Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Author:
Date:
December 20008
Course:
Contents
Summary.......................................................................................................................... 1
1.
2.
Equations of state..................................................................................................... 1
2.1.
2.2.
2.3.
3.
3.1.1.
Nitrogen.................................................................................................... 6
3.1.2.
Argon ........................................................................................................ 7
3.1.3.
Oxygen ...................................................................................................... 7
3.1.4.
3.2.
4.
Mixture contributions...................................................................................... 8
4.2.
Vapour-liquid equilibrium............................................................................. 11
5.
4.2.1.
4.2.2.
4.2.3.
5.2.
6.
Conclusion............................................................................................................... 18
References ...................................................................................................................... 19
A.1.
A.2.
A.3.
A.4.
A.5.
ii
Summary
A Matlab model has been created that can be used to calculate various single phase
thermodynamic properties and vapour-liquid equilibrium conditions of nitrogenargon-oxygen mixtures. The model can be used during the design of cryogenic air
distillation columns. It is based on an empirical equation of state that is explicit in
Helmholtz energy and describes the real mixture properties as an excess
contribution to the Helmholtz energy. The accuracy of the Matlab model is around
0.1% for the relevant single phase thermodynamic properties and around 0.5% for
vapour-liquid equilibrium conditions.
1.
Air distillation
Distillation of air is currently the most commonly used technique for the production
of pure oxygen and nitrogen on an industrial scale. Atmospheric air is a mixture of
nitrogen, oxygen, argon, carbon dioxide, water and other trace elements. In an air
separation process, carbon dioxide and water are typically removed from this
mixture before it enters the distillation unit. The air that is fed to a distillation
column can therefore be properly considered as a mixture of nitrogen, oxygen and
argon only; their mole fractions are given in Table 1 [1].
Table 1: Composition of air
Component
N2
0.7812
O2
0.2096
Ar
0.0092
Total
1.0000
The design of a distillation column depends heavily on the thermodynamic data that
is used as basis. It is therefore desired to use thermodynamic data that is as
accurate as possible. The aim of this essay is to present a model for the calculation of
accurate thermodynamic data for nitrogen-argon-oxygen mixtures.
2.
Equations of state
2.1.
P=
RT
= RT
v
(2.1)
Here, R is the universal gas constant. Other well-known pressure explicit functional
forms are the virial equation, with temperature dependent constants B , C , D , ,
P=
RT B C D
+ + + + ... ,
v v 2 v3 v 4
(2.2)
P=
RT
a
2,
vb v
(2.3)
P=
RT
a (T )
,
v b v (v + b )
(2.4)
P=
RT
a(T )
v b v (v + b ) + b (v b )
(2.5)
RT B0 RT A0 C0 / T 2 bRT a a
c
P=
+
+
+ 6 + 3 2 1 + 2 exp 2 .
2
3
v
v
v
v
vT v
v
(2.6)
The virial equation of state describes the deviation from ideal gas behaviour as an
infinite power series expansion in terms of the volume. The Van der Waals,
Soave-Redlich-Kwong and Peng-Robinson equations of state are also knows as cubic
equations of state, because they describe the non-ideal part with a term that is cubic
in the specific volume.
These equations of state can also be used to describe mixtures, although the results
are generally less accurate than for pure components. For mixtures, the constants
that are used become composition dependent. They can be calculated using certain
mixing rules and additional binary interaction parameters.
2.2.
During the development of the early equations of state, the main aim was to obtain
more accurate expressions without increasing the complexity too much, because
calculations were often still executed without a computer. The wide-spread
availability of computational speed opened they way for more complex and accurate
equations of state, based on an empirical rather than an analytical approach [2][3].
The number of model parameters increased to more than 30, which resulted in the
name multi-parameter equations of state. This name is a bit misleading though,
because the early equations of state also use multiple parameters, although less.
The term empirical equations of state will be used in this work.
Along with the change to empirical equations of state came the change from a
pressure explicit functional form to a Helmholtz energy explicit functional form. The
main advantage of using the Helmholtz energy is that all other thermodynamic
properties can be determined by differentiation only, which makes the equation of
state a fundamental equation. If a pressure explicit expression is used, integration is
required too. When translating this difference into practice, it means that more
complex functional forms can be used if an equation of state is explicit in Helmholtz
energy.
The first developments in the field of empirical equations of state were aimed at
optimising the function form that was used. The goal was to obtain a very high
accuracy while using as less parameters as possible. When developing a high
accuracy equation of state for a wide range of conditions, a large number of
experimental results is required to fit it to. The first efforts were therefore mainly
focused on well-measured substances, and several very high accuracy equations of
state have been developed in this way. Because the accuracy of calculated
thermodynamic properties is within the experimental error of experiments, they can
be used as reference for these properties. This is why these equations of state are
also called reference equations of state.
The functional forms of reference equations of state are rather substance specific.
Because there is also a desire for equations of state for less well-measured
substances, more general functional forms have been being developed as well. This
has been done by optimising the functional form using the experimental data of a
group of relatively similar and well-measured substances, and applying this
functional form to other less well-measured substances of the same group. These
3
equations of state are called technical equations of state. Their accuracy is still high,
but not as high as is the case for the reference equations of state. The number of
parameters used in technical equations of state is typically less than for reference
equations of state.
Another development within field of empirical equations of state is the extension of
its application from pure substances to multi-component mixtures, again using
certain mixing rules and additional binary interaction parameters.
2.3.
In the near past, early equations of state for nitrogen, argon, oxygen and mixtures of
these components still received some attention in literature; see for example the
work of Asami and Ebisu for the pure components [4]-[6], and the work of Hwang for
mixtures [7]. But today, publications on the subject concern empirical equations of
state only. In contrary, most commercial simulation software still uses the early
equations of state as basis for its calculations.
Both technical as well as reference equations of state are available for pure nitrogen,
argon and oxygen [8]-[13]. Technical equations of state are easier to implement in
calculations because the functional form is exactly the same for each substance and
because less parameters are involved. This makes them more attractive than the
reference equations of state. On the other hand, the reason for developing technical
equations of state was to provide an equation of state for substances that are less
well-measured, not because the calculations involved with reference equations of
state are too demanding. So if both are available, one should always choose to use
reference equations of state. They are simply more accurate.
Several empirical equations of state for air and mixtures of nitrogen, argon and
oxygen in general have been published. The first ones modelled the mixture
properties based on an extended corresponding states approach. This means that the
mixture is modelled as if it is a pure substance, using mixing rules and composition
dependent shape factors to relate the mixture to a reference substance [14]. Later
equations of state quantify the deviation of the real mixture properties from the
ideal mixture properties in terms of an excess contribution, again using mixing rules
and composition dependent shape factors.
Some of the most recent equations of state are limited to a fixed ternary mixture
composition [15], to variable compositions for binary mixtures only [16], or to the
liquid phase only [17]. This makes them unsuitable for use in the current work,
since variable ternary mixture compositions for both liquid and vapour phase are
required during distillation column modelling. There are only two publications that
satisfy these criteria; one is focused on nitrogen-argon-oxygen mixtures only [18],
and the other is a more general model that can be used for mixtures of more varied
4
substances [19]. The former one can be interpreted as the reference equation of state
for the nitrogen-argon-oxygen mixture, and the latter as a technical equation of
state for this mixture. It is therefore the former one that is used as basis for the
current work.
3.
Tc ,i
i ( , ) = i
ai (T , )
=
c ,i
RT
(3.1)
Here, Tc ,i is the critical temperature, c ,i the critical pressure and ai the molar
Helmholtz energy of a pure component. The dimensionless Helmholtz energy of a
mixture ( mix ) is built up out of two parts; a part that is related to the ideal mixture
properties and a part that is related to the non-ideal mixture properties:
n
mix = xi ( i + ln xi ) + E
(3.2)
i =1
Here, xi are the mole fractions of the n components. The last term is called the
section 3.2. The dimensionless Helmholtz energy of a pure component is also built
up out of an ideal and a non-ideal part:
i = i0 + ir
Here,
i0
(3.3)
ir
the dimensionless
residual Helmholtz energy of a pure component. They are discussed in more detail in
section 3.1.
The dimensionless Helmholtz energy of a mixture can also be split up into an overall
ideal part, containing the ideal mixture as well as the ideal gas parts, and an overall
real part, containing the residual as well as the excess parts:
i =1
ideal
real
mix = mix
+ mix
= xi ( i0 + ln xi ) + xi ir + E
n
i =1
(3.4)
3.1.
[unit]
Nitrogen
Argon
Oxygen
[g.mol-1]
28.0135
39.9480
31.9988
[K]
126.192
150.687
154.581
[MPa]
Critical temperature
Critical pressure
Critical density
3.396
4.863
5.043
-3
[mol.dm ]
11.184
13.407
13.630
[K]
63.151
83.806
54.361
[kPa]
12.523
68.891
0.1463
-3
24.07
101.5
0.3237
-3
[mol.dm ]
30.957
35.465
40.816
[K]
77.355
87.302
90.188
[mmol.dm ]
3.1.1.
Nitrogen
The equation of state used for nitrogen is described by Span et al. [10]. The
dimensionless ideal gas Helmholtz energy is given by
0 = ln + N1 ln + N 2 + N3 + N 4 1 +
(3.5)
N 5 2 + N 6 3 + N 7 ln (1 exp ( N8 ) )
r = N k i j
k
32
k =1
36
k = 33
ik
k =7
ik
j exp ( l
k
exp k ( 1) k ( k )
jk
(3.6)
3.1.2.
Argon
The equation of state used for argon is described by Tegeler et al. [11]. The
dimensionless ideal gas Helmholtz energy is given by
0 = ln + N1 + N 2 + 1.5ln
(3.7)
r = Nk i j
k
37
k =1
41
k = 38
ik
k =13
ik
j exp ( l
k
exp k ( 1) k ( k )
jk
+
2
(3.8)
3.1.3.
Oxygen
The equation of state used for oxygen is described by Schmidt and Wagner [12]. The
dimensionless ideal gas Helmholtz energy is given by
0 = ln ( / 0 ) + N1 1.5 + N 2 2 + N 3 ln + N 4 +
2
N 5 ln ( exp ( N 7 ) 1) + N 6 ln 1 + exp ( N8 ) + N 9
3
(3.9)
r = Nk i j
k
k =1
N k ik jk exp ( 2 ) +
24
k =14
32
k = 25
ik
j exp ( 4 ) (3.10)
k
Here, N k , ik , and jk are empirical parameters, given in Appendix A.1 in Table 11.
3.1.4.
Ancillary equations
In addition to the equations of state described in sub-sections 3.1.1 till 3.1.3, also
some ancillary equations are given that can be used to determine thermodynamic
properties at phase changes. The values of these pure component properties can be
used to obtain initial guesses for vapour-liquid equilibrium calculations of mixtures.
The ancillary equations for nitrogen and argon are described in the references given
in sub-sections 3.1.1 and 3.1.2; those for oxygen are described by Stewart et al. [13].
The ancillary equations can be used to calculate the saturated vapour pressure ( P ),
the saturated liquid density ( ' ) and the saturated vapour density ( '' ):
P T n
ln = c N k ik
Pc T k =1
(3.11)
' n
ln = N k ik
c k =1
(3.12)
n
'
= 1 + N k i
c
k =1
(3.13)
'' Tc n
ln = N k ik
c T k =1
(3.14)
Equations (3.11) and (3.14) can be used for nitrogen, argon as well as oxygen.
Equation (3.12) should be used for nitrogen and argon only, and equation (3.13) for
oxygen only. Here, N k and ik are empirical parameters, given in Appendix A.1 in
Table 12, Table 13 and Table 14, and further
= 1
3.2.
T
Tc
(3.15)
Mixture contributions
The equation of state for mixtures of nitrogen, argon and oxygen is described by
Lemmon et al. [18]. It is based on equation (3.4), with the dimensionless excess
Helmholtz energy given by
i =1 j = i +1
(3.16)
Here, Fij are empirical mixture parameters, given in Appendix A.1 in Table 15.
Equation (3.16) involves the reduced density and reciprocal reduced temperature.
When considering pure components the calculation of these quantities uses the
critical density and critical temperature. But when considering mixtures,
composition dependent reducing parameters for the density ( red ) and for the
temperature ( Tred ) should be used:
3
Tred = xiTc ,i +
i =1
red
xx
i =1 j = i +1
(3.17)
ij
2
3
3 xi
=
+ xi x jij
i =1
i =1 j = i +1
c ,i
(3.18)
Here, ij and ij are empirical mixture parameters, given in Appendix A.1 in Table
15. Only the real part of equation (3.4) should be evaluated using these reducing
parameters to calculate the reduced density and temperature. The ideal part still
uses the critical densities and temperatures, resulting into
n
Tc ,i
mix = xi i0
i =1
4.
+ ln xi +
c ,i
x
i =1
r
i
Tred
,
T red
E Tred
,
+
T red
(3.19)
Calculation routines
4.1.
The single phase thermodynamic properties that are most relevant when modelling
a distillation column are the pressure, the molar enthalpy ( h ) and the molar entropy
( s ); they are given by
r
P
= 1+
RT
(4.1)
0 r
r
h
=
+
+
+1
RT
(4.2)
0 r
s
0
r
=
+
(4.3)
Other thermodynamic properties that are often used to verify calculation routines
are the molar internal energy ( u ), the isochoric molar heat capacity ( Cv ), the
isobaric molar heat capacity ( C p ) and the speed of sound ( w ). They are given by
0 r
u
=
+
RT
(4.4)
2 0 2 r
Cv
= 2
+
2 2
R
(4.5)
r
2 r
2 0
Cp
2 r
2
=
+
+
2
2
r
2 r
R
2
1 + 2
2
r
2 r
+
2 r
r
w2
2
= 1 + 2
+
2
1
2 0
2 r
2
RT
+
Mw
2 2
(4.6)
(4.7)
at constant composition.
All Helmholtz energy derivatives that are used in equations (4.1) till (4.7) are given
explicitly in Appendix A.2. Table 16, Table 17 and Table 18 contain the relevant
partial derivatives of the ideal gas Helmholtz energies of nitrogen, argon and oxygen
respectively. The relevant partial derivatives of the residual Helmholtz energy can
be calculated from Table 19. If equations (3.6), (3.8) and (3.10) are compared, three
types of terms can be distinguished; simple polynomial terms, terms containing a
simple exponential and terms with more complicated exponentials. The derivatives
of each of these terms can be obtained by multiplying the original terms with their
corresponding expressions given in the table. Table 20 contains the relevant partial
derivatives of the excess Helmholtz energy.
10
4.2.
Vapour-liquid equilibrium
The basic principle on which the distillation concept is built, is that the most volatile
component in the mixture concentrates in the vapour phase and the least volatile
component in the liquid phase. It is therefore essential for the modelling of a
distillation column to know how each of the components gets distributed over the
two phases. In practice this distribution is often expressed using equilibrium
constants ( K i ), they are defined as the ratios between the mole fractions in the
vapour phase ( yi ) and the mole fractions in the liquid phase ( xi )
Ki =
yi
xi
(4.8)
A three component two phase system involves ten variables, namely a temperature,
a pressure and three mole fractions in each of the phases. Since the mole fractions in
a single phase must add up to one, there are only eight independent variables. When
two phases are at chemical equilibrium, this means that the temperature, pressure
and chemical potentials of the components ( i ) are equal in both phases. These five
dependencies leave three degrees of freedom. This derivation is known as the phase
rule for non-reacting systems. It can be used to determine four practical equilibrium
calculations. In each of the calculations, the composition of one phase is fixed in
combination with either the temperature of the pressure. The result of the
calculation is given by the composition of the other phase and the pressure if the
temperature was fixed or the temperature if the pressure was fixed. Each type of
calculation has been given a name based on the quantities that are calculated; an
overview is given in Table 3.
Table 3: Equilibrium calculations
Calculation type
Dew point temperature
Dew point pressure
Bubble point temperature
Bubble point pressure
Flash
Fixed quantities
Calculated quantities
P , yi
T , yi
P , xi
T , xi
T , P , zi
T , xi
P , xi
T , yi
P , yi
yi , xi , f v
As can be seen in Table 3, a fifth common calculation type exists; the flash
calculation. In this calculation, the total mole fractions ( zi ) are fixed in addition to
the temperature and pressure, and the vapour and liquid phase mole fractions are
calculated in combination with the molar vapour fraction ( f v ).
11
4.2.1.
ni T ,V , n
i =
(4.9)
j
All derivatives with respect to the numbers of moles are calculated numerically in
this work, as suggested by Lemmon et al. [18].
Using Newtons method, the numbers of moles of the components are continuously
updated until the chemical potentials are equal. In the case of a flash calculation,
the total composition is given and the distribution of the components over both
phases is solved for. The vector containing the changes in number of moles for each
iteration step ( ni ) is then given by
v l
n = n = iv + li
n
j n j
v
i
l
i
v
i
il )
with i = j = 3
(4.10)
Here, the first right-hand side term is a three-by-three matrix containing the partial
derivatives of the chemical potentials with respect to all components. Similar to the
change vector, the second right-hand side term is a vector with length three.
In the other four types of calculations listed in Table 3 the composition of one phase
is fixed and the composition of the other phase is solved for. If the vapour phase
12
composition is fixed, the changes in number of moles for the liquid phase can be
calculated using
il
n = l
n
j
l
i
l
i
iv )
with i = j = 3
(4.11)
A complication for all calculations is that the total number of moles in one or both
phases changes, which changes the pressures of the affected phases. An obvious way
to obtain or keep the desired pressure is to vary the volumes (or densities) of the
affected phases. In theory it is possible to include the required volume changes in
the iteration scheme as well. The additional partial derivatives that are needed are
those for the chemical potentials and pressures with respect to the volumes.
However, in practice this approach proves to require a very accurate initial guess,
making it unsuitable for the current work. Therefore a different solution strategy
has been used. It is explained in more detail for the case of a bubble point pressure
calculation in sub-section 4.2.2; and it can be applied to all calculations except the
flash. But, with the results of the other calculations, an accurate initial guess can be
made for a flash calculation based on Newtons method with simultaneous updates
in all numbers of moles and volumes, as is discussed in sub-section 4.2.3.
4.2.2.
The aim of a bubble point pressure calculation is to find the pressure and liquid
phase composition that are in equilibrium with a given vapour phase composition at
a given temperature.
The chosen strategy to find the solution is to start by picking two pressures, based
on an estimation of the mixtures saturated vapour pressure. At these two chosen
pressures, Newtons method is used to find the numbers of moles in the liquid phase
that result into chemical equilibrium. Next, the found liquid composition is
normalised and the chemical potentials in the liquid phase are calculated using this
new composition. The difference between these chemical potentials and the fixed
chemical potentials in the vapour phase can then be translated into a quantity that
represents the overall deviation ( d ). At the bubble point pressure this deviation is
zero. It is relatively easy to find the zero point, because the deviation is almost a
perfect linear function of the pressure. A schematic of the calculation routine is
shown in Figure 1, Matlab code for this routine can be found in Appendix A.3.
13
Calculate d =
(1
v
i
/ il )
Estimate P at which d = 0
Find equilibrium nil for P, normalise nil, calculate d
|d| < at P ?
No
Yes
Pbub = P
&
xi = nil
In practise, an accurate bubble point pressure is often found after a single iteration.
The accuracy of this value is checked by calculating the overall deviation at this
estimated pressure. If the accuracy is too low, the first estimation can be used to
obtain a more accurate estimate for the bubble point pressure. The accuracy can
continuously be improved by adding more pressures, but in practise the maximum
number of required iterations for = 104 is two.
For a dew point pressure calculation, the liquid and vapour phase compositions in
Figure 1 should simply be swapped. If a bubble point temperature calculation is
performed, the pressure and temperature should be swapped. And in the case of a
dew point temperature calculation, both liquid and vapour phase compositions, and
pressure and temperature should be swapped.
4.2.3.
Flash calculation
Using routines for bubble and dew point temperature calculations similar to the
routine discussed in sub-section 4.2.2, a flash calculation becomes straightforward. A
schematic of the calculation routine is shown in Figure 2, Matlab code for the entire
routine can be found in Appendix A.4.
14
Stop
Yes
Estimate fv
Estimate niv & nil
Estimate Vv & Vl
Find equilibrium niv, nil, Vv & Vl at P & T
Calculate yi, xi & fv
The first step is to check whether the specified input lies in the two-phase region of
the mixture. This is done by calculating the bubble and dew point temperatures of
the mixture at the specified pressure. If the specified temperature does not lie in
between these two temperatures the calculation is terminated because the specified
conditions lie in a single phase region. Otherwise, an estimation for the vapour
phase fraction is made based on the calculated and specified temperatures, using an
empirical equation
T Tbub
fv =
Tdew Tbub
0.6
(4.12)
Using this vapour phase fraction, estimations for the vapour and liquid phase
compositions are made. The composition of the smallest phase is estimated using the
difference between the overall composition and the bubble or dew point composition.
The composition of the largest phase is then obtained by subtracting the smallest
phase from the total composition. If the vapour phase is the smallest one, the
compositions are estimated using
niv = f v zi + ( yibub zi ) (1 f v )
nil = zi niv
(4.13)
(4.14)
Here it is assumed that the total number of moles equal unity. If the liquid phase is
the smallest phase, the compositions are estimated using
15
nil = (1 f v ) zi + ( xidew zi ) f v
niv = zi nil
(4.15)
(4.16)
Using these estimates for the compositions, estimates for the volumes of the phases
can be calculated via the densities. Then the final step is to find the compositions
and volumes at which the two phases are at equilibrium, using Newtons method for
all the compositions and volumes simultaneously. The result of this iteration step
then simply leads to the desired values of the vapour phase fraction and the vapour
and liquid phase mole fractions.
5.
This chapter compares the results of the Matlab model that has been developed to
execute the calculation discussed in chapter 4 with literature values of the
calculated properties as reported by Lemmon et al. [18]. All calculations concern air.
Sub-section 5.1 compares the results for the single phase thermodynamic properties
and sub-section 5.2 compares vapour-liquid equilibrium results. In order to compare
calculated values with reference values, Lemmon et al. use the average absolute
deviation ( AAD ), which is defined as
X icalc
1 n
AAD = 1 ref
n i =1
Xi
(5.1)
5.1.
The performance of the developed Matlab model has been assessed in the state space
region that is most relevant for air distillation; temperatures between 60 and
16
Table 4: Overview of the average absolute deviations for thermodynamic properties of air
P / MPa
0.101325
0.200000
0.500000
1.000000
calculated
literature
AAD in u
0.0009
0.0007
0.0005
0.0004
0.0006
0.0010
AAD in h
0.0007
0.0006
0.0005
0.0004
0.0005
0.0010
AAD in s
0.0005
0.0005
0.0004
0.0004
0.0004
0.0010
AAD
0.0025
0.0032
0.0036
0.0033
0.0031
0.0048
Table 4 shows the average absolute deviations for each thermodynamic property at
different pressures. Each value is the average over the entire temperature range,
including both liquid and vapour phases. The average over all thermodynamic
properties is given in the last column. At the bottom of the table, the averages over
all pressures are given as well; they can be compared to the shown literature values.
The AAD s of the isochoric heat capacity and the velocity are comparable, but
the AAD s for the other calculated thermodynamic properties are about half of the
literature values. This means that in the worst case the accuracy of the Matlab
model calculations is only slightly lower than the accuracy of the literature values.
5.2.
The performance of the vapour-liquid part of the developed Matlab model has been
assessed by comparing calculated bubble and dew point pressures with literature
values. These literature values are not obtained by a method similar to the method
used in this work; they are obtained using ancillary equations that have been fitted
to experimental data. Table 5 shows a comparison between the literature and
calculated values of the bubble and dew point pressures at various temperatures,
along with the absolute deviations ( AD ).
Table 5: Overview of the bubble and dew point pressures
Pdew / MPa
Pbub / MPa
T/K
AD
literature calculated
literature calculated
60
0.00555
0.00555 0.0008 0.00258
0.00261
65
0.01432
0.01432 0.0003 0.00776
0.00781
70
0.03191
0.03191 0.0000 0.01943
0.01950
75
0.06333
0.06333 0.0000 0.04228
0.04231
80
0.11462
0.11461 0.0001 0.08232
0.08223
85
0.19262
0.19266 0.0002 0.14665
0.14627
90
0.30475
0.30478 0.0001 0.24320
0.24226
95
0.45886
0.45882 0.0001 0.38047
0.37887
100 0.66313
0.66319 0.0001 0.56742
0.56484
105 0.92606
0.92534 0.0008 0.81341
0.80952
110 1.25642
1.25441 0.0016 1.12824
1.12265
115 1.66327
1.65902 0.0026 1.52226
1.51475
120 2.15573
2.14825 0.0035 2.00674
1.99777
AAD
0.0008
-
AD
0.0106
0.0063
0.0036
0.0007
0.0012
0.0026
0.0039
0.0042
0.0046
0.0048
0.0050
0.0050
0.0045
0.0044
17
The absolute deviations averaged over all temperatures are given at the bottom of
the table. The accuracy of the literature values for the bubble point pressures that is
given by Lemmon et al. amounts 0.0057. So the Matlab model, which is based on the
mixture model described in chapter 3, performs well compared to the literature
values, which are obtained from an equation explicitly fitted for the bubble point
pressure of air as a function of temperature.
6.
Conclusion
A modern empirical equation of state for mixtures of nitrogen, argon and oxygen has
successfully been implemented in a Matlab model. The equation of state is explicit in
Helmholtz energy and describes the real mixing properties as an excess Helmholtz
energy. The Matlab model can be used to calculate various single phase
thermodynamic properties and vapour-liquid equilibrium conditions. The accuracy of
the Matlab model is comparable to the accuracy of the literature model that is used
as basis. The average absolute deviation of the relevant single phase thermodynamic
properties is below 0.1% and the average absolute deviation of the vapour-liquid
equilibrium conditions is below 0.5%. The Matlab model is suitable for use during
the design of an air distillation column.
18
References
[1]
[2]
[3]
[4]
[5]
[7]
[10]
Tegeler, Ch.; Span, R.; Wagner, W. A new equation of state for argon
covering the fluid region for temperatures from the melting line to 700 K at
pressures up to 1000 MPa. J. Phys. Chem. Ref. Data. 1999, 28, 779.
[12]
Schmidt, R.; Wagner, W. A new form of the equation of state for pure
substances and its application to oxygen. Fluid Phase Equilibirua. 1985, 19,
175.
[13]
19
[15]
[16]
Estela-Uribe, J.F. A Helmholtz energy model for air and binary mixtures of
nitrogen, oxygen and argon. Fluid Phase Equilibria, 2006, 243, 171.
[17]
[18]
Lemmon,
E.W.;
Jacobsen,
R.T.;
Penoncello,
S.G.;
Friend,
D.G.
20
A.1.
Model parameters
Table 6: Parameters for 0, equation (3.5) for N2, (3.7) for Ar and (3.9) for O2
k
Nk for N2
Nk for Ar
Nk for O2
2.5E+0
8.31666243E+0
-0.740775E-3
-12.76952708E+0
-4.94651164E+0
-0.664930E-4
-0.784163E-2
0.250042E+1
-1.934819E-4
-0.214487E+2
-1.247742E-5
0.101258E+1
6.678326E-8
-0.944365E+0
1.012941E+0
0.145066E+2
26.65788E+0
0.749148E+2
0.414817E+1
Nk
ik
jk
lk
9.24803575275E-01
0.25
-4.92448489428E-01
0.875
6.61883336938E-01
0.5
-1.92902649201E+00
0.875
-6.22469309629E-02
0.375
3.49943957581E-01
0.75
5.64857472498E-01
0.5
-1.61720005987E+00
0.75
-4.81395031883E-01
10
4.21150636384E-01
1.25
11
-1.61962230825E-02
3.5
12
1.72100994165E-01
13
7.35448924933E-03
0.5
14
1.68077305479E-02
15
-1.07626664179E-03
16
-1.37318088513E-02
2.75
17
6.35466899859E-04
0.75
18
3.04432279419E-03
2.5
19
-4.35762336045E-02
20
-7.23174889316E-02
21
3.89644315272E-02
22
-2.12201363910E-02
23
4.08822981509E-03
21
Nk
ik
jk
lk
24
-5.51990017984E-05
25
-4.62016716479E-02
16
26
-3.00311716011E-03
11
27
3.68825891208E-02
15
28
-2.55856846220E-03
12
29
8.96915264558E-03
12
30
-4.41513370350E-03
31
1.33722924858E-03
32
2.64832491957E-04
16
33
1.96688194015E+01
34
-2.09115600730E+01
35
1.67788306989E-02
36
2.62767566274E+03
33
20
325
1.16
34
20
325
1.16
35
15
300
1.13
36
25
275
1.25
Nk
ik
jk
lk
8.8722304990011E-02
7.0514805167298E-01
0.25
-1.6820115654090E+00
-1.4909014431486E-01
2.75
-1.2024804600940E-01
-1.2164978798599E-01
4.0035933626752E-01
0.25
-2.7136062699129E-01
0.75
2.4211924579645E-01
2.75
10
5.7889583185570E-03
11
-4.1097335615341E-02
12
2.4710761541614E-02
0.75
13
-3.2181391750702E-01
14
3.3230017695794E-01
3.5
15
3.1019986287345E-02
22
Nk
ik
jk
lk
16
-3.0777086002437E-02
17
9.3891137419581E-02
18
-9.0643210682031E-02
19
-4.5778349276654E-04
20
-8.2659729025197E-05
10
0.5
21
1.3013415603147E-04
10
22
-1.1397840001996E-02
23
-2.4455169960535E-02
24
-6.4324067175955E-02
25
5.8889471093674E-02
26
-6.4933552112965E-04
27
-1.3889862158435E-02
10
28
4.0489839296910E-01
13
29
-3.8612519594749E-01
14
30
-1.8817142332233E-01
11
31
1.5977647596482E-01
14
32
5.3985518513856E-02
33
-2.8953417958014E-02
14
34
-1.3025413381384E-02
35
2.8948696775778E-03
36
-2.2647134304796E-03
24
37
1.7616456196368E-03
22
38
5.8552454482774E-03
39
-6.9251908270028E-01
40
1.5315490030516E+00
41
-2.7380447449783E-03
38
20
250
1.11
39
20
375
1.14
40
20
300
1.17
41
20
225
1.11
Nk
ik
jk
3.983768749E-01
-1.846157454E+00
1.5
23
Nk
ik
jk
4.183473197E-01
2.5
2.370620711E-02
-0.5
9.771730573E-02
1.5
3.017891294E-02
2.273353212E-02
1.357254086E-02
-4.052698943E-02
2.5
10
5.454628515E-04
11
5.113182277E-04
12
2.953466883E-07
13
-8.687645072E-05
14
-2.127082589E-01
15
8.735941958E-02
16
1.275509190E-01
3.5
17
-9.067701064E-02
5.5
18
-3.540084206E-02
19
-3.623278059E-02
20
1.327699290E-02
21
-3.254111865E-04
8.5
22
-8.313582932E-03
23
2.124570559E-03
6.5
24
-8.325206232E-04
10
5.5
25
-2.626173276E-05
22
26
2.599581482E-03
11
27
9.984649663E-03
18
28
2.199923153E-03
11
29
-2.591350486E-02
23
30
-1.259630848E-01
17
31
1.478355637E-01
18
32
-1.011251078E-02
23
Nk for N2
ik
Nk for Ar
ik
Nk for O2
ik
-6.12445284
-5.9409785
-6.043938
1.26327220
1.5
1.3553888
1.5
1.175627
1.5
-0.765910082
2.5
-0.46497607
-0.994086
-1.77570564
-1.5399043
4.5
-3.456781
3.361499
24
Table 13: Parameters for , equation (3.12) for N2 and Ar, and (3.13) for O2
k
Nk for N2
1.48654237
0.3294
1.5004262
0.334
1.507678
1/3
-0.280476066
2/3
-0.3138129
2/3
0.85810805
2/3
0.089414309
8/3
0.086461622
7/3
0.19035504
-0.119879866
35/6
-0.041477525
ik
Nk for Ar
ik
Nk for O2
ik
Nk for N2
ik
Nk for Ar
ik
Nk for O2
ik
-1.70127164
0.34
-1.70695656
0.345
-1.498431
1/3
-3.70402649
5/6
-4.02739448
5/6
-2.116826
2/3
1.29859383
7/6
1.55177558
-0.905713
-0.561424977
13/6
-2.30683228
13/3
-5.65999
5/3
-2.68505381
14/3
-18.90964
-53.780774
ij [dm3.mol-1]
Fij [-]
ij [K]
N2-Ar
1.121527
-1.237713
-7.6031E-07
N2-O2
-0.85635
-4.1847E-07
Ar-O2
0.597203
-2.115126
4.1232E-07
A.2.
ln + N1 ln + N 2 + N 3 + N 4
N1 + N 3 N 4
2 0
2
N1 + 2N 4
2N 5
+ 6N 5
+ N 5
3N 6
+ 12N 6
+ N 6
3
+ N 7 ln (1 exp ( N 8 ) )
+ N 7 N 8
N 7 N8
2
1
exp ( N 8 ) 1
exp ( N 8 )
( exp ( N ) 1)
2 0
2
ln + N1 + N 2 + 1.5 ln
N 2 + 1.5
1.5
25
ln ( / 0 ) + N1
2 0
2
+ N 3 ln + N 4 +
N 5 ln ( exp ( N 7 ) 1) + N 6 ln (1 + 2 / 3 exp ( N 8 ) ) + N 9
+ N 2
1.5
2
3
4
N1
N1
1.5
1.5
+ 2N 2
+ 6N 2
+ N 3 + N 4 + N 5 N 7
N3 N5 N7
2
exp ( N 7 )
exp ( N 7 ) 1
exp ( N 7 )
( exp ( N ) 1)
N 6 N 8
N6 N8
2
exp ( N 8 )
1 + 2 / 3 exp ( N 8 )
exp ( N 8 )
(1 + 2 / 3 exp ( N ) )
Table 19: Terms needed for the calculation of partial derivatives of r. The derivatives of the
terms in the top row are obtained by multiplying them with the terms in other rows.
jk
2 r
2
ik ( ik 1)
2 r
2
jk ( jk 1)
2 r
ik jk
Nk k
i
jk
ik
N k k
jk
exp
ik lk
lk
Nk k
i
jk
lk
lk
)( i
1 lk
lk
jk 2 k ( k )
lk
)l
2
(i
lk
jk ( jk 1)
(i
ik 2k ( 1)
lk
jk
(i
exp k ( 1) k ( k )
lk
)j
(j
k
(i
2k ( 1) ) ik 2 k
2
2 k ( k ) ) jk 2 k
2
2k ( 1) ) ( jk 2 k ( k ) )
2 3
2 1.4
2 1.5
xi x j Fij ( N1 + N 2 )
i =1 j =i +1
2 3
2 1.4
2 1.5
xi x j Fij ( 2 N1 + 2 N 2 )
i =1 j =i +1
2
3
2 1.4
2 1.5
xi x j Fij ( 1.4 N1 + 1.5 N 2 )
i =1 j =i +1
2 E 2 3
2 1.4
2 1.5
xi x j Fij ( 2 N1 + 2 N 2 )
2
i =1 j =i +1
2 E 2 3
2 1.4
2 1.5
xi x j Fij ( 3.36 N1 + 0.75 N 2 )
2
i =1 j =i +1
2
r
2
3
2 1.4
2 1.5
xi x j Fij ( 2.8 N1 + 3 N 2 )
i =1 j =i +1
26
A.3.
27
28
A.4.
29
30
A.5.
u
h
s
K
60.00
70.00
78.90
81.72
90.00
100.00
110.00
120.00
-3
mol.dm
33.039
31.590
30.220
0.15504
0.13938
0.12446
0.11256
0.10281
-1
J.mol
-6566.7
-6016.0
-5520.9
-241.3
-60.9
153.7
366.1
577.2
-1
-1
-1
J.mol
J.mol .K
-6563.6 26.807
-6012.8 35.298
-5517.6 41.957
412.2 116.094
666.0 119.053
967.8 122.233
1266.3 125.078
1562.8 127.658
-1
-1
J.mol .K
33.74
31.76
30.27
21.35
21.11
20.99
20.91
20.85
-1
w
-1
J.mol .K
54.92
55.31
56.07
31.01
30.39
29.99
29.74
29.57
u
h
s
K
60.00
70.00
80.00
85.39
87.99
90.00
100.00
110.00
120.00
-3
mol.dm
33.044
31.597
30.053
29.165
0.29156
0.28376
0.25111
0.22580
0.20545
-1
J.mol
-6567.5
-6017.0
-5460.6
-5155.9
-148.9
-103.1
120.2
338.5
553.8
-1
-1
-1
J.mol
J.mol .K
-6561.4 26.795
-6010.7 35.283
-5453.9 42.716
-5149.0 46.405
537.0 112.192
601.7 112.919
916.7 116.239
1224.2 119.171
1527.3 121.808
-1
-1
J.mol .K
33.75
31.77
30.11
29.34
21.71
21.59
21.28
21.11
21.00
-1
w
-1
J.mol .K
54.90
55.29
56.18
57.03
32.36
32.04
31.06
30.50
30.13
u
h
s
K
60.00
70.00
80.00
90.00
96.12
98.36
100.00
110.00
120.00
-3
mol.dm
33.060
31.617
30.079
28.401
27.270
0.69445
0.67801
0.59595
0.53467
-1
J.mol
-6569.8
-6020.1
-5464.8
-4896.8
-4537.6
-35.8
5.7
247.9
478.6
-1
-1
-1
J.mol
J.mol .K
-6554.6 26.756
-6004.3 35.239
-5448.2 42.663
-4879.2 49.363
-4519.3 53.232
684.2 106.834
743.2 107.429
1086.8 110.706
1413.8 113.551
-1
-1
J.mol .K
33.76
31.79
30.13
28.75
28.06
22.66
22.47
21.81
21.47
-1
K
mol.dm
60.00 33.085
70.00 31.649
80.00 30.122
90.00 28.461
100.00 26.581
106.22 25.224
108.10 1.3831
110.00 1.3379
120.00 1.1566
-1
J.mol
-6573.6
-6025.3
-5471.8
-4906.6
-4317.0
-3926.7
6.9
63.8
336.6
-1
-1
-1
J.mol
J.mol .K
-6543.4 26.692
-5993.7 35.165
-5438.6 42.575
-4871.5 49.254
-4279.4 55.489
-3887.0 59.294
730.0 102.412
811.2 103.157
1201.2 106.554
-1
-1
J.mol .K
54.87
55.24
56.10
57.91
59.88
36.30
35.67
33.36
32.13
J.mol .K
33.78
31.81
30.16
28.78
27.70
27.21
24.08
23.63
22.46
-1
m.s-1
1044.4
952.8
857.7
804.7
181.5
184.1
195.9
206.8
216.9
w
-1
u
h
s
-3
m.s-1
1043.8
952.2
867.6
177.4
187.3
198.3
208.6
218.3
m.s-1
1046.1
954.7
860.1
760.9
695.8
185.5
187.9
200.9
212.5
w
-1
J.mol .K
54.82
55.16
55.96
57.67
61.18
65.37
43.69
41.90
36.89
m.s-1
1048.9
957.8
864.0
766.1
659.2
584.3
185.8
189.2
204.3
31