Você está na página 1de 13

A generalized correlation for predicting gascondensate relative permeability at near

wellbore conditions
M. Jamiolahmady , M. Sohrabi, S. Ireland, P. Ghahri
Institute of Petroleum Engineering, Heriot-Watt University, Riccarton, Edinburgh EH14 4AS, UK
a b s t r a c t a r t i c l e i n f o
Article history:
Received 20 December 2006
Accepted 2 February 2009
Keywords:
relative-permeability
gascondensate
coupling effect
inertial effect
uid ow
low interfacial uid tension systems
It is a well established nding, both experimentally and theoretically that relative permeability (k
r
) of gas
condensate systems at low interfacial tension (IFT) depends on velocity and IFT. There are a number of
correlations available both in the literature and commercial reservoir simulators that account for the coupling
(i.e., the increase of k
r
as velocity increases and/or IFT decreases) and inertial (i.e., the reduction of k
r
as velocity
increases) effects at near wellbore conditions. These functional forms, which with the exception of Whitson et al.
(Whitson, C.H., Fevang, O., Saevareid, A., Oct. 1999. Gas condensate relative permeability for well calculations, SPE
56476. Proc. of SPE Annual Technical Conference and ExhibitionTexas) are based on saturation, estimate the two
effects separately and include a number of parameters, which should be determined by cumbersome and
expensive k
r
measurements at low IFT and high velocity conditions. We have previously reported the
development of a fractional ow based correlation (Jamiolahmady, M., Danesh, A., Henderson, G.D. and Tehrani,
D.H., Dec. 2006. Variations of gascondensate relative permeability with production rate at near wellbore
conditions: a general correlation, SPE Reserv. Eng. Evalu. J., 9 (6), pp. 688697), which expressed the combined
effect of coupling and inertia simultaneously, but its dependency on uid viscosity limited its use to the range of
IFT values studied. In this paper we present a generalized correlation based on relative-permeability-ratio, which
is closely related to a fractional ow of gas and condensate. The parameters of the new correlation are either
universal, applicable to all types of rocks, or canbe determined fromcommonly measuredpetrophysical data. The
performance of the correlation has been evaluated over a relatively wide range of IFTand velocity variations. This
has been conducted by comparing its predictionwith the gascondensate relative permeability values measured
on porous media with very different characteristics. These data had not been used in developing the correlation.
The results are quite satisfactory conrming the generality of the correlation inproviding reliable information on
gascondensate relative permeability values for near wellbore conditions.
2009 Elsevier B.V. All rights reserved.
1. Introduction
The characteristics of gas and condensate ow, when pressure falls
below dew point, are signicantly different from those of conventional
gasoil systems and accurate determination of gascondensate relative
permeability (k
r
) values, is a major challenge (Jamiolahmady et al., 2000).
The effect of interfacial tension (IFT) on relative permeability at low
IFT values has been known for a considerable time (e.g., Bardon and
Longeron, 1980; Asar and Handy, 1988). The improvement to the
relative permeability of condensing systems due to an increase in
velocity is also a relatively well established experimental nding (e.g.,
Danesh et. al., 1994; Henderson et al., 1997, 2001; Ali et al., 1997; Blom
et al., 1997). This improvement in relative permeability of low IFT
systems as velocity increases and/or IFT decreases, known as the
coupling effect, has been attributed to the simultaneous coupled ow
of the gas and condensate phases with the intermittent opening and
closure of the gas passage by the condensate at the pore level
(Jamiolahmady et al., 2000, 2003). This cyclic two-phase ow pattern,
unique to gas condensate systems, was observed in the micromodel
experiments conducted by authors ((Jamiolahmady et al., 2000). The
developed single pore mechanistic model capturing the competition
betweenviscous and capillary forces, consistent with the experiments,
showedthat for these lowIFTsystems, there is a highly conductive lm
of condensate owing withthe gas. The condensate evolves at the pore
throat andblocks the gas passage, after whichthe owof gas continues
till it overcomes the capillary barrier. As IFT increases the number of
pores promoting this ow pattern gradually diminishes switching to a
conventional Darcy type of ow mechanism used in the channel ow
concept. Hence, when the effect of multiple pore interaction was
included in a network of pores (Jamiolahmady et al., 2003) k
r
values
comparable with experimentally measured values were reported.
There are now several correlations in the literature (e.g., Henderson
et al., 1996; BlomandHagoort, 1998; Popeet al., 1998; Whitsonet al., 1998)
and in commercial reservoir simulators (e.g., ECLIPSE, VIP) to express the
Journal of Petroleum Science and Engineering 66 (2009) 98110
Corresponding author. Tel.: +44 131 4513122; fax: +44 131 4513127.
E-mail addresses: jami.ahmady@pet.hw.ac.uk (M. Jamiolahmady),
mehran.sorabi@pet.hw.ac.uk (M. Sohrabi), mehran.sorabi@pet.hw.ac.uk (P. Ghahri).
0920-4105/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.petrol.2009.02.001
Contents lists available at ScienceDirect
Journal of Petroleum Science and Engineering
j our nal homepage: www. el sevi er. com/ l ocat e/ pet r ol
capillary number (ratio of capillary to viscous forces) dependence of the
two-phase ow of gas and condensate at these low IFT systems.
Henderson et al. (2001), through some steady-state k
r
measure-
ments, reported that at low condensate saturations the relative
permeability values decreased when the ow rate was increased, due
to the dominance of the inertial effect. However, the positive coupling
effect surpassed the inertial effect at higher condensate saturation
resulting in a net increase of relative permeability with velocity. There
are some correlations to express the effect of inertia using a two-phase
inertial factor
g
(e.g., Henderson et al., 2001; Whitson et al., 1999;
Mott et al., 2000; App and Mohanty, 2002). The main difculty in
accounting for the inertial effect in all the reported correlations is the
requirement for estimating
g
. Correlations have been reported in the
literature for calculating
g
(e.g., Narayanaswamy et al., 1999; Dacun,
and Thomas, 2000), which relate
g
mainly to S
g
and/or k
rg
. However,
the application of these correlations to gascondensate systems is
open to question as they have been developed mainly for gaswater
systems, with the immobile water phase (Al-Kharusi, 2000).
We recently proposed a fractional ow based correlation (Jamiolah-
mady et al., 2006) accounting for the combined effects of coupling and
inertia as a function of fractional ow. The choice of fractional owas the
main independent variable rather than saturation was based on our
observation that it could express relative permeability of core samples
withdifferent characteristics more reliably. That is, different core samples
have a different pore size distribution, which could lead to different
distributions of the phases, i.e., different phase saturation, at similar ow
conditions. Furthermore, fractional ow is directly related to uid
composition and pressure at steady-state conditions, which is generally
prevailing near the wellbore, hence, it makes it practically much more
attractive compared to that of saturation which depends on core
characteristics. The condensate relative permeability is also linked to
that of gas by the fractional ow, thereby eliminating the need for a
separate correlation for its calculation. However, the dependency of
relative permeability to uid viscosity and the lack of a proper denition
of the lower limit of the correlation limited its use to the range of IFT
values studied. In other words, as will be discussed in Section 2.3, the
reported base IFTof 0.85 mNm
1
, above whichk
r
is not a function of IFT,
is not a universal value and the threshold capillary number xing the
lower limit of velocity (below which k
r
is not a function of velocity) has
also not been identied. It will be demonstrated that although fractional
ow and relative permeability ratio are closely related, the inclusion of
viscosity in the denition of independent variable makes it difcult to x
the base conditions. In other words, the base k
r
curve reported as a
function of fractional ow rate makes it uid dependent whilst its effect
can easily be accounted for separately when calculating the relative
permeability ratio from the known fractional ow and viscosity values
facilitating its generality. This together with the lack of proper denition
for base capillary number, as will be demonstrated in Section 3, could
result in a highly erroneous estimation of relative permeability if
fractional ow based correlation is used. In this manuscript we have
removed both these limitations and sought simpler mathematical
expressions. This enables us to have a correlation with either universal
parameters, applicable to all types of rocks, or those that can be deter-
mined from commonly measured petrophysical data.
2. Structure of correlation
In a two-phase owsystemDarcy's law, which was developed for a
single-phase ow, is used by replacing the absolute permeability (k)
with the effective permeability (k
e
), which is reported as relative
permeability (k
r
=k
e
/k). In gascondensate systems the relative
permeability could be affected by both the coupling (i.e., the increase
of k
r
as velocity increases and/or IFT decreases) and inertial (i.e., the
reduction of k
r
as velocity increases) effects.
In the newproposed formulation, k
rg
is interpolated between the base
(k
rgb
) andthemiscible-uids curves (k
rgm
) using aninterpolationfunction
Nomenclature
A
j
Parameter, j =1 to 7
C
j
Constant, j =1 to 15
k Absolute permeability, m
2
or mD
k(S
wi
) Permeability at S
wi
, m
2
or mD
k
e
Effective permeability, m
2
or mD
k
r
Relative permeability, dimensionless
k
rgtr
Relative permeability ratio, i.e., k
rg
/(k
rg
+k
rc
),
dimensionless
(k
rgtr
)
Micpoeff
k
rgtr
value at which k
rg
affected by the
presence of micro-pores, dimensionless
L Length, m
P Pressure, Pa or psia
Q Flow rate, m
3
s
1
S
wi
Immobile water saturation, dimensionless
S
w1 m
Wetting phase (air) saturation corresponding to one
micron radius from mercury porosimetry P
c
curve,
dimensionless
u Darcy velocity, m s
1
Y
g
Gas weight function for interpolation of k
r
, dimensionless
(Y
g
)
Main
Gas weight function for majority of data points fol-
lowing main trend, dimensionless
(Y
g
)
Micpoeff
Gas weight functionfor datapoints not following
main trend due to effect of micro-pores, dimensionless
Greek letters
Single-phase inertial factor, m
1
(S
wi
) Single-phase inertial factor at S
wi
, m
1

g
Two-phase inertial factor for gas phase, m
1
Porosity of porous medium, %
Viscosity, kg (ms)
1
or cp
Density, kg m
3
Interfacial tension, IFT, between gas and liquid, N m
1
or mN m
1

r
Ratio of the base IFT of 3.0 mN m
1
to current IFT,
dimensionless
Subscript
ave Average value of the quantity
b Value of the quantity for the base case at highest
measured interfacial tension, IFT=3 mN m
1
, and
lowest velocity
g Gas phase
iner Value of the quantity affected by inertia
j An index
c Condensate
m Miscible case
T Total
Abbreviations
f
g
Gas to total ow rate ratio
IFT Interfacial tension
Nc Capillary number, ratio of viscous to capillary forces
Ncr Ratio of current Nc to base Nc
Operators
| | Absolute value
Difference operator
Divergence operator
Gradient operator
99 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
Y
g
. This simple interpolation approach has been favoured by many
investigators to express the coupling effect on k
r
of gascondensate
systems at different values of capillary number (e.g., Coats, 1980;
Henderson et. al., 1996; Whitson et al., 1999). However, here, similarly to
the fractional ow based correlation (Jamiolahmady et al., 2006) we
modify both k
rgm
and k
rb
for the effect of inertia, which enables us to
express the combined effect of coupling and inertia simultaneously, using
Eq. (1).
k
rg
= Y
g
k
rgb
_ _
iner
+ 1 Y
g
_ _
k
rgm
_ _
iner
: 1
The complete list of variables can be found in the nomenclature
section.
The main independent variable is the relative-permeability-ratio
dened as,
k
rgtr
=
k
rg
k
rg
+ k
rc
=
1
1 +
k
rc
k
rg
_ _: 2
Solving the above equation for condensate relative permeability,
k
rc
, gives:
k
rc
=
k
rg
1 k
rgtr
_ _
k
rgtr
: 3
Therefore, when k
rg
is determined as a function of k
rgtr
this
equation automatically gives the corresponding value of k
rc
for the
same value of k
rgtr
.
k
rgtr
ratio could be calculated in terms of gas fractional ow (f
g
),
and uid viscosities. That is, f
g
is dened as,
f
g
=
Q
g
Q
g
+ Q
c
=
u
g
u
g
+ u
c
=
k
r

_ _
k
r

_ _
c
+
k
r

_ _
g
; 4
where Q is the volumetric ow rate, u is the Darcy velocity, is the
viscosity and subscripts (g) and (c) refers to the gas and condensate
phases, respectively.
The extended form of Darcy's law for two-phase ow, Eq. (5), has
been used in Eq. (4) to relate velocity to relative permeability as,
u=
kk
r
jP

_ _
j
j = g; c; 5
where P is the pressure, k is the absolute permeability and is the
gradient operator.
Solving Eq. (4) for relative-permeability-ratio gives:
k
rc
k
rg
_ _
=
1 f
g
_ _
f
g

g
_ _
: 6
Substituting Eq. (6) into Eq. (4), after some mathematical
manipulation, k
rgtr
can be related to f
g
by,
k
rgtr
=
k
rg
k
rc
+ k
rg
=

g
f
g

g
f
g
+
c
1 f
g
_ _: 7
The use of relative-permeability-ratio based on the summation of
relative permeability values of gas and condensate in the denominator of
Eq. (7), rather than the relative permeability of one of the two phases
(e.g., k
rgcr
, which is the gas to condensate relative permeability ratio)
avoids the mathematical difcultyassociatedwiththe latter (i.e., k
rgcr

when k
rc
0 which is equivalent to condensate ow rate0).
2.1. Miscible-uids relative permeability for gas (k
rgm
)
In the present k
rgtr
based correlation the gas miscible curve is
modied to include the inertial effect with an argument similar to that
presented for the fractional ow based correlation (Jamiolahmady
et al., 2006). This is due to the fact that the formulation for the f
g
-based
correlationwas obtained by momentumbalance, which is not going to
be affected by the choice of variable. Therefore, for f
g
we can write
k
rgm
_ _
iner
=
k
rgtr
1 +
m
k

m
_ _
j uj
sm
_
_
_
_
; 8
where | | refers to the absolute value of the quantity, is the (single-
phase) Forchheimer coefcient sometimes referred to as the non-
Darcy or inertial coefcient and subscript mrefers to the miscible case.
In Eq. (8) the required miscible uid properties (i.e. density,
m
,
and viscosity,
m
) are the arithmetic average of the uid properties of
gas and liquid at any given pressure (in the vicinity of the dew point),
which is a good approximation of the average values at the dew point
that are not easily available.
2.2. Base relative permeability for gas (k
rgb
)
The base relative permeability curve is the measured curve at the
lowest velocity level (above which k
r
is a function of velocity) and the
highest IFT value (below which k
r
is a function of IFT), (k
rgb
)
meas
,
which is then modied for the effect of inertia, (k
rgb
)
iner
when the
velocity is high. The exact denition of (k
rgb
)
meas
will be discussed
later in this section. Here a similar approach to that used for the
miscible curve was adopted, i.e., dening an equivalent single phase,
eliminating the need for the two-phase Forchheimer factor,
g
. In
other words, the inertial pressure drop was calculated by using the
single-phase inertial factor, , the total velocity, u
T
, and the summation
of total momentum inow of gas and condensate. Therefore, k
rgb
that
is modied for the effect of inertia is calculated by
k
rgb
_ _
iner
=
k
rgb
_ _
meas
1 +

ave
k k
rgb
meas
f
g

g
j uj
T
_
_
_
_
_
_: 9
In Eq. (9), the weighted average density value is calculated at the
base conditions and based on the fractional ow of the two owing
phases. Eqs. (8) and (9) correctly extend the correlation to the single-
phase owlimits. That is, if (k
rgm
)
iner
fromEq. (8) or (k
rgb
)
iner
fromEq.
(9) is substituted in Darcy law, Eq. (5), and re-arranged it gives the
Forchheimer Equation (1914):
jP =

k
u + j uj u
_ _
g
10
Due to the presence of the rock properties (k and ) in Eqs. (8) and
(9), there are miscible and base relative permeability versus k
rgtr
curves
for each core at any velocity value. The base and miscible gas relative
permeability curves calculated using Eqs. (8) and (9) for Clashach and
Texas Cream cores at different velocities are shown in Fig. 1a and b,
respectively. It is noted that the gas miscible curve dened by Eq. (8), at
low velocities where the inertial term is insignicant, will approach the
k
rg
=k
rgtr
linebut at highvelocities theslope of the line deviates from45
to lower values depending on the values of velocity and single-phase
inertial factor for the given uid system. Similarly, the base relative
permeability curve at low velocities is close to the corresponding
measuredvalue but at highvelocities it is reduced. This reductionis more
pronounced, at higher k
rgtr
values (higher gas fractional ow rates).
Eqs. (9) and (10) are also valid for S
wi
N0, with k
eg
(S
wi
) and
g
(S
wi
)
replacing k and , respectively, assuming that immobile water acts
like a part of the rock.
2.3. Measured base relative permeability for gas (k
rgb
)
meas
The base-relative-permeability curve in our previous work
(Jamiolahmady et al., 2006) was given at IFT=0.85 mN m
1
, with
100 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
the recommendation for eld application that the base curve should
be measured at the highest applicable value of IFT for the uid under
consideration and the lowest practicable velocity. A series of relative
permeability measurements were made on a Texas Cream limestone
core at IFT values higher than 0.85, as shown in Fig. 2. The objective
was to investigate the effect of increasing IFT on the gas and
condensate relative permeability and to identify the IFT value above
which the relative permeability would not be affected by velocity.
The results indicated that there was a continual reduction in relative
permeability with increasing IFT if k
r
was expressed in terms of fractional
ow rate ratio (f
g
), Fig. 2a. However, if the relative permeability curves
were plottedversus relative permeability ratio, the effect of viscosity, uid
property, was eliminated and the variation of k
r
curves measured at high
IFT was demonstrated to be minimal, Fig. 2b. These experimental ndings
suggestedthat theIFTvalueof 0.85mNm
1
consideredfor thebasecurve
in the previous study is not a universal value and it could have a value in
the range of 1 to 3 mN m
1
, Fig. 2b. The two gures also show that at a
given value of IFT the low velocities in the range of 5 to 50 m d
1
have a
minimal effect on relative permeability, especially at high values of f
g
. In
Fig. 2bit canbenoticedthat thereis a slight decreasein(k
rgb
)
meas
, whichis
more pronounced at higher k
rgtr
values, when IFT increases from 0.85 to
3 mN m
1
. These results highlight that the effect of reduction of IFT can
not be compensated by an equivalent increase in velocity. Later in the
following section it will be shown that the effect of viscous and capillary
forces are independently accounted for using a combination of IFT and
capillarynumber (ratioof viscous tocapillaryforces). Hence, bothbaseIFT
and base capillary number are required for this correlation, which has
universal parameters. After careful examination of the measured data,
accuracy of the measurements and the fact that at higher IFT values and/
or lower velocities the coupling effect is diminishing for all core types, we
selected a base IFT and capillary number of 3 mN m
1
and 1E7,
respectively, for measuring the base relative permeability. This value for
the base capillary number was obtained using Eq. (11) as follows:
Nc =

g
u
g
+ u
c
_ _

; 11
where is the interfacial tension.
Based on the data of Fig. 2 for core samples with measurements at
IFT of 0.85 mN m
1
rather than base IFT of 3 mN m
1
, Eq. (12) was
used to adjust the base curve data.
k
rg
_ _
meas
_ _
IFT=3
k
rg
_ _
meas
_ _
IFT=0:85
= 0:69 + 0:31 k
rgtr
_ _
7:07
12
2.4. Weight function formulation (Y
g
)
To determine the form of the weight function Y
g
, Eq. (1), we used
experimentally measured k
rg
values on selected Clashach, Berea, Texas
Cream and RC2 core samples. Tables 14 include the basic data for the
Fig. 1. Base and miscible gas relative permeability versus relative-permeability ratio at three different velocities calculated using Eqs. (8) and (9), respectively, for a) Clashach core
b) Texas Cream core.
101 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
different sets of measurements conducted on these core samples
including the range of IFT and velocity values at which the relative
permeability measurements were conducted. This comprehensive data
bank included over 1300 relative permeability measurements. A
number of measurements on a number of core samples, i.e., Clashach,
Texas Cream, RC3 and a propped fracture, Table 5, were excluded so that
the evaluationexercise conductedonthe performance of the correlation
included data that were not used for its development. This data bank
included over 200 additional relative permeability measurements.
To determine the functional form of Y
g
, for each k
rg
measured
experimentally at the prevailing conditions, Y
g
was calculated, using
the following equation:
Y
g
=
k
rg
k
rgm
_ _
iner
k
rgb
_ _
iner
k
rgm
_ _
iner
; 13
which is the re-arranged form of Eq. (1). (k
rgm
)
iner
at any velocity
level and k
rgtr
were calculated using Eq. (8) and (k
rgb
)
iner
was the
Fig. 2. Gas relative permeability at two velocities and three IFT values for Texas Cream core sample based on (a) gas to total fractional ow rate ratio (b) gas to total, gas plus
condensate, relative permeability ratio.
Table 1
Basic test data for experiments conducted on Clashach core at different conditions.
Index S
wi
k k(S
wi
) IFT (S
wi
) Velocity
range
(k
rgtr
)
Micpoeff
a
(%) (mD) (mD) (mN m
1
) (m
1
) (m
1
)
1 0 553 0.85 1.035E8 7.3700
2 0 553 0.15 1.035E8 7.3700 0.989
3 0 553 0.036 1.035E8 7.3350 0.992
4 0 600 0.15 5.990E7 7.1699
5 0 600 0.036 5.990E7 7.1699
6 33 553 245 0.85 1.035E8 4.580E8 10.91044
7 21 553 446 0.85 1.035E8 1.395E8
b
9.2885
8 21 553 446 0.15 1.035E8 1.395E8
b
9.2738
9 15 440 400 0.85 1.035E8 1.151E8
b
9.474.9
10 15 440 400 0.15 1.035E8 1.151E8
b
9.474.9
S
wi
is the amount of immobile water saturation and k(S
wi
) and (S
wi
) are the single-
phase permeability and inertia factor, respectively, of the core sample at S
wi
.
IFT refers to interfacial tension.
a
(k
rgtr
)
Micpoeff
refers to k
rgtr
=k
rg
/(k
rg
+k
rc
) value at which the measured k
rg
has
been affected by the presence of micro-pores.
b
Calculated (S
wi
) value using, Eq. (17) of Jamiolahmady et al., (2006).
Table 2
Basic test data for the experiments conducted on Berea core at different conditions.
Index S
wi
k k(S
wi
) IFT (S
wi
) Velocity
range
(k
rgtr
)
Micpoeff
(%) (mD) (mD) (mN m
1
) (m
1
) (m
1
)
11 0 110 0.85 1.870E8 7.1454
12 0 110 0.15 1.870E8 7.1567 0.989, 0.978
13 0 110 0.036 1.870E8 7.1227 0.985, 0.954,
0.925
14 26 110 92 0.85 1.870E8 2.286E8
a
9.273.7
15 26 110 92 0.45 1.870E8 2.286E8
a
9.273.7
16 26 110 92 0.15 1.870E8 2.286E8
a
9.273.7
17 26 110 92 0.036 1.870E8 2.286E8
a
9.273.7
a
Calculated (S
wi
) value using, Eq. (17) of Jamiolahmady et al., (2006).
102 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
experimentally measured base k
rg
value at any given k
rgtr
modied for
the effect of inertia using Eq. (9).
The Y
g
weight function depends on the capillary number ratio, the
ratio of the prevailing capillary number to its base value. Similarly to
the fractional ow based correlation (Jamiolahmady et al., 2006) the
capillary number dened by Eq. (14) was selected to represent the
ratio of viscous and capillary forces, which proved to correlate the data
more consistently.
Nc =
kP
/L
=
kj jPj
/
; 14
where is the porosity and L is the core length.
It is noted that the capillary number expression, in Eq. (14), is
different from that of Eq. (11) used for dening the base capillary
number. However, as it will be discussed later they are related to one
another using Darcy law for two-phase ow, Eq. (5).
The data of Fig. 3 is the plot of calculated Y
g
based on the
experimentally measured relative permeability on two dry (without
initial water saturation) Clashach core samples at different IFT values
based on k
rgtr
. The data in this gure demonstrate that the bulk of data
follows a consistent trend but very scattered. In this gure a group of
data have been marked deviating data' with low Y
g
values that
correspond to data points at low IFT and high k
rgtr
values, Fig. 3. This
different behaviour is attributed to the contribution of a different ow
mechanism because of to the presence of condensate in the micro-
pores of the core at high k
rgtr
values (high gas fractional ow rates).
Micro-pores have been shown to affect the ow behaviour of
conventional gasoil systems. Mcdougall et. al. (2002) had to use
two different pore size distributions to match the P
c
curves of cores
with micro-pores. It was noted that for deviating data of Fig. 3, the
(k
rgb
)
meas
values are much lower than their corresponding k
rg
values
at the same k
rgtr
but at lower IFT values, Fig. 4. That is, in the presence
of these micro-pores at lower IFT values, condensate ow does not
impede the ow of gas, i.e., k
rg
is close to unity. But at higher base IFT
value the condensate owrestricts the owof gas thereby reducing its
k
r
to very low values, which results in very low Y
g
values, Eq. (12). If
the percentage of these small pores is very high, e.g., Texas Cream,
even at base IFT, the ow of gas is not affected by the ow of
condensate and the (k
rgb
)
meas
values are close to unity and to the
corresponding values at lower IFT for the same high k
rgtr
, Fig. 5, hence
Y
g
will not take these low values. Furthermore, it is noted that in the
latter case because of to the small difference between the two
interpolation limits, i.e. (k
rgb
)
meas
and k
rgm
the accuracy of the weight
function, Y
g
, does not affect the interpolated value, k
rg
, signicantly.
For the Clashach and Berea core samples the limit of this effect is
different. Last columns in Tables 1 and 2 include the corresponding
k
rgtr
values for these data points. These data indicate that for the
Clashach core, where there is a very small percentage of the micro-
pores, the contribution is limited to very high k
rgtr
values, Table 1,
Table 3
Basic test data for experiments conducted on Texas Cream at different conditions.
Index S
wi
k k(S
wi
) IFT (S
wi
) Velocity
range
(%) (mD) (mD) (mN m
1
) (m
1
) (m
1
)
18 0 9.1 0.85 3.927E9 6.0286
19 0 11.1 0.15 3.927E9
a
5.644.6
20 22 11.1 8.4 0.15 3.927E9
a
5.933E9
b
7.257.2
21 22 11.1 8.4 0.036 3.927E9
a
5.933E9
b
7.257.2
a
of Texas Cream with k=9.1 mD, has been used due to lack of measurement.
b
Calculated (S
wi
) value using, Eq. (17) of Jamiolahmady et al., (2006).
Table 4
Basic test data for experiments conducted on RC2 at different conditions.
Index S
wi
k k(S
wi
) IFT (S
wi
) Velocity
range
(%) (mD) (mD) (mN m
1
) (m
1
) (m
1
)
22 0 11 0.85 1.623E10 3.4430
23 0 11 0.15 1.623E10 6.7215
24 0 11 0.036 1.623E10 6.7108
25 33 11 7.9 0.85 1.623E10 2.527E10
a
10.0321
26 33 11 7.9 0.15 1.623E10 2.527E10
a
10.0161
27 33 11 7.9 0.036 1.623E10 2.527E10M
a
10.080.3
a
Calculated (S
wi
) value using, Eq. (17) of Jamiolahmady et al., (2006).
Table 5
Basic test data for experiments conducted on a number of cores at different conditions.
Index Core type S
wi
k k(S
wi
) IFT or (S
wi
) Velocity
range
(%) (mD) (mD) (mN m
1
) (m
1
)
28 Clashach 0 140 0.85 7.035E8 8.0770
29 25 140 100 0.85 1.164E8
a
10.71027
30 Texas Cream 0 7.4 0.85 3.927E9
b
6.148.9
31 25 7.4 5.6 0.85 6.498E9
a
8.265.2
32 RC3 0 3.9 0.85 1.550E11 22.8183
33 0 3.9 0.15 1.550E11 22.8183
34 0 3.9 0.036 1.550E11 2.391.3
35 Propped fracture 0 146,000 0.85 3.511E5 2501500
a
Calculated (S
wi
) value using, Eq. (17) of Jamiolahmady et al., (2006).
b
of Texas Cream with k=9.1 mD, has been used due to lack of measurement.
Fig. 3. Y
g
vs. Ncr for different Clashach core tests.
103 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
whilst for the Berea core, which contains a higher share of small pores,
the data at lower k
rgtr
values have also been affected, Table 2. In the
presence of connate water, in these water-wet cores, the small pores
are lled with water and the effect of preferential lling by condensate
is not observed.
Due to the unique nature of this behaviour an extra term was
added to account for this micro-pores effect as,
Y
g
= Y
g
_ _
Main
Y
g
_ _
MicPoeff
; 15
where (Y
g
)
Main
expresses the main trend followed by most of the
data points and (Y
g
)
MicPoeff
termaccounts for the effect of micro-pores.
The mathematical expressionfor (Y
g
)
Main
depends oncoreproperties
(i.e., k, and ), interfacial tension (IFT), pressure gradient and the
capillary number (Nc). Different functional forms with different coef-
cients and exponents were obtained to express these dependencies.
During this exercise care was taken to have a simpler yet more general
correlation. Hence, simpler mathematical expressions compared to
those reported for the fractional ow based correlation (Jamiolahmady,
2006) were sought withrealistic trends for the range not covered by the
available measured data, whilst ensuring that the correlation has either
universal parameters, that are applicable to all types of rocks, or para-
meters that could be determined from commonly available petrophy-
sical data. After careful examination of the measured data the following
(Y
g
)
Main
formulation was obtained:
Y
g
_ _
Main
=
1 + A
1
x
1 + A
1
x + A
2
x
2
; 16
with
x = log
r
Ncr
A
1
=
C
1
+ C
3
A
C
4
3
C
2
+ A
C
4
3
A
2
=
C
1
+ C
5
A
C
6
3
C
2
+ A
C
6
3
;
A
3
=

k
p
where C
1
=8, C
2
=800, C
3
=0.20, C
4
=2.20, C
5
=0.15, C
6
=0.81,
r
refers to the ratio of base IFT (
b
) to current IFT value () and Ncr refers
to the ratio of Nc at any condition to the corresponding value for the
base curve at the samek
rgtr
. The base value of Nc at any k
rgtr
value canbe
Fig. 4. Gas relative permeability (k
rg
) vs. relative permeability ratio (k
rgtr
) for Clashach core at three different IFT values.
Fig. 5. Gas relative permeability (k
rg
) vs. relative permeability ratio (k
rgtr
) for Texas Cream core at two different IFT values.
104 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
obtained by Eq. (17), which requires the measured k
rgb
, uid properties
and the corresponding base Nc of 1E7 dened earlier,
Ncb =
f
g
/ k
rgb
_ _
meas
Ncb =
f
g
/ k
rgb
_ _
meas
1E 7; 17
Eq. (17) has been obtained using Darcy law for two-phase ow,
Eq. (5), and Eq. (11). Furthermore, for both the base Ncb and (k
rgb
)
meas
,
Hermite cubic spline interpolation method is used to cover the whole
range of k
rgtr
variation.
Fig. 6 is the plot of Y
g
calculated frommeasuredk
rg
data using Eq. (13)
for Berea together with the best tted curve calculated using Eq. (16).
Fig. 7 demonstrates that for the measurements affected by micro-
pores, the difference between Y
g
calculated from measured k
rg
data
using Eq. (13) and the corresponding (Y
g
)
Main
values obtained from
Eq. (16), can be expressed by a mathematical expression as,
Y
g
_ _
MicPoeff
=
1 + C
7
x
1 + C
7
x + C
8
x
2
_ _
A
4
A
5
; 18
where x=log[
r
Ncr], C
7
=0.11, C
8
=0.03.
In Eq. (18), A
4
and A
5
terms determine the limit and extent of
(Y
g
)
MicPoeff
and are function of k
rgtr
, IFT and the percentage of micro-pores
in the core. Different options were considered to reect the amount of
micro-pores in the system. Finally, considering the reliability and the
amount of information available, S
w1m
, which is the wetting phase
saturation at capillary pressure (P
c
) corresponding to the radius value of
one micrometer obtained from mercury porosimetry P
c
curve, was
selected. The irreducible water saturation in the cores would ll the
micro-pores and hence, the ow of gas and condensate would not be
affected by these pores for the corresponding measurements for cores
Fig. 6. (Y
g
)
Main
vs.
r
Ncr product for Berea core tests together with best tted curve, Eq. (16).
Fig. 7. (Y
g
)
Mipoeff
vs.
r
Ncr product for data of Clashach and Berea core tests at high k
rgtr
affected by micro-pores together with the best tted curve, Eq. (18).
Table 6
The percentage of micro-pores in the cores reected by S
w1 m
.
Core type k S
w1 m
(mD) (%) (PV)
Clashach 553 17.8 0.032
Clashach 140 16.0 0.050
Berea 110 18.2 0.232
Texas Cream 11.1 23.3 0.430
Texas Cream 7.4 21.0 0.430
a
RC2 11 18.0 0.296
RC3 3.9 6.0 0.191
Propped fracture 146,000 35.4 0.001
k is the absolute permeability, is the porosity, S
w1 m
is the wetting phase saturation at
capillary pressure corresponding to the radius value of one micrometer obtained from
mercury porosimetry P
c
curve.
a
S
w1 m
of Texas Creamwith k=11 mD, has been used due to lack of P
c
measurement.
105 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
with S
wi
N0 provided that S
wi
NS
w1m
, which was the case for all mea-
surements conducted here.
In Eq. (18), A
4
limits the effect of micro-porosity to the correspon-
ding k
rgtr
and IFT values for the cores with low and moderate
percentage of micro-pores (e.g., Clashach and Berea cores, Table 6) by
the following equation:
A
4
=
C
2
9
+ k
rgtr
_ _
A
6
C
9
+ k
rgtr
_ _
A
6
_

_
_

_
A
6
=
A
7
C
9
+ C
10
IFT
r

C
11
C
9
+ IFT
r

C
11
_ _
;
A
7
=
A
9
C
10
+ C
12
S
w1m
_ _
C
13
C
9
+ S
w1m
_ _
C
13
_

_
_

_
19
where C
9
=1E4, C
10
=1200, C
11
=3.8, C
12
=100, C
13
=2.5.
In Eq. (18), A
5
restricts the application of this term to the core types
with highpercentage of micro-pores (e.g., Texas Cream, RC 2, Table 6) and
is expressed by,
A
5
=
1 + S
C
15
w1m
C
14
+ S
C
15
w1m
_
_
_
_
; 20
where C
14
=10,000, C
15
=7.
3. Error analysis of relative-permeability-ratio(k
rgtr
) basedcorrelation
The average absolute percent deviation (AAD%) and standard error
of estimates (SEE) between measured k
rg
values and the correspond-
ing calculated values by the k
rgtr
correlation, for data used for the
development of this correlation, are found in Tables 710. In Tables 7
and 8, the AAD% and SEE values in column four and ve, respectively,
labelled as (1), are those obtained using (Y
g
)
Main
, Eq. (16), for k
rg
estimations. The corresponding AAD% and SEE values in column six
and seven, respectively, labelled as (2), are those obtained using Y
g
,
Eq. (15), which include the (Y
g
)
MicPoeff
term, Eqs. (18) to (20). There
are lower values of AAD% and SEE in column six and seven compared
to those in column four and ve, respectively, for dry Clashach and
Berea cores (S
wi
=0), which have data points at high k
rgtr
affected by
the presence of micro-pores. The difference might seem minimal in
some cases but this is due to the fact that the number of data points
affected by micro-pores is very limited compared to the whole
numbers of measurements. The AAD% and SEE values in Tables 911
are all obtained using Y
g
, Eq. (15).
In this exercise AAD% greater than 30% were obtained for two sets
of measurements conducted on Clashach and Berea, tests indexed 8
and 17. It is expected that a systematic error could partly contribute to
Table 7
Average absolute deviations (AAD) and standard error of estimates (SEE) between
measured and calculated values of k
rg
for Clashach core at different conditions.
Index S
wi
IFT AAD%-k
rg
SEE-k
rg
AAD%-k
rg
SEE-k
rg
(%) (mN m
1
) (1) (1) (2) (2)
1 0 0.85 14.6 0.043 14.6 0.043
2 0 0.15 19.9 0.144 12.7 0.051
3 0 0.036 13.4 0.090 11.2 0.062
4 0 0.15 23.0 0.084 23.0 0.084
5 0 0.036 16.3 0.085 16.3 0.085
Ave. 17.4 0.091 15.6 0.065
6 33 0.85 20.0 0.042 20.0 0.042
7 21 0.85 20.7 0.048 20.7 0.048
8 21 0.15 34.9 0.237 34.9 0.237
Ave. 27.8 0.165 27.8 0.165
9 15 0.85 11.6 0.022 11.6 0.022
10 15 0.15 14.4 0.036 14.4 0.036
Ave. 13.0 0.031 13.0 0.031
All 18.9 0.105 17.9 0.093
AAD% and SEE refer to average absolute percentage deviation and standard error of
estimates, respectively.
(1) Corresponds to k
rg
values obtained using (Y
g
)
Main
, Eq. (16).
(2) Corresponds to k
rg
values obtained using Y
g
, Eq. (15).
Table 8
Average absolute deviations (AAD) and standard error of estimates (SEE) between
measured and calculated values of k
rg
for Berea core at different conditions.
Index S
wi
IFT AAD%-k
rg
SEE-k
rg
AAD%-k
rg
SEE-k
rg
(%) (mN m
1
) (1) (1) (2) (2)
11 0 0.85 13.6 0.055 13.6 0.055
12 0 0.15 14.6 0.093 12.7 0.068
13 0 0.036 16.4 0.134 7.7 0.058
Ave. 14.9 0.097 11.3 0.061
14 26 0.85 10.1 0.013 10.1 0.013
15 26 0.45 8.8 0.026 8.8 0.026
16 26 0.15 13.2 0.049 13.2 0.049
17 26 0.036 48.1 0.168 48.1 0.168
Ave. 2 0.077 20.1 0.077
All 17.8 0.088 16.3 0.070
Table 9
Average absolute deviations (AAD) and standard error of estimates (SEE) between
measured and calculated values of k
rg
for Texas Cream core at different conditions.
Index S
wi
IFT AAD%-k
rg
SEE-k
rg
(%) (mN m
1
) (2) (2)
18 0 0.85 8.7 0.045
19 0 0.15 11.2 0.063
Ave. 10.0 0.052
20 22 0.15 18.2 0.075
21 22 0.036 6.2 0.049
Ave. 12.2 0.064
All 11.1 0.056
Table 10
Average absolute deviations (AAD) and standard error of estimates (SEE) between
measured and calculated values of k
rg
for RC2 reservoir core at different conditions.
Index S
wi
IFT AAD%-k
rg
SEE-k
rg
(%) (mN m
1
) (2) (2)
22 0 0.85 17.5 0.049
23 0 0.15 14.3 0.079
24 0 0.036 11.8 0.097
Ave. 14.5 0.073
25 33 0.85 21.4 0.060
26 33 0.15 12.8 0.079
27 33 0.036 7.7 0.081
Ave. 14.0 0.072
All 14.3 0.072
Table 11
Average absolute deviations (AAD) and standard error of estimates (SEE) between
measured and calculated values of k
rg
for a number of cores not used for the
development of the correlation at different conditions.
Index Core type S
wi
IFT AAD%-k
rg
SEE-k
rg
(%) (mN m
1
) (2) (2)
28 Clashach 0 0.85 20.8 0.056
29 25 0.85 18.8 0.041
All 19.8 0.049
30 Texas Cream 0 0.85 9.6 0.028
31 25 0.85 13.5 0.035
All 11.6 0.029
32 RC3 0 0.85 11.4 0.035
33 0 0.15 14.4 0.100
34 0 0.036 17.4 0.126
All 14.4 0.093
35 Propped fracture 0 0.85 12.2 0.029
106 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
these high error values, which was noticed for all measurements in
these two tests.
The AAD% a of 19%, 18%, 11% and 14% are noted for Clashach, Berea,
Texas Cream and RC2, respectively, Tables 710. The corresponding
SEE values are 0.093, 0.070, 0.056 and 0.072, respectively. These values
are considered to be satisfactory bearing in mind that this correlation
has universal parameters and accounts for the combined effect of
coupling and inertia. Fig. 8 is a plot of measured and calculated k
rg
values for the Berea core sample at two different velocities and two IFT
of 0.036 and 0.15 mN m
1
. These measurements are only affected by
the coupling effect, i.e., there is an the increase in k
r
as velocity
increases and/or IFT decreases. The agreements between the
measured k
rg
and corresponding calculated k
rg
, are reasonably good
for the whole range of k
rgtr
variation.
Fig. 9 is a plot of measured and calculated k
rg
values for the RC2
core sample at two different velocities and IFT of 0.15 mN m
1
. The
gure shows that the correlation correctly captures the competition
between coupling and inertia as demonstrated by the experimental
measurements. At high k
rgtr
values (high gas fractional ow rates) the
relative permeability values decrease when the ow rate increases,
because of the dominance of the inertial effect but the positive
coupling effect surpasses the inertial effect at lower k
rgtr
values (lower
gas fractional ow rate) resulting in a net increase of relative
permeability with velocity. In a separate exercise the standard error
of estimates for the six tting parameters (i.e. C
1
to C
6
) of Eq. (16) was
estimated as 6.55 E-5, 1.12E-5, 1.28E-3, 2.13E-2, 4.12E-3, 3.28E-3,
respectively. These SEE are much smaller than the magnitude of the
parameters, which conrm that there is a decent estimate for the
tting parameters. In this exercise the 140 k
r
data points measured on
RC2 core were used and the procedure described by Harris (1998) for
Microsoft Excel Solver was followed.
The performance of the correlation was then evaluated against
measured data on Clashach sandstone (k=140 mD), Texas Cream
limestone (k=7.4 mD), reservoir sample RC3 (k=3.9 mD) cores and
a propped fracture medium (k=146,000 mD). These measurements
had not been used for the development of the correlation. The AAD%
and SEE between the measured and corresponding predicted values
for these data points are found in Table 11. It predicted the measured
data of these cores with AAD% of 20%, 12%, 14% and 12%, respectively.
The corresponding SEE values are 0.049, 0.028, 0.093 and 0.029,
respectively. These reasonably low deviations conrm the integrity of
the formulation, especially considering that RC3 core and propped
fracture have signicantly different core characteristics compared to
the cores used in the development of the correlation, Tables 710. It is
Fig. 8. Gas relative permeability measured in an experiment and estimated by the correlation vs. relative permeability ratio for Berea core at two IFT and velocity values.
Fig. 9. Gas relative permeability measured in an experiment and estimated by the correlation vs. relative permeability ratio for Texas Cream core at two velocity values.
107 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
important to note that in this exercise k, , and S
w1m
, (obtained
from mercury porosimetry P
c
curve) and one set of (k
rgb
)
meas
were
used to predict relative permeability at different velocity and IFT
values. As mentioned previously the correlations available in the
literature estimate the two effects separately and include a number of
parameters, which should be determined by k
r
measurements at low
IFT and high velocity conditions. This new correlation eliminates the
need for measurements of gas/condensate relative permeability at
simulated near wellbore conditions, which are very demanding and
expensive.
Figs. 10 and 11 showthat the predicted values by the correlation for
a Clashach sandstone core and carbonate Texas Cream core both with
25% initial water saturation are consistent with the corresponding
measured values. That is, both the measured and predicted k
rg
values
demonstrate an improvement in relative permeability with velocity
albeit to a different extent for these two cores.
In the case of low permeability RC3 core, Fig. 12, the predicted k
rg
values by the correlation, similarly to the corresponding measured
values, show that the relative permeability decreases due to the
impact of inertia at higher k
rgtr
values (higher gas fractional ow
rates). However at lower k
rgtr
values (lower gas fractional ow rates)
there is an improvement in k
rg
with an increase in velocity, Fig. 12.
In the case of a propped fracture with permeability of 146 D, k
rg
values are reducing with an increase in velocity due to the dominant
effect of inertia at all k
rgtr
values, Fig. 13. The velocity range worked in
this test is well beyond the range of velocity values used in the
previous core tests, Tables 15. There is also a good agreement be-
tween the measured values and corresponding predicted values by
the correlation for these results.
These consistent trends have been achieved mainly because of the
fact that the new correlation expresses the combined effect of
coupling and inertia when the two upper (k
rgm
) and lower (k
rgb
)
limits of correlation are modied for the effect of inertia before
interpolation is carried out.
As mentioned previously, the base-relative-permeability curve in
our previous work (Jamiolahmady et al., 2006) was given at
IFT=0.85 mN m
1
, with the recommendation for eld application
that the base curve should be measured at the highest applicable value
of IFT for the uid under consideration and the lowest practicable
velocity. In Section 2.3 it was discussed that this denition is not the
Fig. 10. Gas relative permeability measured in an experiment and estimated by the correlation vs. relative permeability ratio for Clashach core with 25% initial water saturation at two
velocity values.
Fig. 11. Gas relative permeability measured in the core experiments and estimated by the correlation vs. relative permeability ratio for Texas Cream core with 25% initial water
saturation at two velocity values.
108 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
correct denition of the base relative permeability curve. That is, it
does not satisfy the conditions that should reect that for IFT higher
than base IFT and velocities lower than base velocity, (k
rgb
)
meas
of gas
condensate systems should no longer be a function of IFT and velocity
switching the conventional gasoil systems.
A sensitivity study was also conducted to demonstrate the ad-
vantage of using the new relative permeability ratio based correlation
presented in this manuscript compared to our previous fractional
ow based correlation (Jamiolahmady et al., 2006). In this exercise
two different conditions at which (k
rgb
)
meas
had been measured, were
considered. Based on these information, the predicted k
rg
values using
these two approaches for some of the measured data on Texas
Cream were then compared. The measured data correspond to
the tests indexed 18 and 19 in this paper and 21 and 22 in our
previous publication. The two base measurements considered in this
study correspond to data of Fig. 2 measured at a) IFT of 9.4 mN m
1
and velocity of 5.2 md
1
and b) IFT of 0.85 mN m
1
and velocity of
6.1 md
1
. In the case of the previous correlation two different set of
(k
rgb
)
meas
, Fig. 2a, and Ncb, Eq. (18) of Jamiolahmady 2006, were
obtained. This resulted in a signicant increase in AAD% between
predicted and measured k
rg
values from a) 6% and 13% (reported in
Table 5 of the previous publication) to b) 90% and 109%, respectively.
In the case of the new correlation the (k
rb
)
meas
measured at IFT of
0.85 mNm
1
were rst transferred to that of the specied base IFTof
3 mN m
1
using Eq. (12) then Ncb was calculated using Eq. (17).
Fig. 2b shows that (k
rb
)
meas
, and hence Ncb, remains the same when
IFT is increased from 3 to 9.4 mN m
1
. Therefore, the corresponding
AAD% values of 8.7 % and 11.2% reported in Table 9 is applicable for
both cases. These results highlight that the use of the previous
correlation is limited only to the range of data used in that study. In
other words, a more general denition of both base IFT and base
capillary number are required for a correlation universal parameters,
as presented in this manuscript.
4. Summary and conclusions
A generalised relative permeability correlation for gascondensate
systems, which is based on relative permeability ratio, has been
developed. It accounts for the combined effects of positive coupling
and negative inertia. In this approach gas relative permeability is
correlated and condensate relative permeability is obtained by the
expression of relative permeability ratio. The k
rg
correlation inter-
polates between k
rgb
and k
rgm
curves, both modied for the effect of
inertia, using a generalized interpolating parameter Y
g
, which
expresses the dependency of the relative permeability to velocity
and interfacial tension including micro-pore effect.
The parameters of the correlation are either universal (constant for
all cores) or can be obtained from commonly available petrophysical
Fig. 12. Gas relative permeability measured in an experiment and estimated by the correlation vs. relative permeability ratio for RC3 core at two velocity values.
Fig. 13. Gas relative permeability measured in an experiment and estimated by the correlation vs. relative permeability ratio for a propped fracture at two velocity values.
109 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110
rock properties (i.e., permeability, porosity, single-phase inertial
factor, capillary pressure curve obtained from mercury porosimetry).
This eliminates the need for difcult and expensive relative perme-
ability measurements at near well conditions.
The comprehensive relative permeability data (measured in our
Laboratory at Heriot-Watt University) on cores with permeability
ranging from 9 mD to 550 mD, with the corresponding basic test data
given in Tables 14 and 6, and different lithology (sandstone and
carbonate) were used to develop the presented correlations. When
the correlations were tested for a 3 mD sandstone, 7 mD carbonate
and 146 D propped fracture core, none of which had been used in
developing the correlation, with the corresponding basic test data
given in Tables 56, the results were very satisfactory showing a
relatively low and uniformly distributed deviations.
These results conrm the generality of the correlation and
reliability of the information obtained from it to great extent as
these measurements covered a reasonably wide range of IFT and
velocity variations for porous media with very different character-
istics. However, more data is being generated to further evaluate the
performance of this correlation for very tight cores. That is, although
the RC3 core sample is a relatively low permeability core but the
evaluation of correlation should also be directed towards cores with
permeabilities in the range of a fraction of milli-Darcy.
Acknowledgements
The above study was conducted as a part of the Gascondensate
Recovery Project at Heriot-watt University. This research project is
sponsored by: The UK Department for Business, Enterprise &
Regulatory Reform (BERR), BP Exploration Company (Colombia) Ltd,
Eni Petroleum Co, Gaz de France, Petrobras, StatoilHydro and Total
Exploration UK plc, which is gratefully acknowledged.
References
Ali, J.K., McGauley, P.J., Wilson, C.J., Oct. 1997. The effects of high velocity ow and PVT
changes near the wellbore on condensate well performance. SPE 38923, Proc. of SPE
Annual Technical Conference and Exhibition, Texas, pp. 823838.
Al-Kharusi, S.B., Jan. 2000. Relative permeability of gascondensate near wellbore, and
gascondensatewater in bulk of reservoir. PhD thesis, Heriot-Watt University.
App, J.F., Mohanty, M., 2002. Gas and condensate relative permeability at near critical
conditions: capillary and Reynolds number dependence. J. Petrol. Sci. Engng. 36 (1),
111126.
Asar, H., Handy, L.L., Feb. 1988. Inuence of interfacial tension on gas/oil relative
permeability in a gascondensate system. SPE 11740, SPERE 3 (1), 264275.
Bardon, C., Longeron, D.G., Oct. 1980. Inuence of very lowinterfacial tension on relative
permeability. SPEJ 20 (3), 391401.
Blom, S.M.P., Hagoort, J., Sept. 1998. How to include the capillary number in gas
condensate relative permeability functions? SPE 49268, Proc. of SPE Annual
Technical Conference and Exhibition, Louisiana, pp. 661671.
Blom, S.M.P., Hagoort, J., Soetekouw, D.P.N., Oct. 1997. Relative permeability near welbore
conditions. SPE 38935, Proc. of SPE Annual Technical Conference and Exhibition,
Texas, pp. 957967.
Coats, K.H., Oct. 1980. An equation of state compositional model. Soc. Pet. Eng. J. 20,
363367.
Dacun, L., Thomas, W.E., May 2001. Literature review on correlations of the non-Darcy
coefcient, SPE 70015. Proc. of the Permian Basin oil and Gas Recovery Conference,
Texas.
Danesh, A., khazam, M., Henderson, G.D., Tehrani, D.H., Peden, J.M., June 1994. as
condensate recovery studies. Proc. of DTI Improved Oil Recovery and Research
Dissemination Seminar, London.
Forchheimer, P.: Hydraulik, Chapter 15, Leipzik and Berlin, 1914, pp. 1168.
Harris, D.C., Jan. 1998. Nonlinear least-square curve tting with Microsoft Excel Solver.
Journal of Chemical Education 75 (1), 119121.
Henderson, G.D., Danesh, A., Tehrani, D.H., Peden, J.M., May 1997. The effect of velocity
and interfacial tension on the relative permeability of gas condensate uids in the
wellbore region. J. Pet. Sci. Eng. 17, 265273 also in Proc. of 8th IOR Symposium
Vienna, 1995, 201208.
Henderson, G.D., Danesh, A., Tehrani, D.H., Peden, J.M., June 1996. Measurement and
correlation of gas condensate relative permeability by the steady-state method.
SPEJ 191201 SPE 31065.
Henderson, G.D., Danesh, A., Tehrani, D.H., 2001. Effect of positive rate sensitivity and
inertia on gas condensate relative permeability at high velocity. Pet. Geosci. 7,
4550.
Jamiolahmady, M., Danesh, A., Tehrani, D.H., Duncan, D.B., Oct. 2000. A mechanistic
model of gascondensate ow in pores. Transp. Porous Media 41 (1), 1746.
Jamiolahmady, M., Danesh, A., Tehrani, D.H., Duncan, D.B., 2003. Positive effect of ow
velocity on gas-condensate relative permeability: network modelling and comparison
with experimental results. Transp. Porous Media 52 (2), 159183.
Jamiolahmady, M., Danesh, A., Henderson, G.D., Tehrani, D.H., Dec. 2006. Variations of
gascondensate relative permeability with production rate at near wellbore
conditions: a general correlation. SPE Reserv. Eng. Evalu. J. 9 (6), 688697 SPE
83960, also in Proc. ofthe SPE OEC, Aberdeen, UK, Sept. 2003.
Mott, R., Cable, A., Spearing, M., Oct. 2000. Measurements and simulation of inertial and
high capillary number ow phenomena in gascondensate relative permeability,
SPE 62932. Proc. of SPE Annual Technical Conference and Exhibition Texas.
Narayanaswamy, G., Sharma, M.M., Pope, G.A., June 1999. Effect of heterogeneity on the
non-Darcy ow coefcient. SPE Reservoir Eval. Eng. 2 (3), 296302.
Pope, G.A., Narayanaswamy, G., Delshad, M., Sharma, M., Wang, P., Sept. 1998. Modelling
relative permeability effects in gascondensate reservoirs. SPE 49266. Proc. of SPE
Annual Technical Conference and Exhibition, Louisiana, pp. 643654.
Mcdougall, S.R., Cruickshank, J., Sorbie, K.S., 2002. Anchoring methodologies for pore-
scale network models: application to relative permeability and capillary pressure
prediction, Petrophysics Houston, VOL 43; PART 4, pp. 365375.
Whitson, C.H., Fevang, O., Saevareid, A., Oct. 1999. Gas condensate relative permeability
for well calculations, SPE 56476. Proc. of SPE Annual Technical Conference and
Exhibition Texas.
110 M. Jamiolahmady et al. / Journal of Petroleum Science and Engineering 66 (2009) 98110

Você também pode gostar