1
Received November 17, 2006; Accepted May 29, 2007
c 2008 Begell House, Inc.
Copyright
Nabipour et al.
NOMENCLATURE
g
k
kri
kro
p
pc
Pcm
Pcf
1. INTRODUCTION
The mathematical modeling of flow in porous media, particularly with the ever-increasing power of
computers, is playing a fundamental and increasingly
important role in the prediction of petroleum reservoir performance, groundwater supply, and subsurface
contaminant migration. A critical underlying problem
in these models is the need to predict the behavior of fluid in stratified structures and heterogeneous
geological formations.
Geological surveys clearly indicate that all the sedimentary rocks beneath the surface of the earth are
made up of different layers, each of which has its
own characteristics such as porosity, permeability, and
wettability. The oil saturation profile during gas-oil
gravity drainage (GOGD) in layered porous media
plays an important role in oil recovery efficiency.
Forecasting the behavior of a reservoir is one of the
most important but complicated tasks for petroleum
engineers. Knowledge of oil content of a reservoir is
essential in planning optimum depletion of a given
field. There are many important forces governing fluid
flow in the reservoir such as gravity, capillarity, and
inertia forces (Richardson and Blackwell, 1971).
GOGD is a recovery process in which gravity acts
as the main driving force and gas replaces the de-
S
t
saturation (%)
time (s)
Greek Symbols
Corey exponent
viscosity (cP)
density (kg/m3 )
porosity (%)
potential
pleted volume. In other words, it is a gas-oil displacement process in which gravity forces are dominating.
It may occur in primary stages of oil production
(gas cap expansion drive or gravity segregation) as
well as in supplemental stages, where gas is supplied
from an external source and injected into the reservoir
(Hagoort, 1980).
Dumord and Schols (1974) discovered that residual
oil saturation after gas invasion in highly permeable
sandstone cores containing connate water could be
extremely low. They also experienced very low residual oil saturations in the sand pack gravity drainage
experiments.
Hagoort et al.s (1984) experiments confirmed that
the gravity drainage mechanism could be a very effective oil recovery process in water-wet, connate waterbearing reservoirs, which lowers the remaining oil saturations in the media. Whether these low saturations
are indeed reached in the lifetime of an oil reservoir
depends on the magnitude of the gravitational force
compared to capillary forces, the functionality of the
oil relative permeability, and the reservoir geometry
and heterogeneity.
Dykstra (1978) presented a good example of oil
production by strictly free-fall gravity drainage in
the Lakeview Pool, Midway Sunset oil field. In a
GOGD Process
field study, King and Stiles (1970) demonstrated a
very high displacement efficiency of 87% by gravity
drainage in the East Texas Hawkins reservoir. The
work by Chatzis and Ayatollahi (1993) revealed that
the GOGD process is a very effective mechanism for
the recovery of waterflood residual oil under immiscible conditions.
Evidence shows that gravity drainage is one of the
most effective mechanisms of an oil field development. Although the gravity drainage mechanism is
important, characterizing and modeling the process
are still a great challenge. Almost all the gravity
drainage models are complicated. Some of the models
do not have analytical solutions and have to be solved
numerically (Bennion et al., 1998).
Reservoir heterogeneity has long been recognized
as an important factor in determining reservoir performance, and description of heterogeneity is a crucial
step toward understanding it. The GOGD process in
layered porous media is unstable under certain conditions. Even for two layers, gas fingers through the
less permeable top layer to reach the more permeable bottom layer very quickly (Correa and Firoozabadi, 1996). Under these circumstances, the lowpermeability zone holds a high portion of the wetting
phase (Chatzis and Ayatollahi, 1995).
Goddin et al. (1966) used a finite difference numerical model to study the effects of viscous and
capillary forces on recovery in a field-scale model of a
two-layer, water-wet sandstone reservoir.
Gharbi et al. (1996) investigated the fluid flow in
a highly heterogeneous reservoir to study the detailed
sensitivity of the displacement performance. The results showed that the degree and structure of the
heterogeneity of the reservoir have significant effects
on the efficiency of immiscible displacement using
horizontal wells. Therefore the long horizontal wells
in highly heterogeneous reservoirs do not necessarily
guarantee improved oil recovery (Gharbi et al., 1996).
Correa and Firoozabadi (1996) have provided a criterion for instantaneous gas fingering toward the more
permeable layer. A nonlinear form of the diffusivity
3
equation was used to study gas-oil drainage in layered
systems. Numerical examples showed that recovery
performance of layered systems, unlike homogeneous
media, is sensitive to capillary pressure.
Virnovsky et al. (1998) found that the capillarytrapped oil saturation has a tendency to increase in the
direction of injection into production wells. The proposed model is useful in the calculation of recoverable
oil content and the optimization of a waterflooding
strategy that minimizes oil entrapment in heterogeneous reservoirs.
On the basis of Darcys law and the film flow
theory, Schechter and Guo (1996) developed a new
mathematical model to describe the free-fall gravity
drainage process. A simple nonlinear governing equation in dimensionless form was formulated and solved
numerically as a function of dimensionless time.
Kewen and Horne (2003) mentioned that mathematical models set up to predict oil production accurately
by gravity drainage have been a few. They tried to
find an analytical model that determines the ultimate
oil recovery by free-fall gravity drainage. Also, an
empirical oil recovery model was proposed accordingly to match and predict oil production, and an
approach was also followed to infer capillary pressure
curves from the oil production data by free-fall gravity
drainage.
Tertiary GOGD and the oil bank movement were
also studied by Chatzis and Ayatollahi (1995) in
stratified media. The experimental results showed no
oil trapping in the vicinity of the low- to highpermeability media for water-wet sand packs due to
the water capillary fringe in that zone. The oil followed by gas fingered down to the bottom of a
high-permeability layer resulted in very low residual oil saturation in the gas-invaded low-permeability
zone on the top.
Nabipour et al. (2007) also investigated the enhanced oil recovery efficiency of the mechanism of
thermally assisted gas-oil gravity drainage for secondary and tertiary oil recovery on a fractured laboratory model.
Nabipour et al.
Furthermore, there were a few citations in the literature about the modeling of GOGD in stratified
media. In addition, by changing the permeability contrasts, it is possible to mimic the horizontal fracturing
in the oil zone and its effects on the oil saturation profile and recovery efficiency during the GOGD process.
In this investigation, we tried to find out the basic
effect of macroscopic heterogeneity, including the horizontal fractures in layered porous media. Therefore
a mathematical model has been derived and solved
numerically.
2. NUMERICAL MODELING
Oil recovery prediction during the GOGD process
in heterogeneous, layered porous media as well as
naturally fractured reservoirs is a challenging problem. It is worth mentioning that most of the oil
reservoirs, especially in the Middle East, are layered
and fractured ones (Hernandez, 2002). One of the
basic methods by which we can model a heterogeneous porous medium is to divide the heterogeneous
part into smaller homogenous portions with different parameters and model the interaction of different
portions (Correa and Firoozabadi, 1996).
In gravity drainage, flow occurs vertically inside the
layers by gravity force downward; however, the capillary force acts in the opposite direction. Depletion
continues as long as gravity forces are dominant and
will be hindered when the two forces, gravity and capillary, balance. The basic rules of fluid flow in porous
media have been used to model this process. Darcys
laws for oil and gas phases are, respectively,
u0 =
kkro o
o z
(1)
ug =
kkrg g
g z
(2)
The continuity aligns for oil and gas, assuming constant density and porosity, are
uo
so
+
=0
z
t
(3)
ug
sg
+
=0
z
t
(4)
so
k
pc
g
kro
+ o
= 0 (5)
t
o z
z
gc
The main governing align is three-dimensional, but
the radial dependence can be ignored as the main
changes are in the gravitational direction, the zdirection, which simplifies the model to a transient
one-dimensional align. A schematic view of a typical
GOGD process is shown in Fig. 1.
The boundary conditions on top and at the bottom
of the homogenous layer during the process are
GOGD Process
so
|z=top = 0
z
(6)
(7)
Kro =
0
Kro
S Sro
1 Sro Srw
(8)
A typical capillary pressure correlation was also utilized for this purpose:
Pc = Pc0
S Sro
1 Sro Srw
k kr p0c t
2z
2 p0c k kr t
(11)
b=
g k kr t
c=
2z
It is assumed here that the pore size distribution ()
and residual saturation have been similar for both
high- and low-permeability layers.
a=
(9)
!
Sin1
n+1
n1
Si+1
a n1 + c Si
Si
!
Sin1
n+1
n1
+ a n1 + c Si
Si1
Si
+ Sin+1 = Sin
b Sin
Sin2
(10)
Nabipour et al.
Table 1
Fluid and media properties
Property
Permeability, low-permeability layer (darcy)
Permeability, high-permeability layer (darcy)
Fracture permeability (darcy)
Oil viscosity (cP)
Oil density (kg/m3 )
Pore size distribution constant
Residual oil saturation (Sro )
Residual water saturation (Srw )
Porosity
kr0
pco
Value
1
10
100
30
0.9
2
0.3
0.15
0.3
6
100
k = 10
4. Saturation profile of double-layered, highpermeability layer on the top of a low permeable one
Figure
GOGD Process
7
The ultimate oil recovery factor for case 3 (highpermeability layer on the top of the low-permeability
one) is comparable to that of case 2 (single lowpermeability layer), where the height of capillary rise
in the lower part was sufficient. However, no capillary
rise was detected in the vicinity of high- to lowpermeability media.
However, for the opposite case (low-permeability
layer on the top), the capillary rise in the top layer
results in lower ultimate oil recovery efficiency compared to case 1.
Figure
5. Saturation profile of double-layered, lowpermeability layer on the top of a high permeable one
Nabipour et al.
location at 15 cm
location at 75 cm
GOGD Process
cluding a single horizontal fracture. The properties of
these models are also shown in Table 1.
The saturation profile in the fracture is significantly
different from that in the matrix; the saturation remained almost the same as in the previous cases.
It is worth mentioning that the capillary continuity
was maintained for these cases, as was stated already.
Comparing the results shown in Figs. 79 suggests
that the location of the fracture affects the saturation
profile. It is shown in Fig. 9 that as the fracture
intervenes with the capillary height zone, the liquid
saturation in the fracture cannot attain its residual
value compared to the results shown in Fig. 7.
This numerical modeling has the capability of managing the layer properties such as capillary pressure
and relative permeability. To test the effects of capillary pressure contrast in the fracture and matrix and
also check the capillary continuity between the layers,
the following capillary pressure correlations (Firoozabadi, 1993; Hernandez, 2002) were employed:
1
So Sro
Pcm =
2.91.17 ln
(12)
101.325
1Sro
Pcf =
1
2
0.2150.0081 ln (So0.1)
(13)
101.325
Figure 10 shows the saturation profile of the wetting phase using capillary pressure contrast in the
matrix and fracture, compared with Fig. 7, where the
capillary pressure of layers is assumed to be the same.
The resulting saturation profiles show that the amount
of trapping of the wetting phase in the vicinity of
matrix to fracture has been increased.
3.2. Double-Fractured Case
To test the multifractured media and the effects of
their location on the saturation history as well as the
recovery efficiency, several tests have been done on
double-fractured models. As was mentioned in the
single-fracture case, the permeabilities of the matrix
and fracture are in the order of 1 and 100 darcy,
respectively. Figures 11 and 12 show the predicted
10
Nabipour et al.
liquid saturation profile in double-fractured cases to
indicate the effect of fracture locations.
Figure 13 indicates that if the capillary continuity
exists in the system, the closer the fractures are to the
top, the more the drainage and ultimate recovery will
be, but when the fractures get close to the bottom, this
trend gets reversed. The capillary continuity is maintained in these tests by utilizing high capillarity for
both matrix and fracture. Therefore the existence of
fractures in a given system result in the enhancement
of recovery rate; that is, in single and nonfractured
cases, the rate of recovery is less than that of the
double-fractured case, as shown in Fig. 13.
In fact, the horizontal fractures can help enhance
the recovery only if the drainage of the medium
is managed so that the connectivity between matrix
blocks is not lost.
Figure 14 shows that the amount of oil recovery decreases as the fracture-matrix capillary pressure differentiates to a significant amount, as was mentioned for
the single-fractured case (Fig. 10). Figure 14 clearly
indicates that the assumption of capillary continuity
discredits as the capillary pressure distinction between
Figure 13. Comparison of recovery efficiency for no fracture, single-fracture, and double-fracture cases in case of capillary
continuity
GOGD Process
11
the matrix and fracture increases as a result and gets
away from our assumption of continuity.
Recovery profiles are given in Fig. 15 for the
case in which the matrix and fracture have different
capillary pressure values. As the capillary pressure
contrast increases, the chance of capillary continuity
decreases, and the recovery decreases, too. This also
happens as the fractures get closer to the bottom
of the model, as shown in Fig. 15. It is worth
remembering that the existence of fractures that do
not support connectivity results in a dramatic trapping
of the wetting phase and hence the reduction of
ultimate recovery.
4. EXPERIMENTAL INVESTIGATION
Figure 15. Comparison of recovery efficiency for no fracture, single-fracture, and double-fracture cases in the case of
capillary discontinuity
12
Nabipour et al.
5. CONCLUSION
GOGD Process
13
gas fingering as well as wetting phase trapping
in the stratified model result in low ultimate
recovery compared with the layered model with
capillary continuity.
REFERENCES
Figure
14
Nabipour et al.