Professional Documents
Culture Documents
by
P AT RI CK C HU N G- N I N MA K
B. A. S c , The Uni versi ty of Bri ti sh Col umbi a, 1985
A THES I S S U B MI T T E D I N P A R T I A L F U L F I L ME N T O F
T H E R E QU I R E ME N T S F OR T H E D E G R E E O F
MA S T E R OF AP P L I E D S CI E NCE
in
T H E F A C U L T Y O F G R A D U A T E S TUDI ES
D E P A R T ME N T O F C H E MI C A L E NGI NE E RI NG
We accept this thesis as conforming
to the required standard
T H E UNI VE RS I T Y O F BRI TI SH C OL U MB I A
February 1988
P AT RI CK C HU N G- N I N MA K , 1988
In presenting this thesis in partial fulfilment of the requirements for an advanced
degree at the University of British Columbia, I agree that the Library shall make it
freely available for reference and study. I further agree that permission for extensive
copying of this thesis for scholarly purposes may be granted by the head of my
department or by his or her representatives. It is understood that copying or
publication of this thesis for financial gain shall not be allowed without my written
permission.
Department of Cti&AiCAL- W SrZ-l"J6
The University of British Columbia
1956 Main Mall
Vancouver, Canada
V6T 1Y3
Date /W - Z~ / f
ABSTRACT
The Lielmezs-Merriman equation of state has been modified in such a wa3'
that it can be applied over the entire PVT surface except along the critical
isotherm. The dimensionless T coordinate has been defined according to the two
regions on the PVT surface as :
T /T - 1
T = for T < T
T
c
/T
b
- 1
and
* T/T - 1
T = for T > T
V
T
b "
1
where T
c
is the critical temperature and is the normal boiling point. The
a-function is now given by the following :
a = 1 + p(T*)
q
for T < T
c
and
a = 1 - p(T )
q
for T > T
c
The two substance-dependent constants p and q are generated from the vapor
pressure data.
The applicability of the proposed modification has been tested by
comparing its predictions of various pure compound physical and thermodynamic
properties with known experimental data and with predictions from the
Soave-Redlich-Kwong and Peng-Robinson equations of state. The proposed equation
ii
is the most accurate equation of state for calculating vapor pressure, and
saturated vapor and liquid volumes. The Peng-Robinson equation is the best for
enthalp}' and entropy of vaporization estimations. The Soave-Redlich-Kwong
equation is the least accurate equation for pressure and volume predictions in the
single phase regions. For temperature prediction, all three equations of state give
similar results in the subcritical and supercritical regions. None of the three
equations is capable of representing all departure functions accurately. The
Peng-Robinson equation and the proposed equation are very similar in accuracy
except in the region where the temperature is near the critical. That is, between
0.95 < T
r
^ 1.05, the proposed equation gives rather poor results. For isobaric
heat capacity calculation, both Soave-Redlich-Kwong and Peng-Robinson equations
are adequate. The Soave-Redlich-Kwong equation gives the lowest overall average
RMS % error for Joule-Thomson coefficient estimation. The Soave-Redlich-Kwong
equation also provides the most reliable prediction for the Joul e-Thomson inversion
curve right up to the maxi mum inversion pressure.
None of the cubic equations of state studied in this work is recommended
for second vi ri al coefficient calculation below T
r
= 0.8. An a-function specifically
designed for the calculation of second vi ri al coefficient has been included in this
work. The estimation from the proposed function gives equal, i f not better,
accuracy than the Tsonopoulos correlation.
iii
TABLE OF CONTENTS
A B S T R A C T ii
LI ST O F F I GURE S vi
LI S T OF T A B L E S x
A C K N O WL E D G E ME N T xii
Chapter 1. I NT RODUCT I ON 1
Chapter 2. L I T E R A T U R E R E V I E W 7
Chapter 3. D E V E L O P ME N T OF A N E Q U A T I O N O F S T A T E 19
3.1. Theoretical Considerations 19
3.2. Li el mezs- Merri man Equati on of State 25
3.3. Modified Li el mezs- Merri man Equati on of State 32
Chapter 4. AP P LI CATI ONS OF T H E CUBI C E Q U A T I O N OF S T A T E 45
4.1. Region I : Saturation 46
4.2. Region II : Subcritical 70
4.3. Region III : Supercritical 89
4.4. Region I V : Compressed Li qui d 106
4.5. Inversion Curve 123
Chapter 5. S E COND VI RI AL COE F F I CI E NT 137
5.1. Introduction 137
5.2. Theoretical Background 139
5.3. A New Correlation 141
5.4. Compari son 145
CONCL US I ONS 150
R E C O MME N D A T I O N S F O R F U R T H E R S T U D Y 153
N O ME N C L A T U R E 154
R E F E R E N C E S 157
AP P E NDI X A 164
1. Cri ti cal Compressibility Factor 164
1.1. The Soave-Redlich-Kwong Equati on 169
1.2. The Peng-Robinson Equati on 169
1.3. Thi s Work 170
2. Constants A and B 170
2.1. The Soave-Redlich-Kwong Equati on 171
2.2. The Peng-Robinson Equation 171
2.3. Thi s Work 172
3. Compressibility Factor Equation 172
i v
3.1. The Soave-Redlich-Kwong Equati on 173
3.2. The Peng-Robinson Equation 173
3.3. Thi s Work 173
4. Fugaci ty Coefficient 173
4.1. The Soave-Redlich-Kwong Equati on 176
4.2. The Peng-Robinson Equation 176
4.3. Thi s Work 177
5. Enthal py Departure Function 177
5.1. The Soave-Redlich-Kwong Equation 179
5.2. The Peng-Robinson Equation 180
5.3. Thi s Work 181
6. Heat Capaci ty Departure Functi on 183
6.1. The Soave-Redlich-Kwong Equati on 189
6.2. The Peng-Robinson Equation 190
6.3. Thi s Work 190
7. Joul e-Thomson Coefficient 191
7.1. The Soave-Redlich-Kwong Equation 191
7.2. The Peng-Robinson Equati on 192
7.3. Thi s Work 192
8. Inversion Curve 192
8.1. The Soave-Redlich-Kwong Equati on 195
8.2. The Peng-Robinson Equati on 195
8.3. Thi s Work 196
9. Vi ri al Coefficients 197
9.1. The Soave-Redlich-Kwong Equati on 199
9.2. The Peng-Robinson Equati on 199
9.3. Thi s Work 200
v
LIST OF FIGURES
Fi gure 1.1 Division of the Pressure-Vol ume surface 6
Fi gure 3.1 Maxwell-rule of equal areas 35
Fi gure 3.2 Al gori thm for determining the substance-dependent constants p
and q 36
Fi gure 3.3 Correlations of p with parameter s and with acentric factor u 37
Fi gure 3.4 Correlations of acentric factor u with parameter s 38
Fi gure 4.1 Al gori thm for calculating saturation properties 61
Fi gure 4.2 Region I : Error distribution curves of methane 62
Fi gure 4.3 Region I : Error distribution curves of n-pentane 63
Fi gure 4.4 Region I : Error distribution curves of 1-propanol 64
Fi gure 4.5 Region I : Error distribution curves of argon 65
Fi gure 4.6 Region I : Enthal py and entropy of vaporization of methane 66
Fi gure 4.7 Region I : Enthal py and entropy of vaporization of n-pentane 67
Fi gure 4.8 Region I : Enthal pj' and entropy of vaporization of 1-propanol 68
Fi gure 4.9 Region I : Enthal py and entropy of vaporization of argon 69
Fi gure 4.10 Region II : Compressibility factors of i-butane (Pr = 0.56) and
water (Pr = 0.31) versus reduced temperature 82
Fi gure 4.11 Region II : Compressibility factors of i-butane (Tr = 0.98) and
water (Tr = 0.96) versus reduced pressure 82
Fi gure 4.12 Region II : Reduced enthalpy departure functions of i-butane
(Pr = 0.56) and water (Pr = 0.31) versus reduced temperature 83
Fi gure 4.13 Region II : Reduced enthalpy departure functions of i-butane
(Tr = 0.98) and water (Tr = 0.96) versus reduced pressure 83
Fi gure 4.14 Region II : Reduced entropy departure functions of i-butane
(Pr = 0.56) and water (Pr = 0.31) versus reduced temperature 84
Fi gure 4.15 Region II : Reduced entropy departure functions of i-butane
(Tr=0. 98) and water (Tr = 0.96) versus reduced pressure 84
v i
Fi gure 4.16 Region II : Reduced Hel mhol tz departure functions of i-butane
(Pr=0. 56) and water (Pr = 0.31) versus reduced temperature 85
Fi gure 4.17 Region II : Reduced Helmholtz departure functions of i-butane
(Tr = 0.98) and water (Tr = 0.96) versus reduced pressure 85
Fi gure 4.18 Region II : Reduced Gibbs energy departure functions of
i-butane (Pr = 0.56) and water (Pr = 0.31) versus reduced temperature 86
Fi gure 4.19 Region II : Reduced Gibbs energy departure functions of
i-butane (Tr = 0.98) and water (Tr = 0.96) versus reduced pressure 86
Fi gure 4.20 Region II : Reduced internal energy departure functions of
i-butane (Pr = 0.56) and water (Pr = 0.31) versus reduced temperature 87
Fi gure 4.21 Region II : Reduced internal energy departure functions of
i-butane (Tr = 0.98) and water (Tr = 0.96) versus reduced pressure 87
Fi gure 4.22 Region II : Fugaci ty coefficients of i-butane (Pr = 0.56) and
water (Pr = 0.31) versus reduced temperature 88
Fi gure 4.23 Region II : Fugacit3' coefficients of i-butane (Tr = 0.98) and
water (Tr = 0.96) versus reduced pressure 88
Fi gure 4.24 Region III : Compressibility factors of n-octane (Pr = 4.03) and
ethanol (Pr=1. 63) versus reduced temperature 99
Fi gure 4.25 Region III : Compressibility factors of n-octane (Tr=1. 76) and
ethanol (Tr=1. 95) versus reduced pressure 99
Fi gure 4.26 Region III : Reduced enthalpy departure functions of n-octane
(Pr = 4.03) and ethanol (Pr=1. 63) versus reduced temperature 100
Fi gure 4.27 Region III : Reduced enthalpy departure functions of n-octane
(Tr=1. 76) and ethanol (Tr=1. 95) versus reduced pressure 100
Fi gure 4.28 Region III : Reduced entropy departure functions of n-octane
(Pr = 4.03) and ethanol (Pr=1. 63) versus reduced temperature 101
Fi gure 4.29 Region III : Reduced entropy departure functions of n-octane
(Tr=1. 76) and ethanol (Tr=1. 95) versus reduced pressure 101
Fi gure 4.30 Region III : Reduced Hel mhol tz departure functions of n-octane
(Pr=4. 03) and ethanol (Pr=1. 63) versus reduced temperature 102
Fi gure 4.31 Region III : Reduced Helmholtz departure functions of n-octane
(Tr=1. 76) and ethanol (Tr=1. 95) versus reduced pressure 102
Fi gure 4.32 Region III : Reduced Gibbs energy departure functions of
n-octane (Pr = 4.03) and ethanol (Pr=1. 63) versus reduced temperature. 103
vn
Figure 4.33 Region III : Reduced Gibbs energy departure functions of
n-octane (Tr=1.76) and ethanol (Tr=1.95) versus reduced pressure 103
Figure 4.34 Region III : Reduced internal energy departure functions of
n-octane (Pr = 4.03) and ethanol (Pr=1.63) versus reduced temperature. 104
Figure 4.35 Region III : Reduced internal energy departure functions of
n-octane (Tr=1.76) and ethanol (Tr=1.95) versus reduced pressure 104
Figure 4.36 Region III : Fugacity coefficients of n-octane (Pr = 4.03) and
ethanol (Pr=1.63) versus reduced temperature 105
Figure 4.37 Region III : Fugacity coefficients of n-octane (Tr=1.76) and
ethanol (Tr=1.95) versus reduced pressure 105
Figure 4.38 Region IV : Compressibility factors of n-heptane (Pr = 0.73) and
methanol (Pr = 0.63) versus reduced temperature 116
Figure 4.39 Region IV : Compressibility factors of n-heptane (Tr = 0.67) and
methanol (Tr = 0.70) versus reduced pressure 116
Figure 4.40 Region IV : Reduced enthalpy departure functions of n-heptane
(Pr=0.73) and methanol (Pr = 0.63) versus reduced temperature 117
Figure 4.41 Region IV : Reduced enthalpy departure functions of n-heptane
(Tr = 0.67) and methanol (Tr = 0.70) versus reduced pressure 117
Figure 4.42 Region IV : Reduced entropy departure functions of n-heptane
(Pr=0.73) and methanol (Pr = 0.63) versus reduced temperature 118
Figure 4.43 Region IV : Reduced entropy departure functions of n-heptane
(Tr=0.67) and methanol (Tr = 0.70) versus reduced pressure 118
Figure 4.44 Region IV : Reduced Helmholtz departure functions of n-heptane
(Pr = 0.73) and methanol (Pr = 0.63) versus reduced temperature 119
Figure 4.45 Region IV : Reduced Helmholtz departure functions of n-heptane
(Tr=0.67) and methanol (Tr=0.70) versus reduced pressure 119
Figure 4.46 Region IV : Reduced Gibbs energy departure functions of
n-heptane (Pr=0.73) and methanol (Pr = 0.63) versus reduced temperature.
120
Figure 4.47 Region IV : Reduced Gibbs energy departure functions of
n-heptane (Tr = 0.67) and methanol (Tr = 0.70) versus reduced pressure. 120
Figure 4.48 Region IV : Reduced internal energy departure functions of
n-heptane (Pr = 0.73) and methanol (Pr = 0.63) versus reduced temperature.
121
Vlll
Fi gure 4.49 Region IV : Reduced internal energy departure functions of
n-heptane (Tr = 0.67) and methanol (Tr = 0.70) versus reduced pressure. 121
Fi gure 4.50 Region IV : Fugaci ty coefficients of n-heptane (Pr = 0.73) and
methanol (Pr = 0.63) versus reduced temperature 122
Fi gure 4.51 Region IV : Fugaci ty coefficients of n-heptane (Tr = 0.67) and
methanol (Tr = 0.70) versus reduced pressure 122
Fi gure 4.52 Inversion curve of methane calculated from the SRK, PR, L M ,
and GCP equations 128
Fi gure 4.53 Inversion curve of propane calculated from the S RK, PR, and
L M equations 129
Fi gure 4.54 Inversion curve of n-butane calculated from the SRK, PR, and
L M equations 130
Fi gure 4.55 Inversion curve of carbon monoxide calculated from the S RK,
PR, L M , and GCP equations 131
Fi gure 4.56 Inversion curve of carbon dioxide calculated from the S RK, PR,
and L M equations 132
Fi gure 4.57 Inversion curve of ethylene calculated from the SRK, PR, L M ,
and GCP equations 133
Fi gure 4.58 Inversion curve of p-hydrogen calculated from the S RK, PR,
and L M equations 134
Fi gure 4.59 Inversion curve of ammoni a calculated from the SRK, PR, and
L M equations 135
Fi gure 4.60 Inversion curve of argon calculated from the SRK, PR, L M ,
and GCP equations 136
Fi gure 5.1 Second vi ri al coefficients of n-butane and benzene versus reduced
temperature 148
Fi gure 5.2 Second vi ri al coefficients of n-octane and argon versus reduced
temperature 149
ix
LIST OF TABLES
Tabl e 3.1 Calculated and Fitted p and q Constants 39
Tabl e 3.2 Vapor Pressure RMS % Error 41
Table 3.3 Acentric Factor from the Proposed, Lee-Kesl er, and Edmi ster
Correlations 43
Table 4.1 Summary of Physi cal Properties 51
Table 4.2 Region I - Vapor Pressure RMS % Error 53
Table 4.3 Region I - Li qui d and Vapor Vol ume RMS % Errors 55
Table 4.4 Region I - Enthalp3
f
and Ent ropy of Vapori zati on : Average
Absolute Deviation 57
Table 4.5 Overal l Average Errors ( RMS % and AAD) for the Four Regions
( NC = number of compounds ; N = number of data points) 59
Table 4.6 Coefficients of Ideal Gas Ent hal py Polynomial 75
Table 4.7 Coefficients of Ideal Gas Ent ropy Polynomial 75
Tabl e 4.8 Region II - Vol ume RMS % Error 76
Table 4.9 Region II - Pressure and Temperature RMS % Errors 77
Table 4.10 Region II - Ent hal py and Entropy Departure Functions :
Average Absolute Deviation 78
Table 4.11 Region II - Helmholtz and Gibbs Free Energy Departure
Functions : Average Absolute Deviation 79
Tbale 4.12 Region II - Internal Energy Departure Functions and Fugaci ty
Coefficient : Average Absolute Deviation 80
Table 4.13 Region II - Isobaric Heat Capaci tj' : Average Absolute Deviation . . . 81
Table 4.14 Region II - Joul e-Thomson Coefficient RMS % Error 81
Table 4.15 Region III - Vol ume RMS % Error 93
Tabl e 4.16 Region I U - Pressure and Temperature RMS % Errors 94
Table 4.17 Region III - Ent hal py and Ent ropy Departure Functions :
Average Absolute Deviation 95
x
Table 4.18 Region III - Helmholtz and Gibbs Free Energy Departure
Functions : Average Absolute Deviation 96
Tbale 4.19 Region III - Internal Energy Departure Functions and Fugacity
Coefficient : Average Absolute Deviation 97
Table 4.20 Region III - Isobaric Heat Capacity : Average Absolute Deviation .. 98
Table 4.21 Region III - Joule-Thomson Coefficient RMS % Error 98
Table 4.22 Region IV - Volume RMS % Error 110
Table 4.23 Region IV - Pressure and Temperature RMS % Errors I l l
Table 4.24 Region IV - Enthalpy and Entropy Departure Functions :
Average Absolute Deviation 112
Table 4.25 Region IV - Helmholtz and Gibbs Free Energy Departure
Functions : Average Absolute Deviation 113
Tbale 4.26 Region IV - Internal Energy Departure Functions and Fugacity
Coefficient : Average Absolute Deviation 114
Table 4.27 Region IV - Isobaric Heat Capacity : Average Absolute Deviation 115
Table 4.28 Inversion Pressure RMS % Error 127
Table 4.29 Maximum Reduced Inversion Pressure and Temperature 127
Table 5.1 Second Virial Coefficient : Average Absolute Deviation (cc/mole) 147
xi
A C K N O W L E D G E M E N T
My sincere thanks to professor Jani s Lielmezs for his guidance, his
patience, and his support in carryi ng out this project and his assistance in the
preparation of this manuscript.
I also wish to thank the Uni versi ty of Bri ti sh Col umbi a for a Graduate
Scholarship. Thanks are also due to the financial aid provided by the Nat ural
Sciences and Engineering Research Council of Canada.
xn
CHAPTER 1. INTRODUCTION
Physical and thermodjmamic properties calculations have always been an
essential part of engineering in the process industries. Many methods have been
proposed and used in past years. The older methods have usually relied on
charts or graphs. For the process engineer, the task has been time consuming
and tedious. Accuracy of the results were usually limited by the graph or chart
used. Due to the advent of computers, routine calculations such as
multicomponent vapor-liquid equilibria based on state equations can now be done
in minutes rather than hours. Hence, the demand for an accurate equation of
state has increased in the past decade.
An equation of state (EOS) refers to the equilibrium relation of state
parameters such as pressure, volume, temperature, and composition. In functional
form this relation is
f ( P , V , T , x ) =0
( L 1 )
Such an equation of state may be applied to gases, liquids, and solids.
Recent advances in computers have permitted wide spread efforts in
finding an equation of state which is simple to use and yet accurate enough for
most engineering calculations. Two approaches become evident in the literature:
theoretical and semi-empirical. When the theoretical approach is still in its
infancy stage, the semi-empirical approach is usually preferred and enjoys the
1
2
greatest success.
The so-called semi-empirical equation of state refers to an equation which
has a limited theoretical framework. Thi s form of the equation is usually based
on the corresponding states principle which states that similar thermodynamic
behavior is possessed by different chemical substances when compared at the
same reduced conditions. The constants of the semi-empirical equation are, in
many cases, obtained from experimental data. There are two types of
semi-empirical equation of state of general importance : multi-parameter and
cubic.
The term "multi-parameter" equation of state implies an equation with
more than five constants. These constants are usually substance-dependent.
Exampl es of this type are the equations of Beattie-Bridgeman [5],
Benedict-Webb-Rubin [6,7], Lee-Kesl er [59], and Starl i ng [114,115]. Thi s type of
equation is usually reserved for highly accurate work. To define the constants,
vast amount of input data are needed for multiproperty regression routine. The
Lee-Kesl er equation is the only exception which has a set of generalized
constants for hydrocarbons. However, no constants have yet been developed for
polar substances for the Lee-Kesl er equation.
Cubi c equation of state, on the other hand, refers to an equation which,
i f expanded, is cubic in volume or cubic in compressibility factor. Many of the
common two-constant cubic equations of state can be expressed by the general
form suggested by Schmidt and Wenzel [103] :
3
RT
p =
V- b
a(T)
V
2
+ubV+wb
2
(1.2)
Fr om here on Eq.(1.2) is known as the generalized cubic equation of state. Wi t h
different u and w values, most of the well-known cubic EOS can be reproduced:
for instance when u = 0 and w = 0, Eq.(1.2) becomes the van der Waal s equation
[125]; with u = l and w = 0, Eq.(1.2) becomes the various versions of the
Redl i ch-Kwong equation (RK) [93], and assigning u = 2 and w = - l , the
Peng-Robinson (PR) [87] equation is reproduced.
One of the reasons for the lack of acceptance of multi-parameter state
equations in volume calculation is the use of a time consuming trial-and-error
procedure. In many instances, the initial guess must be close to the solution in
order to have convergence. The cubic equation, on the other hand, can be solved
analytically, even though numerical techniques are often used. The constants
required are readily attainable from mi ni mal input data. In many of the
well-accepted cubic equations only critical properties and the acentric factor are
necessary to define the constants. The Soave modification of the R K equation
(SRK) [110] and the PR equation are such equations.
Since the introduction of the S RK and PR equations i n 1972 and 1976,
respectively, their capabilities have well been recognized by various industries
[2,16,32,33,43,74,86]. One simple reason for their wide acceptance in calculating
fluid thermodynamic properties is due pri mari l y to their simplicity and generality
combined with reasonable accuracy. However, both of these equations represent
polar substances poorly. Lielmezs and Merri man ( LM) [64] realized this limitation
4
and proposed a modified PR equation. They claimed that their modification has
improved significantly vapor pressure prediction and slightly better saturated liquid
compressibility than the original PR equation. In addition, the LM equation can
be applied to polar compounds. Mak [70] further modified the LM equation so
that it can be used for volume calculations in the subcritical region. The
accuracy of the resulting modification is comparable to that of the
multi-parameter Lee-Kesler equation. However, the modification needed
experimental PVT data as the input data in order to determine the required
constants and therefore limited its application. Although improvements have been
made by Lielmezs and Merriman, their equation cannot be used in the region
outside of the critical isotherm on the PVT surface.
The purpose of this work is, therefore, twofold: 1) to extend the capabilhVy
of the modified PR equation by Lielmezs and Merriman to the entire PVT
surface and 2) to test the modification for various property predictions. Since the
SRK and PR equations are popular in industries, it is of interest to test these
two equations along with the present work in property predictions that have
never been tested before. This work is limited to pure compound properties. It is
important to study pure substances because an equation of state that is unable
to adequately represent properties of a substance in the pure state, cannot, in
general, be expected to handle accurate^ mixtures containing the same
substances.
The PVT surface is divided into four regions: saturation, subcritical,
supercritical, and compressed. The subcritical region refers to the area above the
5
saturated vapor curve and beneath the critical isotherm. The supercritical region
is the area above the critical isotherm and the compressed liquid region is the
area under the the critical isotherm and above the saturated liquid curve. These
regions are shown in Fi gure 1.1. The properties included in the testing for the
saturation region are: vapor pressure, vapor and liquid volumes, and enthalpy
and entrop.y of vaporization. In the subcritical and supercritical regions, the
properties included are: pressure, volume, temperature, departure functions,
fugachty coefficient, isobaric heat capacity and Joule-Thomson coefficent. In the
compressed liquid region, the properties are basically the same as those in the
subcritical and supercritical states except Joul e-Thomson coefficient that has not
been included due to the lack of experimental data. In addition to the above
properties, the inversion curve and the second virial coefficient are also included
in this work.
Tc
Region III : Supercritical
L Critical
VPoint
Region IV : / \ \ Region II :
Compressed/* \ \ Subcritical
Liquid / *
\ /V
Region I : \ \
Saturation Curve \ \
Volume
Figure 1.1 Division of the Pressure-Volume surface.
CHAPTER 2. LITERATURE REVIEW
Ever since van der Waals [125] first proposed his equation of state more
than a centur\' ago, scores of modifications of his equation have been made.
Most of these equations have been empirical and arbitrary with parameters that
are adjustable to fit certain kinds of experimental data such as vapor pressure,
liquid or vapor volume. Literature on this subject has grown to large proportions.
A number of reviews have been published which addressed the merits and
limitations of some of these modifications. Tsonopoulos and Prausnitz [124]
reviewed a number of equations that are important for engineering purposes.
Shah and Thodos [104] pointed out the advantages of cubic equations over
multi-parameter equations. Martin [71] discussed some of the essential features an
equation of state should possess. The following review will be confined to cubic
equation of state of the van der Waals type and is by no means exhaustive.
This review will focus on three areas: 1) what approaches have been adopted in
past j'ears for improving the van der Waals t3'pe equation, 2) what properties
and t3'pes of compounds have been tested and, 3) what are the limitations of
these modifications.
The van der Waals equation may be written as the sum of two terms:
P
"
P
R
+ P
A (2.1)
where P
R
is the repulsion pressure and is the attraction pressure. Practically
all cubic equations of state of the semi-empirical t3'pe uses the van der Waals'
hard sphere equation to represent the repulsion pressure:
8
RT
P =
R
V-b
(2.1a)
while the attraction pressure can be expressed by the following equation:
a(T)
V*+ubV+wb
2
(2.1b)
Most modifications are based on modifj'ing the attraction term by 1) proposing a
different temperature dependent function for a(T) and/or, 2) using different u and
w values. One of the most successful modification of the van der Waal s equation
is the two-parameter equation proposed by Redlich and Kwong in 1949 [93]
where the attraction pressure is expressed as
a(T)
V(V+b)
(2.2)
and the two constants a and b are defined by Eq.(2.2a) to (2.2c)
a(T) = fl
a a
(2.2a)
-0.5
(2.2b)
a = T
r
b(T) = %
(2.2c)
9
The two constants Sl^ and fl^ at the critical point have numeri cal values of
0.42748 and 0.08664, respectively. The derivations of Eq.(2.2a) to (2.2c) can be
found in Appendi x A. Although predictions of vapor phase properties have been
improved, liquid phase representations were still poor.
After the work of Redlich and Kwong, Wilson [127,128] proposed that the
attraction constant "a" be made temperature dependent. The temperature
dependence is established from vapor pressure data at reduced temperature of 0.7
and 1.0. However his proposed equation did not gain much attention due to lack
of interest in cubic equations at that time.
Instead of maki ng the constant "a" temperature dependent, Chueh and
Prausni tz [14,15] proposed a modified R K equation with two sets of
substance-dependent constants Sl
&
and fl^, one set for the vapor phase and
another set for the liquid phase, which can be calculated from volumetric data.
The accuracy from the modified equation in estimating both vapor and liquid
volumetric properties has been increased, but the equation lacks internal
consistence. Joffe et al [50,130] recognized this problem and modified the
Chueh-Prausni tz equation by using only one set of constants which were fitted
wi th saturated liquid volume data. Si mi l ar to its predecessor, the equation is not
a generalized correlation. That is, experimental volumetric data are the necessary
input data in order to establish the two constants.
The acceptance of a cubic instead of a multi-parameter equation of state
for industrial application was due to the efforts of Soave [110] who proposed a
10
modified R K equation. Thi s modified equation was generalized, simple to use and
its range of application exceeded all the other modifications. The necessary input
data for this modification are the critical pressure and temperature and acentric
factor, (j. Soave utilized Wilson's original idea by maki ng the attraction constant
"a" temperature dependent. Vapor pressure data at T =0. 7 and the critical point
were used to obtain the required temperature dependance. The a-function is
generalized as a function of reduced temperature T
f
and acentric factor u. For
vapor-liquid equilibrium calculations, the Soave modification satisfies the equal
fugacity criterion, that is the vapor phase fugacity is equal to the liquid phase
fugacity. The importance of this fugacity equality condition will be discussed
further in the next chapter. Instead of the a defined b}' Eq.(2.2b), Soave defined
a by Eq.(2.3) and (2.3a) as:
a = [ 1 + m( l - T
r
'
5
) ]
2
(
2.
3
)
m = 0. 480 + 1. 574u - 0. 176w
2
in oi
Eq.(2.3) and (2.3a) can only be applied to nonpolar and slightly polar compounds
[110]. However the accuracy in vapor pressure prediction had increased more
than 100 % over the original R K [93] equation for a number of hydrocarbons.
Chaudron [12], Simonet and Behar [107] generalized the Q
&
and 0^
constants with the acentric factor but without incorporating the equal fugacity
criterion. Therefore, their equations do not satisfy the thermodynamic requirement
for phase equilibrium. Hence, for saturated vapor-liquid equilibrium calculation,
these equations give unacceptable results even though volumetric property
representations in the single phase region are reasonable.
Peng and Robinson [87] furthered the work of Soave by fitting the
a-function with vapor pressure data from the normal boiling point to the critical
point. In other words, the a multiplier was used to optimize the vapor pressure
as predicted by the EOS in question. In addition, the u and w values in the
denominator of the attraction pressure term have also been changed to 2 and
-1, respectively. The repulsion pressure term P
R
remains the same as before but
the attraction term is now given by the following:
a(T)
P a
V
2
+2bV-b
2
^
2
'
4
^
The a-function has the same form as Eq.(2.3) but the coefficients of the constant
m generalized in u take on new values:
m = 0.37464 + 1.54226u - 0.26992u
2
(2.4a)
The two constants Q
&
and have values of 0.45724 and 0.07780 [87],
respectively. As the result of their modification, Peng and Robinsion claimed that
their equation is superior to the SRK equation for saturated liquid volume
predictions. It was shown by a number of studies that their claim was indeed
valid [66,67,84,129].
Graboski and Daubert [41,42] adopted the same approach as Peng and
12
Robinson by refitting the a-function of the SRK equation with the entire vapor
pressure curve instead of just two points as was done by Soave [110]. For
hydrocarbons and slightly polar compounds, the form of the a-function is the
same as the SRK equation (Eq.(2.3)) but the coefficients of the constant m
generalized in CJ has been modified [41],
m = 0.48508 + 1.55171w - 0.15613w
2
( 2
. 5 )
In the case of hydrogen, the a-function is no longer written as Eq.(2.3) but is
expressed as [42]:
a = 1.202exp(-0.30228T
r
) (
2
.6)
Hamam et al [47], Leiva [60], and Paunovic et al [84] modified the RK
equation by expressing both constants a and b temperature dependent. They
generalized these two constants with acentric factors. Both vapor pressure and
saturated liquid volume data were used to obtain the two constants, a and b.
These studies have shown that saturated liquid volume predictions have improved
while vapor pressure predictions have remained approximately the same as those
for the SRK and PR equations. However, saturated vapor volume predictions
have drastically decreased in accuracy compared to the SRK and PR equations.
Study by Paunovic et al [84] has shown that by allowing both a and b to be
functions of temperature, derivative properties along the saturated liquid-vapor
equilibrium curve when estimated by this type of equation, have decreased in
accuracy.
13
For polar compounds, none of the above equations of state are sufficiently
suitable for representing saturation properties. Thi s is mai nl y due to the fact
that polar compounds do not follow the usual three-parameter corresponding states
principle proposed by Pitzer [88-90]. Thi s principle states that in addition to
reduced pressure and temperature, a third parameter is necessar}' to account for
the deviation of normal fluid behavior from simple fluid. Pitzer defined normal
fluid as the ones with zero or smal l dipole moment and he proposed that the
acentric factor defined as
u = - l o g ( at T
r
= 0. 7) - 1. 0
{2
.1)
be this third parameter. In simple terms, the acentric factor accounts for the
effects of the size and shape of different molecules. In order to obtain the
acentric factor value, reduced vapor pressure ( P
y
p
r
) at T
r
= 0.7 is required. The
three-parameter corresponding states principle with the acentric factor u as the
third parameter is, - therefore, inadequate to account for the dipole moment found
in polar compounds. Hence, for this reason, a number of investigators have
suggested a fourth parameter to characterize polar compounds. Such parameters
have been proposed, for instance, by Ha l m and Stiel [46], O'Connell and
Prausni tz [81], and Tarakad and Danner [120]. Al though each of the proposed
fourth parameter has correlated, with some degree of success, the thermodynamic
functions, however, none of them as yet has gained wide acceptance. It appears
that in order to use an equation of state for polar compounds, generalized
constants with acentric factor are unlikely to provide good representation of state
behavior; indeed substance-dependent constants seem to be the most logical way
14
to achieve high degree of accuracy.
This route was followed by Soave [111] who realized that his earlier
equation was inadequate for polar compounds. He suggested a different a-function
characterized now by two substance-dependent constants m and n :
These two constants can be calculated from vapor pressure data. Lielmezs et al
(LHC) [63] modified the RK equation by defining a different a-function, also with
two substance dependent constants p and q:
These two constants can also be established from vapor pressure data. Similarly,
Lielmezs and Merriman (LM) [64] applied the same a-function to the PR
equation. The present forms of both the LHC and LM equations are incapable of
representing fluid properties above the critical isotherm because a is undefined,
*
that is, T becomes negative and a negative number to the power of a
non-integer, q is not defined mathematically.
a = 1 + (1 - T
r
)(m + n/T
r
)
(2.8)
a = 1.0 + p(T*)
(2.9)
where
(2.9a)
15
Recently Stryjek and Vera [117,118] modified the PR equation by
introducing a more complicated temperature dependent a-function:
a = [ 1 + K<1 - T
r
0
*
5
) ]
2
( 2 1 0 )
K is a complex function of reduced temperature and the acentric factor u:
K- = + [ Ki + K 2 ( K 3 ~ T
r
) ( l " T
r
'
5
) ]
(1 + T
r
'
5
)(0.7 - T
r
)
( 2
.
1 0 a
)
where c
0
is given by
K
0
= 0.378893 + 1.4897153w - 0.17131848a)
2
+ 0.0196544w
3
(2.10b)
and K
L T
K
2
, and *c
3
are substance-dependent constants. In addition to these
three constants, acentric factor, critical temperature and pressure are required.
Stryjek and Vera provided their own set of acentric factors and they claimed
their values are the most accurate and which should be used in conjuction with
their modification. In' other words, in order to achieve the accuracy claimed, four
constants instead of three are now necessary. Both polar and nonpolar compounds
have been tested and vapor pressure predictions have indeed been improved over
the original PR equation. Second virial coefficient calculation from EOS is also
part of their comparison and the performance is of the typical cubic equation of
16
state. More on this subject will be said in the chapter on second virial
coefficient. In the supercritical region, only the fugacity coefficient has been tested
against experimental data. The accurac}' from- the modification for this property
is about the same as the PR equation.
Peneloux et al [85] proposed a
utilizing the volume-translation technique;
correction constant b are corrected by c.
constant are now V and b', respectively:
V = V - c
b" = b - c
modification of the SRK equation by
the specific volume and the volume
The new volume and volume correction
(2.11)
(2.12)
The equation of state after translation becomes
RT a(T)
P = - (o n i
V - b ' (V'+c)(V'+b
,
+2c)
v
;
Peneloux and co-workers have shown that such translation does not alter the
vapor pressure predictions from the original SRK equation. In fact, this shift in
volume by the amount c changes the location of the isotherm and does not
affect the equilibrium condition. Peneloux et al suggested that this constant c be
established from saturated liquid volume at T
r
= 0.7. They claimed that at such
temperature, the prediction gives a balanced representation of the liquid volume
along the saturation curve. They generalized this constant c with the Rackett
[92,113] compressibility factor,
17
RT
c = 0.40768 (0.29441 - Z ^ )
( 2
. i
4
)
This correction should be subtracted from the volume calculated by the SRK
equation. They tested the predictions from their modification for saturated liquid
volume with experimental data for 233 compounds. Both polar and nonpolar
compounds were included. The results from this correction are encouraging.
However, this correction can have adverse effects on vapor volume predictions
because the isotherm has shifted away from the experimental saturated vapor
curve in order to provide a better representation of the saturated liquid curve.
Soave [112] used the same technique for the van der Waals equation and
obtained the constant c at the reduced normal boiling point instead of T
r
= 0.7.
Up to now, the two most popular cubic EOS, SRK and PR, have been
tested mostly with saturation properties. Rarely do comparisons extend to the
supercritical or compressed liquid region. Recently Tannar et al [119] tested their
modification of the LHC [63] equation with the SRK and Lee-Kesler [59]
equations for volumetric calculations in the subcritical and compressed liquid
regions. Comparison of derived or derivative properties such as enthalpy and
entropy of vaporization or departure functions found in the literature are scarce.
In the original work of Peng and Robinson [87], only the enthalpy departure
functions for five compounds were tested against experimental data. Mohanty et
al [79] tested the PR equation for enthalpy of vaporization predictions for
straight chain hydrocarbons from C
x
to C
2 0
. Mihajlov et al [76] compared the
SRK equation along with three other cubic equations for saturation property
predictions for 14 substances. The properties included were vapor pressure,
18
fugacity coefficient and volumes of both vapor and liquid, enthalpy and entropy
of vaporization. Tarakad et al [121] compared eight cubic equations, including the
S RK EOS , for saturated vapor and superheated gas phase density and fugacity
predictions for polar and nonpolar compounds. Probably the most comprehensive
review yet on E OS is the one by Yu et al [129]. Thei r comparison involved 14
cubic equations, both PR and S RK were included, for eight properties of straight
chain hydrocarbons from d to C
1 0
. Reasonable ranges of pressure and
temperature were covered in their stud3'.
Fr om the above discussion, it is clear that a more elaborate comparison
on cubic equations of state with experimental data is necessary. Thus, the effort
of this work is justified.
CHAPTER 3. DEVELOPMENT OF AN EQUATION OF STATE
In the last chapter a number of different approaches have been discussed
in modifying or improving the cubic equation of state. In the first part of this
chapter, some of the fundamental features one must consider in designing an
equation of state is discussed.
The Li el mezs-Merri man equation has briefly been mentioned earlier. The
present form of the L M equation is not capable of representing compound
properties for the region above the critical isotherm. In order to remove this
limitation, the equation must be modified and the second part of this chapter
will be devoted to the formulation of the L M equation for the supercritical
region.
3.1. THEORETICAL CONSIDERATIONS
Equati on of state is most frequently used in vapor-liquid equilibrium
calculations and the conditions of phase equilibrium for a pure substance are
given by Eq.(3.1) to (3.3)
p l
'
p V
(3.2)
G
"
G
(3.3)
These conditions state that at equilibrium, temperature, pressure and Gibbs free
19
20
energy of the two phases, liquid and vapor, must be identical. Eq.(3.3) can
alternatively be expressed in terms of the chemical potential, M
1 v
u = U
(3.4)
and since fugacit.y is related to chemical potential by Eq.(3.5),
u - u = RTln
(3.5)
condition (3.3) can now be written in terms of fugacit}', f
f
1
= f
v
(3.6)
Frequentl y Eq.(3.6) is used as the third condition instead of Eq.(3.3). In order to
accurately reproduce vapor-liquid equilibrium properties from an equation of state,
these three conditions, Eq. (3. 1), (3.2), and (3.6) must be incorporated into the
EOS . The following discussion will show how such a task can be accomplished.
Gibbs free energy is the sum of the Hel mhol tz free energy, A and the
product of pressure, P and volume, V :"
G = A + PV
(3.7)
Since condition (3.3) requires that the Gibbs free energy be equal for the two
phases, or
G
1
- G
V
= 0
(3.3a)
Substituting Eq.(3.7) into (3.3a) yields
A
1
- A
V
+ P
1
V
1
- P
V
V
V
= 0
(3.8)
But A^ - A
V
can be written as
A
1
- A
V
= /
3A
av
dv
(3.9)
and from the fundamental equations of thermodynamics,
3A
3V
= -P
JT
(3.10)
By combining Eq.(3.2), (3.8), (3.9), and (3.10), condition (3.3) can now be
expressed in terms of pressure and volume.
J , PdV + P(V
1
- V
v
) = 0
V
1
(3.11)
where P is the vapor pressure. Eq.(3.11) is shown graphically in Figure 3.1.
The first term of Eq.(3.11) gives the area underneath the isotherm or
22
V
v
J
x
PdV = area 1-2-3-4-5-6-7-1
( 3 1 2 )
and the second term represents the area bound by the isobar P and the isochors
V
1
(=V
7
) and V
V
(=V
6
) or
P(V
V
- v
1
) = area 1-3-5-6-7-1
( 3 1 3 )
But
area 1-2-3-4-5-6-7-1 = area 1-3-5-6-7-1 - area 1-3-2-1
+ area 3-4-5-3
(3.14)
Combination of Eq.(3.13) and (3.14) gives
area 1-3-2-1 = area 3-4-5-3 (3.15)
The volume V^
-
is the saturated liquid and V
v
is the saturated vapor volume.
Eq.(3.11) is formally known as the Maxwell-rule of equal areas.
Since the Maxwell-rule is derived from the conditions of phase equilibrium,
an alternative way of satisfying the three conditions of equilibrium is to
incorporate the Maxwell-rule into the equation of state. The most commonly used
technique to do so is to fit the constant "a" of the attraction pressure term
with vapor pressure data. The reason is that by fitting the constant "a" with
23
experimental vapor pressure data, it is possible to adjust the two areas such
that their sum equals zero. From the Gibbs phase rule:
F = m + 2 -
(3.16)
where F is the number of degrees of freedom, m is the number of components
and 7r is the number of phases, it is clear that there is one degree of freedom
in the saturation region. Therefore an iterative routine must be used to establish
the constant a. The algorithm used is given in a later section of this chapter.
In addition to the Maxwell-rule of equal areas, two other criteria must
also be considered. These two criteria are 1) an equation of state should reduce
to the ideal gas law when pressure approaches zero or
and 2) the EOS should satisfy the thermodynamic stability criteria at the critical
point. From Figure 1.1, it is clear that the critical point is not only an
inflection point, but it is also a point which has zero slope. In other words, at
the critical point, the first and second derivatives of pressure with respect to
volume at constant temperature are zero:
PV
lim
P>0 RT
= 1
(3.17)
3P
= 0
(3.18)
3V
24
9
2
P
av
2
= 0
(3.19)
JT
Another important aspect of an equation of state that must be considered
is the functional dependence of the constants. When the generalized cubic equation
of state, Eq.(1.2) is expanded in terms of density or volume, the following
results (Appendix A):
Z = 1 +
a
b
RT
1
+
V
b
2
+
uab
RT
b
3
+
(w-u
2
)ab
2
RT
1
V
3
(3.20)
The vi ri al equation:
B C D
Z = 1 + - + + +
V V 2 y 3
(3.21)
where B, C, D etc are the second, third, and fourth virial coefficients.
Compari ng the coefficients of Eq.(3.20) and (3.21), the vi ri al coefficients i n terms
of a, b, u, and w are as follows:
B = b -
RT
(3.22)
C = b
2
+
uab
RT
(3.23)
(w-u
2
)ab
2
D = b
3
+
RT
25
(3.24)
Since virial coefficients are functions of temperature only, the constants a, b, u,
and w can either be constant or temperature dependent.
The above criteria are not restricted to cubic equations of state but are
applicable to equations of state in general. In order to have a theoretically sound
EOS, all of the above aspects should be considered.
3.2. LIELMEZS-MERRIMAN EQUATION OF STATE
Lielmezs and Merriman [64] modified the Peng-Robinson equation of state
by defining a new a term of Eq.(2.2a) but retaining the same u and w values
as the PR equation. To express the a term as a continuous, temperature
dependent function along the vapor-liquid equilibrium curve, the previously
* *
proposed T coordinate system by Lielmezs [34,61] was introduced; T is given
by Eq.(2.9a):
* _
T
c
/ T
'
1
where T , T^, T are the critical temperature, normal boiling point temperature,
*
and the state temperature, respectively. This T coordinate system have been
applied to a wide variety of thermodynamic and transport properties. These
include vapor pressure-temperature relation [62], heat of vaporization [34,101],
self-diffusion coefficient [61], thermal conductivity [49], and surface tension [65]
26
correlations.
*
The a term was written in terms of the proposed T as
o(T*) = 1 + p(T*)3
( 2 9 )
where p and q were characteristic constants of the given pure substance. One
advantage of using substance-dependent constants is that the arbitrarily chosen
third parameter such as the acentric factor is eliminated. Another advantage is
that quantum or polar effects are absorbed by the two constants and the
knowledge of the structure or type of compound is not necessary. These two
constants are determined by fitting experimental vapor pressure data to the
a-function wi th the least-squares routine. For the sake of completeness, the
algorithm given by Merri man [75] is shown in Fi gure 3.2. The list of compounds
studied in this work along with the fitted p and q values are given in Table
3.1. Some of the p and q values are different than the ones given previously
by Lielmezs and Merri man [64]. Thi s is due to the fact that different data sets
are used in the curve-fitting.
When experimental data are not available, use of an independent vapor
pressure equation is recommended. In this work, five vapor pressure equations
have been evaluated. These included the Lee-Kesl er ( LK) [59], Riedel [94],
Riedel-Plank-Miller (RPM) [94], Frost-Kal kwarf-Thodos ( FKT) [94], Gomez-Thodos
( GNT) [95] equations. These are all generalized equations and only critical
temperature and pressure, normal boiling point temperature, acentric factor, and
27
molecular weight of the compound are needed as input data. To evaluate these
five equations, they were used to predict vapor pressure of 36 compounds studied
in this work. The predicted results are compared with the experimental data
used to obtain the p and q constants. The comparison is based on root mean
square % error (RMS %)
RMS % Error =
I (% e r r o r )
2
N
(3.25)
where
Experimental - Calculated
% Error = X 100
Experimental
and the results are shown in Table 3.2.
Of the five equations studied, the GNT and FKT equations seem to give
the most reliable results for both polar and nonpolar compounds. However,
pressure is expressed implicitly in the FKT equation and solution must be
obtained by an iteration routine. On the other hand, the GNT equation is a
pressure explicit equation and is simpler to solve. Therefore, the GNT equation is
recommended when no experimental data are available. The conclusion here is
consistent with that of Reid et al [94].
In the course of evaluating different vapor pressure equations, the
parameter, s [28,77,94] :
T, ln(P /P")
s = -=
l-T.
28
(3.26)
br
which is the negative of the slope between the critical point and normal boiling
point of the vapor pressure equation:
lnP
r
= s 1 -
(3.27)
kept on appearing in the literature [9,28,36-38,77,94,95]. The critical pressure is
expressed in atm and P is one atmosphere. Mi l l er [77] suggested that this
parameter s be treated as a fundamental characteristic parameter similar to the
Riedel factor a
Q
[96], the critical compressibility Z
c
, and the acentric factor o.
Giacalone [35] proposed a correlation for estimating the heat of vaporization at
the normal boiling point with the parameter s:
AH , = sRT
vb c
(3.28)
Edmi ster [28] correlated the acentric factor u with s b3
r
the following equation:
u =
1 1 ^
5
~
1 , 0 ( 3
-
2 9 )
Gomez-Nieto and Thodos [36-38] also used this characterization parameter s in
their vapor pressure equations and later Campbel l and Thodos [9] used it in
their saturated liquid density correlation. Since the constants p and q are so
closely related to vapor pressure, a relation might exist between the constants p
and q and the parameter s.
29
To explore this possibility, 34 of the 36 compounds studied here were
divided into three groups according to the nature of the compounds: nonpolar
(Group 1), polar and slightly polar (Group 2), and inert and quantum (Group 3):
Group 1 : Ethane, Propane, n-Butane, i-Butane, n-Pentane, i-Pentane,
Neopentane, n-Hexane, n-Heptane, n-Octane, Benzene,
Ethylene, Propylene, 1-Butene, Nitrogen, Oxygen.
Group 2 : Carbon Monoxide, Carbon Disulfide, Hydrogen Sulfide, Sulfur
Dioxide, Methanol, Ethanol, 1-Propanol, Tertiarj' Butanol,
Water, Ammonia.
Group 3 : Methane, Deuterium, n-Hydrogen, p-Hydrogen, Neon, Argon,
Krypton, Xenon.
Carbon dioxide and acetylene do not have normal boiling points because their
triple point pressures are above one atmosphere, therefore they are not included.
But if they were included, both of them would belong to Group 1. From Table
3.1, one can see that all the q values are relatively constant within each group.
In light of this fact, the constant q of the compound is taken to be the average
value for the group. For Group 1, the average value of q is 0.83, for Group 2,
q is 0.83, and Group 3 compounds take on the value of 0.78.
For the constant p, values vary from 0.01582 to 0.47769. These p values
for the three groups have been plotted against both the parameter s and the
acentric factor o> in Figure 3.3. Also least squares fits have been used to
express p as functions of s and u. The results are as follows:
Group 1 :
p = -1.1977373 + 0.39942704s - 0.026211814s
2
variance = 0.00004469
(3.30)
p = 0.19011333 + 0.67187907a) - 0.85099293a)
2
variance = 0.00004475
30
(3.31)
Group 2
p = -1.8772891 + 0.55686202s - 0.033057663s
2
( 3 3 2
)
variance = 0.00024869
p = 0.1639189 + 1.1372604a) - 1.0533208o>
2
( 3 3 3
)
variance= 0.00023931
Group 3 :
p = 0.34065806 - 0.2439171s + 0.040416063s
2
(3.34)
variance = 0.00003188
p = 0.18425207 .+ 1.1592480a) + 1.8186255a)
2
(3.35)
variance = 0.00041294
For Groups 1 and 2, the correlations of p with s and a) produce approximately
the same variance. For Group 3, the variance produced by fitting p with s is
ten times smaller than for the fit of p with a). This indicates that the
correlation of p with s is stronger than with a). The advantage is more revealing
as is shown in Figure 3.3. Not only does p correlates better with s, the
determination of s is simpler than is o>. Only T , T^, and P
c
are needed for
calculating s and vapor pressure at T
r
= 0.7, P
c
, and T
1 i . i i .
0.0 0.1 0.2 0.3 0.4
Pr
0.5 0.6 0.7
Figure 4.41 Region IV : Reduced enthalpy departure functions of n-heptane
(Tr = 0.67) and methanol (Tr=0.70) versus reduced pressure.
118
10 0
0.5 0.6 0.7 0.8 0.9 1.0
Tr
0.65 0.70 0.75 0.80 0.85 0.90
Tr
Figure 4.42 Region I V : Reduced entropy departure functions of n-heptane
(Pr = 0.73) and methanol (Pr = 0.63) versus reduced temperature.
11.0
10.9
CO
I
CO
Y 10 8
10.7
0 0
10.6
0.0
\
CO
I
CO
V
15.5
15.0
14.5 h
14.0 h
13.5
13.0
Methanol
0 Data
SRK
PR
-
LM
0 0 0
-
0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Pr
Fi gure 4.43 Region IV :
(Tr = 0.67) and methanol (Tr =
Reduced entropy departure functions of n-heptane
0.70) versus reduced pressure.
119
0.51
0.49
0.47
O
E
OS
\
<
I .
<
0.43 -
0 41
0.39
0.0
n-Heptanc
0 Data
SRK
PR
0
0
L M
0
0
I . I . I .
0.2 0.4 0.6
Pr
o
E
OS
<
I
<
1.28
1.26
1.24
1.22
1.20
1.18
-
- ro
0
0
Methanol
0
0 Data
SRK
PR
LM
* - - -
0.8 0.0 0.1 0.2 0.3 0.4
Pr
0.5 0.6 0.7
Figure 4.45 Region IV : Reduced Helmholtz departure functions of n-heptane
(Tr = 0.67) and methanol (Tr = 0.70) versus reduced pressure.
120
3.0
2.0
O
E- 1.0
O
I
o o.o
-1.0
-2.0
-
-
n-Heptane
0 Data
S R K
PR
1
LM
3.0
2.5
2.0
o
E-
QJ
\
G~* 1.5
O
I
O
10
0.5
0.5 0.6 0.7 0.8 0.9 1.0
Tr
0.0
-
Methanol
Data
SRK
PR
1
LM
1 , i . 1...
0.65 0.70 0.75 0.60 0.85 0.90
Tr
Fi gure 4.46 Region IV : Reduced Gibbs energy departure functions of n-heptane
(Pr = 0.73) and methanol (Pr = 0.63) versus reduced temperature.
o
i
o
-0.10
-0.15
-0.20
-0.25
-0.30
0
0.65
0.60
O
E-
\
0.55
O
I
O
0.0 0.2
0.45
0
0
0
Methanol
0 Data
SRK
PR
-I 1 L
LM
_ ^ 1 . __L . 1 .
0.4
Pr
0.6 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Pr
Fi gure 4.47 Region IV : Reduced Gibbs energy departure functions of n-heptane
(Tr = 0.67) and methanol (Tr = 0.70) versus reduced pressure.
121
6.85
6.80 -
O
\
C
ZD
I
2 - 6.70
I
6.65
6.60
es
0
0
nHeptane
0 Data
SRK
PR
LM
0.0 0.2 0.4
Pr
0.6 0.8
O
I
8.8
8.6
8.4
8.2
8.0
7.8
0
0
0 Q
Methanol
0 Data
SRK
PR
LM
1 , L_ 1 I 1 I [ I 1 1
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Pr
Figure 4.49 Region IV : Reduced internal energy departure functions of n-heptane
(Tr=0.67) and methanol (Tr = 0.70) versus reduced pressure.
122
0.7
n-Heptane
0.6 Data
es
SRK
PR
0.5
LM
0.4
0.3
0.2
0.1
0.0
2d=Li i i
0.5
0.4
0.3
0.2
0.1
0.5 0.6 0.7 0.8 0.9 1.0
Tr
0.0
Methanol
eS Data
SRK
PR
LM
0.65 0.70 0.75 0.80 0.85 0.90
Tr
Figure 4.50 Region IV : Fugacity coefficients of n-heptane (Pr = 0.73) and
methanol (Pr = 0.63) versus reduced temperature.
0.25
0.20
0.15
o.io V
0.05
0.7
0.5
0.4
0.3 \
0.2
0.0 0.2 0.4
Pr
0.0
Methanol
-
eS Data
SRK
PR
\
LM
t 1
0.6 0.8 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Pr
Figure 4.51 Region IV : Fugacity coefficients of n-heptane (Tr = 0.67) and
methanol (Tr = 0.70) versus reduced pressure.
123
4.5. INVERSION CURVE
Liquefaction and refrigeration are used extensively in the cryogenic and
natural gas industries. To liquify gases, Joule-Thomson expansion is often used.
Information concerning conditions at which cooling or heating may occur is
essential. When the Joule-Thomson coefficient, u
u =
9P
(4.15)
JH
is positive, cooling occurs and when negative, heating results. The condition at
which the crossover from cooling to heating occurs, is called the inversion
condition, and is defined as
"
= 0
(4.16)
The temperature at which inversion occurs, is called the inversion temperature,
and the corresponding pressure is the inversion pressure. For each isenthalpic
curve on a T-P diagram, there is one inversion point. These points form a
unique line and this locus of points is called the inversion curve. The object of
this section is to compare the inversion curve calculated from the three equations
of state: SRK, PR, and LM with experimental data. A total of nine compounds
are studied. The experimental data sources have been tabulated in Table 4.28.
Along with the experimental data, the predictions from the three EOS are also
compared with the generalized inversion correlation proposed by Gunn, Chueh, and
Prausnitz (GCP) [45]. Their correlation is based on 89 experimental inversion
points of fluids with acentric factors that are close to zero. These fluids include
argon, xenon, nitrogen, carbon monoxide, methane, and ethylene. The GCP
124
correlation is given below:
P
r
= - 36. 275 + 71. 598T
r
- 41. 567T
r
2
+ 11. 826T
r
3
- 1. 6721T
r
* + 0. 091167T
r
5
( 4 1 ? )
The three inversion curve equations derived from the SRK, PR, and LM
equations are given in Appendix A. For ease of reference, the inversion equation
derived from the generalized cubic equation of state, Eq.(1.2) is given below:
C^r+uS+w)
2
+ C
2
$ ( 2 $ +u ) ( - l )
2
+ c
3
( $- l )
2
( $
2
+u$+w) = 0
(4.18)
where
Ci = % T
r
(4.18a)
2 =
"
fl
a
a
(4.18b)
C
3
= 0.0
a* (4.18c)
(4.18d)
S = V/b
The reduced inversion pressure is given by the following:
RT 1 a 1
o _ r a
p
r " ~ 7777 TT7 77777777 (4.19)
(5-1) 0^ ($
2
+u$+w)
To calculate the inversion pressure at a given reduced temperature, the
dimensionless volume, $ is solved from Eq.(4.18) and is substituted along with
the given reduced temperature into Eq.(4.19).
125
The results of the nine compounds tested are summarized in Table 4.28.
Since the GCP correlation is only applicable for simple compounds, that is
compounds with small acentric factor, no estimations are given for propane,
n-butane, carbon dioxide, p-hydrogen, and ammonia. The comparison is based on
RMS % error. From Table 4.28, it is clear that the SRK equation is more
reliable than the PR or LM equation. The SRK equation gives the lowest RMS
% error for most compounds. The large RMS % errors for carbon monoxide
reported for the PR and SRK equations are due to the points that are near the
high temperature ideal gas region at which the inversion pressure is close to
zero but both the PR and SRK equations give large negative inversion pressures.
There are two points on the inversion curve which can be used to
characterize any fluid. These are the maximum inversion pressure and the
maximum inversion temperature. Miller [78] has suggested a reasonable
approximation for locating these two points. For compounds which follow the
corresponding states principle such as hydrocarbons, the maximum inversion
temperature is about 5 times the critical temperature and the maximum inversion
pressure is about 11.75 times the critical pressure at the reduced temperature of
2.25. The values of the maximum reduced inversion pressure, P
r
(max), and
maximum reduced inversion temperature, T
r
(max), from the three equations of
state have been tabulated in Table 4.29. On the average, the maximum reduced
inversion pressure from the SRK equation is 12.05, PR is 13.40, and from LM
is 14.25. The maximum reduced inversion temperature is 4.22 for the SRK
equation, 5.25 for the PR equation, and 5.80 for the LM equation. Therefore, in
general, the SRK equation tends to give the closest maximum inversion pressure
126
and the PR equation gives the closest maximum inversion temperature. The LM
equation overpredicts both the maximum inversion pressure and temperature for
all compounds considered in this study.
Figures 4.52 to 4.60 show the inversion curves of the nine compounds
calculated from the three EOS. In the case of fluid which has small acentric
factor, the curve calculated from the GCP correlation is also included in the plot.
For the low temperature region, all equations give similar predictions. None of
the EOS, however, is capable of giving reasonable representation at the high
temperature region. It should be pointed out that the LM equation is not a
continuous function because it is not defined at T
r
=1.0. Therefore a discontinuity
appears on all the graphs at T
r
= 1.0. It is safe to say that the behavior of the
three equations is identical at reduced temperature below 1.5. The SRK equation
is probably the most accurate among the three EOS right up to the maximum
inversion pressure. This observation is consistent with the findings of Dilay and
Heidemann [24]. The PR and LM equations are not recommended for inversion
curve predictions above Tr=1.5.
Table 4.28 Inversion Pressure RMS % Error
Compound SRK PR LM GCP Tr Range N Ref .
Methane 3.680 12.58 10.72 1 .242 1 .102-2 .465 27 25
Propane 42 . 37 46 . 50 45.97 0 .811-2 . 189 52 39
n-Butane 24.92 30. 18 30. 13 0. .823-2 .023 52 48
Carbon Monoxide 3629 . 6704 . 119.9 23 . 79 1. . 528-5 .064 14 25
Carbon Dioxide 17 .07 22 . 39 21 .29 0. 832-1 . . 391 18 25
E thy 1ene 8.912 17. 20 13.65 4.900 1. 053-1. 495 6 23
p-Hydrogen 50.98 74.67 74 . 14 0. 849-5 459 16 69
Ammon i a 108.8 106 . 2 101 .0 0. 792-1. . 233 20 25
Argon 20. 12 26.65 25.20 5.615 0. .862-1. 989 18 40
Table 4.29 Maximum Reduced Inversion Pressure and Temperature
SRK PR LM
Compound Pr(max) Tr Tr(max) Pr(max) Tr Tr(max) Pr(max) Tr Tr(max)
Methane 11 . . 79 2 . 15 4 .40 13 .08 2 . 29 5. 38 13 . 34 2 . 58 6 . 04
Propane 12 , . 12 1 .89 3 .61 13 .42 1 . 98 4 . '21 14 . . 53 2 .31 4 . 73
n-Butane 12 . . 24 1 .84 3 .45 13 .55 1 . ,91 3. 99 15 . ,08 2 . 26 4 . 45
Carbon Monoxide 11 .87 2 .06 4 . 14 13 . 16 2 . 18 4 . 97 13 .60 2 . 49 5 . 61
Carbon Dioxide 12 . . 33 1 .80 3 . 34 13 .65 1 . 87 3 . 83 15 .68 2 . 24 4 . 28
E thy 1ene 1 1 . .95 1 .99 3 . 93 13 . 24 2 . 10 4 . 67 13 .82 2 . 4 1 5 . 26
p-Hydrogen 1 1 . 95 3 .09 7 .39 13 .70 3, , 74 10.97 13. .61 3 .83 1 1 . 37
Ammon i a 12 . 41 1 . 77 3 . , 26 13 . 74 1 . ,84 3. 72 15 . 27 2 . 19 4 . 24
Argon 1 1 . 77 2 . 17 4 . .49 13 .07 2 . , 33 5 . 52 13 . . 33 2 . .62 6 . 20
7.0
Methane
o . o I ' i i ' 1 ' 1 i 1 1
1
1
I
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
Reduced Pressure, Pr
Figure 4.52 Inversion curve of methane calculated from the S RK, PR, L M , and
GCP equations.
oo
5.0
0.0 I 1 I > 1 . I ; 1
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
Reduced Pressure, Pr
Figure 4.53 Inversion curve of propane calculated from the SRK, PR, and LM
equations.
to
CO
Figure 4.54 Inversion curve of n-butane calculated from the SRK, PR, and LM
equations.
C O
o
6.0
Carbon Monoxide
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0
Reduced Pressure, Pr
Figure 4.55 Inversion curve of carbon monoxide calculated from the SRK, PR,
LM, and GCP equations.
5.0
Carbon Di oxi de
A Expt.
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0
Reduced Pressure, Pr
Figure 4.56 Inversion curve of carbon dioxide calculated from the SRK, PR, and
LM equations.
co
to
Fi gure 4.57 Inversion curve of ethylene calculated from the S RK, PR, L M , and w
GCP equations.
w
Fi gure 4.58 Inversion curve of p-hydrogen calculated from the S RK, PR, and L M
equations.
co
Fi gure 4.59 Inversion curve of ammoni a calculated from the SRK, PR, and L M
equations.
co
cn
Fi gure 4.60 Inversion curve of argon calculated from the SRK, PR, LM, and w
GCP equations.
0 5
CHAPTER 5. SECOND VIRIAL COEFFICIENT
5.1. INTRODUCTION
To account for vapor phase nonideality, the fugacity coefficient is
necessary in phase equilibria calculations. Often when the pressure is moderate,
density is less than half the critical, the virial equation truncated after the
second term provides excellent estimate of the vapor phase fugacity coefficient.
For normal fluids, probably the most popular and accurate correlation for
calculating the second virial coefficient is the method suggested by Tsonopoulos
[123]. This correlation is a modification of the Pitzer-Curl [90] equation which is
based on the three-parameter corresponding states principle utilizing the acentric
factor as the third parameter for improving the accuracy of the prediction. The
Tsonopoulos correlation gives the reduced second virial coefficient as the sum of
two terms:
= f ( o ) + cjf ( 1 )
(5.1)
where f<> and f'
1
' are given by Eq.(5.1a) and (5.1b)
0.330 0.1385 0.0121 0.000607
f") = 0.1445 -
T
r
rp s
r
(5.1a)
137
138
0.331 0.423 0.008
f'
1
' = 0.0637 +
(5.1b)
The f <
0
) represents the reduced second virial coefficient of a simple fluid which
has zero acentricity and f <
1
> is the correction term for a normal fluid. The
coefficients of f <
0
' and f <
1
' were found by curve-fitting with experimental data.
Argon, krypton, and xenon data were used b3' Pitzer and Tsonopoulos to obtain
the function f <
0
>. However, both authors did not specify which compounds were
used for the f <
1
> function. The data sources for the Tsonopoulos correlation were
taken from the 1969 compilation of second virial coefficient by Dymond and
Smith [26].
Recently, Yu et al [129] and Adachi et al [1] have studied a number of
cubic equations of state for pure compound property predictions. These two
studies have shown that none of the currently popular cubic equations of state
are capable of giving accurate second virial coefficients except the equation
proposed by Kubic [57]. The Kubic equation uses the Tsonopoulos correlation to
determine one of its constants. Although the Kubic equation gives excellent second
virial coefficient prediction, it fails to give reasonable approximations for the more
fundamental properties such as vapor pressure and densities. Martin [73]
proposed a modified Martin [72] equation specifically designed for second virial
coefficient calculation. To determine the three required substance-dependent
constants, the slope of the vapor pressure curve at the critical point,
experimental second virial coefficients, and critical constants must be known.
139
5.2. THEORETICAL BACKGROUND
As it was pointed out in the previous section, most of the cubic equations
of state do not give accurate second virial coefficient predictions for the entire
temperature range. However, it is generally true that predictions from the cubic
equation of state are acceptable for reduced temperature above 0.8. Below this
temperature the results deviate greatly from the experimental data. To
understand why the cubic equation fails to give good representation of the second
virial coefficient, it is essential to look at how the virial coefficients are
represented by the cubic EOS.
The compressibility of a gas can be expanded in Taylor series either in
inverse molar volume or pressure, that is :
B C
2 = 1 +
v
+
^
+
" *
( 5
-
2 )
or
Z = 1 + B'P + CP
2
(5.3)
Eq.(5.2) is called the Leiden form whereas Eq.(5.3) is the Berlin form of the
virial equation. The coefficients B, B'... and C, C... are the second and third
coefficients, respectively. The two sets of virial coefficients, B and B' and C and
C, can be shown [91] to be related to each other by Eq.(5.4a) and (5.4b)
B = B'RT
C = C (RT)
2
+ B
2
140
(5.4a)
(5.4b)
For pure gas, these coefficients are functions of temperature only. If the
generalized cubic equation of state, Eq.(1.2):
RT a(T)
V-b V
2
+ubV+wb
2 ( 1
'
2 )
is expanded in inverse molar volume similar to Eq.(5.2), it can be shown that
the second virial coefficient is given by Eq.(5.5)
a(T)
B = b
RT
where
!Zc
=
_
a (
V
p
c p
RT RT R
2
T
2
C C C
where
(5.5)
a(T) = a(T
c
) a
( 5 6 )
The derivation of Eq.(5.5) is shown in the Appendix A. In reduced form:
(5.7)
F = a/T
r
(5.8)
141
It was shown by Shaw and Lielmezs [105] that for most cubic EOS , the
F-function can be written as a power series in inverse reduced temperature. For
example, the F-function of the Soave-Redlich-Kwong [110] and Peng-Robinson [87]
equations are given by Eq.(5.9)
C
2
C
3
F = Ci + +
7
(5
-
9)
r r
The constants C
1 }
C
2
, and C
3
are functions of the acentric factor. If we
compare the functional dependence of F on reduced temperature with that of the
Tsonopoulos correlation, it can readily been seen why the cubic EOS gives such
poor results in the low temperature region. Below T
r
= 0.8, the slope of the
second virial coefficient increases rapidly with decreasing temperature (see Fi gures
5.1 and 5.2). Wi t h the dependence on reduced temperature of power of -1, it is
not possible to include the rapid changes in the low temperature region.
Since the data sets used by Tsonopoulos were materials found before
1968, and the 1980 compilation by Dymond and Smi t h [27] has included
material published up to earl y 1979, an improved correlation was thought to be
possible with these new and more accurate experimental data and this prompted
the present investigation.
5.3. A NEW CORRELATION
The present proposed correlation is based on the PR form cubic EOS :
142
RT a(T)
p =
(5.10)
V-b V
2
+2bV-b
2
Here, the a(T) function is modified such that it gives an accurate second virial
coefficient. The second virial coefficient from a cubic equation of state is given
earlier by Eq.(5.5)
a(T)
B = b
RT
where
a(T) = a ( T J a(T
r
, o)
R
2
T
2
R
2
T
2
a(T) = Q
Q
= 0.45724
c c
RT RT
b = b(T ) =0. ^ = 0.07780 *
P P
c c
(5.5)
Substituting a(T) and b into Eq.(5.5) and simphfying, the reduced second virial
coefficient is
BP
C
a
= % ~
fl
a (5.11)
Rearrange Eq.(5.11), a is then given by Eq.(5.12)
143
a J
where
(5.12)
B =
r
BP
RT
(5.13)
With experimental B
r
, a can now be calculated at each reduced temperature.
It is proposed here that the a-function for normal fluid be given by the
following expression:
a = a<> + wa'
1
' (5.14)
where a
( 0
' represents a of simple fluids and a
(1
' corrects for the deviation of
normal fluid from the simple fluid. The functional form of Eq.(5.14) is based on
the three-parameter theorem of corresponding states with the acentric factor as
the third parameter. In this work, all acentric factors of simple fluids are
assumed to be zero even though the actual values are slightly different than the
assumed. For example, argon has an acentricity of -0.004.
To obtain the functional dependence of a
(01
on reduced temperature, the
calculated a values of the simple fluids are curve-fitted with a power series. The
minimum variance is the criterion for determining the number of terms in the
power series. Once an expression for a <
0
' is obtained, a
( 1
' can then be
144
calculated from the following expression:
a - a<>
a
( i > =
( 5
CO
where a is calculated from Eq.(5.12). The final expressions for a
( 0 )
and a'
1
'
are given by Eq.(5.16) and (5.17):
14.360017 45.000285 78.907097
a<> = -1.4524905 + +
T T
2
T
3
r r r
79.449258 45.841959 14.078304 1.7835426
+ _ + .
V 4 T
5
T I T 1
r r r r
(5.16)
15.205023 20.874489 12.697209
a'
1
' = -4.3816022 + +
rp m
2
T 3
r r r
2.5851948
T 4
(5.17)
r
A total of 57 experimental data points are involved in the least-squares fit of
a
( 0 )
and the variance resulted is 0.0003175. The compounds included in the
curve-fit are neon, argon, krypton, and xenon. The expression for a <
1
' is
obtained from 124 points with a variance of 0.01041. The compounds included
are propane, n-butane, n-pentane, isopentane, neopentane, n-hexane, n-heptane,
n-octane, benzene, ethylene, and propylene. All experimental data are taken from
145
the Dymond and Smith [27] 1980 compilation of second virial coefficients.
5.4. COMPARISON
The second virial coefficients of 24 normal and simple fluids have been
calculated from the proposed correlation, Eq.(5.14). In order to test its validitj',
the results are compared with the Tsonopoulos correlation and four other cubic
EOS. The cubic EOS included are the Soave-Redlich-Kwong, Peng-Robinson,
Lielmezs-Merriman, and Kubic. The results have been tabulated in Table 5.1 and
it has clearly shown that none of the cubic EOS is capable of representing the
second virial coefficient for a wide range of temperature except the Kubic
equation with reasons mentioned earlier. Figures 5.1 and 5.2 have the results of
n-butane, benzene, n-octane, and- argon. Since the LM equation gives almost
identical results as the PR equation, the LM equation is not included in the
figures. Similarly, the Kubic equation is partly derived from the Tsonopoulos
correlation. It also is not included.
All the cubic EOS considered here give satisfactory results for T
r
greater
than 0.8 but below this temperature, the predictions deviate greatly from the
experimental data. The proposed correlation gives excellent agreement for the
entire range of temperatures studied. For example, the average absolute deviation
of benzene from the proposed correlation is about one-third that of the
Tsonopoulos correlation and for n-heptane, it is about one-half. The average
estimated deviations of the experimental data shown in Table 5.1 are taken from
Dymond and Smith [27]. For the majority of the compounds studied, the
146
deviations from both the Tsonopoulos and the present proposed correlations are
well within the experimental uncertainties. Even though the Tsonopoulos
correlation gives better predictions than the proposed correlation for a number of
compounds, the differences are small. The SRK equation seems to give slight^
better prediction than the PR equation for T
f
greater than 0.8. Figures 5.1 and
5.2 show that the SRK predictions lie right on the experimental data whereas
the predictions from the PR equation are slightty off.
Table 5.1 Second V i r i a l C o e f f i c i e n t : Average Absolute Deviation (cc/mole)
Compound SRK PR LM Kub i c This
Work
Tsono-
polous
Deviation Tr Range
Est imates
N
Methane 6.6 1 1 . 2 12 . 1 1 .6 1 .4 0. 8 2 . 6 0 . 58--3 . . 15 16
E thane 5.9 15 . 4 16 . 6 8.6 5 . 8 5 . 7 2.4 0 . 65--1 .96 15
Propane 18. 1 24. 6 25. 0 10. 3 6 . 1 5.8 12.3 0 . 65--1 , .49 15
n-Butane 74 . 1 69 . 0 67 . 2 9.2 11.8 9.6 18.2 0 . 59- - 1 . 32 17
i-Butane 41. 2 43. 9 44 . 2 22. 4 24 . 7 24 . 5 --- 0 .67- -1 . 25 1 1
n-Pentane 78 . 2 67 . 0 66 . 6 14.2 10.9 11.9 28 . 3 0 .64- -1 . 17 12
i-Pentane 103.3 72. 9 72. 0 15.2 20. 0 21. 3 27 . 1 0. .61- -o. 98 7
Neopentane 33 .6 37 . 3 38 . 1 9.9 3 . 5 3.0 17.0 0. .69- -1 . 27 10
n-Hexane 205 . 3 176 . 5 177 . 2 28. 4 19.0 14.1 27 . 2 0. 59- -0. 89 9
n-Heptane 212.1 180 ' .7 180 .9 46. 4 15.0 29 . 1 --- 0. . 56- -1 . 30 12
n-Octane 376. 3 341 .0 339 .9 84 .4 30. 5 51. 5 0. 53- -1 . 23 12
Benzene 164 . 3 147 .9 144 .8 61.1 17.3 60. 1 16 .9 O. 52- -1 . 07 13
Carbon Monoxide 3.9 5.5 7.6 1 . 1 1 .5 0. 5 --- 2 . 05- -3 . 18 7
Carbon D i s u l f i d e 154 .0 135 .4 131 .3 26. 6 39. 0 28. 8 20. 0 0. 51- -0. 79 9
Hydrogen Sulfide 15.8 21 . 4 21 . 9 16.5 16 . 1 16 . 2 --- 1 . 00- -1 . 32 7
Ethylene 1 .7 15. 9 17. 0 2.6 0. 4 0. 4 1 . 2 0. 85- -1 . 59 9
Propy1ene 6. 1 18 . 8 19 . 4 5.8 1 .5 1 . 1 4. 8 0. 77- -1 . 37 8
1-Butene 13.6 23. 0 23. 6 14.7 6.6 7 . 3 0. 72- -1 . 00 12
N i t rogen 4.8 8.3 8. 0 4. 9 3 . 9 3.9 2.4 0. 60- -5 . 56 14
Oxygen 5.7 9 . 2 9.9 3. 1 2 . 1 2 . 2 3.8 0. 58- -2 . 58 1 1
Neon 1 .6 2.7 4 . 1 1 .5 0. 5 1 . 5 1 .0 1 . 35- -13.51 10
Argon 8. 0 8.9 9. 0 2.8 0. 7 1 . 7 1 . 4 0. 54- -6 . 63 18
Krypton 16. 1 15. 7 15 . 8 2 . 2 1 .9 2. 0 3.2 0. 53- -3 . 34 14
Xenon 12. 1 15. 8 16. 2 3.8 1 . 4 2.2 4. 8 0. 55- -2 . 24 16
148
Figure 5.1 Second virial coefficients of n-butane and benzene versus reduced
temperature.
149
SRK PR
o EXPT THIS WORK __T?P!i
oo
0.4 0.6 0.8 1.0 1.2 1.4
Reduced Temperature, Tr
0.5
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
Reduced Temperature, Tr
Figure 5.2 Second virial coefficients of n-octane and argon versus reduced
temperature.
CONCLUSIONS
The Lielmezs-Merriman equation of state and its modification proposed in
this work have been successfully applied to calculate various physical and
thermodynamic properties of pure compounds in different regions on the PVT
surface. The predictions from the LM equation and the modification of it have
been compared with experimental data and with the predictions from two other
equations of state: Peng-Robinson and Soave-Redlich-Kwong.
Along the saturated vapor-liquid equilibrium curve, it has been found that
the LM equation is the most accurate equation for predicting vapor pressure and
saturated liquid and vapor volumes. However, for the two derivative properties:
enthalpj' and entropy of vaporization, the PR equation is the more accurate
equation.
In the subcritical region, the predictions from the three equations are very
similar and very small differences have been observed. In the region where
T
r
<0.95, the PR and LM equations are practically identical in accuracy for most
of the properties tested. However, the LM equation is not recommended for
calculating derivative properties such as departure functions in the region
0.95<T
r
<1.0.
For the supercritical region, the predictions of pressure and volume from
the PR and the proposed modification of the LM equation are again very similar
in accuracy. The overall average RMS % error from the SRK equation is about
150
151
twice that of the PR or modification of the L M equation. For temperature
prediction, all three equations are similar in accuracy. For derivative properties,
the modification of the L M equation gives comparable prediction to the PR
equation in the region where T
r
>1 . 0 5 . The S RK and PR equation are similar in
accurac.y for enthalpy, entropy, and internal energy departure function estimations.
The S RK equation gives the lowest overall average RMS % error for the
Joul e-Thomson coefficient prediction.
In the compressed liquid region, it has been found that the overall
average deviation or RMS % error is the highest among the three single phase
regions in each of the properties studied. None of the three equations is
adequate for calculating pressure. For temperature and volume predictions, the PR
and L M equations give similar results. These two equations are more accurate
than the S RK equation. For derivative properties, predictions from the three EOS
are comparable.
For inversion curve calculation, the most reliable equation of state is the
SRK. Its estimation closely follows the experimental data right up to the
maxi mum inversion pressure. The predictions from all three equations are alike
up to T
r
= 1.5.
None of the three equations of state tested i n this work is capable of
calculating the second vi ri al coefficient accurately for the region below T
r
= 0.8.
The proposed a-function especially designed for second vi ri al coefficient calculation
has been found to be equal, if not better, in accuracy than the Tsonopoulos
152
correlation. The proposed a-function is based on the latest and the most accurate
experimental data sources. The deviation from the proposed correlation has been
found to be well within the estimated uncertainty given in the data sources.
As shown throughout this work, seldom does an equation of state perfom
equally well in every region. Among the three equation of state studied, the
preferred equation varies, depending on what region or what type of compound is
considered. One equation may be particularly suitable for representing saturated
properties. Another may be superior in representing PVT data in the single
phase region. However, none of the three equations studied in this work can
satisfactorily represent all properties over an extended range of conditions.
RECOMMENDATIONS FOR FURTHER STUDY
1. There remain many compounds such as heavy hydrocarbons and alcohols for
which the p and q constants for the Lielmezs-Merriman equation of state
have not been developed. Further work in the development of these
constants is necessary.
2. Since the Peng-Robinson and Lielmezs-Merriman equations are so similar in
accuracy for a number of properties studied, similar work should be done
for the Soave-Redlich-Kwong and the Lielmezs-Howell-Campbell equations to
see if similar conclusions to those found in this work can be drawn.
3. The volume-translation technique has been steadily gaining popularity since
its introduction by Peneloux et al [85]. Perhaps, such a technique can be
incorporated into the LM equation for improving the volume prediction.
4. The proposed a-function designed for second virial coefficient calculation can
only apply to nonpolar compounds. Further development of the proposed
a-function for polar compounds is essential.
5. Since the proposed a-function for calculating second virial coefficient is based
on the Peng-Robinson equation and it was arbitrarily chosen, a similar
a-function based on the Soave-Redlich-Kwong equation can also be correlated.
6. So far only pure compound properties have been studied, binary and
multicomponent mixtures should be studied in the future.
153
NOMENCLATURE
a parameter in the attraction pressure term of the cubic equation of
state defined by Eq.(2.2a)
A molar Helmholtz free energy
A dimensionless constant defined by Eq.(A.30)
AAD average absolute deviation defined by Eq.(4.3)
b parameter in the cubic equation defined by Eq.(2.2c)
b' translated parameter in the cubic equation defined by Eq.(2.12)
B second virial coefficient
B dimensionless constant defined by Eq.(A.31)
c volume correction constant
C third virial coefficient
Cp isobaric heat capacity
D fourth virial coefficient
f fugacity
F number degrees of freedom
F function defined by Eq.(5.8)
G molar Gibbs free energy
H molar enthalpy
AH
V
enthalpy of vaporization
m constant in Eq.(2.3)
m number of components
MW molecular weight
n constant in Eq.(2.8)
154
155
N number of data points
NC number of compounds
p substance-dependent parameter in the Lielmezs-Merriman equation of
state
P pressure
q substance-dependent parameter in the Lielmezs-Merriman equation of
* state
R universal gas constant
RMS root mean square defined by Eq.(3.25)
s characterization parameter defined by Eq.(3.26)
S molar entropy
AS
y
entropy of vaporization
T absolute temperature
*
T dimensionless temperature coordinate defined by Eq.(2.9a) and (3.39)
u generalized cubic equation of state parameter
U molar internal energy
V molar volume
V translated volume defined by Eq.(2.11)
w generalized cubic equation of state parameter
x mole fraction
Z compressibility factor
Zj ^ Rackett compressibility factor
Subscripts
A attraction
156
b normal boiling point
c crtiical property
r reduced property
R repulsion
Superscripts
1 liquid phase
v vapor phase
reference state: ideal gase state
Greek Symbols
a temperature dependence of the parameter a in the cubic equation of
state
K constant defined by Eq.(2.10a)
u chemical potential
u Joul e-Thomson coefficient
co acentric factor
7r number of phases
O
a
coefficient of a defined by Eq.(2.2a)
coefficient of b defined by Eq.(2.2c)
$ dimensionless volume defined by Eq.(4.18d)
REFERENCES
1. Adachi, Y., Lu, B.C.-Y., Sugie, H., Fluid Phase Equilibria 13, 133 (1983).
2. Adler, S.B., Spencer, C.F., Ozkardesh, H., Kuo, C.-M., Phase Equilibria and
Fluid Properties in the Chemical Industry, ACS Symp. Ser. 60, Ed. Storvick,
T.S., Sandler, S.I., (1977), Chapter 7.
3. Alder, B.J., Young, D.A., Mark, M.A., J. Chem. Physics 56, 3013 (1972).
4. Angus, S., Armstrong, B., de Reuck, K.M., International Thermodynamic
Tables of the Fluid State - 5 Methane, 1st ed., Pergamon Press, Oxford
(1978).
5. Beattie, J.A., Bridgeman, O.C., Proc. Am. Acad. Arts Sci. 63, 229 (1928).
6. Benedict, M., Webb, G.B., Rubin, L.C., J. Chem. Phys. 8, 334 (1940).
7. Benedict, M., Webb, G.B., Rubin, L.C., J. Chem. Phys. 10, 747 (1942).
8. Boublik, T., Molecular Physics 42, 209 (1981).
9. Campbell, S.W., Thodos, G., Ind. Eng. Chem. Fundam. 23, 500 (1984).
10. Canjar, L.N., Manning, F.S., Thermodynamic Properties and Reduced
Correlations for Gases, Gulf Publishing Company, Houston (1967).
11. Chakma, A., Lielmezs, J., Islam, M.R., Thermochimica Acta 111, 311
(1987).
12. Chaudron, J., Asselineau, L., Renon, H., Chem. Eng. Sci. 28, 839 (1973).
13. Chen, S.S., Kreglewski, A., Ber. Bunsen. Ges. 81, 1048 (1977).
14. Chueh, P.L., Prausnitz, J.M., AIChE J. 13, 1099 (1967).
15. Chueh, P.L., Prausnitz, J.M., Ind. Eng. Chem. Fundam. 6, 492 (1967).
16. Conrard, P.G., Gravier, J.F., Oil & Gas J. 78, 77 (1980).
17. Cosner, J.L., Gagliardo, J.E., Storvick, T.S., J. Chem. Eng. Data 6, 360
(1961).
18. Das, T.R., Reed, CO., Eubank, P.T., J. Chem. Eng. Data 18, 244 (1973).
19. Das, T.R., Reed, CO., Eubank, P.T., J. Chem. Eng. Data 18, 253 (1973).
20. Das, T.R., Reed, CO., Eubank, P.T., J. Chem. Eng. Data 22, 3 (1977).
157
158
21. Das, T.R., Reed, CO., Eubank, P.T., J. Chem. Eng. Data 22, 9 (1977).
22. Das, T.R., Reed, CO., Eubank, P.T., J. Chem. Eng. Data 22, 16 (1977).
23. De Groot, S.R., Geldermans, M., Physica 8, 538 (1947).
24. Dilay, G.W., Heidemann, R.A., Ind. Eng. Chem. Fundam. 25, 152 (1986).
25. Din, F., Thermodynamic Function of Gases, Butterworth, London (1961).
26. Dymond, J.H., Smith, E.B., The Virial Coefficients of Gases : A critical
Copmpilation, Clarendon Press, Oxford (1969).
27. Dymond, J.H., Smith, E.B., The Virial Coefficients of Pure Gases and
Mixtures: A Critical Compilation, Oxford University Press, Oxford (1980).
28. Edmister, W.C., Petroleum Refiner 37(4), 173 (1958).
29. Edmister, W.C, Lee, B.I., Applied Hydrocarbon Thermodynamics , Vol. 1,
2nd Ed., Gulf Publishing Company, Houston (1984).
30. Eubank, P.T., Adv. Cryog. Eng. 17, 270 (1971).
31. Eubank, P.T., Fort, B.F., Reed, CO., Adv. Cryog. Eng. 17, 282 (1971).
32. Eveiein, K.A., Moore, R.G., Ind. Eng. Chem. Process Des. Dev. 18, 618
(1979).
33. Firoozabadi, A., Hekim, Y., Katz, D.L., Can J. Chem. Eng. 56, 610
(1978).
34. Fish, L.W., Lielmezs, J., Ind. Eng. Chem. Fundam. 14, 248 (1975).
35. Giacalone, A., Gazz. chim. ital. 81, 180 (1951).
36. Gomez-Nieto, M., Thodos, G., Can. J. Chem. Eng. 55, 445 (1977).
37. Gomez-Nieto, M., Thodos, G., Ind. Eng. Chem. Fundam. 16, 254 (1977).
38. Gomez-Nieto, M., Thodos, G., Ind. Eng. Chem. Fundam. 17, 45 (1978).
39. Goodwin, R.D., Haynes, H.M., Thermodynamic Properties of Propane from 85
to 700 K at Pressure to 70MPa, NBS Monograph 170, U.S. Dept. of
Commerce (1982).
40. Gosman, A.L., McCarty, R.D., Hust, J.G., Thermodynamic Properties of
Argon from the Triple Point to 300 K at Pressures to 1000 Atmospheres,
NSRDS-NBS 27, U.S. Dept. of Commerce (1969).
41. Graboski, M.S., Daubert, T.E., Ind. Eng. Chem. Process Des. Dev. 17, 443
159
(1978) .
42. Graboski, M.S., Daubert, T.E., Ind. Eng. Chem. Process Des. Dev. 18, 300
(1979) .
43. Gray, R.D., Heidman, J.L., Hwang, S.C., Tsonopoulos, C, Fluid Phase
Equilibria 13, 59 (1983).
44. Grigull, TJ., Straub, J., Schiebener, P., Steam Tables in Si-Units
Wasserdampftafeln, 2nd Ed., Springer-Verlag, Berlin (1984).
45. Gunn, R.D., Chueh, P.L., Prausnitz, J.M., Cryogenics 6, 324 (1966).
46. Halm, R.L., Stiel, L.I., AIChE J. 13, 351 (1967).
47. Hamam, S.E.M., Chung, W.K., Elshayal, I.M., Lu, B.C.-Y., Ind. Eng.
Chem. Process Des. Dev. 16, 51 (1977).
48. Haynes, W.M., Goodwin, R.D., Thermophysical Properties of Normal Butane
from 135 to 700 K at Pressures to 70 MPa, NBS Monograph 169, U.S.
Dept. of Commerce (1982).
49. Herrick, T.A., Lielmezs, J., Thermochimica Acta 84, 41 (1985).
50. Joffe, J., Schroeder, G.M., Zudkevitch, D., AIChE J. 16, 496 (1970).
51. Kennedy, E.R., Sage, B.H., Lacey, W.N., Ind. Eng. Chem. 28, 718 (1936).
52. Kreglewski, A., Wilhoit, R.C., Zwolinski, B.J., J. Chem. Eng. Data 18, 432
(1973).
53. Kreglewski, A., Chen, S.S., Equations of State in Engineering and Research,
Adv. Chem. Ser. 182, Ed. Chao, K.C., Robinson, R.L., Chapter 11 (1977).
54. Kreglewski, A., Chen, S.S., J. Chimie Physique 75, 347 (1978).
55. Kreglewski, A., J. chimie Physique 77, 441 (1980).
56. Krone, L.H., Johnson, R.C., AIChE J. 2, 552 (1956).
57. Kubic, W.L., Fluid Phase Equilibria 9, 79 (1982).
58. Law, S., Lielmezs, J., Thermochimica Acta 84, 71 (1985).
59. Lee, B.L, Kesler, M.G., AIChE J. 21, 510 (1975).
60. Leiva, M.A., Sanchez, J., Chemical Engineering Thermodynamics, Ed.
Newman, S.A., Ann Arbor Science Publishers, Michigan, Chapter 19 (1983).
61. Lielmezs, J., Z. Phys. Chem. NF (Frankfurt am Main) 91, 288 (1974).
160
62. Lielmezs, J., Astley, K.G., McEvoy, J.A., Thermochimica Acta 52, 9 (1982).
63. Lielmezs, J., Howell, S.K., Campbell, H.D., Chem. Eng. Sci. 38, 1293
(1983).
64. Lielmezs, J., Merriman, L.H., Thermochimica Acta 105, 383 (1986).
65. Lielmezs, J . , Herrick, T.A., Chem. Eng. J. 32, 165 (1986).
66. Lin, C.T., Daubert, T.E., Ind. Eng. Chem. Process Des. Dev. 19, 51
(1980).
67. Lira, R., Malo, J.M., Leiva, M., Chapela, G., Chemical Engineering
Thermodynamics, Ed. Newman, S.A., Ann Arbor Science Publishers,
Michigan, Chapter 14 (1983).
68. Lu, B.C.-Y., Hsi, C, Poon, D.P.L., Chem. Eng. Prog. Symp. Ser. 70(140),
56 (1974).
69. McCarty, R.D., Hord, J., Roder, H.M., Selected Properties of Hydrogen, NBS
Monograph 168, U.S. Dept. of Commerce (1981).
70. Mak, P.C.-N., B.A.Sc. Thesis, UBC (1985).
71. Martin, J.J., Ind. Eng. Chem. 59, 34 (1967).
72. Martin, J.J., Ind. Eng. Chem. Fundam. 18, 81 (1979).
73. Martin, J.J., Ind. Eng. Chem. Fundam. 23, 454 (1984).
74. Mathias, P.M., Copeman, T.W., Fluid Phase Equilibria 13, 91 (1983).
75. Merriman, L.H., B.A.Sc. Thesis, UBC (1983).
76. Mihajlov, A.N., Stevanovic, M.M., Jovanovic, S.D., Chemical Engineering
Thermodynamics, Ed. Newman, S.A., Ann Arbor Science Publishers,
Michigan, Chapter 20 (1983).
77. Miller, D.G., J. Physical Chem. 68, 1399 (1964).
78. Miller, D.G., Ind. Eng. Chem. Fundam. 9, 585 (1970).
79. Mohanty, K.K., Dombrowski, M., Davis, H.T., Chem. Eng. Commun. 5, 85
(1980).
80. O'Brien, L.J., Alford, W.J., Ind. Eng. Chem. 43, 506 (1951).
81. O'Connell, J.P., Prausnitz, J.M., Ind. Eng. Chem. Process Des. Dev. 6, 245
(1967).
161
82. Pal, A.K., Pope, G.A., Arai, Y., Carnahan, N.F., Kobayashi, R., J. chem.
Eng. Data 21, 394 (1976).
83. Passut, C.A., Danner, R.P., Ind. Eng. Chem. Process Des. Dev. 12, 365
(1973).
84. Paunovic, R., Skrbic, B., Novakovic, M., Petrasinovic, L., High
Temperatures-High Pressures 11, 519 (1979).
85. Peneloux, A., Rauzy, E., Freze, R., Fluid Phase Equilibria 8, 7 (1982).
86. Peng, D.Y., Robinson, D.B., Thermodynamics of Aqueous Systems with
Industrial Applications, ACS Symp. Ser. 133 Ed. Newman, S.A., Chapter 20
(1980).
87. Peng, D.Y., Robinson, D.B., Ind. Eng. Chem. Fundam. 15, 59 (1976).
88. Pitzer, K.S., J. Am. Chem. Soc. 77, 3427 (1955).
89. Pitzer, K.S., Lippmann, D.Z., Curl, R.F., Huggins, CM., Petersen, D.E., J.
Am. Chem. Soc. 77, 3433 (1955).
90. Pitzer, K.S., Curl, R.F., J. Am. Chem. Soc. 79, 2369 (1957).
91. Prausnitz, J.M., Lichtenthaler, R.N., de Azevedo, E.G., Molecular
Thermodynamics of Fluid-Phase Equilibria, 2nd ed., Prentice-Hall Inc.,
Englewood Cliffs (1986).
92. Rackett, H.G., J. Chem. Eng. Data 15, 514 (1970).
93. Redlich, O., Kwong, J.N.S., Chem. Rev. 44, 233 (1949).
94. Reid, R.C, Prausnitz, J.M., Sherwood, T.K., The Properties of Gases and
Liquids, 3rd ed., McGraw-Hill, New York (1977).
95. Reid, R.C, Prausnitz, J.M., Poling, B.E., The Properties of Gases and
Liquids, 4th ed., McGraw-Hill, New York (1987).
96. Riedel, L., Chem. Ing. Tech. 26, 83 (1954).
97. Roder, H.M., Childs, G.E., McCarty, R.D., Anderhofer, P.E., Survey of the
Properties of the Hydrogen Isotopes Below Their Critical Temperatures, NBS
Technical Note 641, U.S. Dept. of Commerce (1973).
98. Roebuck, J.R., Osterberg, H., Physical Rev. 46, 785 (1934).
99. Sage, B.H., Kennedy, E.R., Lacey, W.N., Ind. Eng. Chem. 28, 601 (1936).
100. Sage, B.H., Webster, D.C, Lacey, W.N., Ind. Eng. Chem. 29, 658 (1937).
162
101. Santrach, D., Lielmezs, J., Ind. Eng. Chem. Fundam. 17, 93 (1978).
102. Schmidt, A.X., List, H.L., Material and Energy Balances Prentice-Hall Inc.,
Englewood Cliffs (1962).
103. Schmidt, G., Wenzel, H., Chem. Eng. Sci. 35, 1503 (1980).
104. Shah, K.K., Thodos, G., Ind. Eng. Chem. 57, 30 (1965).
105. Shaw, J.M., Lielmezs, J., Chem. Eng. Sci. 40, 1793 (1985).
106. Simnick, J.J., Lin, H.M., Chao, K.C., Equations of State in Engineering and
Research, Adv. Chem. Ser. 182, Ed. Chao, K.C., Robinson, R.L., Chapter
12 (1979).
107. Simonet, R., Behar, E., Chem. Eng. Sci. 31, 37 (1976).
108. Smith, J.M., Chem. Eng. Prog. 44, 521 (1948).
109. Smith, B.D., Srivastava, R., Thermodynamic Data for Pure Compounds Part
B: Halogenated Hydrocarbons and Alcohols, Elseviver Science Publishing
Company, New York (1986).
110. Soave, G., Chem. Eng. Sci. 27, 1197 (1972).
111. Soave, G., Chem. Eng. Sci. 35, 1725 (1980).
112. Soave, G., Chem. Eng. Sci. 39, 357 (1984).
113. Spencer, C.F., Danner, R.P., J. Chem. Eng. Data 17, 236 (1972).
114. Starling, K.E., Hydrocarbon Processing 50(3), 101 (1971).
115. Starling, K.E., Fluid Thermodynamic Properties for Light Petroleum Systems,
Gulf Publishing Company, Houston (1973).
116. Street, W.B., Sagan, L.S., Staveley, L.A.K., J. Chem. Thermodyn. 5, 633
(1973).
117. Stryjek, R., Vera, J.H., Can. J. Chem. Eng. 64, 323 (1986).
118. Stryjek, R., Vera, J.H., Can. J. Chem. Eng 64, 820 (1986).
119. Tannar, D., Lielmezs, J., Herrick, T.A., Thermochimica Acta 116, 209
(1987).
120. Tarakad, R.R., Danner, R.P., AIChE J. 23, 685 (1977).
121. Tarakad, R.R., Spencer, C.F., Adler, S.B., Ind. Eng. Chem. Process Des.
Dev. 18, 726 (1979).
163
122. TRC Thermodynamic Tables of Hydrocarbons, Thermodynami c Research
Centre, Texas A & M Uni versi ty System College Station (1963).
123. Tsonopoulos, C , AI Ch E. J . 20, 263 (1974).
124. Tsonopoulos, C , Prausni tz, J . M. , Cryogenics 9, 315 (1969).
125. van der Waal s, J . D. , doctoral disseration, Lei den, Hol l and (1873).
126. Vargafti k, N. B. , Handbook of Physical Properties of Liquids and Gases: Pure
Substances and Mixtures, 2nd Ed. , Hemisphere Publishing Company,
Washi ngton, D. C. (1975).
127. Wilson, G. M. , Adv. Cryog. Eng. 9, 168 (1964).
128. Wi l son, G. M. , Adv. Cryog. Eng. 11, 392 (1966).
129. Yu , J . - M. , Adachi , Y. , Lu , B. C. - Y. , Equation of State : Theories and
Applications, ACS Symp. Ser. 300, Ed. Chao, K. C. , Robinson R. L. , Chapter
26 (1985).
130. Zudkevitch, D. , Joffe, J . , AI Ch E J . 16, 112 (1970).
APPENDIX A
In this Appendix all the necessary constants and equations are derived
from the generalized cubic equation of state [103]. The exact equations of all the
properties considered in this work are given at the end of each section for the
three cubic EOS : the Soave-Redlich-Kwong [110], the Peng-Robinson [87], and the
proposed modification. The generalized cubic EOS is given b}' the following:
RT a(T)
P =
V-b V
2
+ubV+wb
2
(A.l)
1. CRITICAL COMPRESSIBILITY FACTOR
Mul ti pl y Eq. ( A. l ) by ( V- b) ( V
2
+ubV+wb
2
) and collect similar terms:
V
3
+
RT
ub - b
P
V
2
+
RT a
wb
2
- ub
2
- ub + -
P P
RT ab
wb
3
+ wb
2
+
P P
= 0
(A.2)
At the critical point,
( V - v
c
)
3
= 0
(A.3)
Expand Eq. (A.3) algebraically,
164
V
3
- 3V
2
V + 3VV
2
- V
3
= 0
c c c
165
(A.4)
Compare the coefficients of Eq. (A.2) and (A.4), and we have the followings:
RT
ub - b
P
= - 3V
(A..5)
RT a
wb
2
- ub
2
- ub + -
P P
= 3V
C
2
(A. 6)
RT ab
wb
3
+ wb
2
+
P P
= V
3
c
(A. 7)
Let V
c
= Xb, then Eq. (A.5), (A.6) and (A.7) become (A.8), (A.9), and (A. 10),
respectively.
RT
b(u - 1 + 3X) =
P
(A.8)
RT a
b
2
(w - u - 3X
2
) = ub
P P
(A.9)
RT a
X
3
b
2
-wb
2
- wb = -
P P
(A. 10)
Substitute Eq. (A. 10) into (A.9) and simplify
RT
b
2
(w - u -3X
2
) = b(u + w) + wb
2
- X
3
b
2
P
Substitute Eq. (A.8) into (A. 11) and simplify
(w - u - 3X
2
) = (u - 1 + 3X)(u + w) + w - X
Collect similar terms and get Eq. (A. 13):
X
3
- 3X
2
- 3(u + w)X + (w - uw - u
2
) = 0
Define Q
&
and as
a P
c
0_ =
b P
c
RT
At the critical point Eq. ( A. l ) becomes
p =
RT^ _ a(T
c
)
Substituting Eq. (A. 16) into (A. 14)
166
(A. 11)
(A. 12)
(A. 13)
a
R 2 T 2
(A. 14)
c
(A. 15)
C
V-b V
2
+ubV +wb
2 ( A
-
1 6 )
c c c
167
a(T ) (a(T ) )
2
a
RT
c
(V
c
-b) R
2
T
c
2
(V
c
2
+ubV
c
+wb
2
) (A. 17)
Differentiate Eq. (A.l) with respect to volume at constant temperature
9P RT a(2V+ub)
= - + (A IR)
3V (V-b)
2
(V
2
+ubV+wb
2
)
2
At the critical point
3P
3V
= 0 and Eq. (A. 18) becomes
RT
C
a(T
c
)(2V
c
+ub)
(V
c
-b)
2
(V
c
2
+ubV
c
+wb
2
)
2
(A. 19)
Rearrange Eq. (A. 19)
RT (V
2
+ubV +wb
2
)
2
V
=
a ( T
-
) =
(A.20)
(V
c
-b)
2
(2V
c
+ub)
Replace V
c
with Xb and substitute Eq. (A.20) into (A. 17) and simplify
= (X
2
-2X-u-w)
a.
(X
2
+uX+w)
2
(X-l)*(2X+u)
2
(A.21)
Similarly, can be written as
(X
2
-2X-u-w)
Q
h
=
(X-l )
2
(2X+u)
168
(A.22)
Divide &
a
with fl^
J2
a
(A.23)
bRT
c
or
T_ =
bRfi
(A.24)
Divide C2 wi th f^
2
a
v " ^
<A
-
25)
or
P =
b
2
f l
(A. 26)
By definition, Z_ is
c
P V
7 - C C Z
c - (A.27)
169
Substitute Xb for v , T c with Eq. (A.24), and Pc with (A.26), Eq. (A.27)
becomes
z
c
= x f l
b (A. 2 8)
Replace with Eq. (A.22) and get
X(X
2
-2X-u-w)
(X-l)
2
(2X+u)
1.1. The Soave-Redlich-Kwong Equation
u = 1
w = 0
X = 3.8473221...
G
0
= 0.4274802...
= 0.0866403...
Z
c
= 0.3333333...
1.2. The Peng-Robinson Equation
u = 2
w = -1
X = 3.9513730...
Q = 0.4572355.
a
(A.29)
= 0.0777960
= 0.3074013
1.3. This Work
u = 2
w = -1
X = 3.9513730.
0 = 0.4572355
3.
= 0.0777960
Z = 0.3074013
c
2. CONSTANTS A AND B
Let's define A and B as
aP
A =
R 2 T 2
bP
B =
RT
where
a = a(T
c
) a
171
b
=
b
(
T
c
> (A.33)
Substituting a and b into Eq. (A.30) and (A.31), A and B can be rewritten as
A =
a
a
(A.34)
B
= % (A.35)
r
2.1. The Soave-Redlich-Kwong Equation
0.5 .2
a = [ 1 + m(l - T
r
) ]
The constant m is
m = 0.480 + 1.574CJ - 0.176w
2
2.2. The Peng-Robinson Equation
0.5 ,2
a = [ 1 + K(1 - T
r
) ]
The constant K is
K = 0.37464 + 1.54226u - 0.26992u
2
2.3. This Work
For T < 1.0
a = 1 + p(T*)
q
(T /T - 1)
T = -
(T /T
b
- 1)
and for T > 1.0
a = 1 - p(T*)
q
_ (T/T, - 1)
(T
c
/T
b
- 1)
The constants p and q are substance-dependent constants.
3. COMPRESSIBILITY FACTOR EQUATION
Multiply Eq. (A.2) by P
3
/(R
3
T
3
) and let
PV
Z =
RT
Collect similar terms and Eq. (A.2) becomes
173
Z
3
+ (uB - B - 1)Z
2
+ (A + wB
2
- uB - uB
2
)Z
- (AB + wB
2
+ wB
3
) = 0 ^
A 3 6
^
3.1. The Soave-Redlich-Kwong Equation
Z
3
- Z
2
+ (A - B - B
Z
) Z AB = 0
3.2. The Peng-Robinson Equation
Z
3
- (1 - B) Z
2
+ (A - 3B
2
- 2B)Z - (AB - B
2
- B
3
) = 0
3.3. This Work
Z
3
- (1 - B)Z
2
+ (A - 3B
2
- 2B)Z - (AB - B
2
- B
3
) = 0
4. FUGACITY COEFFICIENT
Fugacity coefficient for a pure compound can be computed by the following
equation :
174
f 1 P
in - = - ;
P RT 0
RT
V
P
dP
(A. 3 7)
Eq. (A. 3 7) may be rearranged to give
f 1 P lnP
m - = ; vdp - ; dinp
( A 3 8 )
P RT P lnP
0
where P is the reference state pressure of a pure substance in the ideal gas
state with molal volume V at some temperature T. The molal volume V is
given by
RT
The first term of the right hand side of Eq. (A.38) can be integrated by parts,
that is
/ VdP = A(VP) - / PdV
( A 4 0 )
The second term of the right hand side of Eq. (A.40) can be evaluated by
substituting the generalized cubic equation, Eq. (A.l), for P
dV adV
J PdV = RT J /
( A - 4 1 )
V-b V
2
+ubV+wb
2
175
The upper limit of the integral is V and the lower limit is V. Integrating Eq.
(A.41) gives
S PdV = RTln
v - b
V - b 2b6
1
In
V+(9
2
b
V+0
3
b
In
2b6
1
v + e
2
b
v+e
3
b
(A.42)
where
6
X
* =
- w
(A.43)
e
2
=
u
- - 0i
2
(A.44)
6, =
u
2
(A. 45)
Combining Eq. (A.38), (A.40), and (A.42), the fugacity coefficient becomes
In - = (Z - 1) - In
P
V-b
V-b
In
2bRT<9i
V+f?,b
V+0
3
b
a V+0
2
b P
+ In - In
2bRT0
x
V
0
+e9
3
b po
(A. 46)
176
By l'Hospital's rule, the term (V+02 b)/( V+t^b) approaches 1 as P approaches
zero. The fugacity coefficient for a pure compound is then given by Eq. (A.47):
f A
In - = (Z-l) - ln(Z-B) + In
P
Z+t?
2
B
Z+0
3
B
(A. 47)
4.1. The Soave-Redlich-Kwong Equation
0
a
= 0.5
e2 = o.o
e
3
= 1.0
z
Z+B
4.2. The Peng-Robinson Equation
f A
In - = (Z-l) - ln(Z-B) + - In
P B
6
1
= Jl
6
2
= 1 - y/2
e3 = i + /2~
f A
In - = (Z-l) - ln(Z-B) + In
P 2v^B
Z-0.414B
Z+2.414B
177
4.3. This Work
e, = 1/2
e2 = 1 - vT
t 93 = 1 + /2
f A
In - = (Z-l) - ln(Z-B) +
=
I n
P 2\/2B
Z-0.414B
Z+2.414B
5. ENTHALPY DEPARTURE FUNCTION
By definition, enthalpy is given
H = U + PV
(A. 48)
Differentiate Eq. (A.48) with respect to volume at constant temperature
9(PV) " 9H ' 9U
3V
T
9V
+
T
9v
(A. 49)
But
dU = TdS - PdV
(A.50)
Differentiate Eq. (A.50) with respect to volume at constant temperature
178
au
r p
" 3S "
3v
T
X
3v
(A.51)
From Maxwel l relation
" 35 ' " 3P "
3v
T
3T
(A.52)
T
Combining Eq. (A.49), (A.51), and (A.52) gives
r
3H
1
3P 3(PV)
= T
- P +
3V
T
3T
V
3V
(A.53)
T
Integrating Eq. (A.53) gives the isothermal changes in enthalpy
H
2
" H
x
= /
3P
T( >
3T
V
dV + A(PV)
(A. 5 4)
Differentiate Eq. (A.l) with respect to temperature at constant volume
3P
3T
JV
V-b V
2
+ubV+wb
2
da
dT
(A.55)
Substitute Eq. (A.l) and (A.55) into (A.54) and simplify
179
H
2
- H
x
- a - T
da
dT
dV
V
2
+ubV+wb
2
+ A(PV)
(A. 5 6)
Integrating Eq. (A.56) gives
H
2
- H, =
a - T(da/dT)
26yb
V+0
2
b
In
v+e
3
b
+ P
2
V
2
- P
1
V
1
(A. 5 7)
pressure. Eq. (A.57) becomes
H - H = RT(Z-l) +
a - T(da/dT)
2c9ib
be the ideal
Z+t?
2
B
In
Z+t9
3
B
gas at zero
(A.58)
5.1. The Soave-Redlich-Kwong Equation
a = a(T ) [ 1 + m(l - T
r
0
-
5
) ]
:
Differentiate a with respect to T and simplify
da
dT
= - ma(T
c
)
a
TT
C J
da ,
T = - ma(T )(aT )
dT
c r
180
da
a - T
dT
= (1 + m)
A T Z
H -- H = = RT(Z-l) + R(l+m)
B a
0
5
In
Z+B
5.2. The Peng-Robinson Equation
a = a(T
c
) [ 1 + K(1 - T/-
5
) ]
2
Differentiate a with respect to T and simplify
da
= - K3(T )
dT
c
a
TT
c J
da ,
T = - Ka(T_)(oT_)
dT c r
da
a - T
dT
= (1 + K )
H - H = RT(Z-l) +
R(l+c) A T
2v^2 B a"
In
Z-0.414B
Z+2.414B
5.3. This Work
For T < 1.0
r
a = a(T
c
) [ 1 + p(T )
q
]
T =
(T
c
/T - 1)
(T
c
/T
b
- 1)
Differentiate a with respect to T and simplify
da
dT
qa(T
c
)(o-l)
T
r
(T
c
-T)
da
T =
dT
qT
c
a(T
c
)(g-l)
(T
c
-T)
da
a - T
dT
= a
1 +
qT
r
(T
c
-T)
where
4> = 1 +
qT qT,
(T
c
-T) a(T
c
-T)
182
For T > 1.0
r
a = a(T
c
) [ 1 - p(T )
q
]
T =
(T/T
c
- 1)
(T
c
/T
b
- 1)
Differentiate a with respect to T and simplify
where
da qa(T
c
)(a-l)
dT (T-T )
da qa(T_)T(o-l)
T =
dT (T-T )
c
da
T
dT
= a
1 -
cjT qT
(T-T
c
) a(T-T
c
)
= a 4>
1 -
qT qT
(T-T
c
) a(T-T
c
)
The enthalpy departure function is given by the following:
H
H = RT(Z-l) +
RT$ A
2y/2 B
In
Z-0.414B
Z+2.414B
183
6. HEAT CAPACITY DEPARTURE FUNCTION
The constant pressure heat capacity departure function is given by the following
thermodynamic relation:
(C
p
- C
p
") =
9(H-H)
9T
(A.59)
JP
The enthalpy departure function is given by Eq. (A.58) and differentiating Eq.
(A. 5 8) gives
9(H-H) 9(PV) 9(RT)
9T
p
9T
P
9T
JP
9 a - T(da/dT)
( )
9T 26^
V+0
2
b
I n
. P
V+0
3
b
or
a - T(da/dT)
2e
x
b
9 v+e
2
b
( I n
9T V+0
3
b
(A. 60)
(C
p
- C
p
) = P
9V
9T
- R + x
x
l n
V+c?
2
b
V+(9,b
a - T(da/dT)
2c?
x
b
(A.61)
184
where
dT
a - T(da/dT)
26
x
b
(A.62)
Differentiating Eq. (A.62) gives
Xx = -
d
2
a
dT
2
(A.63)
and
X
2
= ( In
9 v
v+e
2
b
V+f9
3
b
9V
9T
(A.64)
Differentiate Eq. (A.64) and simplify, x
2
becomes
26
x
b
X
2
=
V
2
+ubV+wb
2
9V
9T
(A.65)
JP
Substituting Eq. (A. 63) and (A.65) into (A.61) gets
(C
p
- C
p
") = P
9V
9T
- R -
JP
26
x
b
a - T(da/dT)
V
2
+ubV+wb
2
d
2
a V+6
2
b
In
dT
2
V+0
3
b
9v
9T
P
(A.66)
Substitute Eq. (A.l) for P and simplify. Eq. (A.66) becomes
185
(C
p
- C
p
") =
RT T(da/dT)
V-b V
2
+ubV+wb
2
3V
3T
T d
2
a
In
V+t?
2
b
26
x
h dT
2
In
V+t?
3
b
- R
(A.67)
Volume can be written as a function of pressure and temperature:
V = f(P,T)
(A.68)
Take the total differential of Eq. (A.68)
av 3V
dV =
dP +
3P
T
dT
JP
(A.69)
At constant volume
9V 9V
0 =
dP +
3P
T
dT
(A. 70)
Rearranging Eq. (A.70) and gets
186
3v 3V 3P
3T
P
3P
T
3T
(A. 71)
Or
3V
3T
(3P/3T)
(3P/3v)
r
v
(A.72)
(3P/3V)
T
is given by Eq. (A.18) and (3P/3T)
V
is given by Eq. (A.55). Let
X
3
=
T(da/dT)
V
2
+ubV+wb
2
RT
V-b
(A. 73)
Multiply x
3
by (3P/3T)
v
and simplify
3P
3T
R
2
T 2RT(da/dT)
+
(V-b)
2
(V-b)(V
2
+ubV+wb
2
)
da
dT (V
2
+ubV+wb
2
)
2
(A. 74)
The numerator of the first term of the right hand side of Eq. (A. 6 7) is Eq.
(A. 7 4) and the denominator is (3P/3V)
T
. Now multiply this fraction by the
fraction:
(V-b)(V
2
+ubV+wb
2
)
-R
2
Tb
187
and get
Numerator = x:
3P
3T
JV
V
2
+ubV+wb
2
2 da
b(V-b) Rb dT
(V-b)
R
2
b V
2
+ubV+wb
2
da
dT
(A.75)
and
Denominator =
3P
9V
V
2
+ubV+wb
2
Rb(V-b)
a(2V+ub)(V-b)
R
2
Tb(V
2
+ubV+wb
2
)
(A. 76)
Rewrite Eq. (A.75) and (A.76) in terms of A, B, and Z
and
3P
3T
JV
Z
2
+uBZ+wB
2
B(Z-B)
2 da
Rb dT
B da
Rb dT
(Z-B)
B(Z
2
+uBZ+wB
2
)
(A. 7 7)
3P
3V
Z
2
+uBZ+wB
2
A(2Z+uB) (Z-B)
RB(Z-B) RB Z
2
+uBZ+wB
2
(A. 78)
188
Let
Z
2
+uBZ+wB
2
M =
(Z-B)
and
B da
N =
Rb dT
Substituting M and N into Eq. (A.77) and (A.78). The first term of right hand
side of Eq. (A.67) becomes
XOP/ 3 V)
T
R(M-N)
2
( 9 P / 3 V )
t
M
2
-A(2Z+uB)
(A. 79)
The final equation for (C
p
- C
p
) is then given by the following:
R(M-N)
2
T d
2
a
In
Z+6
2
B
M
2
-A(2Z+uB) 2 0 i b dT
2
In
Z+t?
3
B
- R
(A.80)
Or Eq. (A.80) can be written as
C - C
0
<~p l_p
' Cp - Cy " " Cp" " Cy"
R R R R
(A. 81)
189
where
C - C
0
d
2
a Z+0
2
B
In
dT
2
Z+t9
3
B
(A.82)
c
p
c
v
(M-N)
2
M
2
-A(2Z+uB)
(A.83)
= 1
(A.84)
6.1. The Soave-Redlich-Kwong Equation
Differentiate Eq. (A. 3 2) twice with respect to temperature and get
d
2
a
dT
2
a(T
c
)m
2TT
C
5
m
C
M =
Z(Z+B)
(Z-B)
B da
N =
Rb dT
C - C
0
=
d
2
a
dT
2
(Z+B)
In +
R(M-N)
2
M
2
-A(2Z+B)
- R
6.2. The Peng-Robinson Equation
Differentiate Eq. (A.32) twice with respect to temperature and get
d
2
a
dT
2
a ( T j K
2TT
0
5
C
a"
5
K
+
^0 .5 0 . 5
C
M =
Z
2
+2BZ-B
2
Z-B
B da
N =
Rb dT
C - C
0
=
2/2 b
d
2
a
In
Z+2. 414B
dT
2
_
z-o.
414B
R(M-N)
2
M
2
-2A(Z+B)
6.3. This Work
Differentiate Eq. (A.32) twice with respect to temperature and get
For T
r
< 1.0
d
2
a
dT
2
(2T
r
-q-l) da
T
r
(T -T) dT
and for T
r
> 1.0
d
2
a
dT
2
(q-D da
(T-T ) dT
191
M =
Z
2
+2BZ-B
2
Z-B
B da
N =
Rb dT
C - C
0
= U
P *-p
2|/2~ b
d
2
a Z+2 .414B
In
dT
2
Z-0 .414B
R(M-N)
2
+ - R
M
2
-2A(Z+B)
7. JOULE-THOMSON COEFFICIENT
The Joule-Thomson coefficient, u, is given by the relation:
-1
M =
T(3P/3T),
OP/3V).
+ V
(A.85)
(3P/3V)
T
is given by Eq. (A.18), (3P/3T)
V
is given by Eq. (A.55) and (C
p
-
C
p
) is given by Eq. (A.80)
7.1. The Soave-Redlich-Kwong Equation
R 1 da
V-b V(V+b) dT
-RT a(2V+b)
+
(V-b)
2
V
2
(V+b)
2
3P
3T
3P
3V
7.2. The Peng-Robinson Equation
ap
3T
da
V-b V
2
+2bV-b
2
dT
3P
av
JT
-RT 2a(V+b)
+
(V-b)
2
(V
2
+2bV-b
2
)
2
7.3. This Work
ap
3T
JV
da
V-b V
2
+2bV-b
2
dT
ap
av
-RT 2a(V+b)
+
(V-b)
2
(V
2
+2bV-b
2
)
2
8. INVERSION CURVE
The inversion condition, u = 0, is given by the equation
ap ap
T
+ V
= 0
3T
V
av
T
where
3P
3T
JV
RT 1 da
V-b V
2
+ubV+wb
2
dT
193
Since
Let
Therefore
or
where
a = a(T
c
) a
(A.32)
da da
T = a(T ) T
dT dT
(A.88)
da
<j> = T
dT
(A. 89)
3P
9T
RT
V-b V
2
+ubV+wb
2
a(T W
(A. 90)
9P
9T
T
r
P
c
1
a
p
c*
\ <S-l) fl
b
2
(5
2
+u5+w)
(A. 91)
I = V/b
(A.92)
9P
9V
JT
-RTV aV(2V+ub)
(V-b)
2
(V
2
+uV+wb
2
)
2
(A.93)
194
or
9P
9V
-T P
r c
fi P a
a c
S(2$+u)
JT
Ojj ($-l)
2
0^ ($
2
+u5+w)
2
(A. 94)
Adding Eq. (A.91) and (A.94) and simplify
( o - D
:
b
$(2$+u)a - (
2
+u$+w)c6
($
2
+u$+w)
2
(A. 95)
Rearranging Eq. (A. 95) gives the inversion curve equation
C^^+uS+w)
2
+ C
2
$(2$+u)(S-l)
2
+ C,<S-l>
a
(S
a
+u$+w) = 0
where
(A. 96)
C
2
= - R
a
a
C
3
= K<t>
(A. 9 7)
(A. 98)
(A.99)
The reduced inversion pressure is given by the following:
p
r =
RT
r
1 R
a
a 1
($-1) f^
2
(S
2
+u$+w)
(A. 100)
195
8.1. The Soave-Redlich-Kwong Equation
0 = - m(aT
r
)
0
-
5
C
x
= 0.08664T
r
C
2
= - 0.42748 [ 1 + m(l - T
r
0
-
5
) ]
:
C
3
= - 0.42748 m(aT
r
)
0
-
5
The inversion curve is
Ci S CS + l )
2
+ C
2
(2$+l)($-l>* + C,<5+1)<5-1>
2
= 0
The inversion pressure is
T 1 0 a 1
P
r " ~
8.2. The Peng-Robinson Equation
0 = - r c (aT
r
)
0
-
5
C
x
= 0.07780T
r
C
2
= - 0.45724 [ 1 + / c(l - T
r
-
S
) ]
:
C
3
= - 0.45724 *c(aT
r
)
0
-
5
196
The inversion curve is
C1 ( S
2
+2S-1)
2
+ 2C2 S($+1) ( S-1)
2
+ C
3
($
2
+2$-l)($-l)
2
= 0
The inversion pressure is given by
r
% ( S - D V S
2
+2$- l
8.3. This Work
For T
r
< 1.0
a-1
<S> = - q
1-T
r
C
x
= 0.07780T
r
C
2
= - 0.45724 [ 1 + p(T*)
q
]
a-1
C
3
= - 0.45724 q
1-T
r
and for T
r
> 1.0
T (1-a)
* = - q
T
r
-1
C
x
= 0.07780T
r
197
C
2
= - 0.45724 [ 1 ~ p(T*)
q
]
T
r
(l-a)
C
3
= - 0.45724 q
T
r
-1
The inversion curve is
C1 ( S
2
+2S-1)
2
+ 2C2 S( S+D( S-1)
2
+ C
3
($
2
+2$-im-l)
2
= 0
The inversion pressure is given by
T 1 0 a 1
P
= _E 2
Ou (s-i) n
h
2
s
2
+2$-i
9. VIRIAL COEFFICIENTS
The generalized cubic equation is given by Eq. (A.l):
RT
V-b V
2
+ubV+wb
2
The first term, (RT/V-b), can be written as
RT RT 1
V-b V 1-b/V
(A.l)
(A. 101)
or
198
RT RT
V-b V
b b
2
b
3
1 + - + + +
V V
2
V
3
(A. 102)
The second term can be written as
V
2
+ubV+wb
2
V V+ub+(wb
2
/V)
(A.103)
or
V
2
+ubV+wb
2
a
V
1 ub (u
2
-w)b
2
- + +
V V
2
V
3
(A. 104)
Substract Eq. (A. 104) from (A. 102) gets
Z = 1 + b -
RT
b
2
+
uab
RT
1
V
2
b
3
+
(w-u
2
)ab
2
RT V
3
(A. 105)
The virial equation:
B C D
Z = 1 + - + + +
V V
2
V
3
(A. 106)
Therefore the second virial coefficient is
r
a
B = b
RT
199
(A.107)
The third vi ri al coefficient is
uab
C = b
2
+ (A.108)
RT
The fourth virial coefficient is
(w-u
2
)ab
2
D
=
b 3 +
(A. 109)
RT
9.1. The Soave-Redlich-Kwong Equation
a
B = b
RT
ab
C = b
2
+
RT
ab
2
D = b
3
-
RT
9.2. The Peng-Robinson Equation
a
B - b
RT
2ab
C = b
2
+
RT
5ab
2
D = b
3
RT
200
9.3. This Work
a
B = b
RT
2ab
C = b
2
+
RT
5ab
2
D = b
3
RT