connecting centers of the adjacent hypersquares. fc-component azeotrope is equal to 2k.

Then Mn~xcan be obtained as the hypersurface of From the same theorem used in the previous paper1},
this //-dimensional manifold because the hypersurface the sum of indices of singular points on M71"1is equal
consists of (n- l)-dimensional simplices and satisfies to the Euler-Poincare characteristics of AP"1, equal to
the conditions for M71"1 mentioned above.
For instance, the indices of pure materials and The sum of indices of all singular points on Mn~~xis
binary azeotropes in the ternary system are defined
on the surface of an octahedron which consists of four represented as 2]n 2&Sfc, taking account of the degree
original triangular diagrams and four inverse images of multiplicity of pure materials and azeotropes where
and is constructed by connecting centers of adjacent Si and Sk(k> 1), respectively, are the sum of the indices
squares which constitute the surface of a cube15. of pure materials and that of ^-componentazeotropes
Mn~x contains 2n apices and each of them has on the original composition simplex.
2(/z-1) adjacent apices because the n-dimensional Then the topological condition is obtained as
hypercube consists of 2n hypersquares each of which Cond(I*) t l'S^l+i-l)*-1
is adjacent to 2{n-\) hypersquares. The number of
/^-dimensional simplices in Mn~xis given as follows: Cond (1*) and Cond (2) are available for the clas-
2k X,_!C4 X2n+(k+1)=2*+1xnCk+1 sification of multicomponent vapor-liquid equilibria,
the examination of the soundness of azeotropic data
2. Topological Condition for w-Component Systems and the qualitative prediction of the presence and
It is obvious from the procedure of constructing types of multicomponentazeotropes.
Má""1 that a singular point contained in the original Nomenclature
composition simplex repeatedly appears in Mn~x. pi = vapor pressure of the /-th component
The degree of multiplicity of the pure material in Si = sum of indices of pure materials
Mn~xcoincides with that of an apex and the degree of Sk(k>1) = sum of indices of^-component azeotropes
multiplicity of the /^-component azeotrope coincides Xi - liquid composition of the i-ih component
with that of a (k-l)-dimensional simplex. Since yi = vapor composition of the i-th component
Mn~l contains 2kx nCk of (k- l)-dimensional simpli- Ti = activity coefficient for the i-th component
ces and the original compositionsimplex contains nCk Literature Cited
of them, the same (k- l)-dimensional simplex appears 1) Matsuyama, H. and H. Nishimura: /. Chem. Eng. Japan,
2k times in Mn~x. Therefore, the degree of multiplic- 10, 181 (1977).
ity of the pure material is equal to 2 and that of the

ESTIMATION OF THERMAL CONDUCTIVITY OF

LIQUID HYDROCARBONS

Virendra K. MATHUR, Jai D. SINGH and William M. FITZGERALD

Department of Chemical Engineering, University of New Hampshire, Durham, New Hampshire, 03824 U.S.A.

Thermal conductivity is an important physical correlation for predicting thermal conductivity under
property due to its universal significance in heat various conditions of temperature and pressure. A
transfer applications and equipment design. Despite large number of estimation techniques have been
its usefulness, experimental data even at ordinary proposed to calculate thermal conductivity of a pure
temperatures are scarce, mainly due to the fact that liquid. These methods vary in complexity, degree
laboratory measurement of thermal conductivity is of accuracy and generality. Sheffy and Johnson20>
comparatively difficult. There appears to be only proposed an empirical equation which predicts
one solution to this problem: to develop a suitable a linear decrease of liquid thermal conductivity
with temperature. The correlation is unsuitable for
Received August 3, 1977. Correspondence concerning this article should be
addressed to V. K. Mathur. highly polar compounds and gives poor results for
VOL. ll NO. 1 1978 67
 Table 1 a values for various hydrocarbon compounds for engineer.
Ototake and Nezu correlation Ototake and Nezu13) have presented the following
Compound a x 10~4 Compound a x 10~4 equation for predicting liquid thermal conductivity
Methane 1. 430 «-Decane 2. 43 1 in terms of the density of a compound:
Ethane 1. 652 «-Dodecane 2. 278 kL = (TIT.y*la{ VI V0-P(yj Ve)} (1)
Propane 1. 964 w-Tetradecane 2. 634
w-Pentane 2. 1 10 rc-Hexadecane 2. 704 /3 is reported to be equal to 0.76 and independent of
w-Hexane 2. 067 Cyclohexane 2. 802
^-Heptane 2. 173 Benzene 2. 312 the nature, temperature, and pressure of a compound.
«-Octane 2. 278 Toluene 2. 034 The value of a differs for each compoundand was
w-Nonane 2. 228 i7-Xylene 2. 296 calculated for a number of hydrocarbons using ex-
perimental thermal conductivity data. The calculated
values of a are presented in Table 1. The variation
in the values of a for different compounds is a serious
shortcoming and very much restricts the applicability
of Eq. (1). In an effort to overcome this weakness,
the authors of this paper have found that a linear
relationship exists between a and the critical com-
pressibility factor, Zc, for pure hydrocarbon liquids.
A plot of this relationship is shown in Fig. 1. The
scattering in the plot may be due to errors in the ex-
perimental thermal conductivity values used for a
calculations. These errors, in general, are estimated
to be as high as 4%22}. Using the relationship between
a and Zc Eq. (1) can be modified to:
kL = (Try^/{(A -BZc)( Vr - 0J6 Vmr)} (2)
where kL =liquid thermal conductivity, cal/sec à" cm à" °K,
Tr=reduced temperature, Vr=reduced volume, Vmr
=reduced volume at melting point, Zc=compres-
sibility factor, A=5.230X 10\ secà"cmà"°K/cal, B=
ll.454x 104, secà"cmà"°K/cal. The above equation has
no unique constants. Melting point, critical temper-
ature, and volume data for hydrocarbons are availa-
ble^. Liquid volumes at various temperatures and
Fig. 1 a vs. compressibility factor, Zc critical compressibility factors can be found13 or
estimated. Ototake et ah used Lu's9) method for the
branched-chain hydrocarbons, naphthenes, and high- estimation of specific volumes at the melting point of
molecular weight compounds. Kanitkar and the compound and temperature at which thermal
Thodos7) presented a semi-empirical equation for conductivity is desired. We have found that the
predicting thermal conductivity of liquid hydrocarbons Oklahoma State University method10>12) yields more
in terms of molecular weight, critical constants and accurate values of liquid specific volumes and is also
reduced density. The correlation in general shows simple to use. Either the Lu9) or the Rackett15)
high percentage deviations for paraffins, naphthenic method can also be used, probably with some sacrifice
and aromatic compounds. Robbins and Kingrea17) in accuracy. The use of solely critical properties and
proposed an equation based on the original suggestion liquid specific volumes makes simple the application
of Weber17>23) that kL is proportional to the product of Eq. (2) for the estimation of thermal conductivity
of Cpp4/3. This estimation method is reported to be of pure liquid hydrocarbons. The correlation is
applicable to organic liquids at pressures below 500 psi applicable to both light and heavy hydrocarbons.
between reduced temperatures of0.4 and 0.9. Lenoir8) The proposed equation was evaluated for twenty-five
and Missenardn) have proposed correlations for pre- paraffinic, naphthenic and aromatic compoundsusing
dicting thermal conductivity at high pressures, pro- forty-six data points21). Thermal conductivity values
vided data is available at one pressure. In this were predicted with an absolute average percent
investigation, a semi-empirical equation for predicting deviation of 4.36, as shown in Table2. The results
thermal conductivity of pure liquid hydrocarbons of this correlation have been compared with those
(paraffins, naphthenes and aromatics) has been pro- obtained from other most commonlyused methods
posed that meets the requirements of a process design for thermal conductivity estimation, namely those of
68 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
 Table 2 Comparison between experimental and calculated Table 3 Results of the evaluation of the proposed correla-
values of liquid hydrocarbon thermal conductivity tion of predicting liquid thermal conductivity of hydrocar-
Temp. &[*]xlO6 Percent Source bons at high pressures
[°K] Exp. Cal. error ofexp.kL Temp. Press.[psia] &[*]x lO6Cal. Percent Source
323
16)
Compound [°K] Exp. error ofexp.&z,
Propane 277 187 177.5 262.1 266.4

^-Butane -5.07
1.63 2) Propane 310.9 200 204.4 189.3 -7.40 3)
311 228.6 237.6 3.92 2) 1000 217.4 204.8 -5.81 3)
«-Pentane 293 272 267. 3 - 1. 72 16) 2000 232.0 220.0 -5. 17 3)
303 265 260.3 -1.72 16) 3000 243.4 232.7 -4.37 3)
«-Hexane 293 313 292.6 -6. 53 16) 4000 253.0 243.8 -3.62 3)
303 303 285.5 -5.79 16) 5000 261.8 253.8 -3.06 3)
311 295 279.7 -5.18 16) 344.6 1000 184.6 170.9 -7.41 3)
330 275 265.9 -3.30 16) 2000 204.0 193.9 -4.93 3)
333 271 263.7 -2.69 16) 3000 217.2 210.9 -2.94 3)
w-Heptane 293 317 296.5 6.46 16) 4000 227.6 224.5 -1.34 3)
303 299 290.2 2.93 16) 5000 238.9 236.8 -0.90 3)
311 300 285.2 -4.94 16) rc-Butane 310.9 1000 242.0 247.4 2.21 2)
333 277 271.0 -2.17 16) 2000 252.2 257.7 2. 19 2)
350 262 259.8 -0.83 16) 3000 260.4 267.2 2.59 2)
«-Octane 293 326 320.2 -1.77 16) 4000 267.5 276.0 3. 17 2)
311 310 307.9 -0.67 16) 5000 275. 1 284.2 3.31 2)
350 277 281.0 1.44 16) 344.3 1000 213.1 220.2 3.32 2)
«-Nonane 293 339 312.5 -7.81 16) 2000 224.3 233.5 4.15 2)
311 321 301.6 -6.05 16) 3000 233.3 245.1 5.07 2)
350 284 277.5 -2.28 16) 4000 242.0 255.5 5.55 2)
«-Decane 314 315 309.5 -1.73 16) 5000 250. 1 265.0 5.95 2)
349 278 287.2 3.31 16) 1000 190.5 191.0 0.24 2)
«-Dodecane 311 329 296.6 -9.85 16) 377.6 2000 203.1 209.4 3.08 2)
«-Tetradecane 316 328 332.8 1.46 16) 3000 213.7 224. 1 4.83 2)
«-Hexadecane 316 338 337.3 0.21 18) 4000 223.4 236.5 5.86 2)
«-Octadecane 316 349 370.9 6.27 18) 5000 232.3 247.8 6.64 2)
2-Methyl pentane 305 259 263.8 1.87 18) «-Pentan 310.9 200 245.2 256.2 4.49 4)
2,3-Dimethylbutane 305 248 268.9 18) 1000 253.0 264.5 4.55 4)
322 238 258. 1 18) 2000 261.3 274.1 4.89 4)
2,2,5-Trimethyl-hexane 311 257.6 274.9 18) 3000 269.0 283.0 5.20 4)
Cyclohexane 293 297 298.8 19) 4000 275.8 291.5 5.67 4)
311 283 283.7 16) 5000 281.6 299.5 6.35 4)
Methylcyclopentane 293 289 277.0 14) 344.3 200 219.6 231.6 5.49 4)
311 275 268.0 16) 1000 229.4 241.7 5.39 4)
Benzene 323 327 366. 6 16) 2000 239.0 253.0 5.86 4)
Toluene 293 325 293. 6 -9. 67 16) 3000 247.3 263.2 6.43 4)
293 316 298.2 -5.63 16) 4000 254. 8 272. 6 7.00 4)
Ethylb enzene 353 281 270.8 -3.61 16) 5000 261.1 281.4 7.79 4)
o-Xylene 353 283 313.5 10.77 16) 377.6 200 197.6 204.8 3.66 4)
ra-Xylene 293 376 346.6 -7.82 16) 1000 208.8 218.3 4.55 4)
/>-Xylene 293 370 358.9 -2.99 14) 2000 219.8 232.3 5.65 4)
293 325 335.9 3.36 14) 3000 228.9 244.4 6.78 4)
Mesi tylene
rc-Butylbenzene 293 329 306.0 -6.98 14) 4000 236.6 255.1 7.84 4)
311 313 299.2 -4.42 18) 5000 243.9 265.0 8.65 4)
AAPD**=4.36 18) AAPD-4. 8

[cal/cm - sec - °K] * [cal/cm-sec-°K]

Absolute average percent x deviation

=l/n 2- |cal.-exp.|/exp. 100
(AAPD)

Oklahoma State University method10>12) was used for

Sheffy and Johnson16'20), Kanitkar and Thodos7), and estimating liquid volumes in the sub-cooled region.
Robbins and Kingrea16'17). The predicted values from Either the Lu9) or the Yen and Woods24}method can
Eq.(2) have been found to be equal or better in also be used, probably with some sacrifice in accuracy.
accuracy. The calculated thermal conductivities for methane
If the proposed correlation is to find general ac- and ethane from Eq. (2) were about 15 to 30% lower
ceptance, it should be able accurately to calculate than experimental values because the predicted a
liquid thermal conductivities at high pressures. The values from Fig. 1 were considerably higher than actual
method was tested for propane, /2-butane and n- values as given in Table1. However, the use of
pentane at temperatures ranging from 40° to 160°F a values of 14305 and 16518 in Eq. (1) for methane
and pressures from 200 to 5000psi. The errors and ethane, respectively, gave accurate results over
varied from 0.24 to 8.65 %, giving an overall absolute a wide range of temperatures and pressures, as shown
average percent deviation of 4.76 for forty-four data in Table 4. The overall absolute average deviation
points. The results are shown in Table3. The was estimated to be 2.71 %for fifteen data points.
VOL. ll NO. 1 1978 69
 V
=volume [cm 3/g-mol e]
Table 4 Results of the evaluation of the proposed correla-

z
tion for predicting liquid thermal conductivity of hydrocar- = compressibility factor, PV/RT
bons at high pressures (a for methane=14305, a, for a, p = constants
ethane- 16518) p = density [g/cm 3 ]
Temp. Press. &[*]x lO6 &[*]xlO6 Percent Source <Subscripts>
Compound [°K] [psia] Exp. Cal. Exp. Cal. error ofexp.^
Methane 125.5 2961 453.9 453.3 -0.13
c = critical properties
4412 478.8 465.1 -2.87 m = properties at melting point
7352 523.4 487. 1 -6.94 mr = reduced properties at melting point
150.3 1470 348.0 359.2 3.22 r = reduced properties
4416 410.3 392.6 -4.32
5882 437.0 407.4 -6.78 Literature Cited
175.2 1470 272.6 274.7 0.76
4412 345.2 329.7 -4.99
1) "API Technical Data Book-Petroleum Refining", Am.
Pet. Inst. (1970).
5887 373.3 350.8 -6.03
Ethane 277.6 400 202.4 204.6 1.09 2) Carmichael, L. T. and B. H. Sage: /. Chem. Eng. Data, 9,
511 (1964).
1000 220. 8 224. 1 1.52
3) Carmichael, L.T., J.Jacobs and B.H. Sage: /. Chem.
2000 248. 0 247. 5 -0.2
3000 269.4 265.8 -1.35 Eng. Data, 13, 40 (1968).
4000 283.0 281.2 -0.63 4) Idem: ibid., 14, 31 (1969).
5000 296.3 294.8 0.19 5) Carmichael, L.T., V.Berry and B.H.Sage: ibid., 8,
AAPD-2.70 281 (1963).
* [cal/cm-sec-°K] 6) Ikenberry, L.D. and R.A. Stuart: /. Chem. Phys., 39,
1561 (1963).
7) Kanitkar, D. and G. Thodos: Can. J. Chem. Eng., 47, 427
(1969).
Conclusion
8)9) Lenoir, J. M.: Petroleum
Lu, B. C.Y.:
Refiner,
Chem. Eng., 66, (9),
36, 137 (8),(1959).162 (1957).
The proposed equation for the prediction of liquid
thermal conductivity for pure hydrocarbons is simple 10) Maddox, R. N. and V. K. Mathur: Paper presented at
and accurate, and requires only easily available input National AIChE Meeting, Houston (Texas) (1971).
l) Missenard, A.: Rev. Gen. Thermique, 9, 101, 649 (1970).
data. The correlation is found to predict thermal
12) Oklahoma State University Density Program distributed
conductivity at normal as well as at high pressure with by Natural Gas Processors Association, Tulsa, (U. S. A.).
good agreement between experimental and calculated 13) Ototake, N. and M. Nezu: KagakuKogaku, 35, 1109 (1971).
values for twenty-seven hydrocarbons. 14) Pachaiyappan V.: Brit. Chem. Eng., 16, 382 (1971).
15) Rackett, H. G.: /. Chem. Eng. Data, 15, 4 (1970).
Acknowl edgment 16) Reid, R. C. and T. K. Sherwood: "The Properties of Gases
and Liquids", McGraw Hill, New York (1966).
Our thanks are due to Dr. R. N. Maddox, School of Chemical 17) Robbins, L. A., and C. L. Kingrea: Hydrocarbon Processing
Engineering, O.S.U., Stillwater (OK) 74074, for his suggestions. and Petrol. Refiner, 41, (5), 133 (1962).
One of the authors (J.D.S.) is grateful to the University Grants 18) Sakiadis, B. C. and J. Coates: AIChE J., 3, 121 (1957).
Commision, New Delhi, for providing financial support. 19) Idem: ibid., 1, 275 (1955).

20) Sheffy, W. J. and E. F. Johnson: /. Chem. Eng. Data, 6,

Nomenclature 245 (1961).
A,B = constant 21) Singh, J. D.: M. S. Thesis, Banaras Hindu University,
Cp = specific heat [cal/g- °C] Varanasi, India (1973).
kL = liquid thermal conductivity [cal/sec- cm- °C] 22) Tsederberg, N. V.: "Thermal Conductivity of Gases
n
= number of data points and Liquids", M.I.T. Press (1965).
P
23) Weber, H. F.: Ann. Phys. Chem., 10, 103 (1880).
= pressure [psia]
T
= temperature[°K] 24) Yen, L. C. and S. S. Woods: AIChEJ., 12, 95 (1966).

70
JOURNAL OF CHEMICAL ENGINEERING OF JAPAN