Comparative Study of Eight Equations of State For Predicting Hydrocarbon Volumetric Phase Behavior
Comparative Study of Eight Equations of State For Predicting Hydrocarbon Volumetric Phase Behavior
Comparative Study of Eight Equations of State For Predicting Hydrocarbon Volumetric Phase Behavior
Summary. The objective of this study is to present a review of eight equations of state (EOS's) and compare their ability to
predict the volumetric and phase equilibria of gas-condensate systems. Included in the study are the Peng-Robinson (PR), the
Soave-Redlich-Kwong (SRK), the Schmidt-Wenzel (SW), the Usdin-McAuliffe (UM), the Heyen, the Kubic, the Adachi-Lu (AL),
and the Patel-Teja (PT) EOS's. The SW equation exhibits a superior predictive capability for volumetric properties of cOl1densate
systems. The PR equation is found to represent the phase equilibrium behavior of condensate systems accurately. In terms of
compressibility factors, the SW and PT equations give better predictions than other equations.
Introduction
Since the proposal of the van der Waals equation, many EOS's have and
been proposed for the representation of fluid volumetric, ther-
modynamics, and phase equilibrium behavior. These equations, b=nbRTc/pc' ..................................... (6)
many of them a modification of the van der Waals EOS, range in
complexity from simple expressions containing two. or three con- where na and nb are the SRK dimensionless pure-component pa-
stants to complicated forms containing more than 30. Although the rameters. For the unmodified SRK equation, na =0.42747 and
complexity of any EOS presents no computational problem, most nb =0.08664 are independent of temperature, pressure, composi-
apthors prefer to retain the simplicity found in the van der Waals tion, or particular component.
cubic equation while improving its accuracy through modifications. Eq. I can be expressed in terms of the compressibility factor as
Semiempirical EOS's set forth in recent years have retained the
van der Waals thermal repulsive term. Differences exist in the ex- Z3 -Z2 +(A-B-B2)Z-AB=0, ...................... (7)
pression of the attractive pressure term. The most significant mile-
stone in the development of cubic EOS's is the publication by where Z is the compressibility facto~ and A and B are defined by
Soave I of a modification in the evaluation of the parameter a in
the attractive term of the Redlich-Kwong equation. A =aap/(RT)2 ............................ .' ....... (8)
This suggestion by Soave has prompted an enormous increase
in activity on the part of scientists and engineers who are interested and
in the use of EOS's to calculate fluid properties.
The objective of this study is to present a review of developments B=bp/(RT) . ...................................... (9)
in cubic EOS's and to compare their predictive capabilities for volu-
metric and vapor/liquid equilibrium (VLE) predictions. In the VLE calculations for mixtures, Soave suggested the fol-
lowing mixing rules:
Review of Recent Developments in Cubic EOS's
SRK EOS and Modifications. The SRK I EOS has the form
(aa)m = 2: 2:ziz/aiapia)O.5(1-Kij) ............... (10)
p=RT/(V -b) -aa/[V(V +b)], ........................ (I) i j
Graboski 2 suggests that no interaction parameters are required for This expression was later expanded by the investigators to give the
hydrocarbon systems. With nonhydrocarbons present, however, relationship
binary interaction can greatly improve the equilibria predictions.
Vidal 3 and Elliot and Daubert 4 presented a theoretic background m=O.3796+ 1.485",-0.1644",2 +0.01667",3.
to the meaning of the interaction coefficients and the quadratic mix-
ing rules of the Soave equation. Introducing the z factor into Eq. 15 yields
Correlations of optimal Kij proposed by Elliot and Daubert are
for N2 systems, Kij =0.107089+2.9776kij; for CO 2 systems, Kij
=0.08058-0.77215kij -1.8407(kij)2; and for methane systems
with compounds of 10 carbons or more, Kij =0. 17985+2.6958kij The fugacity coefficient of Component j in a mixture is calculated
+ 10. 853(kij )2, where from
with
where V is the molal volume of the hydrocarbon phase (gas or liq- and kl being an adjustable parameter characteristic of each pure
uid) as calculated by Eq. 1. The authors proposed six different compound.
schemes for calculating c;. For petroleum fluids and heavy Ihaveri and Youngren II found in their application of the PR
hydrocarbons, Peneloux et at. suggested that the best correlating equation to reservoir fluids that the error in the prediction of the
parameter for c; is the Rackett 6 compressibility factor ZRA: gas-phase compressibility factors ranged from 3 to 5 % and the error
in the liquid density predictions from 6 to 12 %. Following the proce-
dure proposed by Peneloux et at. ,5 Ihaveri and Youngrcn intro-
duced the volume-correction parameter, c;, to the PR EOS.
Gibbons and Laughton 7 introduced two adjustable substance- Because the proposed third parameter has the same units as the scc-
dependent parameters, X and Y, for the expression a of the SRK ond paramcter of the unmodified PR equation, b i , they defined a
equation: dimensionless shift parameter, s;, as S i = c;l b i' They then presented
values of the shift paramcters for the well-defined lighter hydrocar-
a= I +X(Tr -l)+ Y(Tr 0.5 -1). bons. They also outlined the characterization procedure of the shift
. parameter for thc heptanc-plus fraction. Ihaveri and Youngren con-
Values of X and Yare chosen by minimizing the root-mean-square cluded that a significant improvement in the volumctric predictions
error in the calculated vapor pressures for the whole rangc of thc by thc modified PR equation can be obtained.
liquid/vapor-pressure curve. However, the authors were unable to
generalize a correlation for X and Y. UM EOS. Usdin and McAuliffe 12 introduced a new coefficient,
d, to replace b in the second term of the SRK equation:
PR EOS and Modifications. Peng and Robinson 8 proposed a p =RT/(V -b) -aa/[V(V+d)] . ....................... (18)
modified Redlich-Kwong EOS capable of predicting the liquid den-
sity as well as vapor pressure to improve VLE predictions further. They contended that the two terms of the SRK equation are inter-
Thcir equation is given by connected by cocfficient b, forcing the conclusion that all materials
possess a value of critical compressibility of 0.333. This conclu-
p=RT/(V-b)-aa/[V(V+b)+b(V-b)], ............... (15) sion cffccts a substantial depression in the values of liquid density.
They stated that severing the tie created by the shaping of the coeffi-
with cient band rcplacing it with d would produce accurate liquid den-
sities.
a=0.457235(RTY/pc Defining
and
b = O. 077796RTc/Pc'
11j +(6Z~-1)113+[12(Z1)2 -3Z1]11" + (8Z1-3)(Z1)2 =0, 11 b =0.08664+[(lOO-15l9w)1O- 50ITr -l)], ......... (20)
m = 0 .48049 + 4. 516wZ[ where Vm is the molar volume. Stein also presented the values of
the constants A 1 through A7 in a tabular form. In terms of com-
+ [0.67713(w -0.35) -0.02](Tr -0.7), for Tr 5, 0.7 pressibility factor and fugacity coefficient, Eq. 19 corresponds to
Eqs. 7 and 8, respectively, with if;i and if; defined as
and
D= L,11diZiPr/Tri,
i
where the reduced temperature Tri and reduced pressure Pri are
defined by the expressions Tri=T1Tci and Pri=plPci'
In their verification of the proposed equation and in comparison and
with the SRK EPS, Usdin and McAuliffe suggested that a better
pure-component density prediction could be obtained with accura-
cy similar to that of the SRK equation for VLE computations.
Heyen identified the parameters k, n, m, and 0 for any compo-
Stein-Redlich-Kwong EOS. Stein 13 stated that the deficiency in nent by forcing the EOS to match experimental vapor pressure and
the Redlich-Kwong two-constant equation is that the number ofvari- saturated liquid volume. This led to the following relationships for
abIes in Eq. 19 is less than the number of degrees of freedom; hence these parameters, with w as the correlating factor:
the calculation system is underdefined. The Redlich-Kwong equation
is defined by k=0.49164+0.43882w-0.0882Iw 2 ,
and
with
where nb is the smallest positive real root of the equation and where
where
am = I; I;ZiZj(aia)o.S(l-Ki), ..................... (24)
I J
')'0 =4.275051- 8.8788891Tr + 8.5089321Tr 2
A=ampl(Rn, .................................. (27a) For mixtures, the classic mixing rules, Eqs. 24 through 24, are
suggested. The fugacity coefficient is calculated from
B=bmPI(RT), .................................. (27b)
In </>j= -In{Z-B+[pl(RT)](b/Z-B)} - [pl(RT)]2[2>/;/(Z+C)]
and
+ [pl(RT)]Ac/(Z+ C)2.
C=cmpl(RT). . ................................. (27c) Kubic pointed out that the ratio blVe varies as a function of the
acentric factor in the modified Martin equation, which improves
The expression for the fugacity coefficient is given by the accuracy of the equation over a wider range of fluids and im-
proves its ability to handle asymmetric mixtures. The investigator
In cPi= -In(Z-B)+ {bi/[(Z-B)->/;/RTd]ln(Q+d)/Q-d} concluded that the predictions given by this modification are as good
as or better than those given by the PR EOS for VLE.
+0.05A[(b j +cj)/(Q" -d 2 )] +0.125A[cj(3B+C)
PT EOS. Patel and Teja]9 proposed the following form of EOS: -(1PJd)ln[(Q+d)/(Q-d)] +am(b i +ci)/[2(Q2 -d 2 )]
and
where flc=I-3rc and fla=3r~+3(1-2rJflb+flb+(1-3U,
with rc=pcVcl(RTe)' and where fib is the smallest positive root
of the cubic
Patel and Teja concluded that the proposed EOS was capable of
yielding accurate VLE predictions while reproducing with sufficient
accuracy liquid- and vapor-phase densities.
An approximation value of fib is given by
SW EOS. Schmidt and Wenzel 20 proposed the following form of
fib =O.32429re -0.022005. EOS:
re =0.329032 -0.076799w+0.0211947w 2 .
Eq. 30 can be used for the calculation of mixture properties if An approximation value of (3 c is given by
the constants a, b, and c are replaced by the mixture constants am'
b m , and c m as follows: (3e =0.25989-0.0217w+O.00375w 2 .
If we define m2 =mo+0.71(Tr-0.779)2,
and
and mo=0.5361 +0.9593w, for w>0.3671.
C=cmPI(RT),
For supercritical compounds, the temperature dependence of the
Eq. 30 can be rearranged in terms of Z to produce parameter a is also dependent on w:
Z3 +(C-l)Z2 +(A - 2BC - B- C - B2)Z+(BC + B2 C - AB) =0. a=I-(OA774+1.328w)ln T" for Tr>l.
For mixtures, Sthmidt and Wenzel used the classic mixing rules
(Eqs. 10 and 11) for evaluating am and b m . The authors adopted where
the following mixing rules for the acentric factor: F(Zi) =RT In[RT(Z- B)lp] - [pbJ(Z- B)]
+ (2-.f;JD-am{3JD2)ln(GIH)
+ [(Ei +{3i)/G-(Ei -(3;)IH][a m /(2D)],
•
D EXP
o P-R
+ P-T
)C HEYEN
¢ KUBIC
V s-w
.. • SRI<.
... 1...
l~--------~----------'----------T----------~---------'
• ,...
aCPIBBL OJt DBW .oINT GAS
'I'" " ..
Fig. 1-Swollen volumes vs. cumulative gas injected during sw.elling with lean gas at 20QoF.
S =(iJ2 +4iJ-4)o.s, ~
f ....
G=ZRTlp+(iJ+s)b m I2, JIi
.: n ..
;:
f'"
D=Sb m •
H"
and
'"
Z
f
.... a EXP
o
; '"'
P-l
•t SIU<
ELF
Ie P-T
Schmidt and Wenzel concluded that the proposed EOS can rep- o HEYEN
• CHEVRON
lH.
resent the VLE of mixtures with accuracy similar to that of other • s-w
* PETEX
established EOS's and with superior accuracy for volumetric pre-
dictions.
...
Equations Studied and Data Sources •• HI lH' ,... UN ...
SCP/IIBL 0., DlIW PODfT GA.
The eight equations evaluated in this study are the PR, SRK, PT, .
SW, Heyen, Kubic, AL, and the UM EOS's. The selection ofthese Fig. 2-Dewpoint pressure vs. cumulative gas injected during
equations was influenced by my familiarity with certain equations swelling with lean gas at 200°F.
and interest in evaluating their predictive capability in comparison
with the established EOS's.
Ten condensate-hydrocarbon systems with experimental PVT 4. The predictive capabilities of selected EOS's were tested
measurements were used as a data base. The following procedure against measured PVT data for each condensate system. Predic-
was adopted throughout the study. tion accuracy for PVT properties was calculated for average absolute
1. The Riazi and Daubert 21 correlation was used to calculate deviations (AAD). Graphic presentation of results has been em-
physical properties of the plus-fraction. phasized throughout the study, and tabulations are given where ap-
2. Katz-and-Firoozabadi 22 -tabulated physical properties were propriate .
. assigned for all components between n-C s and the plus-fraction.
3. The interaction coefficient between methane and the heaviest Results and Discussion
component was then adjusted until a match of the measured dew- The following discussion is by no means a comprehensive study
point pressure was obtained. of the broad field of EOS' s as applied to complex multicomponent
• D EXP
0 P-R
A. P-T
+ s-w
X A-L
-.. ,
~
•
~•
" t
..
A P-T 6 P-T
+)( + s-w
.
HEYEN
s-w ~ P-R and A-K
P-T and A-I
S-W and A-K
M .. to
I·
M
i
~ .
= II
= II
11
11
.,+---------,---------,---------~--~--~
• t ...
" .
.....SlID •••••
. .. ...
..
• 1111 ,...
.+---------,---------,---------.---~--~
.... ..
....SSlJlt••••••
Fig. 4-Graphic representation of constant-volume depletion
at 200°F. Fig. 5-Saturation vs. pressure, constant-volume depletion.
TABLE 3-EXPERIMENTAL AND PREDICTED K VALUES AT 700 psig AND 200°F, CASE 1
Z
3.2 1.1 4.5 1.44 1.1 7.7 2.1
eight equations. All equations performed equally well for the pres- is summarized in Table 4. Overall, the PR, PT, and SW equations
sure/volume test and predicted the measured data accurately. give the best predicted VLE. PT and SW equations show excellent
Fig. 4 compares experimental and predicted liquid dropout by compressibility factor prediction with an average deviation of 1.1 %.
various EOS's. With the exception of the SW equation, all equa-
tions showed a large increase in predicted liquid volumes with pres-
sure just below the satUration point. For this spccific case, the SW Case 2-Rich-Gas Condensate: Moses-Wilson Hydrocarbon Sys-
equation exhibited an excellent agreement with the experimental tem. The hydrocarbon system reported in Case 1 was used by Moses
data. An attempt was made to recalculate the liquid-phase volumes and Wilson 25 to study experimentally and to evaluate the effec-
with the Alani-Kennedy24 density equation; the results are shown tiveness of N2 in maintaining reservoir pressure and displacing
in Fig. 5. The Alani-Kennedy equation is seen to yield better pre- condensate from the retrograde gas-condensate reservoir. A swelling
diction results when applied in conjunction with the EOS. test with N2 was conducted on a gas-condensate sample. The dew-
Experimental and predicted equilibrium ratios at 200°F and 714.7 point pressure was observed and the initial portion of the retrograde
psia [93°C and 4.93 MPa] are compared in Table 3 and presented curve was measured for each addition of N 2' The predicted and
graphically in Fig. 6. The K values predicted by various equations measured dewpoint pressures are shown in Fig. 7. The PT and SW
are all quite close to each other and to the experimental data. A EOS's give excellent predictions in comparison with experimental
moderate discrepancy can be observed between the various pre- data.
dicted K values of N2 and the plus fraction. The SW equation was selected to reproduce the measured retro-
For the hydrocarbon system considered in this study and for a . grade curves. As shown in Fig. 8 in this study and Fig. 3 in the
pressure up to 3,500 psia [24 MPa], the comparison in terms of Moses-Wilson 25 report, an excellent agreement with the ex-
the absolute average deviations in VLE and compressibility factors perimental retrograde curves was obtained.
.... ••
.. c
o
£xP. tOO
MIX #1 151 SCF /B8L
.... .,
~
+j/..
MIX 112
MIX #3
Mix 114
395 SCF/BBL
638 SCF /BBL
940 SCF /BBL
e .
"Ii
I 5.
0
I I . .' e
o
EXP
P-R
..:
E
",
§ n .. A
+
x.
P-T
KUBIC
s-w
II
¢ A-L
V SRI(
tI
II. .
, m _ _ _ rn _ _ _ _ ~
,+---,---,----r---r---r--~--_.--_.--_.r__,
PIl••SU••• ,.1,
" .. •••• . ..
Fig. 8-Retrograde liquid condensation for mixtures of con-
densate reservoir fluid and N 2' constant-volume depletion.
.
.... +
. 11
n.'
II
!
M
tr.'
!
5Ie If
::> o EXP
~ 1t.' o P-R o EXP
A WHITSON o P-R
'" II
+
'1-
¢
P-T
KUBIC
s-w
a
+
P-T
KUBIC C7+
¥
... VSRl(
<J
s-w
SRI<
D
+
'.11 •
...
• +-----r----,----~----_r----,_~--~--~
• ".. '1M' nit
,... ..nl
. . .1111".••••••
_HI .... ....I+-------,...----.....----___---r-----.-------.
Fig. 9-Liquid dropout for North Sea gas condensate, Fig. 10-K values of the North Sea fluid at 280°F and 714.7
constant-volume depletion. psia.
Case 3-North Sea Condensate System: Whitson's Hydrocarbon Case 4-Pedersen-Thomassen-Fredenslund Condensate System.
Condensate Gas. This North Sea gas-condensate system was chosen Pedersen et al. 27 presented methods of characterizing heavy frac-
by Whitson 26 to illustrate the application of the two-parameter PR tions in hydrocarbon systems. They also reported the results of pres-
EOS to the hydrocarbon-condensate system. Whitson attempted sure/temperature flash calculations and three PVT experiments
more than 20 different adjustments of the C7+ characterization commonly performed on oil and gas mixtures. The experimental
procedure to improve liquid volume predictions. The author stated results are simulated by the authors with the SRK EOS. Because
that none of these adjustments were helpful in improving liquid the SRK equation produces severe errors for the liquid-phase
dropout predictions. volumes, they used various correlations for the liquid-phase molar
Fig. 9 shows liqUid dropout curves as predicted by the various volumes. Their correlations include: Peneloux et al. .5 Lee and
equations. Whitson's predictions are included for the sake of com- Kesler,28 Standing and Katz,29 and the Alani-Kennedy24 equation.
parison. In Fig. 9, the SW EOS shows a good volumetric match Results of Pedersen et al. 's simulation model for the constant-
with the experimental retrograde curve. The graphic presentation composition expansion test and predictions of selected equations
of the predicted and measured Kvalues is presented in Fig. 10 and (PR, PT, and SW) are presented in Fig. 12. The SW equation shows
tabulated in Table 5. For this hydrocarbon system, the SW equa- an excellent predictive capability for this system.
tion gives consistently higher predicted equilibrium ratios than other Table 7 summarizes the overall AAD in predicted vapor compo-
equations. Table 6 gives a summary of the average etrors in the sitions and compressibility factors for 10 condensate systems with
predicted vapor compositions and compressibility factors. In terms a total of 62 data points for vapor compositions and 69 data points
of AAD, the PT and SW equations give accurate predictions for for compressibility factors. A complete description of the program
the compressibility factors over the range of pressure considered and results of the study can be obtained from the author.
in the study (up to 7,000 psi r48 MPa]). All equations predicted
vapor composition equally well except for the C 7 + component.
Considerable deviation for the predicted molal fraction of C7+
TABLE 6-COMPARISON OF EXPERIMENTAL AND
could be attributed to its incorrect characterization in terms of critical
PREDICTED VAPOR MOLAL COMPOSITIONS AND
properties and acentric factor. COMPRESSIBILITY FACTORS FOR THE
Fig. 11 compares experimental and predicted North Sea gas NORTH SEA FLUID, CASE 3
FVF's as a function of pressure. The graphic presentation shows Table Values Represent Percent AAD*
an excellent match with experimental data for all equations. in Compositions and Z
Z
4.4 2.0 1.9 2.3 3.1
TABLE 5-EXPERIMENTAL AND PREDICTED K VALUES
AT 700 psig AND 280°F, CASE 3 Component PR SW PT SRK Kubic
--
Component Experimental PR PT Kubic SW SRK CO 2 0.50 0.5 0.40 1.2 1.2
-- -- -- -- -- N2 2.0 1.2 1.5 2.7 1.7
CO 2 4.86 4.53 4.75 3.38 5.03 5.74 C1 0.9 0.8 0.7 1.0 1.4
N2 19.41 18.23 19.97 12.82 23.35 21.57 C2 1.3 1.6 1.20 1.2 0.4
C1 7.27 9.12 10.18 6.50 11.54 9.08 C3 1.8 2.7 1.9 1.3 1.7
C2 2.70 3.32 3.61 2.56 4.03 3.30 i-C 4 2.3 2.6 2.5 2 2.7
C3 1.35 1.75 1.84 1.38 2.06 1.66 n-C 4 1.7 3.1 2.0 1.2 3.4
i-C 4 0.85 1.06 1.09 0.84 1.24 1.06 i-C s 2.1 4.0 2.4 2.1 4.9
n-C 4 0.74 0.89 0.90 0.72 1.0 0.84 n-C s 2.30 2.8 2.5 2.7 5.7
i-C s 0.45 0.52 0.52 0.44 0.58 0.52 Cs 3.9 3.9 3.9 5.5 8.1
n-C s 0.41 0.46 0.46 0.39 0.50 0.45 38.3 37.8 32.2 40.7 30.6
C7 +
Cs 0.34 0.25 0.25 0.23 0.27 0.24
C7+ 0.029 0.01 0.01 0.02 0.01 0.01 * Number of data points = 6.
1.12f1 C EXP
o ALL EQCATIO:.lS
1.1:11
I.ml
t.1111
t.""
I .•tst
o"EX>
I.n"
A
'-1.
sn
1.'l1t + r-t
II 'III'ZLOIJI.
<) su. aDd. L-1.
I.Utl ... s-w
• su. a" A-l
:*SUaIl4S-K
1.'Mt
I.'HI
Z
2.4 1.5 1.6 2.3 2.4 4.8 10.7 5.3