Fluid Phase Equilibria 202 (2002) 2947

Vaporliquid equilibrium data for the

propane + 1,1,1,2,3,3,3-heptafluoropropane (R227ea) system at
temperatures from 293.16 to 353.18 K and pressures up to 3.4 MPa
A. Valtz, C. Coquelet, A. Baba-Ahmed, D. Richon
Centre dEnergtique, Ecole Nationale Suprieure des Mines de Paris CENERG/TEP,
35 Rue Saint Honor, 77305 Fontainebleau, France
Received 21 December 2001; received in revised form 1 March 2002; accepted 8 March 2002

Abstract
Isothermal vaporliquid equilibrium (VLE) data for the propane + 1,1,1,2,3,3,3-heptafluoropropane (R227ea)
binary system were measured at 293.16, 303.14, 313.14, 333.15, 343.16 and 353.18 K and pressures up to 3.5 MPa.
The experimental method, used in this work, is of the static-analytic type. It takes advantage of two pneumatic capillary samplers (RolsiTM , Armines patent) developed in the Cenerg/TEP laboratory. The peculiarity
of R227eapropane binary system is to present azeotropic behavior at each studied temperature.
The six sets of isothermal P, x, y data are represented with the SoaveRedlichKwong (SRK) equation of state
(EoS) and several mixing rules involving the NRTL model.
2002 Published by Elsevier Science B.V.
Keywords: Vaporliquid equilibrium; 1,1,1,2,3,3,3-Heptafluoropropane; R227eapropane binary system

1. Introduction
In 1987, the modification of the Montreal protocol has prohibited the use and the production of chlorofluorocarbons (CFCs) in industrialized nations. Accurate knowledge of the thermophysical properties
of mixtures containing HFC and hydrocarbons, which are proposed as alternative refrigerants is of great
importance to evaluate the performance of refrigeration cycles and to determine the optimum composition
of new working fluids.
During the Second Workshop on Refrigerants held at Ecole des Mines in Paris, France, under the
auspices of IUPAC, Dr. J. Morley (DuPont Fluoroproducts, Hemel Hempstead, UK) addressed the question are we near an industry standard for refrigerant properties? He discussed the various important

Corresponding author. Tel.: +33-164694965; fax: +33-164694968.

E-mail address: richon@paris.ensmp.fr (D. Richon).
0378-3812/02/$ see front matter 2002 Published by Elsevier Science B.V.
PII: S 0 3 7 8 - 3 8 1 2 ( 0 2 ) 0 0 0 5 6 - 0

30

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

world-wide activities which were taking place to determine accurate thermophysical properties of candidate alternative refrigerant working fluids.
Knowledge of vaporliquid equilibrium (VLE) data for new mixtures allows to choose those offering
the best suitable thermodynamic properties. The development of models for representation and prediction
of physical properties and phase equilibria, as well as the improvement of current equations of state cannot
be handled seriously without VLE data.
Using an apparatus based on a static-analytic method, the isothermal VLE measurements on the
propane + 1,1,1,2,3,3,3-heptafluoropropane (R227ea) binary mixture were performed at temperatures
between 293.16 and 353.18 K. The obtained data are fitted using the SoaveRedlichKwong (SRK)
equation of state (EoS).
2. Experimental section
2.1. Materials
Propane is from Messer Griesheim with a certified purity higher than 99.95 vol.%. The R227ea used
was purchased from Solvay Fluor und Derivate (Hannover, Germany) and had a certified purity higher
than 99.99 vol.%. No further purification was performed before use.
2.2. Apparatus and experimental procedures
The apparatus used in this work (Fig. 1) is based on a static-analytic method with liquid and vapor
phase sampling. This apparatus is similar to that described by Laugier and Richon [1].
The equilibrium cell is immersed in a thermo-regulated liquid bath, the temperature control is performed
within 0.01 K. In order to perform accurate measurement of temperature in the equilibrium cell and to
check for thermal gradients, two platinum resistance thermometer probes (Pt100) are inserted inside
wells drilled directly into the body of the equilibrium cell at different levels (see Fig. 1) and connected
to an HP data acquisition unit (HP34970A). These two Pt100 are carefully and periodically calibrated
against a 25
reference platinum resistance thermometer (TINLEY Precision Instruments). The resulting
uncertainty is not higher than 0.02 K. The 25
reference platinum resistance thermometer was calibrated
by the Laboratoire National dEssais (Paris) based on the 1990 International Temperature Scale (ITS 90).
Pressures are measured by means of a pressure transducer (Druck, type PTX611, range 00.6 MPa)
connected to the same HP data acquisition unit (HP34970A) as the two Pt100; the pressure transducer
is maintained at constant temperature (temperature higher than the highest temperature of the study) by
means of a home-made air-thermostat, which is controlled by a PID regulator (WEST instrument, model
6100). The uncertainty in the pressure measurements is estimated to be within 0.3 kPa, as a result
of a pressure calibration against a dead weight balance (Desgranges and Huot 5202S, CP 0.340 MPa,
Aubervilliers, France).
The HP online data acquisition system is connected to a personal computer through a RS-232 interface.
This complete system allows real time readings and recording of temperatures and pressures all along
the different isothermal runs.
The analytical work was carried out using a gas chromatograph (VARIAN model CP-3800) equipped
with a thermal conductivity detector (TCD) connected to a data acquisition system (BORWIN ver 1.5,
from JMBS, Le fontanil, France). The analytical column is HaySep T 100/120 Mesh (silcosteel tube,

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

31

Fig. 1. Flow diagram of the equipment: C, carrier gas; EC, equilibrium cell; FV, feeding valve; LB, liquid bath; LS, liquid
sampler; PP, platinum resistance thermometer probe; PrC, propane cylinder; PT, pressure transducer; RC, refrigerant cylinder;
SM, sampler monitoring; ST, sapphire tube; TC1 and TC2 , thermal compressors; Th, thermocouple; TR, temperature regulator;
VS, vapor sampler; VSS, variable speed stirrer; VP, vacuum pump.

length 1.6 m, diameter 1/8 in. from Restek). The TCD was repeatedly calibrated by introducing known
amounts of each pure compound through a syringe in the injector of the gas chromatograph. Taking into
account the uncertainties due to calibration and the dispersion of analyses, accuracy on vapor and liquid
mole fractions is estimated to be within 1.0% over the whole range of concentration.
The experimental procedure is the following: at room temperature, the equilibrium cell and its loading
lines are evacuated down to 0.1 Pa. One thermal compressor is loaded with propane (TC1 ) and the second
with R227ea (TC2 ). At the required equilibrium temperature (equilibrium temperature is assumed to be
reached when the two Pt100 give the same temperature value within their temperature uncertainty for at the
least 10 min), a volume of about 5 cm3 of R227ea is introduced into the equilibrium cell. The vapor pressure
of the heaviest component (R227ea) is then recorded at this temperature. To describe the two-phase
envelope with at least 10 twin points (liquid and vapor), given amounts of the light component (propane)
are introduced step by step, leading to successive equilibrium mixtures. Equilibrium is assumed when
the total pressure remains unchanged within 1.0 kPa during a period of 10 min under efficient stirring.
For each equilibrium condition, at least six samples of both vapor and liquid phases are withdrawn
using the pneumatic samplers RolsiTM [2] and analysed in order to check for measurement repeatability.

32

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

3. Correlations
Isothermal VLE measurements on the system R227eaC3 H8 were performed in the temperature range
from 293.16 to 353.18 K and at pressures up to 3.5 MPa. The critical temperature (TC ), critical pressure (PC ), and acentric factor (), for each of the two pure components are provided in Table 1 [3].
Our experimental VLE data are correlated by means of a home-made software. We have used the SRK
equation [4] of state given below, which gives good results for VLE of either non-polar or slightly polar
mixtures.
R227ea is a polar compound, and the use of the Soave -function results in a systematic deviation between experimental and calculated vapor pressures. Therefore, we prefer to use the Mathias
Copeman -function [5] with three adjustable parameters, which was especially developed for polar
compounds.

 
 2
 3 2



T
T
T

(T ) = 1 + c1 1
(1)
+ c2 1
+ c3 1
TC
TC
TC
where c1 , c2 and c3 are the three adjustable parameters. MathiasCopeman coefficients were evaluated
for our whole temperature domain using a modified simplex algorithm. The objective function is:

100  Pexp Pcal 2

F =
(2)
N
Pexp
where N is the number of data points, Pexp is the measured pressure, and Pcal is the calculated pressure.
In our EoS approach, we need a mixing rule and an activity coefficient model. The first mixing rule
used to correlate the VLE is the van der Waals mixing rule:

a=
(3)
xi xj aij
i


b=
xi bi

(4)

where
aij =

ai aj (1 kij )

(5)

where kij is the binary interaction parameter. This parameter is found by a careful curve fitting involving
our mixture data.
Two other mixing rules have been used to improve the representation of the equilibrium data. First,
we have selected the modified HuronVidal (MHV1) mixing rule proposed by Michelsen [6] where the
Table 1
Critical parameters and acentric factors [3]
Compound

PC (MPa)

TC (K)

Propane
R227ea

4.25
2.98

369.83
375.95

0.1523
0.3632

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

attractive parameter is calculated from:




 ai
GE (T , P , xi )
RT 
bi
xi
xi ln
a=b
+
bi
q1 i
q1
b
i

33

(6)

The reference pressure is P = 0 and b is calculated from Eq. (4). Michelsen recommends q1 = 0.593.
The second approach is that of Wong and Sandler [7], which is based on the HuronVidal approach and
the van der Waals mixing rules.

i
j xi xj (b (a/RT))ij

b=
(7)
1 (( i xi (ai /bi )/RT) + (AE (T , P = , xi )/CRT))


a
a 
=
(8)
b
xi xj b
RT
RT ij
i
j





a 
a 
1 
a 
b
b
(1 kij )
=
+ b
RT ij 2
RT i
RT j

(9)

where kij is a binary interaction parameter. Wong and Sandler have shown that the excess Helmholtz
energy of mixing at infinite pressure, AE (T , P = , xi ), is approximately equal to the excess Gibbs
energy at low pressure. Consequently, they assumed:
AE (T , P = , xi )
RT

GE (T , P = 0, xi )
RT

(10)

The excess Helmholtz energy is calculated using the NRTL [8] local composition model.
GET ,P
RT

  xj exp(j,i (j,i /RT ))

xi
j,i
k xk exp(k,i (k,i /RT ))
i
j

(11)

i,i = 0 and i,i = 0. Here, j ,i , j ,i and i ,j are adjustable parameters. For our system, which belongs
to a given polar mixture type, it is recommended [8] to use j,i = 0.3. The j ,i and i ,j are adjusted
directly to VLE data through a modified simplex algorithm using the objective function:





100  Pexp Pcal 2  yexp ycal 2
(12)
+
F =
N
Pexp
yexp

4. Results and discussion

4.1. Vapor pressures
The values of the SRK EoS MathiasCopeman coefficients fitted to vapor pressures of R227ea are
reported in Table 2. Table 3 compares the calculated pressures and experimental data. In this work, we
have compared our measured propane vapor pressures with the values calculated using McLinden [9]

34

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

Table 2
Adjusted MathiasCopeman coefficients
Coefficients

R227ea

Propane

C1
C2
C3

1.134
1.965
8.690

0.823
1.552
6.097

Table 3
Experimental and calculated vapor pressures of R227ea
T (K)

Pexp (MPa)

Pcal (MPa)

293.17
313.33
318.21
323.21
328.19
333.20
338.19
348.20
353.03

0.3892
0.7076
0.8070
0.9198
1.0429
1.1792
1.3269
1.6667
1.8531

0.3891
0.7079
0.8074
0.9196
1.0425
1.1783
1.3266
1.6670
1.8541

correlation (see Eqs. (13) and (14)). As the relative deviation is less than 0.15%, we trust the McLinden
correlation to determine the MathiasCopeman coefficients.


A3 T0
P (bar) = A2 exp
(13)
+ A4 T0 + A5 T01,89 + A6 T03
1 T0
T0 = 1

T
A1

(14)

with the Ai coefficients reported in Table 4.

MathiasCopeman coefficients for propane are reported in Table 2. Table 5 summarizes the calculated
pressures using the McLinden correlation and the SRKMathiasCopeman. The vapor pressures of both
pure compounds are accurately represented (within 1.0 kPa).
4.2. Vaporliquid equilibrium
The VLE data obtained are listed in Tables 611. At each temperature, the van der Waals mixing rule binary parameter is adjusted and can be represented by a linear temperature dependency Eq. (15) (see Fig. 2).
k12 = 0.017 + 0.00046T

(15)

Table 4
McLinden correlation coefficients for propane
A1

A2

A3

A4 (bar)

A5

A6

369.82

42.5

6.477

0.1583

2.756

4.363

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

35

Table 5
Propane calculated pressures PML (from McLinden equation) along with calculated pressures Pcal from RKS EoS and adjusted
MathiasCopeman coefficients
T (K)

PML (MPa)

Pcal (MPa)

293.17
313.33
318.21
323.21
328.19
333.20
338.19
348.20
353.03

0.8374
1.3751
1.5359
1.7146
1.9073
2.1170
2.3427
2.8505
3.1245

0.8373
1.3753
1.5360
1.7145
1.9071
2.1168
2.3425
2.8507
3.1248

The results of the correlation are reported in Tables 611 and plotted in Fig. 3.
The SRK EoS + MHV1 mixing rules combined with the NRTL model gives a better fit than the SRK
EoS + van der Waals mixing rules. The two NRTL parameters are slightly temperature dependent. Linear
relationships are convenient for these purposes:
12 = 3737.70 2.456T

(16)

21 = 1573.63 3.420T

(17)

The results of the correlation are reported in Tables 611 and plotted in Fig. 4.
The SRK EoS + WongSandler mixing rules and NRTL model gives a better representation of the data
than the SRK EoS + van der Waals mixing rules. Temperature dependence of NRTL parameters is given
by Eqs. (18) and (19).

Fig. 2. The van der Waals mixing rules: binary parameter kij as a function of temperature. () Fitted to isotherms; () Eq. (15).

36

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

37

38

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

39

40

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

41

42

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

Fig. 3. Pressure as a function of propane mole fraction in the propaneR227ea mixture at different temperatures. () 293.16 K;
() 303.15 K; (+) 313.14 K; () 333.15 K; (*) 343.16 K; () 353.18 K. Solid lines: calculated with RKS EoS and van der Waals
mixing rules with kij from Eq. (5).

12 = 4033.40 5.556T

(18)

21 = 188.29 + 3.036T

(19)

In this case the k12 binary parameter is temperature independent; its value is 0.25. The results of this
correlation are reported in Tables 611 and plotted in Fig. 5.
The binary parameters of the various mixing rules used at each condition are reported in Table 12.
Table 12
Binary parameters of various mixing rules used at each temperature
T (K)

van der Waals mixing rules

MHV1 mixing rules

WongSandler mixing rules

293.16
303.15
313.15
333.15
343.15
353.18

0.118
0.122
0.127
0.136
0.141
0.145

3017.70
2993.16
2968.60
2919.48
2894.92
2870.29

2404.60
2349.10
2293.54
2182.42
2126.86
2071.13

571.03
536.86
502.66
434.26
400.06
365.76

1078.32
1108.65
1139.01
1199.73
1230.09
1260.54

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

43

44

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

Fig. 4. Pressure as a function of propane mole fraction in the propaneR227ea mixture at different temperatures. () 293.16 K;
() 303.15 K; (+) 313.14 K; () 333.15 K; (*) 343.16 K; () 353.18 K. Solid lines: calculated with RKS EoS and the MHV1
mixing rules with adjusted parameters from Eq. (11).

The mean relative absolute percentage deviations on pressure (MRDP) and the mean relative percentage
deviations on vapor phase mole fraction (MRDY) are defined by:



100   Ucal Uexp 
(20)
MRDU =


U
N
exp

where U = P or y1 , are listed in Table 13; N is the number of data points.

Table 14
Composition and pressure of the azeotrope at each temperature (MHV1 mixing rules)
T (K)

xaz

P (MPa)

293.16
303.15
313.14
333.15
343.16
353.18

0.8090
0.8012
0.7940
0.7813
0.7754
0.7700

0.8886
1.1482
1.4578
2.2548
2.7582
3.3525

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

45

Fig. 5. Pressure as a function of propane mole fraction in the propaneR227ea mixture at different temperatures. () 293.16 K; ()
303.15 K; (+) 313.14 K; () 333.15 K; (*) 343.16 K; () 353.18 K. Solid lines: calculated with RKS EoS and the WongSandler
mixing rules with adjusted parameters from Eqs. (9) and (11).

Fig. 6. Azeotropic composition as a function of temperature.

46

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

Fig. 7. Azeotropic pressure as a function of temperature.

We also calculated the BIAS listed in Table 13 for all the cases:

100  Uexp Ucal
BIAS =
Uexp
N

(21)

Table 13 shows that the WongSandler mixing rules give a better representation of the experimental
data than the other mixing rules.
4.3. Azeotrope
Table 14 and Figs. 6 and 7 show the coordinates of the azeotrope (pressure, temperature and composition). When the temperature increases, the azeotropic composition of propane decreases.
5. Conclusion
In this paper we present the VLE data for the propaneR227ea system at six different temperatures. We
used a static-analytic method to obtain the experimental data. The SRK EoS, with the MathiasCopeman
-function is chosen to fit these data with different mixing rules. In each case, the data are found consistent
with the models and the best results are obtained with the WongSandler mixing rules. The deviations of
pressures and propane vapor compositions are less than 1.0%.
List of symbols
a
parameter of the EoS (energy parameter)
A
Helmhotz energy

A. Valtz et al. / Fluid Phase Equilibria 202 (2002) 2947

b
C
F
G
kij
P
R
T
x
y
Z

47

parameter of the EoS (co-volume parameter)

numerical constant = ln(1/2)
objective function
Gibbs energy
binary interaction parameter
pressure (MPa)
gas constant (J/(mol K))
temperature (K)
liquid mole fraction
vapor mole fraction
compressibility factor

Greek letters
ij NRTL model parameter (Eq. (11))

deviation
ij NRTL model binary interaction parameter (Eq. (11); J/mol)

acentric factor
Superscript
E
excess property
Subscripts
az
azeotrope
cal
calculated property
C
critical property
exp experimental property
i, j
molecular species

infinite pressure reference state

ML from McLinden equation
1
propane
2
R227ea
References
[1] S. Laugier, D. Richon, Rev. Sci. Instrum. 57 (1986) 469472.
[2] P. Guilbot, A. Valtz, H. Legendre, D. Richon, Analysis 28 (2000) 426431.
[3] M. Huber, J. Gallagher, M.O. McLinden, G. Morrison, Thermodynamic Properties of Refrigerants and Refrigerant Mixtures
Database, REFPROP V.6.01, National Institute of Standards and Technology, Boulder, CO, 1996.
[4] G. Soave, Chem. Eng. Sci. 4 (1972) 11971203.
[5] P.M. Mathias, T.W. Copeman, Fluid Phase Equilib. 13 (1983) 91108.
[6] M. Michelsen, Fluid Phase Equilib. 60 (1990) 213219.
[7] D.S.H. Wong, S.I. Sandler, AIChE J. 38 (1992) 671680.
[8] H. Renon, J.M. Prausnitz, AIChE J. 14 (1968) 135144.
[9] M.O. McLinden, Int. Rev. Refrig. 13 (1990) 149162.