Energy & Fuels 2005, 19, 561-572
561
Predicting Natural Gas Dew Points from 15 Equations of
State
Kh. Nasrifar* and O. Bolland
Department of Energy and Process Engineering, Norwegian University of Science and
Technology, N-7491 Trondheim, Norway
M. Moshfeghian
Kuwait Institute for Scientific Research, Petroleum Research and Studies Center,
13109 Safat, Kuwait
Received July 5, 2004
This paper is a comparative study of 15 cubic equations of state (EoSs) for predicting natural
gas dew points. Two-, three-, and four-parameter EoSs are used to predict the natural gas dew
points. Natural gases contain a large amount of supercritical methane; therefore, the fugacity of
methane, as a test of suitability of the EoSs, is predicted and compared to the recommended
values in International Union of Pure and Applied Chemistry (IUPAC) tables. The vapor pressures
of components that are normally present in natural gas mixtures, as another test of the suitability
of the EoSs, are also predicted and compared with experimental data. The dew points of several
natural gas mixtures then are predicted, using the EoSs, and compared to experimental values.
Results reveal that the dew points of lean synthetic natural gases are predicted best by the
Redlich-Kwong family of EoSs, whereas rich natural gases dew points are described significantly
better by the Patel-Teja family of EoSs.
1. Introduction
Predicting the natural gas dew point is of primary
importance in reservoir engineering, where liquid dropout diminishes production from gas condensate wells.
It is also equally important to consider it in the design
of gas transportation pipelines and custody transfer.
Equations of state (EoSs) are often used to predict
natural gas dew points. Although EoSs are plentiful,
the Redlich and Kwong1 (RK) EoS, a modified version
of it by Soave2 (RKS), and the Peng and Robinson3 (PR)
EoS are usually prescribed for calculating natural gas
dew points.4 The Patel and Teja5 (PT) EoS and a
modified version of it by Valderrama6 (PTV) are also
considered valuable in hydrocarbon processing.7,8 However, many EoSs, including some variants of the popular
PR and RKS EoSs, have been developed, and this trend
seems to be continuing. The question is this: Are the
RKS and PR EoSs still the best for hydrocarbon
processing? To answer this question, the newly devel* Author to whom correspondence should be addressed. Telephone:
+47 735 98462. Fax: +47 735 98390. E-mail: khashayar.nasrifar@
ntnu.no.
(1) Redlich, O.; Kwong, J. N. S. Chem. Rev. 1949, 44, 233-244.
(2) Soave, G. Chem. Eng. Sci. 1972, 27, 1197-1203.
(3) Peng, D. Y.; Robinson, D. B. Ind. Eng. Chem. Fundam. 1976,
15, 59-64.
(4) Valderrama, J. O. Ind. Eng. Chem. Res. 2003, 42, 1603-1618.
(5) Patel, N. C.; Teja, A. S. Chem. Eng. Sci. 1982, 37, 463-473.
(6) Valderrama, J. O. J. Chem. Eng. Jpn. 1990, 23, 87-91.
(7) Danesh, A.; Xu, D.; Todd, A. C. Fluid Phase Equilib. 1991, 63,
259-278.
(8) Danesh, A. PVT and Phase Behavior of Petroleum Reservoir
Fluids; Elsevier Science B.V.: Amsterdam, The Netherlands, 1998.
oped EoSs should be taken into consideration. Clearly,
evaluation of all developed EoSs may not be practical;9
however, it is likely that, based on some criteria, one
can select the appropriate EoSs and study them to
calculate natural gas dew points. The subject of this
study is to evaluate and recommend suitable EoSs.
Natural gases consist of many components. Nitrogen
(N2), carbon dioxide (CO2), and hydrogen sulfide (H2S)
are usually the non-hydrocarbon components. Methane
(CH4), ethane (C2H6), and other hydrocarbons (up to C40
or even C50) might be present in a natural gas mixture.
Nitrogen and methane are always supercritical under
reservoir conditions, whereas the heavy hydrocarbons,
which cause condensation, are present under subcritical
conditions. The temperature and pressure of a hyperbaric reservoir10 could be as high as 450 K and 100 MPa,
respectively. In fact, any EoS capable of predicting
natural gas dew points should be able to characterize
these complex mixtures under reservoir conditions
accurately. The following strategy is then outlined:
Only cubic EoSs are evaluated. These EoSs are empirical or semiempirical, yet simple and robust, and their
application to natural gas mixtures has been proven
useful. Because the RKS and PR EoSs have the best
pressure-volume-temperature (PVT) relationships
among the other two-parameter EoSs, several variants
of them will be studied. Three-parameter EoSs and the
(9) Vidal, J. Fluid Phase Equilib. 1989, 52, 15-30.
(10) Floter, E.; de Loos, Th. W.; de Swan Arons, J. Ind. Eng. Chem.
Res. 1998, 37, 1651-1662.
10.1021/ef0498465 CCC: $30.25 2005 American Chemical Society
Published on Web 02/11/2005
562
Energy & Fuels, Vol. 19, No. 2, 2005
Nasrifar et al.
Table 1. PVT Relations for the Equations of State (EoSs) Used in This Study
PVT relationa
equation of state, EoS
RKS, RKT, RKSS
P)
aCR(Tr)
RT
v - b v(v + b)
PR76, PR78, PRD, PRT, PRF, PRG
P)
aCR(Tr)
RT
v - b v(v + b) + b(v - b)
PT, PTV
P)
aCR(Tr)
RT
v - b v(v + b) + c(v - b)
SW
P)
aCR(Tr)
RT
v - b v2 + (1 + 3)bv - 3b2
GD
P)
aCR(Tr)
RT
v - b v(v + c) + c(v - b)
MMM
P)
aCR(Tr)/xT
RT v + b
v v-b
v(v + Nb)
TBS
P)
aCR(Tr)
RT
v - b v2 + (b + c)v - (bc + d2)
a Calculated using the following parameters: R is the gas constant; T is the temperature; T is the reduced temperature; P is the
r
pressure; v is the molar volume; c, d, N, and are the EoSs parameters; is the acentric factor; b is the molecular co-volume; aC is the
attractive parameter at the critical temperature; and R is the attractive parameter temperature dependence (this latter parameter is also
called the R-function).
Trebble-Bishnoi-Salim11 (TBS) EoS, as the only fourparameter EoS, will also be used to predict natural gas
dew points. van der Waals mixing rules with a geometric
mean average combining rule for the attractive parameter are used to extend the EoSs to multicomponent
mixtures. In most cases, the van der Waals mixing rules
adequately describe natural gas mixtures. However,
because these mixing rules are not perfect, binary
interaction parameters (BIPs) are usually incorporated
to correct the interaction between binary molecules;
these corrections are represented by the parameter kij
in the EoSs. The impact of binary interaction parameters will also be discussed in this study.
First, the EoSs are used to predict the supercritical
behavior of methane and nitrogen, in terms of fugacity,
for wide ranges of pressure and temperature. The EoSs
then are used to predict the vapor pressure of some light
and heavy hydrocarbons at the prevailing temperatures
of the reservoirs. The behavior of components in a
mixture is clearly different than that in a pure state;
however, these comparisons can still give good indications for the suitability of EoSs in predicting natural
gas dew points. Afterward, the accuracies of the EoSs
are compared to predict the dew points of some synthetic
natural gases that were reported recently by a Spanish
group.12,13 Finally, having specified an exponential decay
function for the composition of single carbon number
(SCN) groups in C7+ fractions, the dew points of rich
natural gases are predicted and compared with experimental data. The effect of correlations on determining
the critical properties of the SCN group will also be
discussed.
2. The Equations of State
The EoSs used in this study are classified as two-,
three-, and four-parameter EoSs. The two-parameter
EoSs are solely comprised of the RK and PR families.
From the RK family, the modified versions of Soave2
(RKS), Twu et al.14 (RKT), and Souahi et al.15 (RKSS)
are used. From the PR family, the original Peng and
Robinson (PR76,3 PR7816) EoSs are used, as well as
modified versions of them by Danesh et al.17 (PRD), Twu
et al.18 (PRT), Floter et al.10 (PRF), and Gasem et al.19
(PRG). The three-parameter EoSs include the following: Schmidt and Wenzel20 (SW), Guo and Du21 (GD),
Patel and Teja5 (PT), a modified PT EoS by Valderrama6
(PTV), and an EoS by Mohsen-Nia, Modarress, and
Mansoori22 (MMM). The TBS EoS is the only fourparameter EoS that is used in this study.
The PVT relationships for the previously mentioned
EoSs are given in Table 1. In Table 2, the temperature
dependences for the attractive terms of the EoSs (which
are also called R-functions) are presented. The EoSs are
presented briefly in the following sections, however, the
details can be found in the given references.
2.1. Two-Parameter Equations of State. 2.1.1. The
RKS EoS.2 The RK EoS successfully relates the PVT of
(11) Salim, P. H.; Trebble, M. A. Fluid Phase Equilib. 1991, 65, 5971.
(12) Jarne, C.; Avila, S.; Blanco, S. T.; Rauzy, E.; Otin, S.; Velasco,
I. Ind. Eng. Chem. Res. 2004, 43, 209-217.
(13) Avila, S.; Blanco, S. T.; Velasco, I.; Rauzy, E.; Otin, S. Ind. Eng.
Chem. Res. 2002, 41, 3714-3721.
(14) Twu, C. H.; Coon, J. E.; Cunningham, J. R. Fluid Phase Equilib.
1995, 105, 61-69.
(15) Souahi, F.; Sator, S.; Albne, S. A.; Kies, F. K.; Chitour, C. E.
Fluid Phase Equilib. 1998, 153, 73-80.
(16) Robinson, D. B.; Peng, D.-Y. The Characterization of Heptanes
and Heavier Fractions for the GPA Peng-Robinson Programs, GPA
Research Report No. 28, Gas Processors Association, Tulsa, OK, 1978.
(17) Danesh, A.; Xu, D.-H.; Tehrani, D. H.; Todd, A. C. Fluid Phase
Equilib. 1995, 112, 45-61.
(18) Twu, C. H.; Coon, J. E.; Cunningham, J. R. Fluid Phase Equilib.
1995, 105, 49-59.
(19) Gasem, K. A. M.; Gao, W.; Pan, Z.; Robinson, R. L. Fluid Phase
Equilib. 2001, 181, 113-125.
(20) Schmidt, G.; Wenzel, H. Chem. Eng. Sci. 1980, 35, 1503-1512.
(21) Guo, T.-M.; Du, L. Fluid Phase Equilib. 1989, 52, 47-57.
(22) Mohsen-Nia, M.; Modarress, H.; Mansoori, G. A. Fluid Phase
Equilib. 2003, 206, 27-39.
Predicting Natural Gas Dew Points
Energy & Fuels, Vol. 19, No. 2, 2005 563
Table 2. The r-Functions Used in This Study
R-functiona
equation of state, EoS
RKS, PR76, PR78, PT, PTV, SW, GD, MMM
PRD
PRF
PRG
RKT, PRT
RKSS
TBS
R ) [1 + m(1 - xTr
{
{
)]2
RSubcritical ) [1 + m(1 - xTr)]2
RSupercritical ) [1 + 1.21m(1 - xTr)]2
RSupercritical methane ) a0 + a1Tr + a2T2r + a3T3r
ROthers ) [1 + m(1 - xTr)]2
R ) exp[(A + BTr)(1 - TC+D+E
)]
r
R ) R(0) + (R(1) - R(0))
(M-1) exp[L(1 - TNM)]
R(i) ) TN
r
r
R ) R(0) + (R(1) - R(0))
R(i) ) 1 + [A(1 - Tr) + B(1 - Tr)2 + C(1 - Tr)3 + D(1 - Tr)6]/Tr
R ) 1 + m(1 - xTr) + p(x0.7 - xTr)(1 - xTr)
Calculated using the following parameters: Tr is the reduced temperature; is the acentric factor; and m, a0, a1, a2, a3, A, B, C, D,
E, N, M, L, and p each are either a constant or a function of the acentric factor.
gases; however, it poorly predicts the vapor pressure and
liquid density of pure compounds. Soave2 introduced a
temperature dependence (R) for the attractive term of
the RK EoS, as given in Table 2. This term significantly
improves the accuracy of the EoS to predict vapor
pressure, although the accuracy of the EoS to predict
liquid density remains unchanged.23 As a consequence
of this modification, the RKS can be used successfully
in the fluid phase equilibria of hydrocarbon mixtures.
However, the R-function causes the RKS to predict
anomalous behaviors under extreme conditions.24
2.1.2. The RKT EoS.14 Soave2 correlated the R-function for the RKS EoS by matching the predicted vapor
pressure at a reduced temperature of 0.7 to the experimental value. Although it has been proven to be useful,
the vapor pressure predicted by the RKS EoS usually
gets worse at reduced temperatures of <0.7. Recognizing
this behavior, especially for heavy hydrocarbons, Twu
et al.14 developed an R-function that works well from
the triple-point temperature to the critical-point temperature. By predicting the vapor pressure of heavy
hydrocarbons well, the RKT EoS could be capable of
describing the vapor-liquid equilibria (VLE) of natural
gas mixtures that contain heavy hydrocarbons.
2.1.3. The RKSS EoS.15 Souahi et al.15 developed a
new R-function with the same purpose as Twu et al.;14
however, their R-function seems to be more accurate in
regard to predicting the vapor pressure of pure compounds. However, they did not address the behavior of
the R-function under supercritical conditions.
2.1.4. The PR763 and PR7816 EoSs. Another successful
PVT relation among EoSs is the PR EoS. Compared to
RK family of EoSs, the PR family of EoSs generally
predicts the liquid density of compounds more accurately;23 however, the accuracy is not good enough for
industrial applications. The PR EoS takes advantage
of the Soave-type R-function, hence, showing similar
quality with temperature. However, Peng and Robinson3
used a reduced temperature range (0.7-1) to correlate
the PR R-function. Also, it was first correlated to pure
compounds with an acentric factor of <0.5 (PR76),3 and
later, in 1978, they extended it to compounds with a
larger acentric factor (PR78).16
2.1.5. The PRD EoS.17 The PRD EoS is a revised
version of the PR78 EoS. It takes into consideration the
(23) Nasrifar, Kh.; Moshfeghan, M. Fluid Phase Equilib. 1999, 158160, 437-445.
(24) Segura, H.; Kraska, T.; Meja, A.; Wisniak, J.; Polishuk, I. Ind.
Eng. Chem. Res. 2003, 42, 5662-5673.
supercritical behavior of components in a hydrocarbon
mixture by multiplying the parameter m in the PR
R-function by a factor of 1.21. As reported by Danesh et
al.,17 it works well for calculating the dew points of gas
condensate mixtures.
2.1.6. The PRF EoS.10 The PRF is similar to the PRD
EoS, i.e., it uses the PR78 EoS R-function for all
components except methane under supercritical conditions. Floter et al.10 introduced a third-order polynomial
R-function for methane under supercritical conditions
by matching the fugacity of the PR EoS to the recommended values in IUPAC tables. The PRF EoS is quite
impressive for describing the VLE of the binary asymmetric mixtures.
2.1.7. The PRT EoS.18 The reason for developing the
PRT EoS is similar to that for the RKT EoS. The PRT
EoS is an improved version of the PR EoS, especially
for predicting the vapor pressure of pure compounds at
low temperatures.
2.1.8. The PRG EoS.19 Gasem et al.19 developed a new
R-function in exponential form, having recognized that
the Soave-type R-function used by the PR EoS does not
decrease monotonically to zero with temperature. The
PRG EoS also attempts to improve the predictive
capability of the PR EoS for vapor pressure.
2.2. Three-Parameter Equations of State. 2.2.1.
The PT5 and PTV6 EoSs. The PT and PTV EoSs have
the same PVT relationship and R-functions. The difference between them involves the calculation of the EoSs
parameters, i.e., a, b, and c. In the PTV EoS, the actual
compressibility factor is used, whereas the critical
compressibility factor in the PT EoS is a conventional
parameter. Consequently, the PTV EoS predicts liquid
densities near the critical point more accurately than
the PT EoS.
2.2.2. The SW EoS.20 Schmidt and Wenzel20 recognized that the RKS EoS accurately predicts the thermodynamic properties of fluids with an acentric factor
of almost zero, whereas the PR EoS with an acentric
factor of 0.3. Knowing that, they developed a new EoS
that reduces to the RKS EoS at an acentric factor of 0
and to the PR EoS at an acentric factor of 1/3. In the
SW EoS, the acentric factor is a third parameter. The
SW EoS accurately predicts the liquid density and vapor
pressure of light and moderate compounds.
2.2.3. The GD EoS.21 The GD EoS is a threeparameter EoS that is very similar in form to the PT
EoS.5 It was developed particularly for hydrocarbon
564
Energy & Fuels, Vol. 19, No. 2, 2005
Nasrifar et al.
mixtures. The R-function for the GD EoS is similar to
the Soave-type R-function; however, the parameter m
is defined differently for different compounds, based on
the acentric factor. As shown by Guo and Du,21 the pure
and mixture properties are predicted quite accurately
using the GD EoS.
2.2.4. THE MMM EoS.22 Mohsen-Nia et al.22 did not
use a van der Waals repulsive term for the MMM EoS.
Instead, they used a more accurate empirical repulsive
term, having considered the molecular simulation data
of hard spheres. The MMM EoS is accurate for predicting vapor pressure and liquid density of light and
moderate pure compounds. The R-function in the MMM
EoS is of the Soave type. They also used a temperature
dependence term for the molecular co-volume parameter. As shown by Mohsen-Nia et al.,22 this improves
the capability of the EoS to predict liquid density;
however, as noted by Salim and Trebble,11 this may
cause anomalous behaviors under extreme conditions.
2.3. The Four-Parameter Equation of State: TBS
EoS.11 The TBS EoS is a four-parameter EoS. The PVT
relationship for the TBS is flexible and can be reduced
to other EoSs. Although it is accurate for predicting the
thermodynamic properties of pure compounds, its extension to mixtures requires four mixing rules. Considering that mixing rules are generally empirical, the
application of four mixing rules may diminish the
predictability of the TBS EoS for mixtures.
properties. Nevertheless, usually, compositions up to C6
are available with a heavy end that must be characterized.
It is a common practice to split a C7+ fraction into
several SCN groups and then calculate the critical
properties of each group, using available correlations.26-29
Katz30 introduced a simple decay exponential function
to express the distribution of SCN groups. Pedersen et
al.25 also determined that the compositional distribution
of SCNs in North Sea petroleum fractions is bestdescribed by an exponential function. Recently, Starling31 also suggested an exponential decay function for
splitting heavy ends. The decay functions are similar
to each other. However, the decay function by Pedersen
et al.25 seems to be more general and will be used in
this study:
3. Mixing Rules
CN
The van der Waals mixing rules have proven to be
useful in hydrocarbon processing. The van der Waals
quadratic mixing rule with a geometric combining rule
is used for the attractive parameter of the EoSs:
a)
i j xixjaij
(1)
with
aij ) xaiiajj(1 - kij)
(2)
where kij is the BIP. For the second, third, and fourth
parameters, the following mixing rule is used:
j xjwj
w)
(3)
where w ) b, c, d, and in different EoSs.
4. Heavy Ends
The heavy constituents of a natural gas are usually
considered as a group and reported as C7+. The C7+
fraction is commonly specified by molecular weight and
specific gravity. The C7+ fractions may be comprised of
many components and different families of hydrocarbons. Although it is possible to analyze the C7+
fractions accurately, using new techniques such as gas
capillary chromatography, Pedersen et al.25 discovered
that the specification of components up to C20 may be
adequate for an accurate calculation of thermodynamic
(25) Pedersen, K. S.; Blilie, A. L.; Meisingset, K. K. Ind. Eng. Chem.
Res. 1992, 3, 1378-1384.
Zn ) exp(A + BMn)
(4)
where Zn is the SCN group mole fraction and Mn is the
SCN group molecular weight. The unknowns A and B
are determined for a C7+ fraction using the following
constraints:
CN
ZC7+ )
ZC
(5)
and
ZC MC
C
n
MC7+ )
ZC7+
(6)
where CN is the heaviest SCN to be considered in a C7+
fraction and Cn is a dummy variable. To determine A
and B (and, hence, the SCN distribution), the SCN
molecular weight and volume are needed. We used the
generalized SCN properties, as reported by Whitson.32
5. Results and Discussion
The behavior of components in a multicomponent
mixture can be significantly different than in a pure
state, depending on the difference in size, shape, and
polarity of components. However, in the VLE of natural
gas mixtures, it is valuable to evaluate (i) the supercritical behavior of methane, which comprises a large
portion of natural gases; (ii) the supercritical behavior
of nitrogen, which interacts significantly with hydrocarbons; and (iii) the subcritical behavior of hydrocarbons, which cause condensation. This study is limited
to temperatures of <450 K and pressures of <100 MPa,
because usually reservoir temperatures and pressures
fall within these ranges.
(26) Cavett, R. H. In Proceedings of the 27th API Meeting, San
Francisco, CA, 1962; pp 351-366.
(27) Riazi, M. R.; Daubert, T. E. Ind. Eng. Chem. Res. 1987, 26, 755759.
(28) Twu, C. H. Fluid Phase Equilib. 1984, 16, 137-150.
(29) Lee, B. I.; Kesler, M. G. Hydrocarbon Process. 1980, (July), 163167.
(30) Katz, D. J. Petrol. Technol. 1983 (June), 1205-1214.
(31) Starling, K. E. Presented at the American Gas Association
Operations Conference, Orlando, FL, April 27-30, 2003.
(32) Whitson, C. H. SPE J. 1983 (August), 683-694.
Predicting Natural Gas Dew Points
Energy & Fuels, Vol. 19, No. 2, 2005 565
Table 3. Predicting the Fugacity of Methane and
Nitrogen from Cubic Equations of State
Methanea
Nitrogenb
equation of state,
EoS
%AAD
RMSD
(MPa)c
%AAD
RMSD
(MPa)c
RKS
RKT
RKSS
PR76
PR78
PRD
PRF
PRT
PT
PTV
SW
GD
MMM
TBS
3.77
6.01
7.92
7.68
7.51
22.93
1.74
5.19
1.16
1.46
6.72
2.66
10.65
9.84
2.90
4.68
6.91
6.68
6.54
18.95
1.59
4.68
0.73
1.64
5.45
2.27
8.21
8.02
2.10
2.26
7.30
7.96
7.85
25.99
7.85
5.00
1.58
2.29
3.37
1.33
7.18
12.42
2.86
2.86
11.56
13.45
13.31
34.49
13.31
10.30
4.81
6.10
4.08
3.42
7.17
16.39
a For a temperature range of 220-440 K and pressure range of
1-100 MPa. Data taken from ref 33. Number of points ) 96. b For
a temperature range of 200-450 K and pressure range of 1-100
MPa. Data taken from ref 34. Number of points ) 72. c Root mean
square deviation: RMSD )
x(1/N) Nj (f cald,j-f expl,j)2.
Table 3 presents the average absolute deviation,
expressed by
%AAD )
100
N
|caldj - explj|
explj
(7)
where the cald and expl denote the calculated and
experimental values for predicting the fugacity of
methane and nitrogen for the defined temperature and
pressure limits. Under reservoir conditions, methane
and nitrogen are always supercritical. The accuracy of
an EoS to predict the fugacity of methane leads to the
accurate characterization of natural gas mixtures.
Indicated in Table 3, the PT, PTV, and GD EoSs are
significantly in good agreement with the recommended
values in the IUPAC tables.33,34 The PRF EoS is also
accurate for methane, because the authors matched the
fugacity of the EoS to the IUPAC values. Among the
two families, the RK family is superior to the PR family
and the RKS EoS is remarkably better than the RKT
and RKSS EoSs. In the PR family, the PRF and PRT
describe the fugacity of methane and nitrogen better
than the others. The PRD, MMM, and TBS EoSs are
clearly inferior, with respect to the others. Figure 1
depicts the agreement between the PT EoS, as the best
EoS in predicting fugacity, and the IUPAC values for
the given temperature and pressure ranges.
In Tables 4 and 5, one can see the ability of different
EoSs to predict the vapor pressure of potential compounds in natural gases for a temperature range from
the triple-point temperature to the critical-point temperature. When the coexistence curve temperature
range for a compound was outside of 450 K, the
predictions were limited to temperatures of <450 K. The
RK family is clearly superior to the PR family. Among
the other EoSs, the GD EoS is superior, whereas the
(33) Wagner, W.; de Reuck, K. M. Methane: International Thermodynamic Tables of the Fluid State, Vol. 13; IUPAC Chemical Data
Series No. 41; Blackwell Science, Ltd.: Oxford, U.K., 1996.
(34) Angus, S.; de Reuck, K. M.; Armstrong, B. Nitrogen: International Thermodynamic Tables of the Fluid State, Vol. 6; IUPAC
Chemical Data Series No. 20; Pergamon Press: Oxford, U.K., 1979.
Figure 1. Fugacity coefficients of methane from the PT
equation of state (EoS). (Symbols are recommended values
from IUPAC.33)
MMM is inferior. Overall, the RKT and PRT EoSs
predict the vapor pressure of pure compounds quite
accurately, with the same average absolute deviations.
The RKSS and RKS EoSs are ranked third and fourth,
respectively. Considering these results and the prediction capability of the EoSs for describing the fugacity
of methane, one can conclude that, overall, the RK
family EoSs can best predict the VLE of natural gas
mixtures and are superior to other EoSs. However, this
is not absolutely true, knowing that several other
parameters have a role in describing the dew points of
natural gases by EoSs.
BIP is one of the important parameters affecting the
VLE of mixtures. The BIPs are, at best, empirical,
usually temperature-dependent, and dependent on the
EoS. The BIPs are determined by correlating (curvefitting) the VLE of binary mixtures. Having the BIPs
of all binaries in a mixture, one can predict the dew
points of a mixture and expect a good result. Danesh et
al.17 claimed that, in the VLE of multicomponent
mixtures, it is very important to obtain the EoS parameters accurately; however, they never ignore the importance of BIPs in the VLE calculation. In Figure 2, the
dew points of a synthetic natural gas mixture are
predicted with and without BIPs by the RKS, PR, and
PT EoSs and compared to experimental data. The
(35) Daubert, T. E.; Danner, R. P. Physical and Thermodynamic
Properties of Pure Chemicals: Data Compilation; Hemisphere Publishing Corp.: London, U.K., 1992.
(36) Knapp, H.; Doring, R. Vapor-Liquid Equilibria for Mixtures
of Low Boiling Substances; Berhens, D.; Eckerman, R. Eds.; Dechema
Chemistry Data Series, Part I-Binary System; Dechema: Frankfurt,
1986.
(37) Willman, B. T.; Teja, A. S. AIChE J. 1986, 32, 2067-2078.
566
Energy & Fuels, Vol. 19, No. 2, 2005
Nasrifar et al.
Table 4. Predicting the Vapor Pressure of Pure Compounds, in Terms of %AAD, from the Equations of State for the RK
and PR Families
%AAD for Vapor Pressurea
component
Tr range
RKS
RKT
RKSS
PR76
PR78
PRG
PRT
CO2
H2S
C1
C2
C3
n-C4
i-C4
n-C5
i-C5
neo-C5
n-C6
cyc-C6
CH3-cyc-C5
C6H6
n-C7
CH3-cyc-C6
n-C8
n-C9
n-C10
n-C12
n-C14
n-C16
n-C18
n-C20
average
0.71-1
0.50-1
0.48-1
0.30-1
0.23-1
0.32-1
0.28-1
0.31-0.96
0.25-0.98
0.59-1
0.35-0.87
0.51-0.81
0.25-0.84
0.50-0.80
0.34-0.83
0.26-0.79
0.38-0.79
0.27-0.77
0.39-0.73
0.40-0.68
0.40-0.65
0.41-0.62
0.41-0.60
0.40-0.59
0.44
2.26
2.01
3.47
3.06
2.31
8.92
6.15
2.52
0.85
5.36
0.60
9.41
0.50
5.46
22.1
2.69
2.78
3.84
0.67
6.32
5.89
9.35
8.54
4.81
0.11
3.69
0.26
1.83
7.25
2.99
4.79
2.38
7.81
0.37
2.49
0.68
4.82
1.36
1.05
4.96
4.67
13.86
6.18
2.42
2.49
8.58
5.45
6.91
4.06
0.30
3.29
0.63
0.69
4.26
1.71
8.22
5.04
4.43
0.52
3.92
0.60
3.17
1.26
3.16
12.99
3.20
8.66
4.79
1.99
6.09
2.81
13.48
13.48
4.53
0.52
3.84
0.50
3.47
9.26
4.31
20.09
15.22
11.81
0.28
9.60
1.38
29.26
2.30
13.69
48.29
4.73
16.25
6.87
16.08
28.64
24.71
54.48
67.76
16.39
0.37
3.76
0.52
3.35
9.92
4.97
21.09
16.39
13.19
0.30
10.21
1.64
31.47
2.64
14.36
51.23
4.77
15.86
5.20
10.64
16.66
6.98
25.08
24.12
12.28
0.13
3.37
1.05
1.45
3.47
0.85
10.67
8.67
3.97
0.58
5.62
1.51
10.03
2.29
7.44
23.11
2.25
2.97
3.69
5.27
9.81
3.42
16.72
16.30
6.03
0.11
3.72
0.23
1.62
6.64
2.88
5.34
2.46
7.27
0.41
2.58
0.80
3.93
1.47
0.98
6.32
4.78
13.41
6.41
2.72
2.16
8.48
5.46
6.98
4.05
Experimental values taken from ref 35.
Table 5. Predicting the Vapor Pressure of Pure
Compounds, in Terms of %AAD, from 6 Cubic Equations
of State
%AAD for Vapor Pressure
component
Tr range
PT
CO2
H2S
C1
C2
C3
n-C4
i-C4
n-C5
i-C5
neo-C5
n-C6
cyc-C6
CH3-cyc-C5
C6H6
n-C7
CH3-cyc-C6
n-C8
n-C9
n-C10
n-C12
n-C14
n-C16
n-C18
n-C20
average
0.71-1
0.50-1
0.48-1
0.30-1
0.23-1
0.32-1
0.28-1
0.31-0.96
0.25-0.98
0.59-1
0.35-0.87
0.51-0.81
0.25-0.84
0.50-0.80
0.34-0.83
0.26-0.79
0.38-0.79
0.27-0.77
0.39-0.73
0.40-0.68
0.40-0.65
0.41-0.62
0.41-0.60
0.40-0.59
0.35
3.38
1.21
1.63
5.65
2.96
16.94
13.26
8.94
0.30
8.72
0.99
24.43
1.84
12.45
42.18
3.43
14.55
4.41
12.21
23.70
19.94
49.51
64.55
14.06
PTV
SW
GD
MMM
Table 6. Compositions of Lean Synthetic Natural Gases
Used in This Studya
Composition (mol %)
component
SNG-1
SNG-2
SNG-3
SNG-4
SNG-5
CO2
N2
C1
C2
C3
i-C4
n-C4
i-C5
n-C5
n-C6
n-C7
n-C8
25.908
1.559
69.114
2.62
0.423
0.105
0.104
0.034
0.023
0.110
0.202
0.313
90.483
8.038
0.801
0.081
0.123
0.01
0.0079
0.0047
0.0011
1.7
0.772
84.446
8.683
3.297
0.293
0.589
0.084
0.086
0.05
0.51
6.9
88.1882
2.72
0.85
0.17
0.32
0.085
0.094
0.119
0.0258
0.018
0.284
5.651
83.3482
7.526
2.009
0.305
0.52
0.12
0.144
0.068
0.0138
0.011
TBS
0.23
0.75 1.12
0.15 0.89
2.71
3.88 3.57
3.73 2.86
5.48
0.57 0.78
1.06 1.12
3.03
2.39 2.74
1.53 2.52
5.36
5.76 5.04
6.37 2.7
5.06
2.27 2.06
2.83 1.24
21.54
6.07 6.62
5.57 8.50
23.52
3.82 4.53
2.73 7.13
13.79
4.80 3.97
6.87 1.52
0.76
0.29 0.41
0.41 0.58
11.17
2.90 3.28
3.29 6.09
2.11
0.77 0.69
0.43 0.79
34.02
1.69 3.35
2.90 10.94
2.41
1.53 1.43
1.03 1.40
16.19
2.51 3.09
2.37 9.73
51.13 11.69 13.98
7.54 25.41
5.67
3.68 3.48
3.19 4.12
17.68
7.63 7.66
7.10 3.90
3.32
5.93 4.72
3.45 5.20
7.18 11.29 1.27 13.46 2.95
14.04 26.93 4.90 40.02 3.44
9.17 31.63 7.85 62.74 15.99
13.46 71.08 15.11 142.44 8.82
5.99 109.81 17.97 278.19 16.62
11.46 13.32 4.98 24.97 6.02
compositions of this synthetic natural gas and the others
used in this study are given in Table 6. The BIPs have
been taken from the literature.36,37 As Figure 2 shows,
the use of BIPs significantly improves the prediction
accuracy of the RKS, PR, and PT EoSs for the dew
points of mixture SNG-1. However, in Figure 3, the use
of BIPs slightly impairs the predictions for mixture
SNG-2. It is also shown that the dew points of mixture
SNG-2 have been predicted by the RKA and PTV EoSs.
For these predictions, the BIPs of the RKS EoS have
been used for the RKA EoS and the BIPs of the PT EoS
for the PTV EoS. It may be possible to use the BIPs of
Data taken from refs 12 and 13.
one EoS for the other EoS, having compared the
predictions with several experimental data.7 However,
there are other EoSs to be used in this study and the
BIPs are not available for them. To have a fair comparison, hereafter, we set a zero value for BIPs for all
binaries in EoSs.
Figures 4-6 show the dew points of mixture SNG-3,
as predicted by the RK family, PR family, and the other
EoSs, respectively, and experimental values are also
shown. Clearly, the RK family EoSs are in better
agreement with experimental data than those of the PR
family and the other EoSs. For the RK family EoSs,
regardless of which EoS is used, the predictions are
similar. In the PR family, the PRT EoS is somewhat
superior. It is also shown that the PRD and PRF EoSs
overpredict the upper branch of the dew point curve.
Among the three-parameter EoSs, the SW EoS exhibits
superiority, whereas the MMM exhibits inferiority. The
TBS EoS is comparable to the MMM EoS.
Figures 7-9 illustrate the dew point loci for mixture
SNG-4, as predicted by the EoSs and compared to the
Predicting Natural Gas Dew Points
Energy & Fuels, Vol. 19, No. 2, 2005 567
Figure 2. Experimental (from ref 12) and predicted dew points for mixture SNG-1.
Figure 3. Experimental (from ref 13) and predicted dew points for mixture SNG-2.
Figure 4. Experimental (from ref 12) and predicted dew points from the RK family for mixture SNG-3.
568
Energy & Fuels, Vol. 19, No. 2, 2005
Nasrifar et al.
Figure 5. Experimental (from ref 12) and predicted dew points from the PR family for mixture SNG-3.
Figure 6. Experimental (from ref 12) and predicted dew points from six EoSs for mixture SNG-3.
Figure 7. Experimental (from ref 13) and predicted dew points from the RK family for mixture SNG-4.
experimental data. As can be seen, the cricondentherm
can only be predicted by the RK family EoSs, and the
PRD and SW EoSs; however, except for the PRD EoS,
none of the EoSs are capable of predicting the cricondenbar. The PRD EoS is in good agreement with the
experimental data and considerably superior with re-
Predicting Natural Gas Dew Points
Energy & Fuels, Vol. 19, No. 2, 2005 569
Figure 8. Experimental (from ref 13) and predicted dew points from the PR family for mixture SNG-4.
Figure 9. Experimental (from ref 13) and predicted dew points from six EoSs for mixture SNG-4.
Figure 10. Experimental (from ref 13) and predicted dew points from the best EoS for mixture SNG-5.
spect to the others. However, the RK family EoS, and
the PRT, SW, and GD EoSs are also in agreement with
the experimental data. The predictions from these EoSs
are also compared with the experimental data for
mixture SNG-5 in Figure 10. Except for the PRD EoS,
the agreement with the experimental data is quite good
570
Energy & Fuels, Vol. 19, No. 2, 2005
Nasrifar et al.
Table 7. Specifications of Rich Natural Gases Used in This Studya
Value
specification
A38
B38
composition (mol %)
CO2
N2
C1
C2
C3
i-C4
n-C4
i-C5
n-C5
C6
C7+
molecular weight, MWC7+
specific gravity, SPGRC7+
dew point temperature, Tdew (K)
dew point pressure, Pdew (MPa)
0.82
0.87
64.04
10.57
5.75
1.35
2.37
1.06
0.98
1.01
11.18
186
0.807
394.82
36.053
3.47
0.39
80.19
6.28
2.75
0.43
0.88
0.31
0.35
0.53
4.42
171
0.813
422.04
54.268
C39
82.38
4.28
3.51
1.61
3.03
0.60
0.68
0.99
2.92
125
0.74
277.59
21.339
D38
E38
0.45
0.38
83.00
3.76
1.44
0.89
0.97
0.41
86.16
3.55
1.54
0.46
0.46
0.26
0.20
0.35
5.64
253
0.850
405.93
81.565
4.36
3.08
2.63
106
0.733
288.71
18.278
F40
0.10
95.22
1.68
0.91
0.26
0.33
0.16
0.11
0.25
0.98
122.6
0.723
377.59
23.063
G38
H41
0.18
0.15
86.57
3.83
1.97
0.49
0.72
0.34
0.40
0.57
4.78
200
0.820
379.82
60.329
8.64
0.71
70.85
8.53
4.95
0.75
1.26
0.41
0.40
0.46
3.04
155.3
0.8311
392.04
39.851
I42
91.35
4.03
1.53
0.82
0.34
0.39
1.54
138.78
0.7961
367
26.46
J43
K44
L45
0.61
0.46
68.64
13.90
6.89
0.66
2.66
0.62
0.94
1.14
3.48
152.3
0.7763
363
23.53
2.44
0.08
82.10
5.78
2.87
0.56
1.23
0.52
0.60
0.72
3.10
132
0.774
355.6
28.1
2.17
0.34
70.64
10.76
4.94
3.02
1.35
0.90
5.88
153
0.81
408.7
33.47
Superscripted values next to each of the 12 SNG mixtures denote the reference source.
Figure 11. Deviation plot for natural gas mixture A, showing the percent deviation in predicting dew point, as a function of the
number of single carbon numbers (SCNs).
for the other EoSs. However, the RKSS EoS, as a
representative of the RK family, and the SW EoS
describe the dew point loci significantly better than the
others.
The synthetic natural gases used in this study are
considerably lean in heavy hydrocarbons. A real natural
gas is usually rich in heavy hydrocarbons. All heavy
hydrocarbons are usually grouped together and reported
as C7+. It is then important to split the C7+ fraction into
several SCN groups and calculate the critical properties
of each group if an EoS is to be used to calculate the
natural gas dew point.
Table 7 gives the specifications of real natural gases
(38) Elsharkawy, A. M. Fluid Phase Equilib. 2002, 193, 147-165.
(39) Sage, B. H.; Olds, R. H. T. Soc. Petrol. En. AIME 1947, 170,
156-173.
(40) Reamer, H. H.; Sage B. H. T. Soc. Petrol. En. AIME 1950, 189,
261-268.
(41) Pedersen, K. S.; Thomassen, P.; Fredenslund, A. In Proceedings
of the AIChE Spring National Meeting, New Orleans, LA, March 6-10,
1988.
(42) Hoffmann, A. E.; Crump, J. S.; Hocott, C. R. T. Soc. Petrol. En.
AIME 1953, 198, 1-10.
(43) Donohue, C. W.; Buchanan, R. D. J. Petrol. Technol. 1981, 35,
263.
(44) Firoozabadi, A.; Hekim, Y.; Katz, D. L. Can. J. Chem. Eng.
1978, 56, 610-615.
used in this study. Mixture A, which contains 11.18%
C7+, is used to find the optimum number of SCN groups
to calculate the natural gas dew point using different
EoSs. This evaluation is given in Figure 11 in terms of
%Bias:
%Bias )
Pd,cald - Pd,expl
100
Pd,expl
(8)
Figure 11 shows that, if the C7+ fraction in mixture A
is divided into more than 10 SCN groups, all the EoSs
become stable and predict the dew point of mixture A
uniformly. However, to a large extent, the prediction
ability of EoSs is dependent on the critical properties
of the SCN groups. This is best observed in Figure 12,
where the %Bias of the GD EoS, as a representative of
the EoSs, is plotted against the number of SCNs.
Different correlations give different predictions; however, the correlations of Lee and Kesler29 (LK) and
Twu28 seems to be better than those of Cavett26 and
Riazi and Daubert27(RD). Figure 12 also shows that the
predictions by the GD EoS become improved and
(45) Coats, K. H.; Smart, G. T. SPE Reservoir Eng. 1986, 227-299.
Predicting Natural Gas Dew Points
Energy & Fuels, Vol. 19, No. 2, 2005 571
Figure 12. Deviation plot for natural gas mixture B, showing the percent deviation in predicting dew point using the GD EoS,
as a function of the number of SCNs.
Table 8. Prediction of the Natural Gas Dew Point Pressures, in Terms of %AAD, from the Equations of State
%AAD of the Predicted Dew Point Pressure
SCN group
RKS
RKT
RKSS
PR76
PR78
PRT
PRF
PRG
PT
PTV
A
B
C
D
E
F
G
H
I
J
K
L
average
19.21
8.24
2.24
2.50
31.90
17.23
21.49
26.18
9.17
10.04
2.28
9.90
13.64
15.50
8.90
5.36
7.17
27.92
13.74
18.84
23.39
15.01
15.34
4.08
6.69
13.49
8.39
27.19
5.49
9.45
18.26
8.28
13.20
18.01
22.36
21.37
9.74
3.96
13.81
21.47
2.39
3.92
0.90
38.17
27.67
28.43
31.90
4.54
4.52
7.63
16.02
15.32
19.53
4.80
4.66
0.68
35.68
25.88
26.01
30.36
5.26
5.99
6.16
14.43
14.95
11.01
18.93
2.56
5.25
26.34
18.78
16.63
24.12
11.08
14.86
2.15
7.41
13.26
6.35
35.77
10.51
14.77
21.46
15.37
10.53
20.61
17
20.41
7.71
4.14
15.38
17.61
21.24
2.98
1.35
33.63
24.12
23.94
28.92
5.37
7.99
4.39
12.61
15.34
6.89
8.57
3.24
2.21
19.99
23.83
14.38
28.02
6.25
9.10
3.48
11.07
11.42
8.91
12.21
2.89
4.23
9.67
20.19
11.64
23.62
11.13
14.86
1.96
6.51
10.65
unchanged for 12 or more SCNs, regardless of which
correlation is used to characterize the C7+ fraction.
Considering Figures 11 and 12, we have concluded that
12 SCN groups are suitable for dew point calculations.
Previously, Elsharkawy38 reported a similar conclusion,
having examined the dew point prediction of the RKS
EoS for a natural gas mixture.
In Table 8, the %AADs in predicting the dew points
of natural gases given in Table 7 are presented. In these
calculations, the C7+ fractions were divided into 12 SCN
groups. The values given in Table 8 are, in fact, the
average values. Using each EoS, the dew point of each
mixture has been predicted four times. Each time, one
of the correlations of LK, Cavett, Twu, and RD has been
used to calculate the critical properties and the Edmister46 equation was used to calculate the acentric factor.
The calculated values were then averaged and have
been reported in Table 8. As can be seen, the best five
EoSs are, respectively, PTV, PT, GD, PRT, and RKT.
Among the two families, clearly the RK family is
superior and should be preferred. It is also valuable to
find the best correlations of critical properties for
predicting dew points. Table 9 clearly shows that the
Twu28 correlation is in better agreement with the EoSs
and should be used to calculate the SCN group critical
properties.
(46) Edmister, W. C. Pet. Refin. 1958, 37, 173-179.
SW
12.88
3.55
7.74
11.65
37.83
10.53
30.05
28.33
22.16
6.17
17.09
GD
TBS
14.41
24.52
2.61
1.21
23.30
17.85
14.32
22.46
12.13
10.63
1.86
7.42
12.73
16.14
39
9.81
12.95
12.47
46.77
21.53
34.07
9.67
10.95
12.42
21.53
20.61
Table 9. Prediction of the Natural Gas Dew Point
Pressures, in Terms of %AAD, Using Different
Characterization Methods
%AAD of Dew Point Measurement
equation of state, EoS
LK
Cavett
RD
Twu
RKS
RKT
RKSS
PR76
PR78
PRT
PRF
PRG
PT
PTV
SW
GD
TBS
average
12.99
13.27
11.56
13.66
13.27
12.14
13.06
18.03
10.37
10.97
19.82
10.45
21.52
13.93
14.58
14.53
14.18
17.29
16.35
13.68
17.97
15.21
12.57
11.35
16.78
14.07
21.16
15.36
15.36
16.01
14.07
18.92
17.92
14.26
14.67
16.83
13.46
10.91
16.58
15.22
25.81
16.16
11.62
11.05
16.13
14.90
13.38
13.93
14.08
12.46
9.28
9.51
15.17
11.17
18.21
13.15
6. Conclusions
The dew points of lean and rich natural gas mixtures
have been predicted using 15 equations of state (EoSs).
The supercritical behavior of methane and nitrogen, and
the vapor pressure of the components forming natural
gases, have been studied using these EoSs. It can be
inferred that, if an EoS is capable of predicting the
fugacity of methane under supercritical conditions and
572
Energy & Fuels, Vol. 19, No. 2, 2005
the vapor pressure of natural gas components accurately, especially the heavy ones, one can expect the
EoS to be accurate in regard to predicting the natural
gas dew points. All the EoSs in the RK family and the
GD EoS are good examples.
The impact of binary interaction parameters (BIPs)
on the prediction of EoSs has been studied. It has been
found that, for synthetic natural gases, the use of BIPs
does not necessarily improve the dew point prediction
accuracy of the EoSs. This conclusion suggests that the
critical properties and acentric factor used in this work
to calculate the EoS parameters may not be the same
as those used for optimizing BIPs.
The lean synthetic natural gas dew points have best
been predicted using the EoSs from the RK family, i.e.,
the RKS, RKSS, and RKT EoSs and the SW EoS. The
Nasrifar et al.
PRT and GD EoSs also have been determined to be
promising.
It has been found that, using an exponential distribution devised by Pedersen et al.25, a C7+ fraction must
be divided into at least 12 SCN groups so that the
predictions can become improved and unchanged by any
of the EoSs. The use of the Twu28 correlation to calculate
the critical properties of SCN groups is also recommended.
Generally, it has been found that three-parameter
EoSs can describe natural gases more accurately than
the other EoSs. Accordingly, the PTV, PT, and GD EoSs
are recommended for predicting rich natural gas dew
points. For lean natural gases, however, the RKS,
RKSS, RKT, and SW EoSs are superior.
EF0498465
