à"Short Communications

VISCOSITY OF SF6+N2 AND SF6+He GASEOUS MIXTURES

Yoshiyuki TANAKA, Makoto NAKAJIMA, Hironobu KUBOTAand Tadashi MAKITA

Department of Chemical Engineering, Kobe University, Kobe 657

Viscosity of gaseous mixtures is an important ther-
mophysical property, not only because of the use-
fulness of data themselves, but also because it serves to
test various aspects of moderngas kinetic theory,
empirical correlation and estimation of thermophysical
properties of fluids. The present paper continues the
work on the experimental determination of the viscos-
ity of binary gaseous mixtures and constitutes a sequel
to our previous work24) on N2+He, N2+CC1F3(R13)
and N2+CHC1F2(R22) systems. It is the main pur-
pose of this report to supply accurate viscosity data for
nonpolar binary gaseous mixtures, SF6+N2 and SF6+
He, which are stable dielectrics of great importance in
the electrical industries due to their
strength and good arc-quenching properties.
Measurements have been performed by
ball viscometer at 25.00°C (298.15K),
(323.15 K) and 75.00°C (348.15 K) up to about 8 bar
(0.8 MPa) on the relative basis with respect to the
reference viscosity values of nitrogen and hydrogen.
The uncertainty in the viscosity obtained is estimated to
be within 1 %. The experimental results are analyzed
in terms of the Chapman-Enskog theory10'145, and the
empirical estimation formulas for the viscosity
gaseous mixtures given by Bird-Hirschfelder-Curtiss2)
Wilke26) and Brokaw3).
1. Experimental
Viscosity was determined by measuring the rolling
velocity of a steel ball in an inclined Pyrex glass tube
filled with a sample fluid. The general arrangement of
the instrument and the experimental procedures are
the same as described in the previous work24). The
rolling time was measured within ±0.1 ms by means
of electromagnetic induction coils and an electronic
time-interval counter. The temperature of the vis-
cometer was maintained constant within ±0.01 K in a
liquid thermostat. The pressure was measured by a
Heise Bourdon tube gauge within 0.1%.
cometer was calibrated at each temperature with re-
spect to the reference viscosity values of nitrogen and
Received May28, 1979. Correspondence concerning this article should
be addressed to Y. Tanaka. M. Nakajima is with Ina Seito Co., Ltd., Toko-
name 479.
VOL. 13 NO. 2 1980
hydrogen generated by correlations with the aid of
experimental data presented by Michels and Gibson22\
Kestin and Wang19), Kao and Kobayashi13), and
Golbev and Petrov6). Only the laminar flow region
of fluids around the rolling ball was used for the
viscosity determination. The uncertainty in the
viscosity obtained is estimated to be within 1 %.
Nitrogen, hydrogen and helium were obtained from
commercial cylinders with the specification of mini-
mumpurity of 99.99 %. Research-grade sulfur hexa-
fluoride sample fluid was supplied by Asahi Glass
Co. Ltd., which reported a purity of 99.5%. Mole
fraction of gaseous mixtures was determined by gas
high dielectric chromatography. The density was evaluated with the
aid of the second virial coefficients given by Hamann
a rolling- et al. for SF67) and Michels et al. for He23), together
50.00°C with the mass conservation law.
2. Experimental Results and Discussion
The experimental results are given in Tables 1 and
2 and illustrated in Figs. 1, 2, 3 and 4.
Figures 1 and 2 represent the variation of viscosity
for SF6+N2 and SF6+He systems, respectively, as
of a function of pressure. The experimental data of Hel-
, lemans et al.8) (25°, 50°C) and Dawe et al.4) (50°, 75°C)
for pure SF6 at 1 atm are also plotted for comparison
as well as the reference values of nitrogen. The
present results for SF6 at 1 atm agree well with them
within 0.1 % at 25° and 75°C, although the result at
50°C is 0.93% lower than that of Hellemans et al.8).
The temperature coefficient of viscosity (drj/dT)p for
both pure gases and mixtures are positive, similarly
to other dilute gases.
The measurement of pressure effect on viscosity is
limited to low pressures below about 8 bar, so as to
avoid the occurrence of turbulent fluid flow around the
rolling ball which disturbs precise determination of
viscosity. Therefore, viscosity isotherms for both
The vis- mixtures are substantially independent of pressure
throughout the range of experimental conditions within
the experimental error, although it is known14} that the
initial density dependence of viscosity is positive for
N2 and SF6 and negative for He, respectively, in thfe
*55
 Table 1 Experimental data for the viscosity of SF6+N2system
P v P V P V p ' V p V
25°C jc(SF6)=0. 194 x(SF6)=0. 345 x(SF6)=0. 628 x(SF6)=0. 746 jc(SF6) = 1. 000
1.01 174.2 1.01 170.2 1.01 162.0 1.01 158.6 1.01 152.7
1.22 174.2 1.22 170.7 1.22 162.5 1.22 158.9 1.22 152.8
1.36 174.5 1.36 170.2 1.36 162.5 1.36 158.7 1.36 153.1
1.70 174.8 1.70 170.9 1.70 162.2 1.70 158.8 1.70 152.6
2.39 174.1 2.05 170.3 2.05 162.8 2.05 158.7 2.05 152.6
2.74 175.2 2.39 170.5 2.39 162.4 2.39 158.8
3.08 174.9 2.74 170.0 2.74 161.7 2.74 158.8
3.43 174.9 3.08 170.4 3.08 162.3
3.77 174.4 3.77 169.0
4.12 174.1 4.12 170.0
5.15 176.2
5.84 174.9
6.53 176.2
50°C jc=0. 103 x=0.493 x=0.874 *=1.000
1.01 183.7 1.70 174.7 1.01 165.1 1.01 161.7
1.36 183.6 2.05 174.3 1.36 165.2 1.36 162.3
2.05 183.5 2.74 174.3 2.05 165.0 1.70 162.0
2.39 182.4 3.08 174.7 2.05 162.5
3.08 182.3
3.77 182.6
4.12 182.6
4.46 182.4
5.15 182.4
5.84 183.2
75°C x=0.119 x=0.641 x=0.814 jc=1.000
1.01 197.0 1.01 187.7 1.01 182.6 1.01 176.1
1.70 198.1 1.22 186.5 1.36 182.7 1.22 176.3
2.39 198.6 1.36 186.4 1.70 182.9 1.70 177.3
3.08 197.7 1.70 187.1 2.05 184.1
3.77 198.2 2.05 187.9
4.46 198.9
5.15 198.5
5.84 200.7

The following units are used in Tables 1 and 2: x(SF6), mole fraction of SF6, P, pressure in 105 Pa(=bar) and r], viscosity in
10-7 Pa-s(=10~6 poise).

Fig. 1 Viscosity of SF6+N2mixtures as a func-

tion of pressure and composition at 25° and 75°C
present temperature range.
The composition dependence of viscosity
mospheric pressure is illustrated
SF6+N2 and SF6+He systems, respectively,
with the experimental results of Kestin et aLU) at 25°C,
which are the only experimental data available for
these binary gaseous mixtures. Their data* agree rea-
sonably with our results in both systems with a mean
deviation of about 1.4%. The viscosity of SF6+N2
156
Fig. 2 Viscosity of SF6+Hemixtures as a func-
at at- tion of pressure and composition at 50°C
in Figs. 3 and 4 for
together * It was confirmed through private communications with the
authors that the original paper14) has a printing error in Tables
XIII and XIVon page 249. The error involves the exchange of
the three columns of temperature and viscosity data between the
two tables. The figures in Table XIVfor the viscosity of N2-f
SF6 system should be replaced by the figures in three columns in
Table XIII for Ar+SF6 system.
JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
 Table 2 Experimental data for the viscosity of SF6+Hesystem
P 7] P f) P 7] P 71 P 7}

25°C x(SF6) =0. 000 x(SF6) =0. 201 jt(SF6)=0. 408 jc(SF6) =0. 607 x(SF6)=0. 809
1.01 201.3 1.01 209.6 1.01 187.4 1.01 167.2 1.01 160.5
1.70 201.2 1.36 208.8 1.36 187.0 1.36 167.7 1.36 158.9
2.39 200.9 1.70 209.3 1.70 187.0 1.70 167.7 1.70 159.7
3.08 200.5 2.05 210.1 2.05 186.7 2.05 167.9 2.05 159.4
3.77 200.2 2.39 209.9 2.39 186.9 2.39 167.6 2.39 159.4
5.15 199.5 2.74 209.4 2.74 186.8 2.94 167.7
6.53 199.3 3.08 209.7 3.08 186.9 3.43 168.1
7.91 199.5 4.12 208.8 3.43 186.3 3.77 167.3
9.29 199.6 4.46 209.2 4.12 167.6
10.67 199.3 5.15 208.8
5.84 208.0
6.53 208.1

50°C x=0.000 x-0. 127 jc=0.250 x-0.498 x=0.761

1.01 211.0 1.01 220.8 1.01 205.9 1.01 184.2 1.01 172.8
1.36 211.0 1.36 220.8 1.36 205.9 1.36 183.4 1.36 173.2
3.08 212.1 1.70 220.8 1.70 206.4 1.70 183.9 1.70 172.4
3.77 212.3 2.39 220.2 2.05 206.3 2.05 184.4 2.05 172.9
5.15 211.7 3.08 220.1 2.39 206.4 2.39 184.9 2.39 173.6
6.53 210.7 3.43 220.4 2.74 206.7 2.74 184.9
7.91 211.0 3.77 220.2 3.08 206.8 3.08 185.5
4.12 221.4 3.43 206.7 3.43 186.3
4.46 221.3 3.77 206.6 3.77 185.8
5.15 221.2 4.12 207.2
5.84 220.4
6.53 220.2
7.22 221.9
75°C x=0. 129 jc=0.200 x=0.408 x-0.606 *=0.795
1.01 232.8 1.01 226.3 1.01 213.6 1.01 190.2 1.01 181.1
1.36 232.4 1.36 225.9 1.36 213.1 1.36 190.7 1.36 180.5
1.70 232.7 1.70 226.0 1.70 213.8 1.70 191.5 1.70 181.0
2.05 232.9 2.05 226.4 2.05 215.5 2.05 191.6 2.05 181.8
2.39 233.1 2.39 226.2 2.39 215.0 2.39 191.5
2.74 233.6 2.74 227.0 2.74 214.6 2.74 192.8
3.08 231.4 3.08 226.6 3.08 215.0 3.08 192.8
3.43 232.4 3.43 227.3 3.43 213.9 3.43 193.1
3.77 232.7 3.77 228.5 3.77 213.0 3.77 193.5
4.12 233.8 4.12 228.0 4.12 213.6
4.46 233.8 4.46 229.6 4.46 213.0
5.15 234.2 5.15 229.0 5.15 213.6
5.84 234.6 5.84 227.7 5.84 214.8
6.53 236.1 6.53 227.7 6.53 214.3

Fig. 3 Composition dependence of viscosity for Fig. 4 Composition dependence of viscosity for
SF6+N2 mixtures at 25°, 50° and 75°C at normal SF6+He mixtures at 25° and 75°C at normal
pressur e pressure

VOL. 13 NO. 2 1980 157

 Various empirical formulas21) have been also pre-
Table 3 Interaction parameters adopted in the present
work sented for the prediction of viscosity of gaseous mix-
Kestin et «/.14)
c;[10-10m]
Hirschfelder
£/k[K]
et a/.10>
<7[10-10m] e/k[K]
tures. The most successful expression is given in the
following form by Sutherland : 1°
N2 3.568 113.07 3.681 91.5 7
V2
He 2.556 ll.29 2.576 10.22 <= 1+012(*2/*O ' l+^2l(^lM2)
SF6 5.160 225.36 5.46 155
SF6 +N2 4.267 175.89 4.571* 119.09*
SF6+He 3.869 49.16 4.018* 39.80* In this expression the viscosity of mixtures at any
composition can be calculated from the viscosities of
Interaction parameters for unlike molecules are calculated pure components yl and rj\, when the combination
by the mixing rules, on-ioi+Oi)!! and e12=(ei*£2)1/2-
factors <p12 and <f>2i are evaluated in advance. Several
Table 4 The optimumvalues of combination factors theoretical and empirical methods have been proposed
Mixture Temp. [°C] 012 02i
dev.
Mean
[%]
for determination of theWilke26)
Hirschfelder-Curtiss2),
combination factors by Bird-
and Brokaw3). The
2.262 0.368 0.43
SF6+N2 25
50 2.264 0.372 0.99
interaction parameters and collision integrals for Len-
75 1.915 0.425 0.83 nard-Jones (12-6) potential required for these calcula-
SF6 +He 25 3.630 0.115 0.92 tions are cited from Hirschfelder et al.10).
5.050 0.095
50
75 5.090 0.090
0.57
1.49
The calculated viscosity values are compared with
the experimental results in Figs. 3 and 4. As for SF6+
N2 system, Chapman-Enskog theory10)14) and the
mixtures varies almost linearly with composition of
SF6 at each temperature. The experimental values empirical formula of Wilke26) give better agreement
of viscosity have no definite maximumthroughout the with our results, with a meandeviation of 1%. No
range of composition. On the other hand, the vis- remarkable difference is found between the two kinds
cosity of SF6+He mixtures varies non-linearly with of calculations by Chapman-Enskog theory10)14) men-
composition and reveals a conspicuous maximumat tioned above except for mixtures of low SF6 compo-
nearly 0.1 mole fraction of SF6 at each temperature. sition. Although the expression of Bird-Hirschfelder-
Thus, binary gaseous mixtures containing helium tend Curtiss2) solely predicts the existence of a maximum
to have a maximum on the viscosity vs. composition near 0.1 mole fraction of SF6, no such definite maxi-
isotherm as seen in the systems such as He withAr12>20), mumwas observed in our experimental work.
Kr12), Xe25), O214), N215'17), CO25'16) andCF414). Onthe The rigorous theory10)14) also gives the best agree-
ment with the present experimental results for SF6+
other hand, binary mixtures containing nitrogen gener- He system, with a mean deviation of 1.8%. All
ally do not have any definite maximumexcept mixtures
with He15>17). As far as we know, no minimumhas theories successfully predict the existence of a maxi-
mumnear 0.1 mole fraction of SF6.
ever been observed in any binary gaseous system. The optimumvalues of the combination factors can
The first approximation to the rigorous kinetic
be determined on the basis of experimental viscosity
theory of Chapman and Enskog10) represents success-
fully the transport properties of dilute monatomic values. Table 4 lists the optimum combination factors
gases. However,recent experimental and theoretical obtained by the graphical method of Huck and Thorn-
work by Kestin and coworkers1>9'14>18) has revealed ton11} as well as the meandeviations between calculated
that the theory also produces fairly good results for and experimental values. Suffix 1 denotes nitrogen or
helium and suffix 2 sulfur hexafluoride.
viscosity of mixtures containing monatomicgases and
non-polar polyatomic gases by employing appropriate Nomenclature
interaction parameters ai3- and ei3- and the collision
k
= Boltzman constant [1.308622 x 10-23J K"1]
integrals. They have determined empirically the
P
= pressure [105 Pa=bar]
T
optimumvalues of interaction parameters and uni- = temperature [K]
versal collision integrals as a function of reduced tem- = mole fraction of i-ih component [-]
perature on the basis of most reliable experimental Bij/k = depth of potential minimum for like
data of viscosity and the second virial coefficient. and unlike molecules
7]i = viscosity of z-th component
Therefore, the present calculations by the Chapman- Gij = collision diameter for like and
Enskog theory were performed by using both the
unlike molecules
empirical parameters and collision integrals of Kestin fa = combination factor for viscosity
et al.U) (Chapman-Enskog-Kestin) and those of <Subscripts>
Hirschfelder et al.10) based on the Lennard-Jones (12- m = values for mixture
6) potential and the most popular mixing rules (Chap- 0 = valuesat attenuation
man-Enskog-L-J) as listed in Table 3.
158 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
 14) Kestin, J., H. E. Khalifa, S. T. Ro and W. A. Wakeham:
Literature Cited Physica, 88A, 242 (1977).
1) Abe, Y., J. Kestin, H. E. Khalifa and W. A. Wakeham: 15) Kestin, J., Y. Kobayashi and R. T. Wood: ibid., 32, 1065
Physica, 93A, 155 (1978). (1966).
2) Bird, R. B., J. Q. Hirschfelder and C. F. Curtiss: Trans. 16) Kestin, J. and S. T. Ro: Ber. Bunsenges. Physik. Chem.,
Amer. Soc. Mech. Eng., 16, 1011 (1954). 78, 20 (1974); 80, 619 (1976).
3) Brokaw, R. S.: /. Chem. Phys., 42, 1140 (1965).
17) Kestin, J., S. T. Ro and W. A. Wakeham: /. Chem. Phys.,
56, 4036 (1972).
4) Dawe, R. A., G. C. Maitland, M. Rigby and E. B. Smith:
18) idem: Physica, 58, 165 (1972).
Trans. Faraday Soc, 66, 1955 (1970).
5) DiPippo, R., J. Kestin and K. Oguchi, /. Chem. Phys., 46, 19) Kestin, J. and H. E. Wang: Trans. ASME, 80, ll (1958).
4758 (1967). 20) Maitland, G. C. and E. B. Smith: /. Chem. Soc. Faraday
6) Golubev, I.F. and V.A. Petrov: Trudy GIAP, No. 2, 5 Trans. I, 70, 1191 (1974).
(1953). 21) Makita, T.: "Viscosity and Thermal Conductivity",
7) Hamann, S.D., W.J. McManamey and J.F. Pearse: Baifukan, Tokyo (1975).
Trans. Faraday Soc, 49, 351 (1953). 22) Michels, A. and R. O. Gibson: Proc. R. Soc. London, Ser.
A, 134, 288 (1931).
8) Hellemans, J. M., J. Kestin and S. T. Ro: Physica, 65, 376
(1973).
23) Michels, A. and H. Wouters: /. Chem. Reference Data,
9) idem: ibid., 71, 1 (1974). 2, 923 (1970).
10) Hirschfelder, J. O., C. F. Curtiss and R. B. Bird: "Molec- 24) Tanaka, Y., H. Kubota, T. Makita and H. Okazaki: /.
ular Theory of Gases and Liquids", John Wiley and Chem. Eng. Japan, 10, 83 (1977).
Sons, Inc., New York (1954); Rev. Ed. (1964). 25) Thornton, E.: Proc. Phys. Soc, 16, 104 (1960).
ll) Huck, R.J. and E. Thornton: Proc Phys. Soc, 92, 244 26) Wilke, C. R.: /. Chem. Phys., 18. 517 (1950).
(1967).
12) Kalelkar, A. S. and J. Kestin: /. Chem. Phys., 52, 4248
(Presented in part at the 12th Autumn Meeting of The Soc.
(1970).
of Chem. Engrs., Japan, at Okayama, October 10, 1978.)
13) Kao, J. T. F. and R. Kobayashi: ibid., 47, 2836 (1967).

PREDICTION OF LIQUID-LIQUID EQUILIBRIA FOR TERNARY

SYSTEMS BY THE ASOG METHOD

Katsumi TOCHIGI, Mayumi HIRAGAand Kazuo KOJIMA

Department of Industrial Chemistry, Nihon University, Tokyo 101

A knowledge of liquid-liquid equilibria is essential the ASOGparameters.

for the design of liquid-liquid extraction and azeotropic 1. ASOGMethod
distillation devices. Several attemps have been made
to predict multicomponent liquid-liquid equilibria In the ASOGmethod the activity coefficient of com-
from binary data using activity-coefficient equations ponent i in liquid solution is given by following Eqs.
based on the concept of local composition, such as (1) to (3),
NRTL19), modified Wilson26) and UNIQUAC1'25
lnri=lnrf*+lnr? (1)
equations. At the same time, prediction of liquid-
liquid equilibria by group contribution methods5'23'245 l
n.rfH=]n(vfHlz
\ I 3=1 xpfH)+l-v**/x
/ / j=l xtfH (la)
has been reported. Fredenslund et al.b) have pre-
dicted liquid-liquid equilibria for a large number of
lnr?=Z v*i(ln r,-lnni») (lb)
systems using UNIFAC,though they presented the k
predicted results concerning solubility data only. The
authors11} have determined, for 31 groups, the group ln/\ =-ln 2 Xlahn+\-
I j: I UialJj:
\ I Xmal/m) /
m

pair parameters necessary for predicting activity coef- (2)

ficient by ASOG(Analytical Solutions of Groups)
method. This work discusses prediction of liquid- Xk=H xtVki 2 XtJ^Vki (3)
liquid equilibria for 31 ternary systems (25°C-40°C) by
Received August 6, 1979. Correspondence concerning this article should where vfH is the number of atoms (other than hydrogen
be addressed to K. Kojima.
atoms) in molecule /, i>ki is the number of atoms (other
VOL. 13 NO. 2 1980 159