Three-Parameter Modification of the

Peng-Robinson Equation of State To

Improve Volumetric Predictions
Bharat s. dhaverl, SPE, Areo Oil & Gas Co.
Gary K. Youngren, SPE, Areo Oil & Gas Co.

Summary. The Peneloux-Rauzy-Freze (PRF) method of improving volumetric predictions by introducing a third parameter into a
two-parameter equation of state (EOS) is applied to the Peng-Robinson EOS (PR-EOS). The modified PR-EOS is evaluated for
application to hydrocarbon fluids. A method is developed for characterizing the third parameter for the heptanes-plus fractions. The
usefulness of the modified PR-EOS in improving volumetric predictions is illustrated by applying the equation to several crude-oil
and gas-condensate systems from the literature.
Introduction
In recent years, two-parameter cubic EOS's-e.g., the PR-EOS and
the Soave-Redlich-Kwong EOS (SRK-EOS)-have been commonly
used by the petroleum industry for predicting the phase behavior
and volumetric properties of hydrocarbon fluid mixtures. Once the
heptane-plus fraction of the hydrocarbon fluid is properly characterized into a mixture of pseudocomponents, these equations predict
the vapor/liquid equilibrium conditions with a reasonable accuracy.
However, the volumetric estimates obtained through these twoparameter EOS's are not as accurate. In our experience with the
application of the PR-EOS to reservoir fluids, we found that the
errOr in the prediction of gas-phase Z factors ranged from 3 to 5 %
and the error in the liquid density predictions ranged from 6 to 12 %.
Recently, Peneloux et al. 1 developed a method of improving the
volumetric predictions by introducing a third parameter into a two
parameter cubic EOS. This method is particularly attractive because
the third parameter does not change the vapor/liquid equilibrium
conditions determined by the unmodified, two-parameter equation,
but modifies the phase volumes by effecting certain translations
along the volume axis. Thus, if a given reservoir fluid is already
characterized for use in some two-parameter EOS, the application
of the PRF method to this fluid requires characterization of only
the third parameter.
In this work, we apply the PRF method to the PR-EOS. Some
background material on the PRF method is given in the next section.
For the modified three-parameter PR-EOS, the section ThirdParameter Characterizations for the PR-EOS presents the thirdparameter values for some lighter hydrocarbons and develops a correlation for characterizing the third parameter for the heptane-plus
fractions of reservoir fluids. To apply the modified, three-parameter
PR-EOS to calculate the phase and volumetric behavior of reservoir
fluids, the section on Applications develops a novel, two-step
procedure for characterizing all three parameters for the heptane
plus fractions. The first step characterizes the two parameters of
the unmodified PR-EOS with the phase-behavior data derived from
the analysis of conventional laboratory experiments. The second
step adjusts the correlation coefficients of the third-parameter correlation mentioned earlier, using the heptane-plus density data at
standard conditions. Also presented in this section is the application
of the modified PR-EOS to several crude-oil and gas-condensate
systems..
PRF Method
Consider one mole of a mixture of n components of composition
Zi, at temperature T and pressure p, obeying an EOS of the form
E(p,T, V,Zi)=O,
Copyright 1988 Society of Petroleum Engineers

SPE Reservoir Engineering. August 1988

(1)

where V is the molar volume. Assuming that at the thermodynamic

equilibrium, the mixture may split at most into two distinct phases,
we can determine the phase properties by solving the following wellposed system of (2n+3) equations in (2n+3) unknowns.
The equations are

fLXi+(I-fL)Yi=Zi,

"

(2)

xi=l,

(3)

i=1

xicPf(p,T,Xi)=YicPf(p,T,Yi), ........................ (4)

E(p,T,Vf,Xi)=O,

(5)

and

E(p,T, Vg,yi)=O,

(6)

where the unknowns arefL, Xi' Yi' Ve, and Vg:

HerefL denotes the mole fraction of liquid phase. (Xi'V e) and
(Yi' Vg ) denote the composition and the molar volume of the liquid
and gas phases, respectively. Eqs. 2 and 4 apply to each Component
i and thus pose 2n equations in total. Subscripts and superscripts
and g denote the liquid and gas phases, respectively. Eq. 4 is the
thermodynamic-equilibrium condition for each Component i and
equates the fugacity of each component in equilibrium phases. cPi
is the fugacity coefficient for Componenti and can be evaluated
from the functional form of the Eq. I EOS. Eq. 2 is the materialbalance equation for each component, while Eq. 3 represents the
overall material balance. Eqs. 5 and 6 related the molar volume
of each equilibrium phase to its composition, temperature, and
pressure through the Eq. I EOS.
Two-parameter cubic EOS's mentioned earlier form Eq. 1 as a
third-degree polynomial in molar volume Vand have two mixture
parameters, a and b, related to the component parameters, ai and
bi' through the following mixing rules:

a=

E E(1--0ij)a/"'a/hziZj
i

(7)

and

b= Ebiz i

(8)
1033

Here, cPjj is .the cqmponent-pair interaction coefficient. aj and bj

are related to the component's critical temperature, Te , critical
pressure, Pe' and acentric factor, w, and Zj is the mole fraction of
Component i. The two-parameter EOS, Eq. 1, can be written as
E2(T,p, V,zj,a,b)""O . ......................... (9)

The PRF method introduces' a third mixture parameter, c, into the

two-parameter EOS, Eq. 9, by effecting the following translation
along the volume axis:

V =V-c.

each well-defined component in the fluid, as well as for each pseudocomponent characterizing the heptane-plus fraction. For welldefmed components, the PRF method calculates the third parameter,
c, by calculating the saturated liquid molar volume, V, at the component's reduced temperature of 0.7 with the two-parameterEOS
and effecting the translation to match that volume with the experimental volume, VE (For experimental volumes, we take the
Spencer and Danners correlation2 for saturated liquid volumes.)
Thus,

VE=V =V-c

(16a)

.................................... (10)

or
The mixture parameter, c, is related through a linear mixing rule
to a third parameter, Cj, for each Component i:
C= Ec/zj

(11)

Substituting the volume-translation relation, Eq. 10, into Eq. 9 results in the modified three-parameter EOS:
_
E 3 (T,p,V,z/,a,b,c)-O

(12)

The fugacity coefficient, ;j, can now be evaluated from the translated three-parameter EOS, Eq. 12. Please refer to Ref. 1for details.
;;;j can be represented in terms of cP/ as
_
.( Cjp)
tPi""cP/(T,p,z/)exp - RT

(13)

The thermodynamic-equilibrium conditions,Eq. 4, appropriate to

the modifiea EOS can then be written as
-i

-g

x/cP/(T,P,Xi)=Y/tP/ (T,p,yj),

(C/p)

(CiP)

XjtP/ (T,p,x/)exp - RT =YjrPj (T,p'Yi)exp RT '

or
xirPf (T,P,Xi) =YjcPf(T,P,Yi)'

Thus, the equilibrium conditions derived from the modified threeparameter equation are the same as those derived from the unmodified two-parameter EOS. Consider once a.gain the system of
(2n+3), Eqs. 2 through 6, based on the two-parameter EOS, Eq.
9. If (!L,Xj,Yj,vi, Vg) constitute the solution of this system,then
the solution of the system of equations bas~ o,g the modified EOS,
Eq. 12, can be represented as (!L,x/,Yj, V i, V g), where
n

V i= Vi-

CjXj

(14)

j=1

and

(16b)

Peneloux et at. 1 applied their method to the SRK equation to calculate the third parameter for several well-defined light
hydrocarbons (lighter than n-clecane). Although the third parameter
was calculated by matching the saturated liquid volume at only one
datum point (Tr =0.7), they concluded that the application of the
modified equation results in superior volumetric predictions fot' pure
fluids at other conditions as well as for mixtures, except for pure
components in the neighborhood of critical points. To apply their
method to petroleum fluids, Peneloux et at. calculated the third parameter for some well-defined hydrocarbons to represent paraffinic,
naphthenic, and aromatic compounds in the heavier hydrocarbon
fractions. They took n-alkanes, n-alkylcyclohexanes, and nalkylbenzenes, respectively, to represent paraffms, naphthenes, and
aromatics, obtained their critical properties and acentric factors from
the correlations by Robinson and Peng, 3 and calculated the third
parameter by matching the molar volume of these .hydrocarbons _
at 60F [15.5C] and 1 atm [0.1013 MPa]. From--their earlier
conclusion-that once the third parameter is calculated by matching
the saturated liquid molar volume at Tr =0.7, it can improve volumetric predictions even at other conditions-it can be justified that
the third parameter can be calculated by matching the experimental
volume at any other condition except in the critical region. Here
they used standard conditions (60F and 1 atm [15'sC and 0.1013
MPa]) to match, because the molar volumes of these hydrocarbons
at standard conditions were availableas correlations in Ref. 3. Next,
they correlated the third parameter as a sixth-degree polynomial
in the carbon'number of these components. Application of this correlation to calculate the .third parameter for a pseudocomponent requires information about the pseudocomponent's carbon number
and its paraffinic, naphthenic, and aromatic content.
This section presented the background information on the PRF
method of modifying a two-parameter EOS and calculating the third
parameter. In the following sections we apply the PRF method to
the PR-EOS.
Thlrd.parameter Characterizations
for the PHEOS
For well-defined \)ydrocarbons lighter than n-heptane, we calculated the third parameter with the PRF method and matched the
saturated-liquid volumes at Tr =0.7. Because the third parameter,
c, has tl1e same units as the second parameter, b, of the unmodified
PR equation, we define a dimensionless shift parameter, s, as
C
S=- ...............................

n
V g=Vg -

C=V-VE

c/Yi'

(15)

(17)

i=1

Implementation of the three-parameter modification into an existing

phase-behavior solution algorithm based on any two-parameter EOS
is very simple. The two-parameter equation-based algorithm is executed first to obtain the solution (!LoX/,Y;, Vi, Vg)' The phasebehavior part ofthe solution (!LoX/,Yi) does not change. The phase
volumes (Vi, Vg) are then modified to (V i' V g) with the component's third-parameter values, as in Eqs. 14 a.nd 15.
The application of the three-parameter modified equation to
hydrocarbon fluids requires the value of the third parameter for
1034

The shift parameter for the well-defined lighter hydrocarbons is

presented in Table 1. Note that the shift parameter for these components is negative, implying that the volume calculated by the twoparameter equation is underestiinated and requires a negative translation to match the experimental volume. Note also that the magnitude of this parameter decreases with the increasing carbon
number.
To characterize the shift parameter for pseudocomponents
describing the heptane-plus fraction of hydrocarbon fluids, we
proceed like Peneloux et a1. alld calculate first the shift parameter
SPE Reservoir Engineering, August 1988

TABLE 1-SHIFT PARAMETER FOR HYDROCARBONS

Component
Methane
Ethane
Prop~ne

Isobutane
Normal butane
Isopentane
Normal pentane
Normal hexane

~!,---------------------,

s
- 0.1540
-0.1002
-0.08501
-0.07935
-0.06413
-0.04350
-0.04183
- 0.01478
o

for well-defined hydrocarbons heavier than n-heptane by matching

the molar volume at the standard condition. The shift parameter
for n-alkanes, n-alkylcyclohexanes, and n-alkylbenzenes is plotted
vs. the component's molecular weight in Fig. 1. Instead of following
P6neloux et ai. and developing a polynomial correlation in carbon
number requiring many correlation coefficients, we explore here
some simple way of correlating the shift parameter with the
molecular weight, M. Examining Fig. 1, we observe that the shift
parameter is negative for lower molecular weights and decreases
in magnitude with increasing molecular weight until it becomes zero.
Then the shift parameter is positive and increases with increasing
molecular weight. Thus, s can be represented as a power function
in M as

s= I-diMe,

(18)

where d and e are positive correlation coefficients. The shiftparameter correlation, Eq. 18, was fit to the curves in Fig. I and
the correlation coefficients, d and e, were determined with
regression analysis. These coefficients are given in Table 2. The
average-percent-deviation column in Table 2 is defined as

nd

I VE-Vc I xlOO

i~l
VE
D=---------,

(19)

where V is the molar volu!!1e at standard conditions and nd is the

number of hydrocarbons. D< 0.3 % for three hydrocarbon groups
and thus justifies the mathematical form of the shift-parameter correlation, Eq. 18.
Having correlated the shift parameter for the well-defined heavier
hydrocarbons with their molecular weight, we extend the applicability of correlation, Eq. 18, to pseudocomponents characterizing
the heptane-plus fraction of hydrocarbon fluids. Let M, (i = I to
np ) represent the molecular weight of Pseudocomponent i, with a
total of np pseudocomponents describing the heptane-plus fraction.
Then, the pseudocomponent's shift parameter si (i= I to np ) is correlated to the molecular weight, M" through the use of correlation
(Eq. 18). The characterization of the third parameter for the heptaneplus fraction thus requires evaluation of only two correlation coefficients, d and e in the correlation (Eq. 18).
A general method for calculating these two coefficients from the
laboratory density data on the heptane-plus fraction is discussed next.

Applications
Characterization Procedure. To apply the modified threeparameter PR-EOS to calculate the phase behavior and volumetric
properties of reservoir fluids, we need to characterize the three parameters (ai, bi' and Ci) for each Pseudocomponent i making up
the heptane-plus fraction. We assume that a reasonable distribution
(Zi' M i; i= I to np ) of the heptane-plus fraction 'into np pseudocomponents can be derived from the laboratory true boiling point, simulated distillation of the heptane-plus fraction, or from some
probability distribution function. 4 We use the two-step procedure
described below for characterizing all three parameters for the pseudocomponents.
Step 1. The first step consists of characterizing the parameters
at and b i of the unmodified, two-parameter PR equation and
SPE Reservoir Engineering. August 1988

LEGEND
0=

n-alkanes

=n....olkyl-cyclohexones
o = n-alkyl-benzene.

?----:r-----.------.-----..-----I
70

120

170

220

270

MOLECULAR WEIGHT

Fig. 1-Shlft parameter for welldeflned hydrocarbons.

TABLE 2-SHIFTPARAMETER CORRELATION

COEFFICIENTS FOR HYDROCARBONS
HEAVIER THAN HEXANE

Component
Type
nalkanes
nalkylcyclohexanes
nalkylbenzenes

Correlation Coefficient
-d- -e2.258
0.1823
3.004
0.2324
2.516
0.2008

i5
(%)
0.19
0.28
0.24

specifying the nontrivial interaction coefficient of each pseudocomponent with the well-defined components. The parameters ai and
b i are expressed by the EOS in terms of the component's physical
properties: the critical temperature, the critical pressure, and the
acentric factor. A priori estimate of these properties can be obtained
from several physical property correlations available in the literature. 5 Correlations for nontrival interaction coefficients of each
pseudocomponent with the well-defined components are also
available,3,5,6 The pseudocomponent's physical properties and the
nontrivial interaction coefficients with the well-defined components
present in the reservoir fluid are set a priori with some reasonable
correlation. In the usual characterization procedure, some of these
parameters are adjusted by matching the two-parameter EOS predictions with the laboratory data on the reservoir fluid. Some of
the standard laboratory experiments performed on the reservoir fluid
are the constant-composition expansion, the differential liberation,
and the constant-volume-depletion experiments conducteq at the
reservoir temperature. Some of the measured experimental data
contain information that is a combination of the phase-behavior properties (fL, X;, and Yi) and the volumetric properties (Ve and Vg ).
For example, the liquid saturation, S, in the constant-compositionexpansion experiment can be expressed as
S=/L Ve/[fL V e+(1-!L)Vg]'

(20)

If such data with mixed-phase behavior and volumetric information

are used directly in the matching procedure, some of the pseudocomponent parameters being adjusted may converge to nonphysical
values to account for errors in volumetric predictions. We take a
novel approach by analyzing the laboratory experiments and separate
the phase-behavior information (fv Xi, and Yi) from the volumetric
information (Ve and Vg ) (see the Appendix). Then the pseudocomponent parameters are adjusted to match the phase-behavior information only. The first step completes the characterization of ai'
bi, and
parameters of the reservoir fluid.
Step 2. The second step is the characterization of the third parameter, Ci' Because ci is expressed as ci=Sibi (Eq. 17), and b i
is already characterized in Step 1, this step consists of characterizing
the dimensionless shift parameter. For well-defined components,
the shift parameter is given in Table 1. For pseudocomponents,
the characterization of the shift parameter requires the specification

oij

TABLE 3-COMPOSITION OF RESERVOIR FLUIDS

Mole Percent in Reservoir Fluid

Component
CO 2
N2
Methane
Ethane
Propane
Isobutane
Normal butane
Isopentane
Normal pentane
Hexanes
C7+

1A

18

60.88
7.38
5.03

67.69
8.21
5.59

2.78

3.09

1.96
1.84
20.13

2.18
2.04
11.20

2
2.37
0.31
73.19
7.80
3.55
0.71
1.45
0.64
0.68
1.09
8.21

2.44
0.08
82.10
5.78
2.87
0.56
1.23
0.52
0.60
0.72
3.10

9.11

-- -- -45.58
5.11
3.03
2.18
1.92
1.54
31.53

C 7 + properties
132
270
191
191
184
Molecular weight
Specific gravity
0.8363 0.8363 0.8160 0.7740 0.8935
Reservoir
180.5
200
temperature, OF
190
190
280

of the correlation constants (d and e) in Eq. 18. In principal, the

two coefficients can be obtained by matching some laboratory
density data on the heptane-plus fraction at standard conditions (60F
and 1atm [15'soC and 0.1013 MPa]). In the examples that follow,
we have used the following type of data.
1. When the laboratory true-boiling-point distillation experiment
on the heptane-plus fraction reports the density ofvarious cuts, we
calculate the pseudocomponent densities with a mixing role and use
the densities to adjust d and e.
2. Some laboratory experiments, like the constant-volume
depletion and the differential liberation, report the density of the
heptane-plus fraction in the original reservoir fluid, as well as in
the fluid (liquid phase) depleted to a lower pressure. These two
data points are sufficient to adjust the two coefficients.
3. In the absence of this information, one datum is normally
available: the density of the heptane-plus fraction in the reservoir
fluid. Then, we take the power coefficient, e, to be 0.2051, averaged
from the coefficients for three groupsin Table 1, and adjust only
the coefficient d to match the heptane-plus density.

In the parameter-adjustment procedure, we take the a priori estimate for d=2.5927 and e=0.2051, averaged from ibeirvalues
for the three groups in Table 1. Also, we specify the bounds on
these coefficients from the maximum and the minimum values of
the group coefficients:
2.2<d<3.2

(21a)

and
0.18<e<0.25

(21b)

Examples. We have applied the modified PR-EOS to two crudeoil and two gas-condensate systems. The first example is of a crudeoil system from a Chitwood-Boatwright Sand Unit in Grady County,
OK, and was used earlier in the experimental investigations by Evans
and Harris. 7 Two mixtures of the reservoir fluid were reconstituted
from the separator streams in a fixed GOR but with different
amounts of heptane-plus moles in the resultant mixtures (called
Mixtures 1A and 1B here). The mixtures were subjected to a
constant-composition expansion and, at each pressure, the equilibrium phases were analyzed for composition and the phase molar
volumes were measured and reported as z factors. Because the data
were reported distinctly as phase-behavior data (Xi and Yi) and
volumetric data (z and Zg), no analysis of the experiment was necessary to separate information. The second and third examples are
gas-condensate systems. Example 2 is a rich North Sea gas condensate that Whitson and Torp8 chose to illustrate the application
of the two-parameter PR-EOS to gas-condensate systems. The third
reservoir fluid is an Iranian (Pazanan) gas condensate considered
by Firoozabadi et at. 9 The laboratory data from the constantvolume-depletion experiments performed on both of these systems
were available. These data were analyzed by techniques in the Appendix and converted to the phase-behavior part (/L = molar liquid
yield) and the volumetric part (z and Zg). Example 4 is a reservoir
crude oil. The laboratory data from a constant-composition expansion and a differential-liberation experiment performed on this
system were convertedto/L, z' and Zg with the analysis given in
the Appendix. Table 3 reports the composition of the reservoir
fluids, the heptane-plUS properties, and the (reservoir) temperature
at which the laboratory experiments were performed.

TABLE 4-HEPTANE-PLUS CHARACTERIZATION FOR RESERVOIR FLUID 1

Pseudocomponent

Mole Fraction

MoleCUlar
Weight

1
2
3
4
5

0.3727
0.2497
0.1682
0.1165
0.0929

108.9
151.0
212.7
310.1
440.0

To

JJ5L
577.2
649.3
727.3
816.2
954.3

(atm)

Methane-Interaction
Coefficient

27.71
21.87
16.99
12.82
9.23

0.3163
0.4275
0.5769
0.8111
1.049

0.0730
0.0856
0.0971
0.1077
0.1327

Po

TABLE 5-HEPTANEPLUS CHARACTERIZATION FOR RESERVOIR FLUID 2

Pseudocomponent

Mole Fraction

Molecular
Weight

1
2

0.3954
0.2608
0.1391
0.1187
0.0860

106.1
152.7
209.2
281.5
462.3

3
4
5

To

JJ5L

(atm)

Methane-Interactlon
Coefficient

545.7
644.2
713.9
783.1
856.5

26.85
21.58
17.10
13.71
12.24

0.3075
0.4305
0.5671
0.7389
1.1920

0.0366
0.0429
0.0429
0.0525
0.0747

Po

TABLE 6-HEPTANEPLUS CHARACTERIZATION FOR RESERVOIR FLUID 3

Pseudocomponent Mole Fraction

1
2
3
4
1036

0.6379
0.2180
0.0897
0.0544

MoleCUlar
Weight
106.3
144.2
187.6
293.0

To

JJ5L

(atm)

w
_.-

Methane-Interaction
Coefficient

546.2
639.1
698.2
784.4

28.00
22.62
18.64
13.89

0.3085
0.4110
0.5173
0.7792

0.0371
0.0434
0.0480
0.0741

Po

SPE Reservoir Engineering, August 1988

TABU:; 7-LIQUIDPHASE zFACTOR AT 600F AND

1 atm FOR PSEUDOCOMPONENTS CHARACTERIZING
HEPTANE-PLUS FRACTION OF RESERVOIR FLUID 1

:!.,---------------------.

L1quidPhase i: Factor
From Distillation From ThreeParameter
Pseudocomponent
Data
PR-EOS
1
0.006158
0.006317
.2
0.008037
0.008087
3
0.01075
0.01047
. 0.01497
0.01391
4
0.01926
0.02033
5
LEGEND
0=

TABLE 8-LIQUID-PHASE z-FACTOR AT 600F

AND 1 atm FOR PSEUDOCOMPONENTS
CHARACTERIZING HEPTANEPLUS FRACTION
OF RESERVOIR FLUIDS 2 AND 3

Reservoir Fluid 2
Depleted Fluid 2
Reservoir Fluid 3
Depleted Fluid 3

data

= 3-porameter
D

= 2"'parameter

d'00:::0--""'":'2:r:----30"'='00:-'"'---.0.,..00----'0,..00----1
00 0

6000

PRESSURE(PSIA)

Fig. 2-L1quld-phase z factor for Fluid 1A.

Liquid-Phase z Factor
Data
Three-Parameter PREOS
0.009518
0.009644
0.010820
0.010670
0.007199
0.007213
0.007941
0.007925

TABLE 9-SHIFTPARAMETER VALUES AND

CORRELATION COEFFICIENTS FOR HEPTANE.PLUS
FRACTIONS OF RESERVOIR FLUIDS
Shift Parameter
Pseudocomponent
1
2
3
4
5
Correlation Coefficient
d
e

1
0.01667
0.09311
0.1669
0.2411
0.3041

Reservoir Fluid
2
-0.6166
0.0055685
0.06047
0.1094
0.1854

3
0.01813
0.D7058
0.1136
0.1819

--LEGEND
0=

data

'. = 3'-pCitometer

o
d+--__.,..1000

2000

r3000

.,..4000

D=2"'parameter

--,-f

!)OOO

6000

PRESSURE(PSIA)

3.141
0.2476

2.458
0.1800

2.271
0.1800

We applied Step 1 of the characterization procedure described

earlier to these example fluids. The heptane-plus fraction was distributed first into a system of pseudocomponents. For Reservoir
Fluid 1, some of the pseudocomponent parameters were adjusted
to match k-value data (kj =Yi1xj) on Mixtures lA and lB. For GasCondensate Systems 2 and 3, we matched the molar liquid yields.
For the crude oil of Example 4, we matched the!L data derived
from the laboratory experiments. The heptane-plus characterization
parameters for Reservoir Fluids 1 through 3 are given in Tables
4 through 6.
The characterization procedure, Step 2, was then applied to adjust
the correlation coefficients (d and e) for the heptane-plus fractions.
For Reservoir Fluid 1, we adjusted both coefficients and matched
the pseudocomponent's liquid-phase z factors at standard conditions
derived from the distillation data. For the gas condensates, Fluids
2 and 3, we adjusted the two coefficients and matched the liquidphase z factors at standard conditions for the heptane-plus fractions
in the reservoir fluids and in the fluids depleted 71S psia [4.93 MPa].
For Crude Oil 4, we matched the density of heptane-plus fraction
in the reservoir fluid by adjusting only Coefficient d. The match
between the data and predictions for Reservoir Fluids 1 through
3 is shown in Tables 7 and 8. The shift-parameter values for the
pseudocomponents and their correlation coefficients are given in
Table 9. Note that a good match is possible with the coefficient
values within the bounds given in Eq. 21.
Having obtained the complete fluid characterization for use in
the modified equation, we simulated the laboratory experiments on
these fluids to evaluate the improvement in the volumetric predictions resulting from the third parameter. In Fig. 2, we compare
the experimental-liquid-phase z factors for Reservoir Fluid lA with
SPE Reservoir Engineering, August 1988

Fig. 3-Gas-phase zfactor for Fluid 1A.

the predictions through the two-parameter and the modified threeparameter equations. Except for the first two points at higher
pressures, where the experimental data seem questionable, the thirdparameter predictions are in excellent agreement with the data. Fig.
3 compares the gas-phase z factors for the same fluid. There is also
a significant improvement in the gas z factors. We make two important observations here that also hold for the three other examples.
Note that the two-parameter equation overestimates the liquid-phase
volumes and underestimates the gas-phase volumes. The thirdparameter correction, however, consistently improves the volUmetric estimates for both phases by providing a positive translation
for the liquid phase and a negative translation for the gas phase.
This is possible because the liquid is richer in heavier hydrocarbons
that have positive-shIft-parameter values, which results in an overall
positive translation. The gas is richer in lighter hydrocarbons that
have negative-shift-parameter values, which results in an overall
negative translation. The second observation is that although the
shift-parameter values for the lighter hydrocarbons were obtained
by matching the saturated liquid volumes at Tr =0.7 and the shiftparameter values for the heptane-plus pseudocomponents were obtained by matching the heptane-plus density data at 6QF and 1 atm
[IS.SoC and 0.1013 MPa], the improvement in volumetric predictions is significant at higher (reservoir) temperature and higher
pressures.
For the second example of the North Sea gas condensate, we
compare the experimental-volume-percent liquid with the predictions through the two-parameter and the modified equation, in Fig.
4, and the gas-phase z factors in Fig. S. The same comparisons
for Example 3 of Iranian gas condensate can be found in Figs. 6
and 7. We see once again the improvement in volumetric predic1037

iil...- - - - - - - - - - - - - - - - - - - . . . . ,

. . . ---,

~-r-----------------LEGEND
0;;;;

doto

== 3-porameter
0:::; 2-parameter

LEGEND
0:=

deta

.::::. 3--parameter

o = 2-porameter
o-!---_ _.,......_ _.,....._ _,........._ _........_ _........_ _......._......-l
500

1500

25003500

45005500

6500

7500

::l

'+0-0--,."""00--":'2."'0-0---,."'0:"'0........- '1"00---.,..00--........00--'""'7600

PRESSURE(PSIA)

PRESSURE(PS1A)

Fig. 4-Llquld drop-out for North Sea gas condensate (Reservoir Fluid 2).

Fig. 5-Gas-phase z factor for North Sea gas condensate

(Reservoir Fluid 2).
'

tions. The improvement in predictions resulting from the introduction of the third parameter is shown in a tabular form for
all four examples in Table 10. The average percent (leviation is
defined by Eq. 19 for all quantities except the volume-percent liquid
dropout for the gas-condensate systems, where

hydrocarbon fluids. The correlation calculates the shift parameter

from the pseudocomponent's molecular weights and two correlation
coefficients. The correlation coefficients for three groups of welldefined hydrocarbons (the paraffinic, the aromatic, and the
naphthenic) are calculated with the experimental density data.
4. A novel two-step method of characterizing all three parameters of the heptane-plus pseudocomponents is presented. The first
step characterizes the two parameters of the unmodified equation
with the phase-behavior data derived from the analysis of laboratory
experiments. The second step calculates the shift-parameter correlation coefficients by matching the heptane-plus density data at
standard conditions.
5. In conjunction with this characterization method, the modified,
three-parameter PR equation was applied to two crude-oil and two
gas-condensate systems. Although the shift-parameter values were
calculated with the density data at Tr =0.7 for lighter hydrocarbons
and at standard conditions for the heavier hydrocarbons, the
modified equation improved volumetric prediction"s even at the
higher (reservoir) temperature and higher pressures. For all four
examples, the improvement was significant for both the liquid- and
gas-phase volumes.
6. With such consistent and significant improvements in the volumetric predictions, the modified, three-parameter PR-EOS is recommended for calculating the phase and volumetric behavior of
hydrocarbon fluids.

D=

.... ,.,

'

(22)

Here S is the volume percent liquid in the cell and nd is the number
, of data points. For the crude oil in Example 4, the reduction in
the oil density error from 16.8 to 0.7% is remarkable. For all four
examples, we observe that the volumetric predictions with the
modified, three-parameter equation are remarkably superior to those
obtained with the unmodified equation.

Conclusions
1. The PRF method orintroducing a third parameter into a twoparameter EOS is applied to modify the PR-EOS. A dimensionless
shift parameter is defined by making the third parameter dimensionless with the volumetric parameter b of the unmodified equation.
2. The shift parameter for some well-defined hydrocarbons lighter
than n-heptane is calculated with the PRF method.
3. A correlation was developed to characterize the shift parameter
for the pseudocomponents describing the heptane-plus fraction of

"-r----------------------.

Nomenclatu,re

= first (energy) parameter of EOS

b = second (volume) parameter of EOS

...---------------------.
LEGEND

= dolo

3-porometer
o = 2....porameter

0"

::J

~.

..,~ .
3o=-

>N

LEGEND

a = dolo
.:::; 3-parameter
D

.00

=2-parameter
tlOI)

1700

o
.,O-!---_
_.,.................,................,-2300

2900

3500

<l100

4700

PRESSURE(PSIA)

Flg.6-Llquld dropout for Iranian gas condensate (Reservoir Fluid 3).

1038

600

1100

1700

2300

.,......__..._..-__..._..-_........-I
2iOO

'3500

4''10

4700

PRESSURE(PSIA)

Fig. 7-Gas.phase z factor for Iranian gas condensate (Reservoir Fluid 3).
SPE Reservoir Engineering, August 1988

TABLE 10-AVERAGE PERCENT DEVIATION, 6, BETWEEN EXPERIMENTAL DATA AND

PREDICTIONS FROM TWO-PARAMETER AND MODIFIED THREE-PARAMETER PR-EOS
Reservoir
Fluid
Experiment
Constant1
composition
expansion

Quantity
liquid-phase
z factor

11

9.0

1.3

11

3.4

1.7

Volume-percent
liquid

1.8

1.1

Gas withdrawn
Gas-phase
z factor

6
6

3.7
3.5

1.6
1.3

Volume-percent
liquid

0.8

0.5

Gas withdrawn
Gas-phase
z factor

5
5

1.9
.2.8

1.4
1.0

Constantcomposition
expansion

Relative volume

18

2.12

0.24

Differential
liberation

Relative volume

11

2.32

0.84

Gas withdrawn
Gas-phase z factor
Oil density

11
11
11

12.80
2.50
16.80

2.28
1.21
0.67

Gas-phase
z factor

Constantvolume
depletion

Constantvolume
depletion

c = third (volume) parameter of modified EOS

d = correlation coefficient of third parameter
jj = average deviation, %
e = correlation coefficient of third parameter
E = experimental values
!L = mole fraction of liquid phase
Gwd = cumulative gas withdrawn
k = k-value data
M = molecular weight
n = number of components
nd = number of data points
np = number of pseudocomponents
p = pressure
R = gas constant
s = shift parameter (=c/b)
S = liquid saturation
T = temperature
Ii' = molar volume
V = translated molar volume
Vr = relative volume
x = liquid composition
y = gas composition
z = overall system composition or phase compressibility,
depending on subscript
{) = interaction coefficient
cP = fugacity coefficient
C$ = translated fugacity coefficient
OJ = acentric factor

Subscripts
c
C
E
g
i
j
f

Number of
Data
Two-Parameter Three-Parameter
Points
Equation
Equation

=
=
=
=

critical
correlation
at experimental condition
gas-phase identifier
= component identifier
= component identifier
= liquid-phase identifier

SPE Reservoir Engineering, August 1988

p = predicted
r = at reduced condition
s = at saturation pressure

Superscripts
g =

gas-phase identifier
f == liqUid-phase identifier

Acknowledgment
We would like to acknowledge the management of Arco Oil & Gas
Co. for permission to publish this paper.

References
1. Peneloux, A., Rauzy, E., and Freze, R.: "A Consistent Correction
for RedlichKwong-Soave Volumes," Fluid Phase Equilibria (1982)
8,7-23.

2. Spencer, C.F. and Danner, R.P.: "Prediction of Bubble Point Density

of Mixtures," J. Chern. Eng. Data (1973) 18, 230-33.
3. Robinson, D.B. andPeng, D.-Y.: "The Characterization of Heptanes
and Heavier Fractions for the GPA-Peng-Robinsoti. Programs," GPA
Research Report, Tulsa (1978) RR-28.
4. Whitson, C.H.: "Characterizing Hydrocarbon Plus Fractions," paper
BUR 183 presented at the 1980 Europec Meeting, London, Oct. 21-24.
5. Whitson, C.H.: "Effect of Physical Properties Estimation on Equationof-State Predictions," SPEl (Dec. 1984) 685-96.
6. Katz, D.L. and Firoozabadi, A.: "Predicting Phase Behavior of Condensate/Crude-Oil Systems Using Methane Interaction Coefficients,"
JPT (Nov. 1978) 1649-55.
7. Evans, R.B. and Harris, D.: "Equilibrium Vaporization Ratios for
Hydrocarbon Mixture Containing Two Concentrations of Heptanes and
Heavier Fraction," J. Chern. Eng. Data (1956) 1,45-50.
8. Whitson, C.H. and Torp, S.B.: "Evaluating Constant-Volume Depletion
Data," JPT (March 1983) 61O~20.
9. Firoozabadi, A., Hekim, Y., and Katz, D.L.: "Reservoir Depletion
Calculation for Gas Condensates Using Extended Analysis in the PengRobinson Equation of State," Cdn. J. Chern. Eng. (1978) 56, 610-15.

AppendiX-Analysis of Laboratory Experiments

The following laboratory experiments are performed in a PVT cell
at the reservoir temperature. A given sample of the fluid is charged
to the cell, and the fluid saturation pressure, Ps' and the z factor
1039

at saturation pressure, zs' are determined. Then the fluid is expanded in stages by decreasing the pressure. In the constantcomposition expansion, after expansion to a lower pressure, no
fluids are removed from the cell. The experimental data consist
of the following quantities at each stage pressure,p, where S=liquid
saturation, Vr=relative volume=total phase volume/volume at
saturation pressure, and Zg=gas-phase Z factor.
In the differential-liberation experiment, after expansion to a lower
pressure, all the equilibrium gas phase is withdrawn. At each stage,
the following quantities are reported: Gwd=cumulative gas withdrawn, moles per mole of initial fluid, Vr=relative volume=
volume of oil phase/volume at saturation pressure, and Zg = gasphase Z factor
In the constantvolume-depletion experiment, after expansion to
a lower pressure, the cell volume is returned to the original volume
at the saturation pressure by withdrawing enough equilibrium gas.
The experiment reports the equilibrium-gas Z factor, Zg' the cumulative gas withdrawn, Gwd' and the liquid saturation, S, all as
defined earlier.
We analyze these data to separate the phase-behavior information
from the volumetric information. Prom one mole of the fluid mixture
at the saturation pressure, we calculate the following quantities for
all three experiments at each stage pressure, p: h = moles of equilibrium liquid and ze=liquidphase Z factor.
The gas Z factor is the only quantity reported explicitly as volumetric information in the earlier laboratory data. The two other
quantities can be easily calculated for each experiment. They are
related to experimental data as follows.

and

Differential-Liberation Experiment.
IL = I-G wd

h=I-Vr(I-S)zsp/ZgPs

1040

(A-I)

(A-3)

and
(A-4)

Ze= VrZsp/hps'

Constant-Volume-Depletion Experiment.
h=(I-Gwd)~(l-S)ZsP/ZgPs

(A-5)

and
(A-6)

ze=SZsP/ILPs'

SI Metric Co.,verslon Factors

atm x 1.013 250*
E+05
0p (OP-32)/1.8
0p (OP+459:67)/1.8
psi x 6.894 757
E+OO
'Conversion faotor Is exact.

Constant-Composition-Expansion Experiment.

(A-2)

Ze= VrSzsp/hS. .

Pa
C
K
kPa
SPERE

Original SPE manuscript received for review Sept. 16. 1984. Paper accepted for publication
May 1,1987. Revised manuscript reoelved Nov. 20, 1987. Paper (SPE 13118) first presented
at the 1984 SPE Annual Technical Conference and Exhibition held in Houston, Sept. 16-19.

SPE Reservoir Engineering. August 1988