Waterflood With No Explicit Gravity Effect. A
fracture gridblock (may be a fractional number), and Nb is the quarter of a five-spot waterflood problem1 is used to
number of matrix subgrids. No-flow boundary conditions for investigate the effect of subgridding. In this case, water is
component mole consevation equations in the fracture system injected into a quarter-five-spot model at a rate of 200 STB/D
are considered. The boundary condition for matrix blocks is and water and oil are produced at a constant bottomhole
continuity of all phase pressures. pressure of 3900 psia. The reservoir is 600 ft long, 600 ft
wide, and 30 ft thick. The fracture media is discretized into
Case Studies 8x8 uniform gridblocks in the x and y directions, respectively,
In this section, we present several cases to investigate the and has one 30-ft-thick gridblock in the z direction. Zero
effect of subgridding using this developed dual porosity capillary pressure is used for both fracture and matrix media,
model. In the first series of runs, we present a 1D waterflood and the relative permeability curves in the fracture and matrix
case in which the matrix block heights are smaller than media are shown in Fig. 6. The input parameters are presented
fracture gridblock heights (explicit gravity effect) and show in Table 2. Note that in this case, the matrix block heights are
the effect of vertical and horizontal subgridding on the oil the same as the fracture gridblock heights of 30 ft. Hence,
recovery, oil production, and water cut. The second series of gravity has no explicit effect on the fluid exchange between
runs present a 2D waterflood case in which the matrix block the matrix and the fracture. Figures 7 through 9 compare the
heights are the same as the corresponding fracture gridblock oil recovery, oil production rate, and water cut for different
4 IPTC 11195
subgridding cases. As expected, the 2x4 and 2x6 runs show a subgrids in the vertical and horizontal directions to
higher oil recovery compared to the base case results of 4x4 obtain accurate results.
run, while the 6x2 run shows a lower oil recovery. In this case, 4. Based on the simulation studies carried out in this work,
where gravity has no explicit effect on the fluid exchange when capillary imbibition recovery mechanism is
between the matrix and the fracture, increasing the number of dominant in naturally fractured reservoirs, using a small
vertical subgrids keeping horizontal subgrids the same will number of subgrids seems to be sufficient for typical
overestimate the oil recovery whereas increasing the number cases.
of horizontal subgrids keeping vertical subgrids the same will
underestimate the oil recovery. Another interesting Acknowledgement
observation is that increasing the number of vertical subgrids The financial support of the Reservoir Simulation Joint
from 4 to 6 had no significant impact on oil recovery and oil Industry Project in the Center for Petroleum and Geosystems
and water production rates which can be attributed to the Engineering at The University of Texas at Austin is gratefully
absence of explicit gravity effect. acknowledged.
5. Saidi, A.M.: “Simulation of Naturally Fractured Reservoir,” Table 2. Input parameters used for 2D waterflood.
paper SPE 12270 presented at the 1983 SPE Reservoir
Description Fracture Matrix
Simulation Symposium, San Francisco, CA, Nov. 16-18.
6. Pruess, K. and T.N. Narasimhan: “A Practical Method for Number of gridblocks 8x8x1 4x4 and 16x16
Modeling Fluid and Heat Flow in Fractured Porous Media,”
paper SPE 10509 presented at the Sixth SPE Symposium on Size of gridblocks 75x75x30 ft 10x10x30 ft
Reservoir Simulation, New Orlean, LA, Februrary 1982, Soc.
Pet. Eng. J., Vol. 25, No. 1 (Feb. 1985) 14-26. Porosity 0.01 0.19
7. Beckner, B.L., H.M. Chan, A.E. McDonald, S.O. Wooten, and
Permeability 500 md 1.0 md
T.A. Jones: “Simulating Naturally Fractured Reservoirs Using a
Subdomain Method,” paper SPE 21241 presented at the 1991 Initial water saturation 0.0001 0.25
Society of Petroleum Engineers Symposium on Reservoir
Simulation, Anaheim, CA, Feb. 17-20. Water viscosity 0.5 cp 0.5 cp
8. Aldejain, Abdulaziz A.: “Implementation of a Dual Porosity in a
Chemical Flooding Simulator,” Ph.D. dissertation, The U. of Oil viscosity 2.0 cp 2.0 cp
Texas, Austin (1999). Residual oil saturation 0.0 0.3
9. Naimi-Tajdar, Reza: “Development and Implementation of a
Naturally Fractured Reservoir Model into a Fully Implicit, Oil endpoint relative permeability 1.0 0.92
Equation-of-State Compositional, Parallel Simulator,” Ph.D.
dissertation, The U. of Texas, Austin (2005). Corey exponent for oil 2.15 1.8
10. Gilman, J.R.: “An Efficient Finite-Difference Method for
Residual water saturation 0.0 0.25
Simulating Phase Segregation in Matrix Blocks in Double
Porosity Reservoirs,” Soc. Pet. Eng. Res. Eng. (July 1986). Water endpoint relative perm 1.0 0.2
11. Delshad, M., G.A. Pope, and K. Sepehrnoori: “A Compositional
Simulator for Modeling Surfactant Enhanced Aquifer Corey exponent for water 1.46 1.18
Remediation, 1 Formulation,” J. Contamin. Hydrol. (1996) 303.
12. Naimi-Tajdar, R., C. Han, K. Sepehrnoori, T.J. Arbogast, and Initial reservoir pressure 4000 psia 4000 psia
M.A. Miller: “A Fully Implicit, Compositional, Parallel Water injection rate 200 STB/D (constant)
Simulator for IOR Processes in Fractured Reservoirs,” paper
SPE 100079 presented at the 2006 SPE/DOE Symposium on Production well pressure 3900 psia (constant)
Improved Oil Recovery, Tulsa, OK, 22-26 April 2006.
Single
matrix Idealized
Matrix block gridblock
subgridding
25
50
20
40
4x4 Subgrids
2x6 Subgrids
Oil Recovery (%)
10 2x4
20
2x2 Subgrids
5 10
0 0
0 10 20 30 40 50 60 70 80 90 100 0 200 400 600 800 1000 1200
Time (days) Time (days)
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids 2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
Fig. 3. Effect of subgridding on oil recovery vs. time for case 1 Fig. 7. Effect of subgridding on oil recovery vs. time for case 2
(with explicit gravity effect). (no explicit gravity effect).
300 225
125
150
100
100 75
50
50
25
0 0
0 10 20 30 40 50 60 70 80 90 100 0 200 400 600 800 1000 1200
Time (days) Time (days)
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
Fig. 4. Effect of subgridding on oil production rate vs. time for Fig. 8. Effect of subgridding on oil production rate vs. time for
case 1. case 2.
1
1
0.8
0.8
Water Cut (fraction)
0.6
0.6
0.4 0.4
0.2 0.2
0 0
0 10 20 30 40 50 60 70 80 90 100 0 200 400 600 800 1000 1200
Time (days) Time (days)
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids 2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
Fig. 5. Effect of subgridding on water cut vs. time for case 1. Fig. 9. Effect of subgridding on water cut vs. time for case 2.
1
50
0.9
0.8 40
Kro
0.7
Relative Permeability
0.6 30
0.5
0.4 20
0.3
Krw
0.2 10
0.1
0
0
0 200 400 600 800 1000 1200
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (days)
Water Saturation
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
Matrix media Fracture media
Fig. 10. Effect of subgridding on oil production rate vs. time for
Fig. 6. Fracture and matrix relative permeabilities used in case 2.
case 3 (K = 10 md).
IPTC 11195 7
225
100
200
175 80
Oil Production Rate (STB/Day)
150
100
40
75
50 20
25
0
0
0 200 400 600 800 1000 1200
0 200 400 600 800 1000 1200
Time (days)
Time (days)
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
Fig. 11. Effect of subgridding on oil production rate vs. time for Fig. 14. Effect of subgridding on oil production rate vs. time for
case 3. case 4 (with capillary pressure).
225
1
200
0.8 175
0.6
125
100
0.4
75
50
0.2
25
0
0
0 200 400 600 800 1000 1200
0 200 400 600 800 1000 1200
Time (days)
Time (days)
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
Fig. 12. Effect of subgridding on water cut vs. time for case 3. Fig. 15. Effect of subgridding on oil production rate vs. time for
case 4.
4
1
3.5
0.8
3
Capillary pressure (psi)
2.5
Water Cut (fraction)
0.6
1.5 0.4
0.2
0.5
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 200 400 600 800 1000 1200
Water saturation Time (days)
2x2 Subgrids 2x4 Subgrids 4x2 Subgrids 2x6 Subgrids 6x2 Subgrids 4x4 Subgrids
Fig. 13. Capillary pressure curve used for case 4.
Fig. 16. Effect of subgridding on water cut vs. time for case 4.