You are on page 1of 10

DOI: 10.1002/ente.

201300155

Scaling of Recovery by Cocurrent Spontaneous Imbibition


in Fractured Petroleum Reservoirs
Abouzar Mirzaei-Paiaman*[a, b] and Mohsen Masihi[a]
Cocurrent spontaneous imbibition (COCSI) of an aqueous
phase into matrix blocks arising from capillary forces is an
important mechanism for petroleum recovery from fractured
petroleum reservoirs. In this work, the analytical solution to
the COCSI is used to develop the appropriate scaling equations. In particular, the backflow production of the nonwet-

ting phase at the inlet face is considered. The resulting scaling equations incorporate all factors that influence the process and are found in terms of the Darcy number (Da) and
capillary number, (Ca). The proposed scaling equations are
validated against the published experimental data from the
literature.

Introduction
Spontaneous imbibition of an aqueous phase (i.e., the wetting phase, WP) into matrix blocks arising from capillary
forces is an important mechanism to displace the nonwetting
phase (NWP) from fractured reservoirs.[1] This can be either
countercurrent or cocurrent displacement. In countercurrent
spontaneous imbibition (COUCSI) all the permeable faces
of a rock saturated with NWP are brought into contact with
a WP and the WP flows in the opposite direction to the expelled NWP.[2, 3] However, in cocurrent spontaneous imbibition (COCSI) only a portion of the permeable surfaces is in
contact with a WP and, while keeping the remaining permeable surfaces covered by the NWP, both phases flow in the
same direction.
In a typical COCSI experiment, matrix surfaces covered
by WP and NWP are kept constant during the process.[414] In
this case, level of the WP in the fracture system is kept constant at a certain level and there is no viscous force in the
fracture system. We use the same COCSI case in this study.
There are also other cases in which the recovery performance of a matrix block or a stack of matrix blocks under
advancing WP level in the fracture system are studied.[1520]
Despite the fact that the COCSI can take place in fractured reservoir,[19, 21, 22] most attention has been received on
scale-up of COUCSI.[2325] Use of the scaled-up results of
COUCSI experiments leads to pessimistic forecasts regarding
the rate of recovery and final recovery.[10, 18, 26] It has been emphasized that COCSI is more efficient in terms of both final
recovery and displacement rate than COUCSI.[7, 911, 1820, 26, 27]
Due to the significant differences between the recovery performances in these processes, the corresponding scaling
equations cannot interchangeably be used. Several studies
show that the scaling equations developed for the COUCSI
process fail to scale up the COCSI data.[8, 10, 2830]
A few scaling equations have been proposed for recovery
prediction of COCSI.[3133] They have been applied to one-dimensional displacement as no characteristic length has been
defined for the COCSI. These scaling equations do not incor-

166

porate all of the factors influencing the process; consequently, they cannot properly describe the process. The scaling
equation proposed by Rapoport[31] was developed by using
the inspectional analysis of the main governing equations,
making some simplifying assumptions; including that the
prototype WP/NWP viscosity ratio must be duplicated in the
model tests, initial fluid saturations in the prototype must be
duplicated in the model tests, the relative permeability functions must be the same for both the model and the prototype, and the capillary pressure functions for the both the
model and the prototype must be related by direct proportionality. Several studies have attempted to develop the scaling equation of Rapoport (e.g., Ref. [34, 35]) to COUCSI
data only. The scaling equations of Li[32] and Bourbiaux[33]
were derived based on a restricting approximate solution to
the main governing equations. They take the assumption of
piston-like displacement which is valid only for a few particular cases. However, according to Mirzaei-Paiaman and
Masihi,[25] the development of these two scaling equations is
not consistent with common scaling practices.
The main purpose of this study is to present universal scaling equations for one-dimensional COCSI based on the
recent finding of Schmid[36] who notices that the analytical
solution to unidirectional displacement given by McWhorter
and Sunada[37] applies to COCSI with no artificial boundary
conditions. We consider the consistency between the development of the new scaling equations and common practices
as was considered for COUCSI in Mirzaei-Paiaman and
Masihi.[25] These new scaling equations are rewritten in terms
[a] Dr. A. Mirzaei-Paiaman, Dr. M. Masihi
Department of Chemical and Petroleum Engineering
Sharif University of Technology
P.O. Box 11365-9465, Azadi Ave., Tehran (Iran)
E-mail: Mirzaei1986@gmail.com
[b] Dr. A. Mirzaei-Paiaman
Department of Petroleum Engineering
NISOC, Ahvaz (Iran)

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Energy Technol. 2014, 2, 166 175

Recovery in Fractured Petroleum Reservoirs


of two physically meaningful dimensionless numbers: the
Darcy number (Da) and the capillary number (Ca). Then the
new scaling equations are validated by using experimental
data from the literature. Furthermore, we simplified the scaling equations, which can be useful in some certain applications.

The Analytical Solution


The partial differential equation for one-dimensional horizontal displacement of two immiscible, incompressible fluids
can be expressed by:[37]

uw f Sw  ut  DSw

@Sw
@x

This equation in combination with the material balance condition gives:[37]



@Sw
@f Sw @
@S

DSw w
ut
@x
@x
@t
@x

in which the subscript w denotes the WP, f is the porosity, S


is the saturation, t is the time, ut(=uw+unw) is the total volumetric flux (ut = 0 for countercurrent flow and ut > 0 for unidirectional displacement), the subscript nw denotes the
NWP, x is the spatial coordinate, f(Sw) is the fractional flow
in the absence of capillary pressure defined as

f Sw

krw mnw
krw mnw krnw mw

dP

in which kr is the relative permeability, m is the dynamic viscosity, k is the absolute permeability, and Pc is the capillary
pressure.
The appropriate initial and boundary conditions can be defined as,[37]
Sw x; 0 Swi

Sw 1; t Swi

uw 0; t At1=2

The initial condition in Equation (5) states that at time


zero the porous medium is at initial WP saturation, Swi. The
medium is assumed as a semi-infinite host at Swi at the far
boundary for all times [Eq. (6)]. Equation (7) is an imposed
inlet boundary condition defined by McWhorter and
Sunada[37] to solve the problem given in Equation (2).
McWhorter and Sunada[37] define the positive parameter A
as,
Energy Technol. 2014, 2, 166 175



ut
in which fi is the WP fraction of the total efflux, R uw 0;t

is the ratio of the total volumetric flux to the volumetric flux


of the WP at the inlet (R = 0) for countercurrent flow, and
R = 1 for unidirectional displacement, Sw,BC is the saturation
of WP at the inlet open boundary, fn is the normalized f, and
F is the fractional flow function in the presence of capillary
effects.[37]

fn

f  fi R
1  fi R

9
R Sw;BC bSw Db

Sw
F bfn b db
F Sw 1  R Sw;BC
Sw Swi DSw
Swi
F Sw fn Sw dSw

10

The exact solution to Eqation (2) is given implicitly by


McWhorter and Sunada[37] as:

xSw ; t

2A1  fi R 0
F Sw t1=2


11

in which F is the derivate of F with respect to Sw defined as:


R Sw;BC Db
S
F bf b db
F 0 Sw R Sw;BCw Sw Swi DnSw
Swi
F Sw fn Sw dSw

12

and D(Sw) is the capillary diffusion function defined as,


c
DSw f Sw k mrnw
nw dSw

Z Sw;BC

Sw  Swi DSw
dSw
A
21  fi R Swi
F Sw  fn Sw

The solution expressed in Equation (11) can be used by


first prescribing Sw,BC and calculating F(Sw) from Equation (10), and then finally computing A from Equation (8).
However, the use of Equation (10) is indirect, as it has the
form of an implicit functional equation, from which F(Sw)
has to be extracted. Therefore the computation of F(Sw)
from the integral Eq. (10) is performed by an iterative procedure. A convenient first trial is F(Sw) = 1 as suggested by
McWhorter and Sunada.[37]
By combining Equation (1) with the definition of R and
the self-similar variable l xt1=2, the volumetric flux of the
WP at the inlet boundary can be written as:[36]


D Sw;BC
dSw


uw 0; t 
jl0 t1=2 At1=2
1  f Sw;BC R dl

13

Using this equation, Schmid[36] noticed that the inlet boundary condition imposed by McWhorter and Sunada[37] to
solve the original problem is redundant for the case of both
COUCSI and COCSI and the solution given in Equation (11) describes the standard situation found in the laboratory experiments. Schmid and Geiger[23, 24] and Mirzaei-Paiaman and Masihi[25] use this finding to develop the appropriate scaling equations for COUCSI cases. Mirzaei-Paiaman
et al.[38] used the analytical solution for the COUCSI case to

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

167

A. Mirzaei-Paiaman and M. Masihi


propose a mathematically based index for characterizing the
wettability of reservoir rocks. In an another modeling approach, Cai et al.[13, 14] proposed analytical expressions for
characterizing a spontaneous cocurrent imbibition process of
a wetting fluid into gas-saturated porous media based on the
fractal character of the porous media. In their work,[13, 14] the
mass of imbibed liquid is expressed as a function of the fractal dimensions for pores and for tortuous capillaries, the minimum and maximum hydraulic diameters of pores, and the
ratio for minimum to maximum hydraulic diameters as well
as the porosity, fluid properties, and the fluidsolid interaction.

Application of the Analytic Solution to COCSI


Process
[37]

McWhorter and Sunada considered a special case of unidirectional displacement that could be realized in laboratory
by using a semipermeable membrane that is permeable to
only the WP (i.e., R = 1). However in practice if a NWP is initially present and is being displaced by a WP, depending on
the magnitude of the forced injection rate, three flow regimes may occur in a linear flow domain.[39, 40] If there is no
forced injection at the inlet boundary, there will be countercurrent flow (with respect to the NWP) at the inlet face implying R < 1 (regime 1). As the injection rate increases up to
some critical value, the region of countercurrent flow continually decreases but the R < 1 condition remains (regime 2).
Countercurrent flow ceases as the injection rate at the inlet
reaches this critical value and R = 1 (regime 3). Unidirectional displacement only occurs if the injection rate equals or exceeds this critical value.[39, 40] In the COCSI case considered
in this study there is no forced injection at the inlet, and
therefore regime 1 is expected to take place. Several studies
confirm the existence of some backflow production of NWP
in the inlet in regime 1.[7, 9, 10, 18] Thus, the solution presented
by McWhorter and Sunada[37] when applied to COCSI
should be slightly modified to account for the backflow production of NWP at the inlet boundary. This can be achieved
by assigning a value not much smaller and not very close to
1 (e.g., 0.95) to R and proceeding with the analytical solution. The reason for the choice of such a value comes from
the experimental observations.[7, 9, 10]
The main obstacle of the iterative approach proposed by
McWhorter and Sunada[37] is that to obtain the Sw,m values
from Sw,BC (in which Sw,m is the WP saturation reached after
forced displacement) and R = 1, difficulties can be encountered for computing the solution integrals, because the
values of F(Sw) and fn(Sw) approach each other. This causes
the iterative process to become unstable.[41] Our strategy to
account for the contribution of the backflow production by
assuming R
6 1 avoids the occurrence of any instability.
As the analytic solution presented by McWhorter and
Sunada[37] can be applied to COCSI with no artificial boundary condition,[36] after consideration of the contribution of
the backflow production, it sufficiently incorporates all the
factors influencing the COCSI process.

168

Developing Scaling Equations for One-Dimensional COCSI Case


Equation (7) can be integrated to yield the cumulative recovery of the NWP and Q at any time as:

QW

uw dt 2WAt1=2

14

in which W is the surface area perpendicular to flow direction. Similarly the ultimate cumulative recovery, Q1, defined
as the cumulative recovery of NWP up to the time t* at
which the imbibition front reaches the end of the linear
system L can be derived as:

Q1 W

uw dt 2WAt* 1=2

15

Combining two last equations gives:


 t 1=2
Q

Q1
t*

16

To obtain the reference time t* we can assign:


xSwi ; t* L

17

By combining Equations (11) and (17), t* can be derived as:

t*

L
2AF 0 Swi 1  fi R

2

18

Inserting Equation (18) into Equation (16) yields:


Q
2AF 0 Swi 1  fi R 1=2

t
Q1
L

19

Equation (18) can be inserted in Equation (15) to give the


relationship between different volumes as:

Q1

Vp
Vi

F 0 Swi 1  fi R F 0 Swi 1  fi R1  Swi

20

in which Vp(=fWL) is the pore volume and Vi the volume of


the NWP initially presented in the medium. The expressions
for the recovery, normalized by Vp or Vi, can then accordingly be derived as:
Q 2A 1=2

t
Vp L

21

Q
2A

t1=2
Vi L1  Swi

22

As emphasized by Mirzaei-Paiaman and Masihi,[25] for the


development of any scaling equation, its consistency with

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Energy Technol. 2014, 2, 166 175

Recovery in Fractured Petroleum Reservoirs


common scaling practices should be considered. In practice,
to scale imbibition data, recovery curves are normalized by
the reference volume which can either be the ultimate recovery, pore volume, or initial the NWP in place. If the recovery
in the scaling practice is normalized by the ultimate recovery,
then the appropriate dimensionless time would be in the
form of
0

tD

2AF Swi 1  fi R 1=2


t
L

23

If normalizing recovery to pore volume Vp is required, the


corresponding dimensionless time tD;Vp should be defined as,
tD;Vp

2A 1=2
t
L

24

Similarly, for the case of normalizing the recovery to the


initial NWP in place(Vi), the corresponding dimensionless
time tD;Vi should be defined as,

q
k

R Swi

dJ

w
f Sw mnw
krnw dSw
Sw;BC F Sw fn Sw

db

30

q
R Swi f Sw mm krnw dSdJ
k
 s
Q
Sw;BC F Sw fn Sw db

Vi
L vw mw 1  Swi 1  fi R

31

 s
Q

Vp
L vw mw

1  fi R
w

nw

The generalized Darcy number (Da) defined as the ratio


of two characteristic lengths (of the pore and domain)[25, 43]
can be written as :

Da

k
L2

32

The generalized capillary number Ca representing the relative effect of viscous forces versus interfacial tension can further be written as:
R Sw;BC Swi Sw f Sw mm krnw dSdJ
dSw
vw mw Swi
F Sw fn Sw

2
s R Sw;BC f Sw mm krnw dSdJ
Swi
F Sw fn Sw dSw
w

nw

CaQ1

nw

tD;Vi

2A
t1=2
L1  Swi

25
CaVp

db

vw mw 1  Swi 1  fi R
s R Swi f Sw mm krnw dSdJ
Sw;BC F Sw fn Sw db
w

nw

The above scaling equations can also be generalized by


presenting them in terms dimensionless numbers. Inserting
Equation (7) into Equation (19) yields:

Sw;BC F Sw fn Sw

CaVi

33

vw mw
1  f i R
s R Swi f Sw mm krnw dSdJ
nw

Moreover, these scaling equations can easily be extended


to other wetting conditions by using the method proposed by
Schmid and Geiger.[24]

34

35

The right-hand sides in Equations (29), (30), and (31) can


Da1=2

Q
2A2 F 0 Swi 1  fi R

Q1
Luw

26

Equation (20) gives the relationship between the Darcy velocity uw and the linear velocity vw as:

uw

vw 
t1=2
F 0 Swi 1  fi R

27

Capillary pressure can be related to the Leverett J function


J(Sw) as:[42]
r

Pc Sw s
J S
k w

28

Inserting Equations (8), (12), and (27) into Equation (26)


and combining with Equations (4) and (28) yields:
q
R
2
m
dJ
Sw;BC f Sw m krnw dS
k
Swi
F Sw fn Sw dSw
 s
Q

Q1
L vw mw R Sw;BC Swi Sw f Sw mm krnw dSdJ
dSw
Swi
F Sw fn Sw
w

nw

nw

29

Similarly, using the relationship between different volumes,


we can write:
Energy Technol. 2014, 2, 166 175

be rewritten as Ca implying that during COCSI, the recovery is controlled by the Darcy and Capillary numbers.

Validation of the New Scaling Equations


A comprehensive survey of the related literature was performed and eight sets of strongly water-wet experimental
data from Hamon and Vidal,[6] Bourbiaux and Kalaydjian,[7]
Akin et al.,[8] and Standnes[10] were found to fulfill the requirements of this study. Hamon and Vidal[6] performed four
experiments on a synthetic porous medium by using water
(WP) and a purified refined oil (NWP). In the experiment
by Bourbiaux and Kalaydjian,[7] the porous medium was
sandstone and brine and Soltrol 130 were used (WP and
NWP, respectively). In the liquidliquid and gasliquid experiments performed by Akin et al.[8] the porous medium
was diatomite with an initial zero saturation of the WP. In
both liquidliquid and gasliquid experiments, the WP fluid
was water and the NWP fluids were n-Decane and air, respectively. In the experiment by Standnes[10] chalk was the
porous medium, and water (WP) and n-Decane (NWP) were
used with an established initial WP saturation of zero. These
experiments are summarized in Table 1.
The ability of the new equations to scale the recovery experiments is compared to the scaling equations of Ma
et al.,[34] Mason et al.,[35] and Li.[32] An appropriate scaling

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

169

A. Mirzaei-Paiaman and M. Masihi


Table 1. Summary of data for experiments performed by Hamon and Vidal[6] (HV1, HV2, HV3, and HV4), Bourbiaux and Kalaydjian[7] (BK1), Akin et al.[8] (AK1 and AK2), and Standnes[10] (ST1).
Experiment Swi

L
k [m2]
[cm]

HV1
HV2
HV3
HV4
BK1
AK1
AK2
ST1

9.7
19.9
49.8
84.8
29
9.5
9.5
5.0

0.35
0.35
0.36
0.36
0.39
0
0
0

1.8  10-15
2  10-15
2  10-15
1.6  10-15
1.37  10-13
7  10-15
7  10-15
2  10-15

mnw
s [mN m1] End End Maximum F(Swi) A [ps]
mw
[mPa s] [mPa s]
point point Pc [kPa]
krw
krnw

0.27
0.28
0.28
0.27
0.23
0.69
0.69
0.43

1
1
1
1
1.2
1
1
1

11.5
11.5
11.5
11.5
1.5
0.0182
0.84
0.95

49
49
49
49
35
72
51.4
46

0.14
0.12
0.10
0.12
0.06
0.09
0.15
0.30

equation should reduce the observed scatter in the recovery


curves (Figure 1) to an acceptable limit. To use the scaling
equations of Ma et al.[34] and Mason et al.[35] only the commonly measured core and fluid property information are
needed, as summarized in Table 1. However, to use the scaling equation of Li[32] the measured endpoint relative permeability and capillary pressure data at initial WP saturation are
needed. Comprehensive studies reporting all required
COCSI recovery data and capillary pressure and relative permeability information are very rare in the literature. In this
work we therefore assume that for water-wet materials,
the relative permeability and capillary pressure information
from a certain rock type is representative for a given material. The relative permeability and capillary pressure data reported by Graue et al.[44] and Schembre and Kovscek[45] are
used. However to use the new scaling equations, the relative
permeability and capillary pressure data over the entire saturation range are needed (i.e., kr and Pc vs. Sw data are
needed). Use of the aforementioned approach to run the
model of Li[32] may not provide reliable estimates for F(Swi)
and A. As the analytical solution presented by McWhorter
and Sunada,[37] after consideration of backflow production of
NWP, describes exactly the COCSI, numerical values of
F(Swi) and A can be obtained by fitting the analytical solution to experiments by using Equations (19) and (21) or
Q
Equations (19) and (22). Based on Equation (21), a plot of Vp
2 1=2
versus L t gives a straight line with slope A that can be
computed by using a regression analysis. The numerical
Q
value of A can also be obtained by using plot of Vi versus
2
1=2
and computing the slope of the resulting straight
L1Swi t
line [Eq. (22)]. After determination of A, the numeral value
Q
2A1f R
of F(Swi) can be computed by plotting Q1 versus L i t1=2
and determining slope of the resulting straight line
[Eq. (19)]. In the slope analyses, there may be some observations deviating from linearity, particularly at early times,
which besides the experimental reading errors can be related
to heterogeneities at the pore level (i.e., pore shape and
pore-level roughness) and/or some geometrical effects that
result in the violation of the one-dimensional flow assumptions such that the spatial gradient of the capillary pressure
was not linear.[8, 46] Buoyancy effects are assumed negligible
for the strongly water-wet small-size rock samples. There is

170

0.98
0.97
0.95
0.95
0.47
1.00
1.00
1.00

480.1
461.2
458.5
512.8
12.5
428.6
200.4
541.5

4.7
5.2
5.7
5.4
4.3
1.0
1.6
1.4

9.5  10-6
1.7  10-5
2.0  10-5
1.7  10-5
2.6  10-5
4.8  10-4
4.6  10-5
5.4  10-5

also some deviation from linearity at very late times when


the imbibition front reaches the
far boundary. Obviously, the
analytical solution considered
in this study is not valid anymore when the WP front contacts the far boundary. Data
points related to the nonlinear
portions should therefore all be
excluded from the regression
analyses. The procedure used to
determine the numeral values

Figure 1. Recovery in terms of different reference volumes: a) ultimate recovery, b) pore volume, and c) initial NWP in place versus time for the experiments collected from the literature.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Energy Technol. 2014, 2, 166 175

Recovery in Fractured Petroleum Reservoirs


of the general principles of dimensional analysis by Rapoport.[31] This approach causes
the effects of some of the factors influencing the process be
systematically neglected. However the main reason may be
that these scaling equations
have not been derived in a consistent way according to
common scaling practices.[25]
The ability of the scaling
equation from Li[32] to plot the
recovery data normalized by
different reference volumes is
shown in Figure 5 versus this
scaling equation. The scaling
result is not satisfactory and
there still exists a nontrivial
scatter in scaling plots. One
should note that this scaling
equation has been specifically
proposed for COCSI. There
may be several reasons for such
poor quality of the scaling relation. This scaling equation has
been derived based upon a restricting approximate solution
to the main governing equation.
Furthermore the development
of this scaling equation is not
consistent with common scaling
practices as highlighted by Mir0
Figure 2. The procedure used to compute the numeral values of A and F Swi from the experimental data: a) plot
zaei-Paiaman and Masihi.[25] In
Q
2 1=2
of V versus L t gives a straight line with slope A that can be computed by using a regression analysis, b) the nuQ
2A1f R 1=2
0
Figure 5, the experiments remeral value of F Swi can be computed by plotting Q versus L t and determining slope of the resulting
straight line.
ported by Hamon and Vidal[6]
are not shown because the scaling equation of Li[32] predicts
negative values if the viscosity
of the NWP (11.5 mPa s, in these experiments) becomes large
of A and F(Swi) is shown in Figure 2 and the summarized recompared to viscosity of WP (1 mPa s, in these experiments).
sults are given in Table 1.
The ability of the new scaling equations to scale the exAs the scaling equations of Ma et al.[34] and Mason et al.[35]
periments is shown in Figure 6, in which the recovery data
are based on the application of general principles of dimennormalized by different reference volumes are plotted
sional analysis by Rapoport,[31] rather than by a mathematical
against the appropriate scaling equations. Depending on the
treatment, establishing an exact relationship between the vertype of normalization on the y-axis, the corresponding scaling
tical and horizontal axes in scaling plots is not possible.[25]
equation should be put on the x-axis. In each plot the curve
Thus, in scaling plots, the vertical axis may be recovery norgiven by the analytical solution is also included. The ability
malized by any of the reference volumes. Figure 3 and
of the new equations to scale the experiments is much better
Figure 4 show, respectively, the performance of the scaling
than the existing scaling equations. However, for some exequations of Ma et al.[34] and Mason et al.[35] after normalizperiments, some scatter around the analytical solution curve
ing the recovery data by the ultimate recovery, pore volume,
is noticeable which may be due to the quality of the reported
or initial oil/gas in place, which is plotted against these scalexperimental data. To be able to compare among the scaling
ing equations. These figures show that these equations
equations (Figures 1, 3, 4, 5, 6, and 7) we use the same
cannot scale recovery curves as there is significant scatter in
8 cycles on their horizontal axes.
all plots. There may be several reasons for the poor scaling
performance of the scaling equations of Ma et al.[34] and
Mason et al..[35] These were derived based on the application
p

Energy Technol. 2014, 2, 166 175

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

171

A. Mirzaei-Paiaman and M. Masihi

Figure 3. Recovery in terms of different reference volumes: a) ultimate recovery, b) pore volume, and c) initial NWP in place versus the scaling equation
of Ma et al.[34]

Derivation of a Simple Scaling Equation for


Water-Wet Systems
As the presented scaling equations incorporate the capillary
pressure and relative permeability parameters, they are applicable to systems with a wide range of wetting conditions.
However, from a practical perspective, this information is
generally not easily known, so there is still need for some
utility in simpler scaling groups. By simple scaling groups we
mean scaling equations that do not incorporate capillary
pressure and relative permeability information and are
mostly applicable to the systems with the same wettability

172

Figure 4. Recovery in terms of different reference volumes: a) ultimate recovery, b) pore volume, and c) initial NWP in place versus the scaling equation
of Mason et al.[35]

conditions (e.g., all water-wet). These equations usually


contain only simple-to-measure rock and fluid parameters
such as porosity, absolute permeability, interfacial tension,
geometrical dimensions, and the wetting and nonwetting
phase viscosities. With the exception of scaling purposes,
there exists significant interest in using simple equations,
mainly in comparative studies, for the objective of studying
the effect of aging time on the wettability alteration,[47, 48] the
effect of water adsorption and resulting microfractures in organic shale rocks,[49, 50] or the spontaneous imbibition characteristics of different porous media.[8, 12]

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Energy Technol. 2014, 2, 166 175

Recovery in Fractured Petroleum Reservoirs

Figure 5. Recovery in terms of different reference volumes: a) ultimate recovery, b) pore volume, and c) initial NWP in place versus the scaling equation
of Li.[32]

Inserting Equations (3), (4), (8), (12), and (28) into the dimensionless time equations brings them into the new forms
as:
v
u q
u2s k
t
 1=2
36
t
tD tD;Vp tD;Vi G
mnw L2
in which G, a group of variables, differs (Table 2).
Our objective is to extract a simple scaling equation from
Equation (36). We note that G is a function of wettability, initial wetting phase saturation, fluid viscosity, and pore structure. Therefore the exclusion of G from the scaling equation
r
p !
[Eq. (36)] makes this equation
Energy Technol. 2014, 2, 166 175

2s

k


mnw L2

t1=2

independent

Figure 6. Recovery in terms of different reference volumes: a) ultimate recovery, b) pore volume, and c) initial NWP in place versus the corresponding
new scaling equations.

of wettability and thus applicable to water-wet systems.


However, some parameters in the G variable are dependent
on fluid viscosity, initial wetting phase saturation, and pore
structure, and their overall effect should be considered, even
after the exclusion of G. In the case of the fluid viscosity dependence, this can be performed by replacing mnw in the simplified equation by an appropriate argument, called the viscosity group. To do this, several forms of the viscosity group
were used and the scaling performance of the resulting simplified scaling equations was checked. The best scaling group
p
was mnw + mw mnw . Therefore the resulting simplified scaling
equation can be written as:

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

173

A. Mirzaei-Paiaman and M. Masihi

Table 2. Expressions for G for different dimensionless time equations.


Dimensionless time
tD

Corresponding expression for G

R S f S k dSdJ
dSw
F S f S
S

G q
R S S S f S k dSdJ
w

w;BC

wi

w;BC

wi

rnw

F Sw fn Sw

Swi

dSw

s
R S S S f S k dSdJ

tD;Vp

w;BC

wi

rnw

F Sw fn Sw
1fi R2

Swi

dSw

s
R S S S f S k dSdJ

tD;Vi

w;BC

tD;simplified

rnw

v
q
u
k
u
2s

t
1=2

p 2 t
mnw mw mnw L

Swi

wi

rnw

F Sw fn Sw
1Swi 2 1fi R2

dSw

37

We should note that because this simplified scaling equaq


k
tion contains the  term, the dependence on the pore structure may be assumed to be relaxed. However, the initial wetting phase saturation dependence cannot be removed. Therefore the simplified scaling equation presented above may not
present very good scaling performance for systems with different initial saturations of the wetting phase. As this scaling
equation is a simplified form of the general scaling equations,
the vertical axis in scaling plots can be recovery normalized
by any reference volume. The ability of this equation to scale
data is shown in Figure 7, which shows an improvement as
compared to Figure 3, Figure 4, and Figure 5. Moreover,
Figure 7 (in comparison to Figure 6) shows acceptable accuracy for the new simplified equation as compared to the new
general scaling equations.

Conclusions
The following conclusions can be drawn from this work:
* As the analytical solution to unidirectional displacement
given by McWhorter and Sunada[37] applies to COCSI,[36]
suitable scaling equations for one-dimensional COCSI can
be found by using this solution.
* Backflow production of the NWP at the inlet open boundary is inherent to COCSI and its contribution to the process should be taken into account when using the analytical solution given by McWhorter and Sunada[37] and when
presenting new scaling equations.
* The strategy to account for the contribution of the backflow production by assuming R
6 1 avoids the occurrence
of the possible instabilities in computing the integrals in
the McWhorter and Sunada[37] solution.
* The new scaling equations presented in this study are universal, incorporating all factors influencing the process, as

174

Figure 7. Recovery in terms of different reference volumes: a) ultimate recovery, b) pore volume, and c) initial NWP in place versus the new simplified
scaling equation.

the exact analytical solution to the problem without any


assumption are used.
Consistency between the development of the new scaling
equations and common practices should be considered to
obtain reliable results.
The new scaling equations can be rewritten in terms of
two physically meaningful dimensionless numbers, Da1/2/
Ca (Da: Darcy number, Ca: capillary number).
The ability of the new equations to scale the experiments
was found to be much better than the existing scaling
equations.

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Energy Technol. 2014, 2, 166 175

Recovery in Fractured Petroleum Reservoirs


*

For the literature scaling equations derived based on the


application of the general principles of dimensional analysis, establishing an exact relationship between the vertical
and horizontal axes in scaling plots is not possible. Thus,
in scaling plots the vertical axis may be recovery normalized by any of the reference volumes.
The scaling equations developed for COUCSI fail to scale
up COCSI data.
The former scaling equations proposed for COCSI, derived based on assumption of piston-like displacement,
yield poor scaling results.
The general scaling equations presented in this study can
be used to present a simple scaling equation for waterwet systems. The ability of the new simplified scaling
equation to scale water-wet data was found to be better
than other existing scaling equations and also acceptable
in comparison to new general scaling equations.

Acknowledgements
The authors thank Sharif University of Technology and the
Research and Technology Departments of NIOC and NISOC
for permission to publish this paper.

Keywords: fluid dynamics fractured reservoirs petroleum


scaling equations wetting phase
[1] G. Mason, N. R. Morrow, J. Pet. Sci. Eng. 2013, 110, 268 293.
[2] A. Mirzaei-Paiaman, M. Masihi, D. C. Standnes, Transp. Porous
Media 2011, 89, 49 62.
[3] A. Mirzaei-Paiaman, M. Masihi, D. C. Standnes, Energy Fuels 2011,
25, 3053 3059.
[4] L. L. Handy, Pet. Trans. AIME 1960, 219, 75 80.
[5] R. Iffly, D. C. Rousselet, J. L. Vermeulen, Paper SPE 4102, presented
at the Annual Fall Meeting, San Antonio, TX, October 8 11, 1972.
[6] G. Hamon, J. Vidal, Paper SPE 15852, Proceedings of the European
Petroleum Conference, vol. 15852, London, October 20 22, 1986,
pp. 37 49.
[7] B. Bourbiaux, F. Kalaydjian, SPE Reservoir Eng. 1990, 5, 361 368.
[8] S. Akin, J. M. Schembre, S. K. Bhat, A. R. Kovscek, J. Pet. Sci. Eng.
2000, 25, 149 165.
[9] M. Pooladi-Darvish, A. Firoozabadi, SPE J. 2000, 5, 3 11.
[10] D. C. Standnes, Energy Fuels 2004, 18, 271 282.
[11] S. S. L. Guen, A. R. Kovscek, Transp. Porous Media 2006, 63, 127
146.
[12] K. Li, R. N. Horne, Transp. Porous Media 2010, 83, 699 709.
[13] J. Cai, B. Yu, M. Zou, L. Luo, Energy Fuels 2010, 24, 1860 1867.
[14] J. Cai, X. Hu, D. C. Standnes, L. You, Colloids Surf. A 2012, 414,
228 233.
[15] R. W. Parsons, P. R. Chaney, SPE J. 1966, 237, 26 34.
[16] J. Kleppe, R. A. Morse, Paper SPE 5084, presented at the Annual
Fall Meeting, Houston, TX, October 6 9, 1974.
[17] H. Kazemi, L. S. Merrill, SPE J. 1979, June, 175 182.
[18] M. Pooladi-Darvish, A. Firoozabadi, J. Can. Pet. Technol. 2000, 39,
31 42.
[19] H. Karimaie, O. Torster, M. R. Esfahani, M. Dadashpour, S. M. Hashemi, J. Pet. Sci. Eng. 2006, 52, 297 304.

Energy Technol. 2014, 2, 166 175

[20] H. Karimaie, O. Torster, J. Pet. Sci. Eng. 2007, 58, 293 308.
[21] A. Firoozabadi, J. Can. Pet. Technol. 2000, 39, 13 17.
[22] F. H. Qasem, I. S. Nashawi, R. G. Muhammad, I. Mir, J. Pet. Sci. Eng.
2008, 60, 39 50.
[23] K. S. Schmid, S. Geiger, Water Resour. Res. 2012, 48, http://
dx.doi.org/10.1029/2011WR011566.
[24] K. S. Schmid, S. Geiger, J. Pet. Sci. Eng. 2013, 101, 44 61.
[25] A. Mirzaei-Paiaman, M. Masihi, Energy Fuels 2013, 27, 4662 4676.
[26] D. Zhou, L. Jia, J. Kamath, R. A. Kovscek, J. Pet. Sci. Eng. 2002, 33,
61 74.
[27] E. R. Rangel-German, R. A. Kovscek, J. Pet. Sci. Eng. 2002, 36, 45
60.
[28] A. Kantzas, M. Pow, K. Allsopp, D. Marentette, Paper no. 97 181,
Seventh Petroleum Conference of the South Saskatchewan Section,
The Petroleum Society of CIM, Regina, October, 19 22, 1997.
[29] K. Li, K. Chow, R. N. Horne, SPE Western Regional/AAPG Pacific
Section Joint Meeting, 20 22 May Anchorage, Alaska, 2002,
DOI:10.2118/76727-MS.
[30] C. U. Hatiboglu, T. Babadagli, J. Pet. Sci. Eng. 2010, 70, 214 228.
[31] L. A. Rapoport, Trans. AIME 1955, 204, 143 150.
[32] K. Li, J. Contam. Hydrol. 2007, 89, 218 230.
[33] B. Bourbiaux, International Petroleum Technology Conference, 7 9
December, Doha, Qatar, 2009, DOI:10.2523/13909-MS.
[34] S. Ma, X. Zhang, N. R. Morrow, Annual Technical Meeting of Petroleum Society of Canada, Jun 7 9, Calgary, Alberta, 1955,
DOI:10.2118/95-94.
[35] G. Mason, H. Fischer, N. R. Morrow, D. W. Ruth, J. Pet. Sci. Eng.
2010, 72, 195 205.
[36] K. S. Schmid, PhD thesis, Institute of Petroleum Engineering, HeriotWatt University, 2012, Available online at http://www.ros.hw.ac.uk/
bitstream/10399/2547/1/SchmidKS_0312_pe.pdf.
[37] D. B. McWhorter, D. K. Sunada, Water Resour. Res. 1990, 26, 399
413.
[38] A. Mirzaei-Paiaman, M. Masihi, D. C. Standnes, Energy Fuels 2013,
27, 7360 7368.
[39] Z.-X. Chen, G. S. Bodvarsson, P. A. Witherspoon, Water Resour. Res.
1992, 28, 1477 1478.
[40] Z.-X. Chen, G. S. Bodvarsson, P. A. Witherspoon, SPE Annual Technical Conference and Exhibition, 23 26 September, New Orleans,
Louisiana, 1990, 10.2118/20517-MS.
[41] R. Fuck, J. Mikyska, M. Benes, T. H. Illangasekare, Vadose Zone J.
2007, 6, 93 104.
[42] J. Bear, Dynamics of Fluids in Porous Media, Dover, New York,
1972.
[43] B. Zappoli, R. Cherrier, D. Lasseux, J. Ouazzani, Y. Garrabos,
arxiv.org/abs/cond-mat/0601196 [cond-mat.stat-mech], 2006.
[44] A. Graue, T. Bognoe, R. Moe, B. Baldwin, E. A. Spinler, D. Maloney,
SCA paper 9907, International Symposium of the Society of Core Analysts, 1999.
[45] J. M. Schembre, A. R. Kovscek, Transp. Porous Media 2006, 65, 31
51.
[46] M. A. Fern, A. Haugen, S. Wickramathilaka, J. Howard, A. Graue,
G. Mason, N. R. Morrow, J. Pet. Sci. Eng. 2013, 101, 1 11.
[47] X. M. Zhou, N. R. Morrow, S. X. Ma, SPE J. 2000, 5, 199 207.
[48] X. Xie, N. R. Morrow, Proceedings of the International Symposium of
the Society of Core Analysis Meeting, Abu Dhabi, UAE, October,
2000.
[49] H. Dehghanpour, H. A. Zubair, A. Chhabra, A. Ullah, Energy Fuels
2012, 26, 5750 5758.
[50] H. Dehghanpour, Q. Lan, Y. Saeed, H. Fei, Z. Qi, Energy Fuels 2013,
27, 3039 3049.

Received: October 25, 2013


Revised: December 5, 2013
Published online on February 10, 2014

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

175

You might also like