You are on page 1of 30

25-1

K
i
= equilibrium ratio,
y
i


x
i
L = ratio of moles of liquid to moles of total mixture
N = mole fraction in the total mixture or feed
w = acentric factor
P = absolute pressure, kPa (abs)
P
k
= convergence pressure, kPa (abs), psia
SECTION 25
Phase Equilibria
FIG. 25-1
Nomenclature
P
*
= vapor pressure, kPa (abs)
R = universal gas constant, (kPa (abs) m
3
) / (kmole K)
T = temperature, K or C
V = ratio of moles of vapor to moles of total mixture
x
i
= mole fraction of component i in the liquid phase
y
i
= mole fraction of component i in the vapor phase
Subscripts
i = component
VAPOR-LIQUID EQUILIBRIA
The equilibrium ratio (K
i
) of a component i in a multicompo-
nent mixture of liquid and vapor phases is defined as the ratio
of the mole fraction of that component in the vapor phase to
that in the liquid phase.
K
i
=
y
i


x
i
Eq 25-1
For an ideal system (such as ideal gas and ideal solution), this
equilibrium ratio becomes the ratio of the vapor pressure of
component i to the total pressure of the system.
K
i
=
P
i
*

P Eq 25-2
This section presents an outline procedure to calculate the
liquid and vapor compositions of a two-phase mixture in equi-
librium using the concept of a pseudobinary system and the
convergence pressure equilibrium charts. Discussion of CO
2
sep-
aration, alternate methods to obtain K values, and equations of
state follow.
K-DATA CHARTS
These charts show the vapor-liquid equilibrium ratio, K
i
, for
use in example and approximate flash calculations. The charts
do not give accurate answers, particularly in the case of nitro-
gen. They are included only for illustrative purposes and to sup-
port example flash calculations and quick estimation of K-val-
ues in other hand calculations. These charts should not be used
in design calculations.
Previous editions of this data book presented extensive sets
of K-data based upon the GPA Convergence Pressure, P
k
, meth-
od. A components K-data is a strong function of temperature
and pressure and a weaker function of composition. The conver-
gence pressure method recognizes composition effects in pre-
dicting K-data. The convergence pressure technique can be
used in hand calculations, and it is still available as computer
correlations for K-data prediction.
Availability of computers, coupled with the more refined K-
value correlations in modern process simulators, has made the
previous GPA convergence pressure charts outdated. Complete
sets of these charts are available from GPA as a Technical Pub-
lication, TP-22.
Data for N
2
-CH
4
and N
2
-C
2
H
6
show that the K-values in
these system have strong compositional dependence. The com-
ponent volatility sequence is N
2
-CH
4
-C
2
H
6
and the K-values are
functions of the amount of methane in the liquid phase. For
example, at 123C and 2070 kPa (abs), the K-values depend-
ing upon composition vary from:
N
2
CH
4
C
2
H
6
10.2 0.824* 0.0118
3.05 0.635 0.035*
where * indicates the limiting infinite dilution K-value. See ref-
erence 5 for the data on this ternary.
The charts retained in this edition represent roughly 12% of the
charts included in previous editions. These charts are a compro-
mise set for gas processing as follows:
a. hydrocarbons 3000 psia P
k
[20 700 kPa (abs)]
b. nitrogen 2000 psia P
k
[13 800 kPa (abs)]
c. hydrogen sulfide 3000 psia P
k
[20 700 kPa (abs)]
The pressures in a. through c. above refer to convergence
pressure, P
k
, of the charts from the Tenth Edition of this data
book. They should not be used for design work or related activi-
ties. Again, they are in this edition for illustration and approxi-
mation purposes only; however, they can be very useful in such
a role. The critical locus chart used in the convergence pressure
method has also been retained (Fig. 25-8).
The GPA/GPSA sponsors investigations in hydrocarbon sys-
tems of interest to gas processors. Detailed results are given in
the annual proceedings and in various research reports and
technical publications, which are listed in Section 1.
25-2
Example 25 -1 Binary System Calculation
To illustrate the use of binary system K-value charts, as-
sume a mixture of 60 kmols of methane and 40 kmols of ethane
at 87C and 345 kPa (abs). From the chart on page 25-10, the
K-values for methane and ethane are 10 and 0.35 respectively.
Solution Steps
From the definition of K-value, Eq 25 -1:
K
C1
=
y
C1


x
C1

= 10
K
C2
=
y
C2


x
C2

= 0.35
Rewriting for this binary mixture:
1 y
C1


1 x
C1
= 0.35
Solving the above equations simultaneously:
x
C1
= 0.0674
y
C1
= 0.674
Also by solving in the same way:
x
C2
= 0.9326
y
C2
= 0.326
To find the amount of vapor in the mixture, let v denote
kmols of vapor. Summing the kmols of methane in each phase
gives:

kmols C
1
+ C
2
= 100 kmols
kmols C
1
+ kmols C
1

= 60 kmols
in vapor in liquid
(y
C1
v) + (x
C1
[100 v]) = 60 kmols
(0.674

v) + (0.0674 [100 v]) = 60 kmols
v = 87.8 kmols
The mixture consists of 87.8 kmols of vapor and 12.2 kmols
of liquid.
FLASH CALCULATION PROBLEM
The problem below illustrates the calculation of multicom-
ponent vapor-liquid equilibrium using the flash equations and
the K-charts in detail. The variables are defined in Fig. 25-1.
Note that the K-value is implied to be at thermodynamic equi-
librium.
A situation of reproducible steady state conditions in a piece
of equipment does not necessarily imply that classical thermo-
dynamic equilibrium exists. If the steady composition differs
from that for equilibrium, the reason can be the result of time-
limited mass transfer and diffusion rates. This warning is made
because it is not at all unusual for flow rates through equip-
ment to be so high that equilibrium is not attained or even close-
ly approached. In such cases, equilibrium flash calculations as
described here fail to predict conditions in the system accurate-
ly, and the K-values are suspected for this failurewhen in fact
they are not at fault.
Using the relationships

K
i
=
y
i

x
i
Eq 25-3
L + V = 1.0 Eq 25-4
By writing a material balance for each component in the
liquid, vapor, and total mixture, one may derive the flash equa-
tion in various forms. A common one is,
x
i
=

N
i


L + VK
i

= 1.0

Eq 25-5
Other useful versions may be written as
L =
N
i


1 + (V/L) K
i

Eq 25-6
y
i
=
K
i
N
i

L + VK
i
Eq 25-7
At the phase boundary conditions of bubble point (L = 1.00) and
dew point (V = 1.00), these equations reduce to
K
i
N
i
= 1.0 (bubble point) Eq 25-8
and
N
i
/K
i
=1.0 (dew point) Eq 25-9
These are often helpful for preliminary calculations where
the phase condition of a system at a given pressure and tem-
perature is in doubt. If K
i
N
i
and N
i
/K
i
are both greater than
1.0, the system is in the two phase region. If K
i
N
i
is less than
1.0, the system is all liquid. If N
i
/K
i
is less than 1.0, the system
is all vapor.
Example 25-2 A typical high pressure separator gas is used
for feed to a natural gas liquefaction plant, and a preliminary
step in the process involves cooling to 30C at 4140 kPa (abs)
to liquefy heavier hydrocarbons prior to cooling to lower tem-
peratures where these components would freeze out as solids.
Solution Steps
The feed gas composition is shown in Fig. 25-3. The flash
equation 25 -5 is solved for three estimated values of L as shown
in columns 3, 4, and 5. By plotting estimated L versus calcu-
lated x
i
, the correct value of L where x
i
= 1.00 is L = 0.030,
whose solution is shown in columns 6 and 7. The gas composi-
tion is then calculated using y
i
= K
i
x
i
in column 8. This correct
value is used for purposes of illustration. It is not a completely
converged solution, for x
i
= 1.00049 and y
i
= 0.99998, columns 7
and 8 of Fig. 25-3. This error may be too large for some applica-
tions.
Example 25-3 Dew Point Calculation
A gas stream at 40C and 5500 kPa (abs) is being cooled in a
heat exchanger. Find the temperature at which the gas starts to
condense.
Solution Steps
The approach to find the dew point of the gas stream is sim-
ilar to the previous example. The equation for dew point condi-
tion (N
i
/K
i
= 1.0) is solved for two estimated dew point tem-
peratures as shown in Fig. 25-4. By interpolation, the
temperature at which N
i
/K
i
= 1.0 is estimated at 41.4C.
Note that the heaviest component is quite important in dew
point calculations. For more complex mixtures, the character-
25-3
ization of the heavy fraction as a pseudocomponent such as hex-
ane or octane will have a significant effect on dew point calcula-
tions.
Carbon Dioxide
Early data on CO
2
systems used to prepare K-data (P
k
= )
charts for the 1966 Edition were not consistent. Later, experi-
ence showed that at low concentrations of CO
2
, the rule of
thumb

K
CO
2
= K
C
1
K
C
2
Eq 25-10
could be used with a plus or minus 10% accuracy. Developments
in the use of CO
2
for reservoir drive have led to extensive inves-
tigations in CO
2
processing. See the GPA research reports (list-
ed in Section 1) and the Proceedings of GPA conventions. The
reverse volatility at high concentration of propane and/or bu-
tane has been used effectively in extractive distillation to effect
CO
2
separation from methane and ethane.
23
In general, CO
2

lies between methane and ethane in relative volatility.
Separation of CO
2
and Methane
The relative volatility of CO
2
and methane at typical operat-
ing pressures is quite high, usually about 5 to 1. From this stand-
point, this separation should be quite easy. However, at process-
ing conditions, the CO
2
will form a solid phase if the distillation
is carried to the point of producing high purity methane.
Fig. 25-5 depicts the phase diagram for the methane-CO
2

binary system.
21
The pure component lines for methane and
CO
2
vapor-liquid equilibrium form the left and right boundaries
of the phase envelope. Each curve terminates at its critical
point; methane at 83C, 4604 kPa (abs) and CO
2
at 31C, 7382
kPa (abs). The unshaded area is the vapor-liquid region. The
shaded area represents the vapor-CO
2
solid region which ex-
tends to a pressure of 4860 kPa (abs).
Because the solid region extends to a pressure above the
methane critical pressure, it is not possible to fractionate pure
methane from a CO
2
-methane system without entering the solid
formation region. It is possible to perform a limited separation
FIG. 25-2
Sources of K-Value Charts
Component
Charts available from sources as indicated
Convergence pressures, kPa (abs) [psia]
Binary
Data
5500 6900 10 300 13 800 20 700 34 500 69 000
[800] [1000] [1500] [2000] [3000] [5000] [10,000]
Nitrogen *
Methane *
Ethylene
Ethane *
Propylene
Propane *
iso-Butane
n-Butane *
iso-Pentane
n-Pentane *
Hexane *
Heptane *
Octane
Nonane x
Decane
Hydrogen sulfide
Carbon dioxide
** Use K
CO
2
= K
C
1
K
C
2
* Binary data from Price & Kobayashi; Wichterle & Kobayashi; Stryjek, Chappelear, & Kobayashi; and Chen & Kobayashi
Drawn for 1972 Edition based on available data
Reused from 1966 Edition
Reused from 1957 Edition
Prepared for Second Revisions 1972 Edition or revised
** Limited to CO
2
concentration of 10 mole percent of feed or less
Note: The charts shown in bold outline are published in
this edition of the data book. The charts shown in the
shaded area are published in a separate GPA Technical
Publication (TP-22) as well as the 10th Edition.
25-4
of CO
2
and methane if the desired methane can contain signifi-
cant quantities of CO
2
.
At an operating pressure above 4860 kPa (abs), the methane
purity is limited by the CO
2
-methane critical locus (Fig. 25-6).
For example, operating at 4930 kPa (abs), it is theoretically
possible to avoid solid CO
2
formation (Fig. 25-7 and 16-36). The
limit on methane purity is fixed by the approach to the mixture
critical. In this case, the critical binary contains 6% CO
2
. A
practical operating limit might be 10-15% CO
2
.
One approach to solve the methane-CO
2
distillation problem
is to use extractive distillation (See Section 16, Hydrocarbon
Recovery). The concept is to add a heavier hydrocarbon stream
to the condenser in a fractionation column. About 10 GPA re-
search reports present data on various CO
2
systems that are
pertinent to the design of such a process.
CO
2
-Ethane Separation
The separation of CO
2
and ethane by distillation is limited
FIG. 25-3
Flash Calculation at 4140 kPa and 30C
Component
Column
1 2 3 4 5 6 7 8
Feed Gas
Composition
30C
4140 kPa
Trial values of L Final L = 0.030
L = 0.020 L = 0.060 L = 0.040L
L + Vk
i
Liquid Vapor
N
i
K
i
N
i

L + V K
i
N
i

L + V K
i
N
i

L + V K
i
x
i
=
N
i

L + V K
i
y
i
C
1
0.9010 3.7 0.24712 0.25466 0.25084 3.61900 0.24896 0.92117
CO
2
** 0.0106 1.23 0.00865 0.00872 0.00868 1.22310 0.00867 0.01066
C
2
0.0499 0.41 0.11830 0.11203 0.11508 0.42770 0.11667 0.04783
C
3
0.0187 0.082 0.18633 0.13642 0.15751 0.10954 0.17071 0.01400
iC
4
0.0065 0.034 0.12191 0.07068 0.08948 0.06298 0.10321 0.00351
nC
4
0.0045 0.023 0.10578 0.05513 0.07249 0.05231 0.08603 0.00198
iC
5
0.0017 0.0085 0.06001 0.02500 0.03530 0.03825 0.04445 0.00038
nC
5
0.0019 0.0058 0.07398 0.02903 0.04170 0.03563 0.05333 0.00031
C
6
0.0029 0.0014 0.13569 0.04730 0.07014 0.03136 0.09248 0.00013
C
7
+* 0.0023 0.00028 0.11334 0.03817 0.05712 0.03027 0.07598 0.00002
TOTALS 1.0000 1.17121 0.77714 0.89834 1.00049 0.99998
C
7
0.00042
C
8
0.00014
* Average of nC
7
+ nC
8 properties

**

K
C1
K
C2
FIG. 25-4
Dew Point Calculation at 5500 kPa (abs)
Component
Column
1 2 3 4 5
Feed Estimated T = 45C Estimated T = 40C
N
i
K
i
N
i



K
i
K
i
N
i



K
i
CH
4
0.854 2.73 0.313 2.75 0.311
CO
2
0.051 0.866 0.059 0.910 0.056
C
2
H
6
0.063 0.275 0.229 0.300 0.210
C
3
H
8
0.032 0.070 0.457 0.080 0.400
1.000 1.058 0.977

K
CO2
calculated as

K
C1
K
C2
Linear interpolation: T
dew
= 40 [40 (45)]
1.000 0.977
= 41.4C
1.058 0.977
Alternative iterate until N
i
/K
i
= 1.0
25-5
by the azeotrope formation between these components. An azeo-
tropic composition of approximately 67% CO
2
, 33% ethane is
formed at virtually any pressure.
24
Fig. 25-7 shows the CO
2
-ethane system at two different
pressures. The binary is a minimum boiling azeotrope at both
pressures with a composition of about two-thirds CO
2
and one-
third ethane. Thus, an attempt to separate CO
2
and ethane to
nearly pure components by distillation cannot be achieved by
traditional methods, and extractive distillation is required.
26

(See Section 16, Hydrocarbon Recovery)
Separation of CO
2
and H
2
S
The distillative separation of CO
2
and H
2
S can be performed
with traditional methods. The relative volatility of CO
2
to H
2
S
is quite small. While an azeotrope between H
2
S and CO
2
does
not exist, vapor-liquid equilibrium behavior for this binary ap-
proaches azeotropic character at high CO
2
concentrations
25
(See
Section 16, Hydrocarbon Recovery).
FIG. 25-5
Phase Diagram CH
4
-Ch
2
Binary
21
FIG. 25-6
Isothermal Dew Point and Frost Point Data for Methane-Carbon Dioxide
32
25-6
PHASE EQUILIBRIA METHODS
Numerous procedures have been devised to predict phase
equilibria (K-values) and the corresponding physical properties
of the associated phases. These include:
Equations of state (EOS)
Activity coefficient models
Electrolytic models
Combinationsof equations of state with liquid theory or
with tabular data
Corresponding states correlations.
A number of methods can be used for the purpose of phase
equilibria and thermodynamic property prediction. In modern
times, calculations are not typically executed by hand, but in-
stead are solved by the use of thermodynamic simulation soft-
ware (commercial or proprietary). This section describes several
of the more popular procedures currently available. It does not
purport to be all-inclusive or comparative.
Equations of State (EOS)
Equations of state have appeal for predicting thermodynam-
ic properties because they provide internally consistent values
for all properties in convenient analytical form. The section be-
low discusses the basic capabilities of EOS, historical develop-
ment, and recent advances.
EOS Capabilities The following summarizes the basic
capabilities and describes the applicability for some of the more
commonly used EOS methods.
Although originally developed to describe simple gases,
EOS have proven reliable for property prediction of most
hydrocarbon-based fluids.
The simple cubic EOS are generally limited to pre-
diction of thermodynamic properties and phase equilib-
ria for ideal or slightly non-ideal systems; they are not
suitable for representation of highly non-ideal systems
(e.g., methanol/water systems).
They typically are applied only to hydrocarbon mix-
tures with relatively low concentrations of non-polar or
slightly polar fluids.
Recent advancements have made cubic EOS suit-
able for handling high concentrations of CO
2
, H
2
S, and
N
2
.
Applicable for prediction of phase equilibria for pure com-
ponents (VLE) and mixtures (VLE and VLLE) and for
prediction of all thermodynamic properties for vapor and
liquid phases.
Originally developed for handling of pure compo-
nents, but inclusion and use of various mixing rules,
which incorporate binary interaction parameters, have
allowed the extension of use to binary and multicompo-
nent mixtures.
Useful over wide ranges of temperature and pres-
sure, including subcritical, supercritical, and retro-
grade regions.
Require minimal pure component data. Experimental bi-
nary data can be used to tune binary interaction pa-
rameters, usually by regression of experimental data.
Major EOS types include cubic, virial, corresponding
states, and multi-parameter. Descriptions of the more
commonly used cubic and virial types are included below:
Cubic EOS (e.g., van der Waal, Redlich-Kwong,
Soave-Redlich-Kwong, Peng Robinson)
Explicit in pressure (P) with respect to tempera-
ture (T) and volume (V). They have separate terms
to correct ideal gas predictions for attraction and
repulsion forces between the molecules (correcting
the real vapor pressure and volume predictions, re-
spectively). When considering the pressure and
temperature fixed, the EOS can be algebraically re-
arranged to give a relationship for V that is a cubic
(3rd order) polynomial.
These EOS will include other parameters, spe-
cific to each chemical species that are generally de-
termined from the critical properties, P
c
and T
c
, for
the chemical species. Additional temperature-de-
pendent functions can be added to more accurately
match pure component behavior (i.e., a tempera-
ture dependent function correlated to the accentric
factor (w) is normally used to better match a pure
components vapor pressure versus temperature be-
havior).
Multicomponent mixtures are treated with the
same EOS parameters that are determined for the
pure components present in the mixture. The equa-
tions used to blend the pure component values are
referred to as mixing rules, which often include
binary interaction parameters to account for non-
ideal interactions between pairs of unlike mole-
cules.
The EOSs are generally not tuned to pure
component liquid density data, so they give poor
representations of liquid molar volume/liquid den-
FIG. 25-7
Vapor-Liquid Equilibria CO
2
-C
2
H
6
21
25-7
sity (There are techniques to improve this, such as
introducing a volume translation term; see Recent
EOS Advancement section below).
Not generally accurate for polar compounds or
long chain hydrocarbons (There are techniques to
improve this, such as introduction of a higher order
temperature dependency on the attraction param-
eter or asymmetric mixing rules; see Recent EOS
Advancement section below).
Virial EOS (e.g., BWR, BWRS, Lee-Kessler-Plocker)
Explicit in pressure (P) with respect to tempera-
ture (T) and volume (V). Expanded in terms of vol-
ume raised to powers much higher than 3
rd
order.
Because of the larger number of parameters
that can be tuned to pure component data, these
EOS can be more accurate than cubic EOS when
calculating liquid densities.
For pure components, these EOS can give a
more accurate representation than cubic EOS for
all thermodynamic data. However, the determina-
tion of these parameters is more complex, requires
more experimental data, and may require a compli-
cated procedure for fitting that experimental data.
When applied to multicomponent mixtures, the
large number of pure component parameters must
be blended, perhaps each with their own mixing
rules. Inaccuracies associated with the application
of these mixing rules may make the mixture prop-
erties no more accurate than what one would get
from a simpler cubic EOS.
Not usually applicable for polar systems.
Computerized corresponding states methods
may be based on virial EOS for reference fluids.
The corresponding states methods then provide the
framework to blend the properties of the reference
fluids to give values for a multicomponent mixture.
Other applicable EOS types
EOS for Associating Systems
EOS that include terms for the physical
forces (attraction/repulsion) and an associating
term that takes into account hydrogen bond-
ing; see Kontogeorgis and Folas
34
for reviews of
associating EOS.
The SAFT (Statistical Associate Fluid The-
ory) family of EOS are based on Wertheims
perturbation theory and can be applied to a
wide range of fluids including long chain com-
ponents and hydrogen bonding (e.g., hydrocar-
bon-alcohol-water systems).
CPA (Cubic-Plus-Association) EOS com-
bines a cubic equation of state (e.g., SRK) for
describing physical interactions with an asso-
ciation term similar to SAFT.
Highly Accurate Pure Component EOS
Typically apply only to utility systems
within a facility, not to the main processing
and separation trains.
Examples include NBS Steam Tables, Span
and Wagner EOS (CO
2
), Wagner and Pruss
EOS (Water), and heat transfer fluid models.
Historical Development of EOS for Phase Equilibria
Two popular state equations for K-value predictions are the
Benedict-Webb-Rubin (BWR) equation and the Redlich-Kwong
equation.
The original BWR equation
17
uses eight parameters for each
component in a mixture plus a tabular temperature dependence
for one of the parameters to improve the fit of vapor-pressure
data. This original equation is reasonably accurate for light
paraffin mixtures at reduced temperatures of 0.6 and above.
8

The equation has difficulty with low temperatures, non-hydro-
carbons, non-paraffins, and heavy paraffins.
Improvements to the BWR include additional terms for tem-
perature dependence, parameters for additional compounds,
and generalized forms of the parameters.
Starling
20
has included explicit parameter temperature de-
pendence in a modified BWR equation that is capable of pre-
dicting light paraffin K-values at cryogenic temperatures.
The Redlich-Kwong equation has the advantage of a simple
analytical form which permits direct solution for density at spec-
ified pressure and temperature. The equation uses two parame-
ters for each mixture component, which in principle permits pa-
rameter values to be determined from critical properties.
However, as with the BWR equation, the Redlich-Kwong
equation has been made useful for K-value predictions by em-
pirical variation of the parameters with temperature and with
acentric factor
11, 18, 19
and by modification of the parameter-com-
bination rules.
15, 19
Considering the simplicity of the Redlich-
Kwong equation form, the various modified versions predict K-
values remarkably well.
Interaction parameters for non-hydrocarbons with hydro-
carbon components are necessary in the Redlich-Kwong equa-
tion to predict the K-values accurately when high concentra-
tions of non-hydrocarbon components are present. They are
especially important in CO
2
fractionation processes, and in con-
ventional fractionation plants to predict sulfur compound dis-
tribution.
The Chao-Seader correlation
7
uses the Redlich-Kwong equa-
tion for the vapor phase, the regular solution model for liquid-mix-
ture non-ideality, and a pure-liquid property correlation for effects
of component identity, pressure, and temperature in the liquid
phase. The correlation has been applied to a broad spectrum of
compositions at temperatures from 50F to 300F and pressures
to 2000 psia. The original (P,T) limitations have been reviewed.
12
Prausnitz and Chueh have developed
16
a procedure for high-
pressure systems employing a modified Redlich-Kwong equa-
tion for the vapor phase and for liquid-phase compressibility
together with a modified Wohl-equation model for liquid phase
activity coefficients. Complete computer program listings are
given in their book. Parameters are given for most natural gas
components. Adler et al. also use the Redlich-Kwong equation
for the vapor and the Wohl equation form for the liquid phase.
6
The corresponding states principle
10
is used in all the proce-
dures discussed above. The principle assumes that the behavior
of all substances follows the same equation forms and equation
parameters are correlated versus reduced properties and acen-
tric factor. An alternate corresponding states approach is to re-
fer the behavior of all substances to the properties of a reference
25-8
substance, these properties being given by tabular data or a
highly accurate state equation developed specifically for the ref-
erence substance. The deviations of other substances from the
simple critical-parameter-ratio correspondence to the reference
substance are then correlated. Mixture rules and combination
rules, as usual, extend the procedure to mixture calculations.
Leland and co-workers have developed
9
this approach exten-
sively for hydrocarbon mixtures.
Shape factors are used to account for departure from simple
corresponding states relationships, with the usual reference sub-
stance being methane. The shape factors are developed from PVT
and fugacity data for pure components. The procedure has been
tested over a reduced temperature range of 0.4 to 3.3 and for pres-
sures to 4000 psia. Sixty-two components have been correlated
including olefinic, naphthenic, and aromatic hydrocarbons.
The Soave Redlich-Kwong (SRK)
13
is a modified version of
the Redlich-Kwong equation. One of the parameters in the orig-
inal Redlich-Kwong equation, a, is modified to a more tempera-
ture dependent term. It is expressed as a function of the acen-
tric factor. The SRK correlation has improved accuracy in
predicting the saturation conditions of both pure substances
and mixtures. It can also predict phase behavior in the critical
region, although at times the calculations become unstable
around the critical point. Less accuracy has been obtained when
applying the correlation to hydrogen-containing mixtures.
Peng and Robinson
14
similarly developed a two-constant
equation of state in 1976. In this correlation, the attractive
pressure term of the semi-empirical van der Waals equation
has been modified. It predicts the vapor pressures of pure sub-
stances and equilibrium ratios of mixtures.
In applying any of the above correlations, the original criti-
cal/physical properties used in the derivation should be inserted
into the appropriate equations. It is common for one to obtain
slightly different solutions from different computer programs,
even for the same correlation. This can be attributed to differ-
ent pure component and binary parameters, iteration tech-
niques, convergence criteria, and initial estimation values,
among other items as described in the Recent EOS Advance-
ments sub-section below. Determination and selection of a par-
ticular equation of state and interaction parameters must be
done carefully, considering the system components, the operat-
ing conditions, etc.
Recent EOS Advancements While some of the funda-
mental, basic equation of state forms are included at the end of
this section, there have been many advancements in the predic-
tion of phase equilibria and thermodynamic properties since
the last update of this section (pre-1990). As a result, and due to
the extensive use of commercial simulation tools, results which
differ somewhat will likely be obtained for the same Equation of
State, depending on the software chosen, and even options se-
lected within the software. In addition to those items listed in
the Historical development of EOS section above, this is large-
ly due to the advancements made in application of EOS meth-
ods. The following is a brief summary of some basic reasons for
these differences from one software package to another, along
with a general description of advanced applications of EOS
methods.
Improved mixing rules
A number of different mixing rules can be applied to
an EOS, some much more complex than others. In gen-
eral, more complex mixing rules allow for the range of
applicability of an EOS extended further beyond the
available experimental data; however, more experi-
mental data is required to allow for a proper fit of the
mixing equation.
Application of more complex mixing rules can make
EOS methods adequate for polar/non-ideal systems.
Specifically, Activity Coefficient methods have
been used directly in some mixing rules to more ac-
curately predict binary interactions of mixtures
with polar and non-polar components at high pres-
sure, despite the Activity Coefficient method only
being fit to available low pressure experimental
data (i.e., Wong-Sandler).
Enhanced binary interaction parameters
Group contribution methods have been developed to
estimate binary interactions (e.g. Predictive SRK) and
greatly improve predictions especially for mixtures
with polar and non-polar components.
Interaction parameters are typically fitted to ex-
perimental data for each specific EOS and mixing rule
combination. In turn, more quality experimental data
in the pressure, temperature, and compositional region
of a particular application of interest allows for en-
hanced binary interaction parameters and improved
EOS predictions. However, fitting interaction parame-
ters to different sets of data will result in inconsistent
predictions from one tool to another.
Binary interaction parameters are often tempera-
ture dependent, and may be fit by differing tempera-
ture dependency forms, for which proper choice can
impact EOS performance. The ability for a user to spec-
ify binary interaction parameters is included in many
of commercially available simulation products. A tool
that allows for non-constant specification (e.g., includes
temperature dependence) will result in improved re-
sults.
Additional equation terms
Addition of extended or advanced alpha functions
(intermolecular attraction) to improve fitting of vapor
pressure, which can improve the ability of the EOS to
handle polar/non-ideal systems. Addition of volume
translation parameters allow for better prediction of
liquid densities for the EOS.
Liquid phase property handling
Modification of the handling of the term describing
real volume of molecules/intermolecular repulsion al-
low for better prediction of liquid densities for the
EOS.
Solids handling (e.g., ice, hydrate, solid CO
2
, solid hydro-
carbon)
While EOS are used to represent the fugacity of
components in a fluid phase (vapor and liquid), they
can be combined with models representing the fugacity
in the solid phase to model VSE and LSE. See the sec-
tion titled Equations of State and the Solid Phase be-
low for more detailed discussion.
There are a number of multi-parameter equations (i.e.,
GERG
35
), that currently exist and are able to model systems to
within experimental error. However, due to the complexity and
computing power required for these, they are not often used in
25-9
facility design simulations. One obvious use of an equation of
this nature is to generate pseudo-experimental data from which
new binary interaction parameters can be regressed for a spe-
cifc system (P, T, and composition), and in turn used in an EOS
to improve its reliability.
BASIC EQUATION OF STATE FORMS
Refer to original papers for mixing rules for multicomponent
mixtures.
van der Waals
30
Z
3
(1 + B) Z
2
+ AZ AB = 0

A =
aP

R
2
T
2

B =
bP

RT

a =
27 R
2
T
2
c

64 P
c

b =
R T
c

8 P
c
Redlich-Kwong
28
Z
3
Z
2
+ ( A B B
2
) Z AB = 0

A =
aP

R
2
T
2.5

B =
bP

RT

a =

0.42747
R
2
T
2
c
.5

P
c


b =

0.0867
R T
c

P
c

Soave Redlich-Kwong (SRK)


13
Z
3
Z
2
+ (A B B
2
) Z AB = 0

A =
aP

R
2
T
2

B =
bP

RT
a = a
c
a

a
c
=

0.42747
R
2
T
2
c

P
c

a
1

2
= 1 + m ( 1 T
r
1

2
)
m = 0.48 + 1.574 w 0.176w
2

b =

0.08664
R T
c

P
c

Peng Robinson
31
Z
3
(1 B ) Z
2
+ (A 3B
2
2B ) Z (AB B
2
B
3
) = 0

A =
aP

R
2
T
2

B =
bP

RT

a =

0.45724
R
2
T
2
c
a
P
c

a
1

2
= 1 + m (1 T
r
1

2
)
m = 0.37464 + 1.54226 w 0.26992 w
2

b =

0.0778
R T
c

P
c

Benedict-Webb-Rubin-Starling (BWRS)
20, 29

P =
R T
+

B
o
R T A
o

C
o
D
o
E
o
1

V T
2

+
T
3

T
4
V
2

+

bRT a
d 1
+ a

a +
d 1

T V
3
T V
6

+
c 1
1 +

v
2

V
3
T
2
V
2

Note: w, the acentric factor is defined in Section 23.
Other Phase Equilibria Methods
Activity Coefficient Models Another common method
used for the purpose of phase equilibria and thermodynamic
property prediction is the use of Activity Coefficient models.
The following is a brief summary of the basic capabilities and
describes the applicability for some of the more commonly used
Activity Coefficient methods.
Activity coefficient models are the best method for repre-
sentation of highly non-ideal and/or polar systems (i.e.,
aqueous systems, amines, NH
3
, caustic, CO
2
, H
2
S) and
are therefore typically used in the chemicals industry.
While these models are generally only applicable to pre-
diction of phase equilibria for binary and multicompo-
nent mixtures (VLE and LLE), they are not for phase
equilibria of pure components. However, they do require
high quality pure component property predictions (e.g.,
vapor pressure). Depending on the specific Activity Coef-
ficient method, it may not always allow for LLE predic-
tion because tuning the models to VLE specific data or
LLE specific data may result in drastically different pa-
rameters. For this reason, VLLE predictions must also be
used with caution.
Applicable for prediction of thermodynamic properties
for the liquid phase only. Vapor properties are unreliable
and must be calculated using another method; histori-
cally this has been done using ideal gas assumptions for
the vapor phase, but commonly includes more advanced
EOS methods, as described in the Mixed Models section
below.
Limited to systems within the pressure and temperature
ranges of the experimental data it is correlated against.
These models are only suitable for low to moderate
pressure systems, typical in the chemicals industry, be-
cause activity coefficients depend on temperature, but
are independent of pressure, while mutual solubilities
are in fact dependant on pressure in high pressure LLE
systems.
At typical operating pressures, the use of a vapor
pressure is not appropriate for light gases above the
critical point and instead these light gases are treated
as Henrys components, where Henrys law coefficients
are derived from experimental gas solubility data.
25-10
Extrapolation outside the experimental data range
is not recommended.
Requires a separate model to determine liquid density.
The Poynting correction is used to account for pressure
dependence of a components liquid phase activity.
Some examples of common Activity Coefficient models in-
clude: Chien Null, NRTL, Margules, UNIFAC, UNIQUAC, van
Laar, and Wilson.
Electrolytic Models Electrolytic models are a sub-set of
Activity Coefficient models. The general purpose of Electrolytic
models is to handle systems where dissociation of components
is important. In general, these components do not directly par-
ticipate in VLE (i.e., ions or solids that do not dissolve or vapor-
ize), but they often influence activity coefficients of other spe-
cies by reaction or interaction and in turn, indirectly participate
in impacting the phase equilibria of the system. These models
generally require high quality experiment data for the specific
application they are being applied to (e.g., amine gas treating
systems).
Mixed Models Mixed models are commonly used in or-
der to combine the strengths of the various methods described
above. These models commonly use an EOS method for phase
equilibria and prediction of vapor phase thermodynamic prop-
erties, but use an Activity Coefficient or Electrolyte method for
determination of thermodynamic properties and/or density of
the liquid phase(s). However, because of the use of multiple
methods, these models do not always produce consistent predic-
tions, especially at or near the critical point.
Practical Application of Phase
Equilibria Methods
General Considerations While phase equilibria of typ-
ical hydrocarbon systems is modeled very well with currently
available commercial computer simulation tools, there are
many areas where careful attention to method selection is need-
ed. Some systems, due to the thermodynamic complexity, are
difficult to model using the basic methods described above, even
with more and/or better quality experimental data. Some ex-
amples of these complex systems in the gas processing industry
are shown below. The equilibria and thermodynamic property
results obtained using a specific method for these systems
should be carefully evaluated and compared to commercial ex-
perience when used in a process design:
Complex Systems to Model
Operating conditions that cause divergence from ideal
fluid behavior
Self associating/dimerizing (i.e., aqueous solutions)
Dissociating (i.e., amine gas treating, sour water)
Sterically hindered dissociating compounds (i.e.,
certain amines)
Near critical and supercritical fluids (i.e., supercrit-
ical natural gas compression, supercritical H
2
S/CO
2

compression, hydrocarbon systems near critical)
Combinations of non-polar and polar compounds
(i.e., hydrocarbon systems with sour or produced water,
dehydration or dew point depression with glycols or
methanol)
Hydrogen bonding (i.e., methanol-water-hydrocar-
bon systems)
Natural Gas Dew Point Calculations
Other considerations that impact phase equilibria
Reactive (i.e., amine gas treating, caustic treating)
Mass transfer limited (i.e., tertiary and hindered
amines)
Absorption (i.e., CO
2
and H
2
S adsorption in physical
or chemical solvents)
Overall, it is important to understand the capabilities and
limitation of each method of representing phase equilibria and
prediction of thermodynamic properties, and each methods
specific applicability to the gas processing industry for proper
choice and use. However, it should be noted that tuning of a
method to quality experimental data in the region of operation
or interest is perhaps more important than the method choice
itself, specifically, the choice of mixing rules and quality of bi-
nary interaction parameters, which can typically be readily
modified in commercial software.
More specific information relative to phase equilibria meth-
ods can be found in Goodwin et al
36
and Kontogeorgis and Fo-
las.
34
Dew Point Calculation A thermodynamic dew point is
defined as that point where liquid first appears from the gas
phase. An EOS model actually calculates this point with ex-
actly zero liquid dropout. In reality, for natural gas this point
cannot directly be measured from experimental methods, but
can be estimated from PVT data taken very near the dew point
envelope by extrapolating liquid dropout data to 0.0 volume
percent liquid.
In pipeline operations, one can consider a practical dew
point that represents a small volume of liquid condensation
which does not impact pipeline performance, usually represent-
ing a trace of liquid on the pipe wall. This practical dew point is
what is actually measured by the Bureau of Mines chilled mir-
ror device.
In recent work for the GPA by Bullin, et. al., (RR-213), the
practical dew point was defined as 0.00027 m
3
liquid per 1000
m
3
gas. As a natural gas gets leaner, the difference between the
EOS predicted dew point of 0.0 volume percent liquid and the
practical dew point of 0.00027 m
3
liquid per 1000 m
3
gas can
increase to a much as 5.6C. The practical dew point can be
represented by the temperature in a Bureau of Mines chilled
mirror device where droplets begin to form. Thus, when EOS
models are used, both the thermodynamic dew point of 0.0 vol-
ume percent liquid and the practical dew point of 0.00027 m
3

liquid per 1000 m
3
gas should be considered when adjusting the
EOS parameters to match the experimental values, and also to
better determine the conditions that impact plant/pipeline per-
formance. For lean gases where there is more than a 2.2 to 5C
difference in the two values, it may be necessary to only fit to
the practical dew point value to evaluate plant/pipeline perfor-
mance.
This also points to the fact that when fitting an EOS model
to an experimentally obtained dew point value, it should not be
assumed that the reported dew point (experimental) represents
the thermodynamic dew point of 0.0 volume percent liquid, un-
less of course it has been confirmed to be extrapolated from
multiple experimental points within the two phase envelope.
25-11
It cannot be overemphasized that If the EOS is fit to the
wrong dew point condition for some natural gas compositions
(thermodynamic versus practical), it can significantly affect the
amount of liquids that are predicted to condense in plant opera-
tions or in pipelines, affecting overall designs. Also for these
gases with a significant difference in the dew point value, the
volume percent liquid data will display an upward curvature as
the thermodynamic dew point is neared. This may result in a
requirement for a more expanded composition to be used in the
EOS model, perhaps even above C
12
+.
Amine Treating Amine treating is one of the most com-
mon unit operations in the gas processing industry. There are a
variety of amine solution types (i.e. primary, secondary tertiary,
sterically hindered, mixtures, with additives), and solution
strengths, all of which possess characteristics that cause diver-
gence from ideal fluid behavior (self associating/dimerizing, dis-
sociating and potentially sterically hindered, and combinations
of polar and non-polar compounds). In addition they are also
complicated systems to model being reactive, sometimes mass
transfer limited, absorption/desorption systems. This combina-
tion of system attributes presents challenges for predicting
equilibrium and thermodynamic properties accurately. All
thermodynamic phase equilibrium models depend on experi-
mental thermodynamic equilibria data. Many times technology
suppliers, and or/simulation support companies, will use real
plant data to verify a model in area of interest. Some of the most
common models used to describe the phase equilibria of amine
systems are described below.
37
Choice of the appropriate model
should be considered carefully.
Kent-Eisenburg (1976)
38
Model Based on defining chemi-
cal reaction equilibria in the liquid phase. Assumes all ac-
tivity and fugacity coefficients are unity (i.e., ideal solutions
and ideal gases) and forces a fit between experimental and
predicted values by treating two of the reaction equilibrium
constants as variables (e.g., the amine dissociation reaction
and the carbamate formation reation). This approach is
typically only applicable to systems with single acid gases
(CO
2
or H
2
S, but not both) and unless a modified approach
is used, to primary and secondary amines. The model
should not be extrapolated as its thermodynamic basis is
weak and does not support extrapolation.
Deshmukh and Mather (1981)
39
Method Based on De-
bye-Huckel theory and is thermodynamically rigorous.
Similar to the Kent-Eisenburg Model in use of chemical
reactions, but activity coefficients are estimated on the
basis of ion-ion, ion-molecule, and molecule-molecule in-
teraction parameters. Uses either the Suave-Redlich-
Kwong or Peng-Robinson EOS for the vapor phase. An
activity coefficient description of Henrys law is used to
describe the physical solubility of acid gases in the amine
solvents. Binary interaction parameters must be adjust-
ed in order to fit final VLE model to experimental data.
Li and Mather (1994)
40
Method Based on Pitzers Gibbs
excess energy equations
41
and accounts for the various ionic
and molecular species in the liquid phase when acid gas is
dissolved into a mixed amine solution. This is a more mod-
ern rendition of an activity coefficient approach.
Austgen et al. (1991)
42
Electrolytic NRTL based model
This is also an activity coefficient model in the vein of
those discussed in the Elecrolytic and Mixed model sec-
tions above. It is one of the most thermodynamically rig-
orous models, but requires use of binary and ternary in-
teraction parameters in regressing the model to
experimental data.
The Deshmukh and Mather, Li and Mather, and Austgen
methods all use activity coefficient models for the liquid phase,
and EOS models for the vapor phase. Model parameters must
be found for each by regressing the models to experimental
data. The primary limitation is not the models themselves, but
the highly variable, and sometimes unknown quality of the ex-
perimental data. Because the experimental data is often the
limitation and not the model, this may allow any of the activity
coefficient models to be just as reliable as the others for a given
application.
Sour Water Stripping H
2
S, CO
2
, and ammonia, from sour
water is performed in many sour natural gas treating plants.
Common applications include condensed water (H
2
S and CO
2
),
sulfur plant tail gas quench tower blowdown (H
2
S ,CO
2
, and
caustic or ammonia) and to a more limited extent, stripping of
sour produced water (brine, including NaCl and many other
salts, with H
2
S, and CO
2
). Much of the literature describes de-
sign and operation of refinery sour water strippers, which handle
(H
2
S CO
2
, ammonia, cyanides, phenols, and organic acids).
Guidelines specific to gas treating are more limited.
Sour water serves as a good example of the strong effect of pH
on ionic dissociation, and of the smaller, though sometimes sig-
nificant, effect of dissolved salt concentrations on gas solubility.
A principal ionization equilibrium for H
2
S in aqueous solu-
tion,

H
2
S H
+
+ HS

Eq 25-11
is greatly affected by pH. The un-dissociated fraction of the to-
tal H
2
S is close to 1 for a pH of less than 5, about 0.33 at a pH of
7.0, and less than 0.01 at a pH of 9.0
43
. Since the vapor pressure
of H
2
S is produced by its un-dissociated fraction, pH in a strip-
ping tower has a great effect on the stripping efficiency. As H
2
S
gas is stripped from solution, the dissociated ions re-combine to
provide more un-dissociated H
2
S which revives the partial
pressure above the solution. It should be noted that the pH of
the solution changes as CO
2
, is stripped from the water. In some
services, with H
2
S, CO
2
, NH
3
, and low salts, the pH must be
controlled to (typically to 7-8) to insure that both components
can be stripped. Stripping is, of course, enhanced by higher
temperatures.
Salt concentration of tens of thousands of ppm, as is com-
mon in produced waters, reduces the solubility of H
2
S and thus
enhances stripping but this is a smaller influence than tem-
perature, and much smaller than pH. The pH of produced water
is typically above neutral to start, even with the dissolved salts,
because of the natural buffers in the water. The pH is further
elevated, as the CO
2
is stripped from the brine, sometimes re-
sulting in fouling of the mass transfer medium within the strip-
per. Therefore, acid is sometimes added in front of the stripper
to lower the pH and prevent fouling.
A common method to calculate equilibrium of sour water
systems is the Wilson-API-Sour method, and variants, which
use a modification of Van Krevelens approach to account for
the ionization of H
2
S, CO
2
and NH
3
in the aqueous water phase
(See API Publication 955
44

and GPA RR-52
45
). This type of mod-
el is valid up to about 50 psig, with limited other ionic (salt)
species present. To extend the range of the sour equilibrium
prediction, mixed models using a combination of the API-Sour
model, and an EOS (e.g., Peng Robinson) have been developed.
Alternate approaches which account for the influence of other
ionic salts present in the solution are electrolytic and mixed
electrolytic/EOS models.
25-12
A discussion on design of sour water stripper systems is pre-
sented in Section 19. Stevens and Mosher
46
provide a broad re-
view of variations and options, and potential problems areas in
design and operation of sour water systems. Bechok
47
is an old
but traditional reference for sour water equilibria and stripper
design.
EQUATIONS OF STATE
AND THE SOLID PHASE
As will be discussed in some detail later, equations of state
are limited in their ability to predict equilibria involving solid
phases. That makes it desirable to have a way of checking such
predictions made in simulations. Good experimental data, if
available, is the best check. Some ways of checking and/or cor-
recting simulated results are discussed below, after an intro-
duction to the general topic of solid phases.
The Phase Rule, developed by Gibbs back in the 1870s,
still serves as a trusty background to multi-phase equilibria.
For non-reactive compounds, the equation is:
F = C + 2 P Eq 25-12
where C is the number of components or species in the mixture,
P is the number of phases present, and F is the number of de-
grees of freedom of the system. In situations where solids may
be of concern in gas processing, there are usually both liquid
and vapor phases also present. Three numerical examples for
the Phase Rule are shown in the table below, followed by dis-
cussion of their consequences when all three phases co-exist.
C, Number of
Components
P, number
of Phases
Degrees
of
Freedom
Consequence
1, single component 3, V, L, and S 0 Unique Triple Point
2, two components 3, V, L, and S 1 Triple point locus line
3, three components 3, V, L, and S 2
No locus line; much
more complex equilibria
Single-Component Triple Points
The first example, for a single component, is best known. A
unique triple point can be found, for example, on the P-H chart
for CO
2
in this Section. This unique point on such a chart is
useful to better understand the presence of various mixtures of
phases in that area of the chart. However, the numerical values
of the triple point pressure and temperature are not very rele-
vant to process operations. It is highly unlikely that any ther-
modynamic path for a fluid (say, in blowdown) would pass
through the triple point (see two such lines on the examples of
uses for P-H charts, also in this Section). Thus, quantifying a
unique triple point is rarely of help in checking a simulated
prediction.
Two-Component Triple Point Locus Lines
In contrast, in a few industrially important cases, knowl-
edge of the triple point locus line for a two-component system
can provide a very useful check on simulator predictions. The
example below is from a recently published study by Walter et
al
48
that checked, and made corrections to, simulations of blow-
down of a high pressure acid gas line containing (essentially)
only CO
2
and H
2
S.

0
50
100
150
200
250
300
350
400
450
500
-100 -90 -80 -70 -60 -50
P
r
e
s
s
u
r
e
,

k
P
a

(
a
b
s
)
Temperature, C
Tripl e Point l ocus line 44% H2S, 56% CO2 Temperature Correction
In the above diagram, the continuous line marked with tri-
angles is the triple point line for solids deposition in the pres-
ence of both liquid and vapor phases from Sobocinski and Ku-
rata
25
with the data for the line given in the reference document.
For the range of pressures and temperatures shown here, the
solid phase is CO
2
ice. At much lower pressures and tempera-
tures (< 28 kPa (abs) and < 96C) it is possible for solid H
2
S to
precipitate; that region, being of less industrial importance, will
not be discussed here. Sobocinski and Kurata
25
cover this scien-
tifically interesting region (including a eutectic at 87.5 mol %
H
2
S) in detail.
The other two lines coming in from top right are lines for
isenthalpic blowdowns. The dotted line marked with Xs is the
result of a simulation without the benefit of any program exten-
sion to allow solids prediction (see later discussion of such ex-
tensions). This simulation predicted the blowdown would inter-
cept the triple point locus but then go straight through (since
the simulator had no knowledge of the locus) into a region to the
left, which is thermodynamically impossible. Thus, the simula-
tor predicted (a) no solids and (b) falsely low temperatures.
Knowledge of the triple point locus allows a simple manual cor-
rection to the simulated path, by forcing the blowdown to follow
the triple point locus as it must, so long as the three phases
co-exist.
The dashed line marked with circles shows a second, and
smaller, correction that is discussed by Walter et al
48
and mer-
its only brief mention here; very accurate vapor-liquid equilib-
rium data allowed a manual correction of the simulators isen-
thalpic predictions, by adjustment of the vapor/liquid
percentages along the path ahead of the interception of the tri-
ple point locus line.
As pressure reduction and cooling takes place along the lo-
cus line, precipitation of solid CO
2
progressively enriches the
L/V two-phase mixture with H
2
S. It is not intuitively obvious
that a single triple point locus line correctly shows all such P/T
pairs on the graph, regardless of the starting mixture composi-
tion, but that is the case. For clarification of this point, refer to
Sobocinski and Kurata
25
.
This method of using the triple point locus line to check on a
process simulation could be applied to a few other industrially-
important two-component systems, e.g., methane/CO
2
, and
methane/benzene. However, any significant presence of a third
component negates the use of this tool, as discussed next.
25-13
Multi-Component (3 or More) Triple Points
Once a third component is added in significant quantity, the
number of degrees of freedom increases to 2 or more. That
means there are no locus lines that can be drawn on a P-T
graph, but rather a multi-dimensional mass of data that does
not lend itself to simple analysis. Addition of a few percent of
H
2
S or ethane, to a methane/CO
2
mixture, makes for such com-
plexity in many real separation cases.
Simulation Tools for Predicting Solids
Many commercially available process simulators have spe-
cial program extensions that will in some cases give precise pre-
diction of solids formation, or at least a warning that solids-for-
mation threatens in the proposed working region. However,
most predictions are limited to the existence of clathrate hy-
drates. A commercially available extension has recently been
developed, and incorporated into at least one simulator, to pro-
vide a warning of potential CO
2
solids formation. At the time of
writing, one simulator includes solid CO
2
ice formation quanti-
tatively within its normal simulation calculations. Work is on-
going to cover solid species other than CO
2
, but has not yet been
published or commercialized.
Like most equations of state, the Peng-Robinson EOS in its
basic form is intended solely for vapor-liquid-equilibrium calcu-
lations, and is not capable of modeling solids. Program exten-
sions are available for the simulators which, based on the condi-
tions predicted by the simulator EOS, can be used to inform the
engineer whether they are within solids forming conditions.
Such predictions are usually reported by indicating at what
temperature the current stream will form solids if the tem-
perature is colder than the simulator temperature, the process
is outside solids forming conditions, and conversely if the tem-
perature is warmer than the simulator temperature. The engi-
neer can then design to keep a safe margin away from solids
forming conditions. It is important to remember that these cal-
culations rely on the base thermodynamics of the EOS utilized
within the process simulator to give accurate temperatures and
compositions, in order for the calculation predictions to be ac-
curate.
Walter et al report that, when the solids-prediction capabil-
ity of one simulator was utilized, it then accurately predicted
the point of formation of the CO
2
solids, and quantified them
48
.
But even without such simulator capacity, the triple point locus
line from Sobocinski and Kurata
25
had already been used by
hand to make the primary correction (as described by Walter et
al
48
). This could be done for other two-component systems by
the same method, if sufficient triple point data were available.
Systems with three or more components would require much
more complex analysis.
REFERENCES AND BIBLIOGRAPHY
1. Wilson, G. M., Barton, S. T., NGPA Report RR-2: K-Values in
Highly Aromatic and Highly Naphthenic Real Oil Absorber Sys-
tems, (1971).
2. Poettman, F. H., and Mayland, B. J., Equilibrium Constants for
High-Boiling Hydrocarbon Fractions of Varying Characteriza-
tion Factors, Petroleum Refiner 28, 101-102, July, 1949.
3. White, R. R., and Brown, G. G., Phase Equilibria of Complex
Hydrocarbon Systems at Elevated Temperatures and Pressures,
Ind. Eng. Chem. 37, 1162 (1942).
4. Grayson, H. G., and Streed, C. W., Vapor-Liquid Equilibria for
High Temperature, High Pressure Hydrogen-Hydrocarbon Sys-
tems, Proc. 6th World Petroleum Cong., Frankfort Main, III,
Paper 20-DP7, p. 223 (1963).
5. Chappelear, Patsy, GPA Technical Publication TP-4, Low Tem-
perature Data from Rice University for Vapor-Liquid and P-V-T
Behavior, April (1974).
6. Adler, S. B., Ozkardesh, H., Schreiner, W. C., Hydrocarbon Proc.,
47 (4) 145 (1968).
7. Chao, K. C., Seader, J. D., AIChEJ, 7, 598 (1961).
8. Barner, H. E., Schreiner, W. C., Hydrocarbon Proc., 45 (6) 161
(1966).
9. Leach, J. W., Chappelear, P. S., and Leland, T. W., Use of Mo-
lecular Shape Factors in Vapor-Liquid Equilibrium Calculations
with the Corresponding States Principle, AIChEJ. 14, 568-576
(1968).
10. Leland, T. W., Jr., and Chappelear, P. S., The Corresponding
States Principle A Review of Current Theory and Practice,
Ind. Eng. Chem. 60, 15-43 (July 1968); K. C. Chao (Chairman),
Applied Thermodynamics, ACS Publications, Washington, D.
C., 1968, p. 83.
11. Barner, H. E., Pigford, R. L., Schreiner, W. C., Proc. Am. Pet.
Inst. (Div. Ref.) 46 244 (1966).
12. Lenoir, J. M., Koppany, C. R., Hydrocarbon Proc. 46, 249 (1967).
13. Soave, Giorgio, Equilibrium constants from a modified Redlich-
Kwong equation of state, Chem. Eng. Sci. 27, 1197-1203 (1972).
14. Peng, D. Y., Robinson, D. B., Ind. Eng. Chem. Fundamentals 15
(1976).
15. Spear, R. R., Robinson, R. L., Chao, K. C., IEC Fund., 8 (1) 2 (1969).
16. Prausnitz, J. M., Cheuh, P. L., Computer Calculations for High-
Pressure Vapor-Liquid Equilibrium, Prentice-Hall (1968).
17. Benedict, Webb, and Rubin, Chem. Eng. Prog. 47, 419 (1951).
18. Wilson, G. M., Adv. Cryro. Eng., Vol. II, 392 (1966).
19. Zudkevitch, D., Joffe, J., AIChE J., 16 (1) 112 (1970).
20. Starling, K. E., Powers, J. E., IEC Fund., 9 (4) 531 (1970).
21. Holmes, A. S., Ryan, J. M., Price, B. C., and Stying, R. E., Pro-
ceedings of G.P.A., page 75 (1982).
22. Hwang, S. C., Lin, H. M., Chappelear, P. S., and Kobayashi, R.,
Dew Point Values for the Methane Carbon Dioxide System,
G.P.A. Research Report RR-21 (1976).
23. Price, B. C., Looking at CO2 recovery, Oil & Gas J., p. 48-53
(Dec. 24, 1984).
24. Nagahana, K., Kobishi, H., Hoshino, D., and Hirata, M., Binary
Vapor-Liquid Equilibria of Carbon Dioxide-Light Hydrocarbons
at Low Temperature, J. Chem. Eng. Japan 7, No. 5, p. 323 (1974).
25. Sobocinski, D. P., Kurata, F., Heterogeneous Phase Equilibria of
the Hydrogen Sulfide-Carbon Dioxide System, AIChEJ. 5, No. 4,
p. 545 (1959).
26. Ryan, J. M. and Holmes, A. S., Distillation Separation of Carbon
Dioxide from Hydrogen Sulfide, U.S. Patent No. 4,383,841
(1983).
27. Denton, R. D., Rule, D. D., Combined Cryogenic Processing of
Natural Gas, Energy Prog. 5, 40-44 (1985).
28. Redlich, O., Kwong, J. N. S., Chem. Rev. 44, 233 (1949).
25-14
29. Benedict, M., Webb, G. B., Rubin, L. C., An Empirical Equation
for Thermodynamic Properties of Light Hydrocarbons and Their
Mixtures, Chem. Eng. Prog. 47, 419-422 (1951); J. Chem. Phys.
8, 334 (1940).
30. van der Waals, J., Die Continuitat des Gasformigen und Flus-
sigen Zustandes, Barth, Leipzig (1899).
31. Peng, D. Y., Robinson, D. B., A New Two-Constant Equation of
State, Ind. Eng. Chem. Fundamentals 15, 59-64 (1976).
32. RR-76 Hong, J. H., Kobayashi, Riki, Phase Equilibria Studies
for Processing of Gas from CO
2
EOR Projects (Phase II).
33. Case, J. L., Ryan, B. F., Johnson, J. E., Phase Behavior in High-
CO
2
Gas Processing, Proc. 64th GPA Conv., p. 258 (1985).
34. Kontogeorgis, G. M., Folas, K. F., Thermodynamic Models for In-
dustrial Applications: From Classical and Advanced Mixing
Rules to Association Theories, John Wiley and Sons Ltd (2010).
35. Ruhr-Universitat Bochum, Reference Equations of State GERG-
2004 and GERG-2008 for Natural Gases and Other Mixtures,
(2011).
36. Goodwin, A. R. H., Sengers, J. V., Peters, C. J., Applied Thermo-
dynamics of Fluids, RSC Publishing, International Union of Pure
and Applied Chemistry (2010).
37. Kohl, A. L., Nielsen, R. B., Gas Purification, Fifth Edition, pp.
87-91, Gulf Publishing Company, Texas (1997).
38. Kent, R. L., Eisenberg, B., Better Data for Amine Treating, Hy-
drocarbon Processing, Vol. 55, No. 2, February 1976, pp. 87-90.
39. Deshmukh, R. D., Mather, A. E., A Mathematical Model for
Equilibrium Solubility of Hydrogen Sulfide and Carbon Dioxide
in Aqueous Alkanolamine Solutions, Chemical Engineering Sci-
ence, Vol. 36, pp. 355-362 (1981).
40. Li, Y.-G., Mather, A.E., Correlation and Prediction of the Solu-
bility of Carbon Dioxide in a Mixed Ethanolamine Solution, In-
dustrial and Engineering Chemistry Research, Vol. 33, pp. 2006-
2015 (1994).
41. Pitzer, K. S., Ion Interaction Approach Theory and Data Corre-
lation, in Activity Coefficients in Electrolyte Solutions, 2nd Ed.,
Pitzer, K. S., Ed., CRC Press, Boca Raton, FL, Chapter 3, Ap-
pendix I (1991).
42. Austgen, D. M., Rochelle, G. T., Chen, C.-C., Model of Vapor-
Liquid Equilibria for Aqueous Acid Gas-Alkanolamine Systems,
2. Representation of H2S and CO2 Solubility in Aqueous MDEA
and CO2 Solubility in Aqueous Mixtures of MDEA with MEA or
DEA, Industrial and Engineering Chemistry Research, Vol. 30,
p. 543 (1991).
43. Betz Handbook of Industrial Water Conditioning, Inc Betz Labo-
ratories 9th Edition (1991).
44. API Publication 955, A New Correlation of NH3, CO2, and H2S
Volatility Data from Aqueous Sour Water Systems, (1978).
45. Owens, J. L., Cunningham, J. R., Wilson, G. M., Vapor-Liquid
Equilibria for Sour Water Systems with Inert Gases Present,
GPA RR-52 (1982).
46. Stevens, D. K., Mosher, A., Fundamentals of Sour Water Strip-
ping, Brimstone Sulfur Symposia, Vail, CO, September 2008.
47. Beychok. M., Aqueous Wastes from Petroleum and Petro-
chemical Plants, John Wiley and Sons, New York and London
(1967).
48. Walter, F. B., Miller, T. Anderson, S., Prediction of Liquids and
Solids Formation for High Pressure Acid Gas Blowdown, 90th
Annual GPA Convention (2010).
Additional References
See listing in Section 1 for GPA Technical Publications (TP) and Re-
search Reports (RR). Note that RR-64, RR-77, and RR-84 provide exten-
sive evaluated references for binary, ternary, and multicomponent sys-
tems. Also as a part of GPA/GPSA Project 806, a computer data bank is
available through the GPA Tulsa office.
Extensive tabulation of references only is available from Elsevier
Publishers of Amsterdam for the work of E. Hala and I. Wichterle of the
Institute of Chemical Process Fundamentals, Czechoslovak Academy of
Sciences, Prague-Suchdol, Czechoslovakia.
Hiza, M. J., Kidnay, A. J., and Miller, R. C., Equilibrium Properties
of Fluid Mixtures Volumes I and II, IFI/Plenum, New York, 1975. See
Fluid Phase Equilibria for various symposia.
K-DATA CHARTS FOLLOW
AS LISTED BELOW
Methane-Ethane Binary
Nitrogen P
k
2000 psia (13 800 kPa)
Methane through Decane P
k
3000 psia (20 700 kPa)
Hydrogen Sulfide P
k
3000 psia (20 700 kPa)
FIG. 25-8
Critical Locus as Developed for Convergence Pressure
(Formerly used for Convergence Pressure for Hydrocarbons)
25-15
25-16
25-17
25-18
25-19
25-20
25-21
25-22
25-23
25-24
25-25
25-26
25-27
25-28
25-29
25-30

You might also like