Joule Thomson Coefficient

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

Prediction of Joule-Thomson inversion curves

for pure fluids and one mixture by molecular


simulation

Jadran Vrabec ∗ Gaurav Kumar Kedia Hans Hasse


Institut für Technische Thermodynamik und Thermische Verfahrenstechnik,
Universität Stuttgart, D-70550 Stuttgart, Germany

Abstract

The predictive power of a set of molecular models, which have been adjusted to
vapor-liquid equilibria only, is validated. For that purpose, Joule-Thomson inver-
sion curves were determined by molecular simulation for 15 pure fluids, i.e. Argon,
Methane, Oxygen, Nitrogen, Carbon Dioxide, Ethylene, Carbon Monoxide, R11,
R23, R41, R124, R125, R134a, R143a, R152a, and for Air. Comparison of the sim-
ulation results with reference equations of state shows an excellent agreement.

Key words: Joule-Thomson coefficient (C), Thermodynamics (C), Nitrogen (B),


Oxygen (B), Gas mixtures (B)

1 Introduction

The Joule-Thomson (JT) coefficient µJT is defined as derivative of the temper-


ature T with respect to the pressure p at constant enthalpy h
!
∂T
µJT = , (1)
∂p h

∗ Corresponding author, Tel.: +49-711/685-6107, Fax: +49-711/685-6140


Email address: vrabec@itt.uni-stuttgart.de (Jadran Vrabec).
URL: www.itt.uni-stuttgart.de (Jadran Vrabec).

Preprint submitted to Elsevier Preprint 2 July 2004


or equivalently, from standard thermodynamic relations
!
1 ∂h
µJT = − , (2)
cp ∂p T

where cp is the isobaric heat capacity. Depending on state conditions, µJT may
be positive or negative. Positive values imply a cooling of the fluid as it passes
through an adiabatic throttle. The curve connecting all state points where µJT =

0 is the JT inversion curve.

The experimental determination of a fluid’s JT inversion curve is demanding


very precise measurements of volumetric or caloric properties at conditions up

to five times its critical temperature and twelve times its critical pressure. Ex-
perimental JT inversion curve data are therefore scarce, often unreliable [1], and
mostly available only for pure fluids. For a relatively small number of pure flu-
ids, reference equations of state (EOS) have been developed based on thousands

of experimental data points on different thermodynamic properties, e.g. [2] by


Wagner and Pruß which is approved by the International Association for the
Properties of Water and Steam. These reference EOS yield precise results in

the fluid region over a large temperature and pressure range including the JT
inversion curve and are used here for comparison.

Fifteen fluids, i.e. Argon, Methane, Oxygen, Nitrogen, Carbon Dioxide, Ethylene,

Carbon Monoxide, R11, R23, R41, R124, R125, R134a, R143a, R152a, and the
mixture Air are considered. Their intermolecular interactions are described either
by dipolar 2CLJD [8] or quadrupolar 2CLJQ models [9]. In order to calculate the
JT inversion curve for a given molecular model by molecular simulation, several

methods have been proposed in the literature [10,11]. It was not the aim of this
work to optimize the simulative calculation of µJT , so that a straightforward

2
method was used (for more details see below). The main objective of this paper
is to assess the predictive power of molecular models which have been adjusted

exclusively to experimental vapor-liquid equilibria.

2 Molecular models and simulation method

Molecular modeling and simulation offers an interesting approach for predicting

thermodynamic properties over a large range of state points. Most authors who
studied JT inversion curves by molecular simulation investigate the spherical
Lennard-Jones (LJ) model [3–5], which is appropriate only for simple molecules
like Methane and the noble gases. The work of Chacı́n et al. [6] deals with a

more complex fluid, i.e. Carbon Dioxide. They successfully used the two center
Lennard-Jones plus pointquadrupole (2CLJQ) molecular model of Möller and
Fischer [7] which is of the same type as the models sudied here.

In the present work, effective state independent pair potentials based on rigid
Lennard-Jones dumbbells are used. The 2CLJD pair potential is composed of
two identical Lennard-Jones sites a distance L apart (2CLJ) and a pointdipole

with momentum µ placed in the geometric center of the molecule oriented along
the molecular axis

u2CLJD (r ij , ω i , ω j , L, µ) = u2CLJ (r ij , ω i , ω j , L) + uD (rij , ω i , ω j , µ), (3)

wherein u2CLJ is the contribution of the four Lennard-Jones interactions

2 X
2
" 12 6 #
σ σ
X 
u2CLJ (r ij , ω i , ω j , L) = 4 − , (4)
a=1 b=1 rab rab

3
and uD is the dipolar contribution, as given by Gray and Gubbins [12]

1 µ2
uD (r ij , ω i , ω j , µ) = · fD (ω i , ω j ) . (5)
4π0 |r ij |3

In the 2CLJQ pair potential, the dipole is replaced by an axial pointquadrupole

with momentum Q. Thus, in equation (3) the dipolar contribution uD is replaced


by the quadrupolar contribution uQ , as given in [12]

3 Q2
uQ (r ij , ω i , ω j , Q) = fQ (ω i , ω j ) . (6)
4 |r ij |5

Herein r ij is the center-center distance vector of two molecules i and j, rab is one
of the four Lennard-Jones site-site distances; a counts the two sites of molecule i,

b counts those of molecule j. The vectors ω i and ω j represent the orientations of


the two molecules i and j. fD and fQ are two different trigonometrical functions
depending on these molecular orientations, cf. [12]. The Lennard-Jones parame-
ters σ and  represent size and energy, respectively. So 2CLJD and 2CLJQ models

have four model parameters: σ, , L, and µ or Q. These parameters have been


adjusted exclusively to experimental vapor-liquid equilibria for 80 pure fluids in
previous work of our group [8,9]. The spherical LJ fluid, which is not polar, is a
limiting case where L = 0, µ = 0, and Q = 0. Table 1 summarizes the parameters

of the molecular models for all 15 pure fluid models investigated here.

In order to perform simulations of Air, a molecular mixture model has to be


available. Here only the three main components, i.e. Nitrogen (0.7818 mol/mol),
Oxygen (0.2095 mol/mol), and Argon (0.0092 mol/mol) are considered. On the
basis of existing models for pure fluids, molecular modeling of mixtures reduces to

modeling the interactions between unlike molecules. Following prior work [13,14],
a modified Lorentz-Berthelot combining rule with one adjustable binary interac-

4
tion parameter ξ is used for each unlike Lennard-Jones interaction

σA + σ B
σAB = , (7)
2

AB = ξ· A · B . (8)

Table 2 gives the three binary interaction parameters needed for the Air model.
The dipolar and quadrupolar interactions are treated in a physically straightfor-
ward way, without the use of binary parameters.

In molecular simulations of LJ based fluids, thermodynamic properties are re-


duced with the LJ parameters σ and , i.e. reduced temperature T ∗ = T kB /,

where kB = 1.38066 · 1023 J/K is the Boltzmann constant, and reduced pressure
p∗ = pσ 3 /. These reduced properties are used in Figures 1 and 2.

Molecular dynamics simulations were performed in the isobaric-isothermal (NpT )


ensemble, using Anderson’s barostat [15] and isokinetic velocity scaling [16]. After
10 000 equilibration time steps, the residual enthalpy

N X
N
hres =<
X
u(rij , ω i , ω j ) > +p < v > −kB T, (9)
i=1 j>i

was averaged over 200 000 time steps, where the first term indicates the simula-
tion average over the intermolecular potential energy and < v > is the average
volume. In order to calculate the statistical uncertainty of hres a block averaging

method following [17] was used.

Depending upon the density of the state point, the reduced membrane mass

parameter for the barostat M m/σ 4 was chosen from 10−3 to 10−7 , where m is
the molecular mass. The intermolecular interactions were evaluated explicitly up

5
to a cut-off radius of rc = 5σ and standard long range corrections, employing
angle averaging as proposed by Lustig [18], were used.

Initially, a total number of 500 molecules were placed in a fcc lattice configuration
in a cubic simulation box. The initial density of the system was chosen close to

that expected from an EOS, if available, otherwise estimates were used.

To calculate one JT inversion curve pressure for a given temperature, a series

of simulations, generally from 5 to 10, were made around the expected pressure,
covering typically a rather large pressure range of 20 MPa. These isothermal
enthalpies were fitted with a weighted polynomial of third degree as a function
of pressure. The inversion point was simply assumed to be the minimum of that

function.

In order to estimate the statistical uncertainty of the inversion pressure at a

given temperature, two additional polynomials of third degree were fitted to


both the upper and lower bounds of the error bars of the simulated enthalpies.
The difference in the minimum pressures of upper and lower polynomial was
considered to be the statistical uncertainty of the JT inversion pressure. This

was done for Argon as an unpolar fluid, Carbon Monoxide as a dipolar fluid, and
Oxygen as a quadrupolar fluid at three different temperatures. The maximum
uncertainty for the JT inversion pressure by this estimation was in the range of
3 - 4%.

3 Results

In this chapter the present simulation results of JT inversion curves for 15 pure
fluids and Air are presented in pressure-temperature diagrams. The simulation

6
results are compared with reference EOS if the JT inversion curve lies inside
the range of their validity. In general, the predictions from simulation agree

quantitatively with respective EOS data. Unfortunately, for some fluids the EOS
is not valid in the full span of the JT inversion curve. In Figures 2 to 7 also the
critical points are given to illustrate the upper bound of the experimental data
used for the parametrization of the molecular models.

3.1 Spherical, nonpolar fluids

In Figure 1, the present JT inversion curve of the Lennard-Jones fluid is shown


and compared with the several available molecular simulation data sets from
the literature and two physically based EOS for the Lennard-Jones fluid. For
low temperatures almost up to the maximum pressure, an excellent agreement

between all results is found. The EOS of Mecke et al. [19] and Kolafa and Nezbeda
[20] yield almost indistinguishable results for the JT inversion curve. Only near
the maximum a significant deviation of 5% between these EOS and the present
data is found. Also most of the other simulation data yield similar deviations.

The results of Kioupis et al. [5], with both the extended and constraint method,
are in good agreement with the present results and the EOS also on the high
temperature side.

Figure 2 compares the JT inversion curve of the Lennard-Jones fluid with those
of Argon and Methane. The data for these two fluids is obtained from reference
EOS by Tegeler et al. [21] for Argon and Setzmann and Wagner [22] for Methane.
This data is reduced to the usual LJ units using the parameters given in Table 1

taken from Vrabec et al. [9], cf. Section 2. Since these parameters were obtained
by an adjustment to experimental vapor-liquid equilibria only, Figure 2 shows

7
an extrapolation over a large region. It can be seen that the JT inversion curves
of the two real fluids are almost identical in the reduced units. The agreement

between the LJ fluid and the real fluids is excellent on the low temperature
side but at the maximum, systematic deviations of 5% are found. On the high
temperature side again good agreement is observed. Thus Figure 2 proves that
the predictions of the two LJ models for Argon and Methane are very reliable,

and underlines the appropriateness of that model.

3.2 Polar fluids

Figure 3 shows the JT inversion curve simulation data for Carbon Dioxide to-
gether with results from the reference EOS of Span and Wagner [24] and simula-
tion data of Chacı́n et al. [6]. Up to 800 K, the agreement between the simulation

results and the EOS is excellent, but for higher temperatures systematic devi-
ations of up to 11% are observed. The molecular simulation results of Chacı́n
et al. [6] show basically same course as those from the present work, but with
a larger scatter. It has to be noted that the deviations between simulation and

EOS systematically increase as the limit of the validity of the EOS is approached.

Figure 4 presents the JT inversion curve data for three different quadrupolar
fluids Oxygen, Nitrogen, and Ethylene and the dipolar fluid Carbon Monoxide.

They are compared with the reference EOS of Schmidt and Wagner [25], Span
et al. [26], Smukala et al. [27], and Lemmon and Span [28], respectively. The
agreement between the simulation data and the results from the EOS is almost
perfect on the low temperature side of the JT inversion curve for all fluids. Slight

deviations occur in some cases on the high temperature side, but even there they
hardly exceed 5%.

8
Figure 5 shows the JT inversion curves of three Methane derivatives, i.e. R41,
R23, and R11. The present simulation data is compared to EOS (R41 [29], R23

[30], and R11 [31]), which are valid only for a limited part of the low temperature
side. In that region an excellent agreement is found. The present simulation data
sets are smooth and cover the full span of the JT inversion curves.

Figures 6 and 7 depict the simulation results for JT inversion curves of five
different Ethane derivatives together with the according EOS, i.e. R143a [32],

R134a [33], R125 [34], R152a [35], and R124 [36]. The EOS are valid only for a
limited part of the low temperature side, where they agree excellently with the
present simulation data.

3.3 Mixtures

The predictive power of molecular models is validated by the results of the JT


inversion curve of the mixture Air. The simulation results presented in Figure
8 agree excellently with the EOS results from REFPROP [37]. The maximum
systematic deviation is found to be in the range of 2%.

4 Conclusions

In this work, the predictive power of molecular models, fitted to experimental


vapor-liquid equilibria [8,9], is assessed by predicting the JT inversion curves of
15 pure fluids and one mixture, i.e. Air. The comparison of the present results
with reference EOS shows an excellent agreement throughout the full span of JT

inversion curves. Thus the investigated molecular models can be used reliably
for the simulation of molecular scale phenomena, where both thermal and caloric

9
properties are crucial. The models are also valid for mixtures like Air.

5 Acknowledgment

We gratefully acknowledge financial support by Deutsche Forschungsgemein-


schaft, Sonderforschungsbereich 412, University of Stuttgart.

References

[1] Hiza MJ, Kidnay AJ, Miller RC. Equilibrium Properties of Fluid Mixtures. New
York: IFI/Plenum, 1975.

[2] Wagner, W, Pruß, A. The IAPWS Formulation 1995 for the Thermodynamic
Properties of Ordinary Water Substance for General and Scientific use. J. Phys.
Chem. Ref. Data 2002;31:387-535.

[3] Heyes DM, Llaguno CT. Computer simulation and equation of state study of the
Boyle and inversion temperature of simple fluids. Chem. Phys. 1992;168(1):61-68.

[4] Colina CM, Muller EA. Molecular Simulation of Joule-Thomson Inversion Curves.
Int. J. Thermophys. 1999;20(1):229-235.

[5] Kioupis LI, Arya G, Maginn EJ. Pressure-enthalpy driven molecular dynamics
for thermodynamic property calculation II. Applications. Fluid Phase Equilibria
2002;200(1):93-110.

[6] Chacı́n A, Vázquez JM, Muller EA. Molecular Simulation of Joule-Thomson


Inversion Curve of carbon dioxide. Fluid Phase Equilibria 1999;165(2):147-155.

[7] Möller D, Fischer J. Determination of an effective intermolecular potential for


carbon dioxide using vapor-liquid phase equilibria from NpT + test particle
simulations. Fluid Phase Equilibria 1994;100:35-61.

[8] Stoll J, Vrabec J, Hasse H. A set of molecular models for Carbon monoxide and
halogenated hydrocarbons. J. Chem. Phys. 2003.

[9] Vrabec J, Stoll J, Hasse H. A set of molecular models for symmetric quadrupolar
fluids. J. Phys. Chem B 2003;105(48):12126-12133.

[10] Colina C, Muller EA. Joule-Thomson inversion curves by molecular simulation.


Molec. Sim. 1997;19:237-246.

10
[11] Kioupis LI, Maginn EJ. Pressure-enthalpy driven molecular dynamics for
thermodynamic property calculation I. Methodology. Fluid Phase Equilibria
2002;200(1):75-92.

[12] Gray CG, Gubbins KE. Theory of molecular fluids. Fundamentals. Oxford:
Clarendon Press, 1984.

[13] Stoll J,Vrabec J, Hasse H. Vapour-Liquid equilibria of mixtures containing


nitrogen, oxygen, carbon dioxide, and ethane. AICHE J. 2003;49(8):2187-2198.

[14] Vrabec J, Stoll J, Hasse H. Molecular models of unlike interactions in mixtures. J.


Phys. Chem. B 2004; submitted.

[15] Andersen HC. Molecular dynamics simulations at constant pressure and/or


temperature. J. Chem. Phys. 1980;72(4):2384-2393.

[16] Allen MP, Tildesley DJ. Computer Simulation of Liquids. Oxford: Clarendon
Press, 1987.

[17] Fincham D, Quirke N, Tildesley DJ. Computer simulation of molecular


liquid mixtures. I. A diatomic Lennard-Jones model mixture for CO 2 /C2 H6 .
J. Chem. Phys. 1986;84(8):4535-4546.

[18] Lustig R. Angle-average for the powers of the distance between two separated
vectors. Molec. Phys. 1988;65:175-179.

[19] Mecke M, Müller A, Winkelmann J, Vrabec J, Fischer J, Span R, Wagner W.


An accurate van der Waals type equation of state for Lennard-Jones fluid. Int. J.
Thermophys. 1996;17(2):391-404 and 1998;19(5):1493.

[20] Kolafa J, Nezbeda I. The Lennard-Jones fluid: an accurate analytic and


theoretically based equation of state. Fluid Phase Equilibria 1994;100:1-34.

[21] Tegeler Ch, Span R, Wagner W. A new equation of state for argon covering the
fluid region for temperatures from melting line to 700 K at pressures up to 1000
MPa. J. Phys. Chem. Ref. Data 1999;28(3):779-850.

[22] Setzmann U, Wagner W. A new equation of state and tables of the thermodynamic
properties for methane covering the range from the melting line to 625 K at
pressures up to 1000 MPa. J. Phys. Chem. Ref. Data 1991;20(6):1061-1155.

[23] Lotfi A, Vrabec J, Fischer J. Vapour liquid equilibria of the Lennard-Jones fluid
from the NpT plus test particle method. Molec. Phys. 1992;76:1319-1333.

[24] Span R, Wagner W. A new equation of state for carbon dioxide covering the fluid
region from the triple-point temperature to 1100 K at pressures up to 800 MPa.
J. Phys. Chem. Ref. Data 1996;25(6):1509-1596.

[25] Schmidt R, Wagner W. A new form of the equation of state for pure substances
and its application to oxygen. Fluid Phase Equilibria 1985;19(3):175-200.

11
[26] Span R, Lemmon EW, Jacobsen RT, Wagner W, Yokozeki A. A reference equation
of state for the thermodynamic properties of nitrogen for temperatures from 63.151
to 1000 K and pressures to 2200 MPa. J. Phys. Chem. Ref. Data 2000;29(6):1361-
1433.

[27] Smukala J, Span R, Wagner W. New equation of state for ethylene covering the
fluid region from the melting line to 450 K at pressures up to 300 MPa. J. Phys.
Chem. Ref. Data 2000;29(5):1053-1121.

[28] Lemmon EW, Span R. Preliminary equation 1997, in [37].

[29] Outcalt SL. MBWR equation of state as reported in Haynes W M. Thermophysical


properties of HCFC alternatives, National Institute of Standards and Technology,
Boulder, Colorado, Final Report for ARTI MCLR 1996; Project Number 660-
50800, in [37].

[30] Penoncello SG, Lemmon EW, Shah Z, Jacobsen RT. An equation of state for the
calculation of the thermodynamic properties of trifluoromethane (R-23). Ashrae
Transact. 2000;106:739-756, in [37].

[31] Jacobsen RT, Penoncello SG, Lemmon EW. A fundamental equation for
trichlorofluoromethane (R-11). Fluid Phase Equilibria 1992;80:45-56, in [37].

[32] Lemmon EW, Jacobsen RT. An International Standard Formulation for the
Thermodynamic Properties of 1,1,1-Trifluoroethane (HFC-143a) for Temperatures
from 161 to 450 K and Pressures to 50 MPa. J. Phys. Chem. Ref. Data
2000;29(4):521-552, in [37].

[33] Tillner-Roth R, Baehr HD. An international standard formulation of the


thermodynamic properties of 1,1,1,2-tetrafluoroethane (HFC-134a) covering
temperatures from 170 K to 455 K at pressures up to 70 MPa. J. Phys. Chem.
Ref. Data 1994;23(5):657-729, in [37].

[34] Lemmon EW, Jacobsen RT. Preliminary formulation 2002, in [37].

[35] Outcalt SL, McLinden MO. A modified Benedict-Webb-Rubin equation of state


for the thermodynamic properties of R152a (1,1-difluoroethane). J. Phys. Chem.
Ref. Data 1996;25(2):605-636, in [37].

[36] de Vries B, Tillner-Roth R, Baehr HD. Thermodynamic Properties of HCFC


124, 19th International Congress of Refrigeration, The Hague, The Netherlands,
International Institute of Refrigeration 1995;IVa:582-589, in [37].

[37] REFPROP, NIST Standard Reference Database 23, Version 7.0:2002.

12
Table 1
Parameters of the pure fluid molecular models, taken from [8,9].

Fluid σ/Å (/kB ) /K L/Å µ/D Q/DÅ

Ar 3.3952 116.79 - - -
CH4 3.7281 148.55 - - -
O2 3.1062 43.183 0.9699 - 0.8081
N2 3.3211 34.897 1.0464 - 1.4397
CO2 2.9847 133.22 2.4176 - 3.7938
C2 H4 3.7607 76.950 1.2695 - 4.3310
CO 3.3009 36.897 1.1405 0.7378 -
R11 (CFCl3 ) 4.0213 224.15 3.3377 2.7009 -
R23 (CHF3 ) 3.2643 123.56 2.5670 2.1607 -
R41 (CH3 F) 3.0382 137.64 2.4530 1.8850 -
R124 (CHFCl−CF3 ) 3.8852 192.25 3.8852 3.2190 -
R125 (CHF2−CF3 ) 3.6861 162.77 3.6861 2.8245 -
R134a (CH2 F−CF3 ) 3.6138 175.12 3.6138 3.0214 -
R143a (CH3−CF3 ) 3.5960 165.04 3.5395 2.7470 -
R152a (CH3−CHF2 ) 3.5168 182.01 3.3125 2.7354 -

Table 2
Binary interaction parameters, taken from [13,14].

Mixture ξ

Ar + O2 0.988
N2 + Ar 1.008
N2 + O 2 1.007

13
List of Figures

1 Joule-Thomson inversion curve of the Lennard-Jones fluid in

reduced units. Simulation: • present results,  Colina and Muller


[4], 4 Kioupis et al. [5] extended method, 5 Kioupis et al. [5]
constraint method, EOS: — Mecke et al. [19], - - - Kolafa and
Nezbeda [20]. 16

2 Joule-Thomson inversion curves of Argon and Methane in

reduced units. Simulation: • present results, EOS: — Argon,


Tegeler et al. [21], - - - Methane, Setzmann and Wagner [22],
critical point: ◦ Lotfi et al. [23]. 17

3 Joule-Thomson inversion curve of Carbon Dioxide. Simulation:


• present results, 4 Chacı́n et al. [6], EOS: — Carbon Dioxide,
Span and Wagner [24], critical point: ◦ calculated from EOS. 18

4 Joule-Thomson inversion curves of Oxygen, Nitrogen, Ethylene,


and Carbon Monoxide. Present simulation: • Oxygen, 
Nitrogen, N Ethylene, H Carbon Monoxide, EOS: — Oxygen,
Schmidt and Wagner [25], Nitrogen, Span et al. [26], Ethylene,

Smukala et al. [27], - - - Carbon Monoxide, Lemmon and Span


[28], critical points: empty symbols calculated from EOS. 19

14
5 Joule-Thomson inversion curves of three Methane derivatives.
Present simulation: • R41, N R23, H R11, EOS: — R41, Outcalt

[29], R23, Penoncello et al. [30], R11, Jacobsen et al. [31], critical
points: empty symbols calculated from EOS. The dashed lines
are guides to the eye. 20

6 Joule-Thomson inversion curves of two Ethane derivatives.


Present simulation:  R143a, H R134a, EOS: — R143a, Lemmon
and Jacobsen [32], R134a, Tillner-Roth and Baehr [33], critical

points: empty symbols calculated from EOS. The dashed lines


are guides to the eye. 21

7 Joule-Thomson inversion curves of three Ethane derivatives.


Present simulation: • R152a, N R125,  R124, EOS: R125,
Lemmon and Jacobsen [34], R152a, Outcalt and McLinden

[35], R124, de Vries et al. [36], critical points: empty symbols


calculated from EOS. The dashed lines are guides to the eye. 22

8 Joule-Thomson inversion curve of Air. Present simulation: •,


EOS: — REFPROP [37]. 23

15
Fig. 1. Vrabec et al.

16
Fig. 2. Vrabec et al.

17
Fig. 3. Vrabec et al.

18
Fig. 4. Vrabec et al.

19
Fig. 5. Vrabec et al.

20
Fig. 6. Vrabec et al.

21
Fig. 7. Vrabec et al.

22
Fig. 8. Vrabec et al.

23

You might also like