Journal of Physics: Conference Series

PAPER • OPEN ACCESS

Thermophysical properties of fluids: dynamic viscosity and thermal

conductivity
To cite this article: G Latini 2017 J. Phys.: Conf. Ser. 923 012001

View the article online for updates and enhancements.

This content was downloaded from IP address 191.96.250.195 on 21/11/2017 at 00:25

35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

Thermophysical properties of fluids: dynamic viscosity and

thermal conductivity
G Latini
Professor Emeritus, Dipartimento di Ingegneria Industriale e Scienze Matematiche
Università Politecnica delle Marche, Via Brecce Bianche n.10, 60131 Ancona, Italy

E-mail: [email protected]

Abstract. Thermophysical properties of fluids strongly depend upon atomic and molecular
structure, complex systems governed by physics laws providing the time evolution.
Theoretically the knowledge of the initial position and velocity of each atom, of the
interaction forces and of the boundary conditions, leads to the solution; actually this
approach contains too many variables and it is generally impossible to obtain an acceptable
solution. In many cases it is only possible to calculate or to measure some macroscopic
properties of fluids (pressure, temperature, molar volume, heat capacities…). The ideal gas
“law,” PV = nRT, was one of the first important correlations of properties and the deviations
from this law for real gases were usefully proposed. Moreover the statistical mechanics leads
for example to the “hard-sphere” model providing the link between the transport properties and
the molecular size and speed of the molecules. Further approximations take into account the
intermolecular interactions (the potential functions) which can be used to describe attractions
and repulsions. In any case thermodynamics reduces experimental or theoretical efforts by
relating one physical property to another: the Clausius-Clapeyron equation provides a classical
example of this method and the PVT function must be known accurately. However, in spite of
the useful developments in molecular theory and computers technology, often it is usual to
search for physical properties when the existing theories are not reliable and experimental
data are not available: the required value of the physical or thermophysical property must be
estimated or predicted (very often estimation and prediction are improperly used as
synonymous). In some cases empirical correlations are useful, if it is clearly defined the range
of conditions on which they are based. This work is concerned with dynamic viscosity  and
thermal conductivity  and is based on clear and important rules to be respected when a
prediction or estimation method is proposed.

1. Introduction
An acceptable estimation method of a given thermophysical property must be characterized as
follows:
- the values of the thermophysical property must be reliable for pure substances and for mixtures at
any given temperature, pressure and composition;
- the phase (solid, liquid, or gas) must be clearly indicated;
- the number of input data has to be minimized;
- the ranges of the probable errors must be clearly indicated;
- the computation time must be minimized.
Obviously very few of the methods satisfy these conditions, and among the different molecular
theories useful for data correlation, a theory is particularly helpful: the law of corresponding states or
the corresponding-states principle, originally presented by macroscopic arguments, but successively
grounded on a molecular basis. The law of corresponding states (van der Waals [1]) is based on the
hypothesis that equilibrium properties are linked in a universal way with the critical properties, and
allows the development of correlations and estimation methods: two-parameter corresponding states

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

theory asserts that if pressure, volume, and temperature are divided by the corresponding critical
properties, the function relating reduced pressure to reduced volume and reduced temperature is the
same for all substances. Of course small molecules with spherical symmetry are satisfactorily covered
by a two-constant law of corresponding states, but complex, non-spherical and weakly polar
molecules do not follow the law as well, and a third parameter, for example the acentric factor  has
to be added. Summarizing, good correlations exist for nonpolar fluids, but analogous correlations are
often not available for polar fluids or show very low reliability.
Another useful method has to be presented: a macroscopic property clearly depends upon the
molecular structure and can be estimated by the group contributions method (generally by an
algebraic operation that sums the contributions from the molecule‟s parts).
In any case reliable experimental data are preferred with respect to results obtained by estimation
methods and several data banks are available containing experimental and calculated data. Some
estimation methods are based on theory with empirical corrections, others are based on empirical
generalizations and totally empirical correlations are also useful when theory or general principles are
not acceptable.
Several estimation methods require parameters that are characteristic of the single pure compound
or of the constituents of a mixture and classical examples are critical temperature Tc, normal freezing
point Tf and the normal boiling point Tb. Pure component properties include for example the vapor-
liquid critical properties, atmospheric boiling and freezing temperatures, dipole moments, liquid molar
volume, heat capacities, acentric factor and so on. These properties are calculated by several methods
and often can be compared with the corresponding experimental values. The properties of a molecule
are usually calculated by the contributions of its elements (atom contributions, bond contributions,
group interaction contributions…). Some examples are presented in this work.

1.1. Critical Prtoperties: Method of Joback [2,3]

The relations for the critical properties are as follows:

( ) ⟦ *∑ ( )+ *∑ ( )+ ⟧ (1)

( ) ⟦ ∑ ( )⟧ (2)

( ) ∑ ( ) (3)

where tck, pck and vck are referred to the contributions. The group identities and Joback‟s values for
contributions to the critical properties are tabulated in [4]. Tc needs the value of the normal boiling
point Tb. The Joback method for critical properties is reliable for Tc (errors usually less than 5%), but
there is a significant increase in error when estimated (and not experimental) values of Tb are used. For
the calculated values of Pc the reliability is less. For Vc, the average error is generally greather than
5%.

1.2. Critical Properties: Method of Constantinou and Gani [5]

The method of Constantinou and Gani for critical properties is based on an advanced group
contribution method ( “second order”level):

( ) ⟦∑ ( ) ∑ ( )⟧ (4)

( ) ⟦∑ ( ) ∑ ( ) ⟧ (5)

( ) ⟦∑ ( ) ∑ ( )⟧ (6)

2
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

In this case Tc does not depend on the value ofr Tb. Nk is the number of first-order groups of type
k in the molecule; the labels 1 and 2 are referred to the first-order and second-order group; Mj is the
number of second-order groups of type j in the molecule; the value of W is 0.0 and 1.0 respectively
for first-order and second-order calculations.
The first-order contributions are are usually sufficient. The second- order contributions are useful,
but in some cases they are very small and can worsen the results. Several other estimation methods
for critical properties exist in the literature and can be divided into two classes: the first class is based
on so-called factor analysis (correlations built from data of measurable, macroscopic properties such
as densities, molecular weight, boiling temperature, etc), the second class is based on so-called
molecular descriptors which are not normally measurable.

1.3. Normal boiling point Tb and normal freezing point Tf

Some estimation methods need Tb as input information for the calculation of Tc and obviously the
errors can increase (greater than 10%).

1.3.1.-Method of Joback [3, 4]

Joback, using a group contribution scheme with several new functional groups, proposes for Tf and Tb
the following relations:

∑ ( ) (7)

∑ ( ) (8)

The labels tƒk and tbk are for the contributions to Tf and Tb respectively. The errors are usually
greater than 10%.

2. Viscosity of gases and liquids

Dynamic viscosity  is a measure of the internal fluid friction; molecular theories for the gases were
satisfactorily developed by non equlibrium statistical mechanics and kinetic theory, but the
corresponding theories for the liquids are far away from the desired accuracy in the results. In this
paper the discussion is concerned with newtonian fluids, that is with fluids in which the viscosity does
not depend upon either the magnitude of the shearing stress or velocity gradient.
Estimation methods are presented able to:
- correlate viscosities with temperature and pressure;
- estimate viscosities when no experimental data are available;
- estimate the viscosities of mixtures.

2.1. Low-pressure gas viscosity, simple theoretical approach and

estimation methods
The different gas viscosity estimation methods are based on the Chapman-Enskog theory [6] or on the
law of corresponding states. In a very simple way a gas can be modeled to obtain general relations
among viscosity, temperature, pressure, and molecular size. The Chapman-Enskog model shows:

( )( )⁄
(9)

where  is in P, M is the molecular weight in g/mol, T the temperature in K,  the so-called hard-
sphere diameter in Å and a particular collision integral, different for each given intermolecular
potential model.

2.1.1-Method of Chung et al [7,8]

3
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

Chung et al start from eq (9) and obtain:

( ) ⁄
⁄
(10)

where  is in P, M is the molecular weight in g/mol, T is the temperature in K, Vc is the critical
volume in cm3/mol, the collision integral and Fc a factor necessary to take into account the
molecular shapes and polarities of dilute gases. The average absolute error appears to be about 1.9%.

2.1.2. Corresponding states methods

Starting from eq(9) Lucas [9] obtains:

( )
⟦ ⟧ (11)
( )
⁄
where  is in P, Tr is the reduced temperature T/Tc, =0.176( ) , M is the molecular weigth in
kg/kmol, Pc is the critical pressure in N/m2 . are correction factors introduced to take into
account polarity and quantum effects.
Reichenberg [10,11] has suggested an other corresponding states relation for low-pressure gas
viscosity of organic compounds which appears as follows:

⁄ ( )
⁄
(12)
⟦ ( ⁄ )⟧⟦ ( ⟧

 is in P, T is the temperature in K, Tr is the reduced temperature T/Tc, , M is the molecular weigth
in kg/kmol, r is a particular reduced dipole moment and a* depends upon the groups of the different
compounds. The errors are usually in the range 0.5% to 1.5% for nonpolar compounds and 2% to
4% for polar compounds.

2.2. The pressure dependence of the viscosity of pure gases

Usually, at reduced temperatures Tr > 1.5 and reduced pressures Pr < 2, the viscosity shows a small
dependence on the pressure; on the contrary if 1 < T r < 1.5 and P > Pc , pressure has a a very large
effect on viscosity. Near the critical point Tc , the viscosity diverges and its value is larger than
expected, even if this effect (the so-called critical enhancement) is smaller than the analogous
enhancement for thermal conductivity. The thermal conductivity usually increases by a factor of two
near the critical point, while the increase in viscosity is about 1%.
At high reduced pressures the viscosity behavior appears to simulate a liquid state, and an increase
in temperature results in a decrease in viscosity. Finally, at very high-reduced temperatures, again
pressure has little effect and viscosities increase with temperature. This behavior strongly suggests
that density is a variable to take into account for a correct descrition of viscosity at high pressures.
Very few theoretical studies were developed to estimate the dependence of the viscosity on the
pressure; Chapman and Cowling [6] , following the theory due to Enskog, faced the problem for dense
gases assuming that the gas consists of dense hard spheres and behaves as a low-density hard-sphere
system when the events occur at a faster rate owing to the higher rates of collision.

2.3. Viscosities of gas mixtures at low pressures

The kinetic theory developed by Chapman and Enskog [6 ] can be extensively used to calculate the
viscosity of low-pressure multicomponent mixtures The formulas are very complicated and are rarely
used without strong simplifications. Some methods are presented in [4] and usually they are
interpolative (the viscosities of the pure compounds must be available).

4
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

Lucas [9] and Chung et al. [7,8] proposed estimation methods over the entire range of composition.
The errors by these two methods are higher than those of the interpolative techniques, but usually less
than ±5%. Such errors could be reduced if pure component viscosity data are available and employed
by a simple linear correction scheme.

2.4. Viscosities of gas mixtures at high pressures

The methods due to Lucas [9] and Chung et al [7,8] show smaller accuracies when applied to gas
mixtures at high pressures. The accuracies are very poor in the critical region or at densities close to
those of a liquid at the same temperature. The lack of accurate high-pressure gas mixtures
experimental viscosity data does not allow accurate tests, but reasonable absolute average deviations
of 8% to 9% for both polar and nonpolar dense gas mixtures can be supposed. In any case it is
possible, when using the above cited methods, to program the same relations for low-pressure pure
gases, low-pressure gas mixtures, and high-pressure pure gases. One program can cover all those cases
and high-pressure gas mixtures.

2.5. Liquid viscosity

In many cases it is necessary to determine the liquid dynamic viscosity from experimental data when
such data exist. The viscosities of liquids are much larger than those of gases at the same temperature.
For example, the viscosity of liquid benzene near the normal boiling point (353.4 K) is more than 30
times the vapor viscosity, and at lower temperatures, this ratio increases. Moreover, below Tc, low-
pressure gas viscosities vary in a nearly linear way with the temperature; on the contrary low-pressure
liquid viscosities show a strong curvature which can be eliminated if the logarithm of the viscosity is
plotted as a function of the reciprocal absolute temperature. Usually the normal boiling point for
organic compounds would be at a value of Tr-1  1.5. For temperatures below the normal boiling point
(Tr-1> 1.5), the logarithm of the viscosity varies linearly with Tr-1, but above the normal boiling point,
this is not verified. In the nonlinear region, several corresponding states estimation methods have been
suggested. In the linear region most of the corresponding states methods are not reliable, and generally
the estimation methods are used to emphasize the effects of the chemical structure. At comparable
reduced temperatures, viscosities of polar fluids are higher than those of nonpolar liquids. For
engineering purposes it is preferable to use the ratio of the dynamic viscosity to the density, that is the
kinematic viscosity, normally expressed in m2/s or in stokes. The kinematic viscosity  decreases with
increasing temperature so that that ln is nearly linear in temperature for both the saturated liquid and
vapor.
Summarizing, pure liquid viscosities at high reduced temperatures are normally correlated with
some extension of the law of corresponding states. At lower temperatures, several methods are
empirical in nature and based on a group contribution approach. Liquid mixture correlations are
essentially mixing rules relating pure component viscosities to composition.

2.5.1. Liquid viscosity at high pressure

Lucas [12] has suggested the following equation:

( )
(13)

 = viscosity of the liquid at pressure P

sL = viscosity of the saturated liquid at Pvp

Pr = (P – Pvp)/Pc

5
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

 = acentric factor

A = 0.9991-[4.674 10-4/(1.0523T-0.03877 – 1.0513]

D = [0.3257/(1.0039 – Tr2.573)0.2906] – 0.2086

C = -0.07921+ 2.1616Tr – 13.4040Tr2 + 44.1706Tr3 – 84.8291Tr4 +

+96.1209Tr5 – 59.8127Tr6 + 15.6719Tr7

Lucas tested 55 polar and non polar liquids and claimed errors less than 10%.

2.5.2. Liquid viscosity as function of temperature

The viscosities of liquids decrease with the increasing temperature in isobaric conditions and along
the saturation line. From the freezing point to close the normal boiling point, it is a good
approximation to assume a linear relation between ln  and the reciprocal of the absolute
temperature; i.e.:

(14)

This simple equation is usually referred to as the Andrade equation [13,14]. Several modifications
of Eq. (14) have been proposed to improve its accuracy by including some function of the liquid molar
volume or introducing the use of a third constant. In conclusion, from the freezing point to close the
normal boiling point, Eq. (14) is a reliable function correlating the liquid viscosity with the
temperature even if two datum points are required. At low temperature, in order to estimate the liquid
viscosity, structural-sensitive parameters are often employed which are valid only for certain
homologous series or are found from group contributions. These methods often use some extension of
Eq. (14) and are acceptable for reduced temperatures less than about 0.75. In literature are also
described techniques emploing corresponding states methods. These methods are not particularly
reliable.

2.5.3. Liquid viscosity estimation at high temperatures

Low-temperature viscosity correlations usually assume that ln is a linear function of reciprocal
absolute temperature. Because above Tr near to 0.7 this relation is not accurate, from about Tr  0.7
to near the critical point, some estimation methods based upon corresponding states principle are
proposed which are similar to those used for gases. In particular Sastri [4] proposed:

⟦ ⟧ ( ) (15)
( )

where  is in nPa s, b is the viscosity at Tb, the value of  is 0.1175 for alcohols and 0.248 for other
compounds and  = (1 – Tr)/(1 – Tbr).
Sastri claimed average deviations of 10% for Tr> 0.9 and 6% for Tbr <Tr< 0.9. The problem of
joining high-temperature and low-temperature estimation methods for liquid viscosities should be
faced.

2.5.4. Viscosity of liquid mixtures

All the correlations for liquid mixture viscosity are generally accurate at reduced temperatures (of the
pure components) less than 0.7, when pure liquid and liquid mixture viscosities strongly depend upon
the structure of the constituent molecules. The trend of the curve of viscosity as a function of
composition can be nearly linear for so-called ideal mixtures, but mixtures containg alcohols and/or

6
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

water sometimes show a maximum and/or a minimum. Substantially the methods to estimate liquid
mixture viscosities need the knowledge of the values of the viscosities of the pure components and so
the methods are interpolative. A simple example of a correlation for the low-temperature liquid
mixture viscosity is given by the correlation of Grunberg and Nissan [15]:

∑ ∑ ∑ (16)

where x represents le liquid mole fractions and Gij is an interaction parameter depending on the
components and on the temperature.

3. Thermal conductivity of gases and liquids

Some estimation techniques for thermal conductivities  of gases and liquids are covered in this
section with the effects of temperature and pressure. Estimation techniques for gas and liquid mixtures
are also discussed. The units used for thermal conductivity are W/(m K).

3.1. Theory of thermal conductivity of monoatomic and poliatomic gases

A detailed treatment for monatomic gases, without rotational or vibrational
degrees of freedom, yields:

( ) ⁄
(17)

where:
T : the temperature in K

k : the Boltzmann constant = 1.3805 10-23 J/K

M : the molecular weight in kg/mol

 : the characteristic dimension of the molecule in m

v : the collision integral, dimensionless

Eucken [4] proposed some modifications in Eq. (17) to take into account the polyatomic gases by
introducing translational and internal degrees of freedom. Generally the methods useful for estimating
the pure gas thermal conductivity at atmospheric pressure use a particular so-called Eucken [4] factor
(M)/(Cv) which usually depends upon Tr, and ; in any case independent values of the gas
viscosity and heat capacity are necessary. Some methods require only critical temperature and
pressure and are based on the group contribution method in order to take into account the internal
degrees of freedom. Summarizing, the Eucken factor should decrease with temperature, but, from
experimental measures, in several cases the factor appears to increase. Generally all the proposed
methods show errors near 10%.

3.2. The temperature dependence of thermal conductivities of gases at

low-pressure
Thermal conductivities of low-pressure gases increase with temperature at low-pressure. The exact
function (T) is difficult to propose in analitic form: some other temperature-dependent parameters
are necessary in the equations. The range of d/dT is from 4 x 10-5 to 1.5 x 10-4 W/(m K2): the
complex and polar molecules show the larger values. Some power laws relating  with T have been
proposed without acceptable results.

7
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

3.3. The thermal conductivities of pure gases as function of pressure

The thermal conductivities of all gases increase with pressure, and three regions can be usefully
described [4]:

3.3.1. Very Low Pressure

Below about 103 bar (the mean free path of the molecules is comparatively larger with respect to the
dimensions of a measuring cell), the gas thermal conductivity can be treated as proportional to the
pressure. The term zero-pressure is usually used.

3.3.2. Low Pressure

From about 103 to 10 bar the gas thermal conductivity increases about 1% or less per bar and the 1-bar
value or the “zero-pressure”extrapolated value can be used.

3.3.3.High Pressure
The gases thermal conductivity increases with pressure, and close the critical point the effect of
pressure is very high. When the temperature increases at low pressures the gas thermal conductivity
becomes larger, but at high pressure the opposite effect is shown. Similar behaviour is evident for the
region below Tc, where  for liquids decreases with temperature, while for gases there is an increase of
 with T. Substantially pressure effects are small below Tc, exception doing for the behavior of 
near the critical point; in this region  can increase by a factor of six (the so-called enhancement in the
critical region). This enhancement is bigger for the thermal conductivity with respect to the analogous
one evidenced for viscosity.

3.4. Thermal conductivities of gas mixtures

When between the molecules of the mixture there is a great difference in polarity, usually the mixture
thermal conductivity is larger with respect to the thermal conductivity predicted from a mole fraction
average; for nonpolar molecules, the opposite trend is evident and is very high if the difference in
molecular weights or sizes of the constituents is high. It is very difficult to modify monatomic mixture
correlations to cover polyatomic molecules. The proposed techniques proposed are essentially
empirical and the corresponding states methods for low-pressure were proposed with poor results.
Very few experimental data [4] for gas mixtures at high pressures are presented in literature and
generally the researches are concerned with simple gases and light hydrocarbons: nitrogen-carbon
dioxide, ethylene mixtures with nitrogen and carbon dioxide, rare gases and binaries containing carbon
dioxide, nitrogen, and ethane, methane-carbon dioxide and methane-carbon tetra-fluoride binaries. The
proposed estimation methods are substantially modifications of procedures developed for low- and
high-pressure pure gas thermal conductivities: they are tested by the available experimental data and
average errors of about 7% are found.

4. Thermal conductivities of liquids

The thermal conductivities of organic compounds in the liquid phase are usually about 10 to 100 times
larger than those of the low-pressure gases at the same temperature. The dependence on pressure is
generally small (at pressures less than about 50 bar) and the increase of the temperature produces a
decrease of the thermal conductivity. The values of  for the common organic liquids vary in the
range 0.100 to 0.200 W/(m K) at temperatures below the normal boiling point, exception doing for
some compounds as for example water, ammonia and other highly polar molecules.

4.1- Thermal conductivityof pure liquids as function of temperature

Almost all estimation methods for pure liquids thermal conductivity are empirical, even if they appear
reliable below the normal boiling point. Two methods in particular are selected and tested in [4]:

8
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

4.1.1. Latini, et al. Method [4,16]

The method of Latini et al is very simple and the suggested correlation is as follows:

( )
⁄ (18)
where:

 = liquid thermal conductivity in W/(m K)

(19)

Tb = normal boiling temperature (at 1 atm), K

Tc = critical temperature, K

M = molecular weight, g/mol

Tr = T/Tc

The parameters A*, ,  and  are proposed for several organic families (Saturated Hydrocarbons,
Olefins, Cycloparaffins, Aromatics, Alcohols, Organic Acids, Ketones, Esters, Ethers, Refrigerants);
errors are usually less than 10%.

4.1.2. Sastri Method [4]

The Sastri method is represented by the following correlation:

=bam (20)

where:

(21)
( )

b is the thermal conductivity at the normal boiling point and is calculated by the group contribution
values and corrections; Tr and Trb are the reduced temperature and the reduced normal boiling point;
for alcohols and phenols a=0.856 and n=1.23, for other compounds a=0.16 and n=0.2. Sastri claims an
absolute average deviation of 8%.

4.2. Thermal conductivities of pure liquids as function of pressure

At pressures up to 50 to 60 bar, the dependence of the liquid thermal conductivity on the pressure is
negligible (exception doing for the range close to the critical point, where the so-called critical
enhancement occurs). At lower temperatures  increases with pressure.
A useful way (Lenoir [17]) of estimating the effect of pressure on  is given by the relation:

(22)

9
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

2 and 1 refer to liquid thermal conductivities at T and pressures P2 and P1; 2 and 1 are functions
of the reduced temperature and pressure, as shown in particular graphs. A test with experimental data
for 12 liquids, both polar and nonpolar, shows errors of only 2% to 4%.
Missenard [18] and Latini and Baroncini [19] proposed other simple correlations extending the
calculations to very high pressures.

4.3. Thermal conductivities of liquid mixtures

Several organic liquids mixtures show thermal conductivities generally less than those predicted by a
mole or weight fraction average. The methods present in the literature are usually proposed for
temperatures near 20°C.

4.3.1. Filippov correlation [20,21]

Filippov proposed the following equation:

( ) (23)

where w1 and w2 are the weight fractions of components 1 and 2; 1 and 2 are the thermal
conductivities of the pure components. Obviously 2>1. The constant 0.72 may be adjustable if
binary mixture data are available. The correlation is not suitable for multicomponent mixtures.

4.3.2 Jamieson et al. Correlation [22]

Jamieson et al proposed the follozing equation:

⁄
( )⟦ ( ) ⟧ (24)

where w1 and w2 are weight fractions and 2>1. The parameter  is equal to unity if mixture data are
unavailable. The authors claim an average error within about 7% for all types of binary mixtures.
Also in this case the correlation is not suitable for multicomponent mixtures.

4.3.3 Baroncini et al. [23,24,25] correlation

The correlation of Baroncini et al is an extension of eq(18) to binary liquid mixtures and is given by
the following equation:

( )
, . / - ⁄
(25)

where x1 and x2 are the mole fractions of the components 1 and 2. The parameters A1 and A2 can be
estimated from Eq.(19) or can be calculated from pure component thermal conductivities. The mixture
reduced temperature is Trm = T/Tcm, where:

(26)

with Tc1 and Tc2 the pure component critical temperatures. The choice between components 1 and 2 is
made with the criterion A1  A2. The test of the correlation with over 600 datum points for 50 binary
systems including those with highly polar components produced an average error of about 3%. The
method is not suitable for multicomponent mixtures.
The methods suggested in this section for estimating m have been tested by binary mixture data,
they require the thermal conductivities of the pure components and so they are interpolative in nature.

5. An up-to-date review

10
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

In the scientific and technical literature are available a lot of experimental and estimated dynamic
viscosity and thermal conductivity data for organic gases and liquids at the atmospheric pressure in the
temperature range from the normal freezing point to the normal boiling point (and over) at saturation
pressures. A complete and updated source of data can be found in the DIPPR 801 database [26]. For
data elaboration a large collection of data is also due to Vargaftik et al [27]. Both the collections are
presented with the accuracies for the experimental data; when experimental data are not available,
suggested accuracies are presented and often they are greater than 25%. Useful reviews of the
researches on thermophysical properties are due to Assael et al [28], Millat et al [29] and Horvath [30].
A last estimation method for liquid thermal conductivity can be considered and is due to
Gharagheizi et al [31] : the method is based on a group contribution method and concerns 1635 pure
chemical compounds with about 19,000 liquid thermal conductivity data extracted from the DIPPR
801 database at different temperatures. The parameters of the model cover 49 chemical substructures
plus temperature. This method is cited for its applicability in terms of number of compounds and large
temperature ranges. The authors claim an average absolute relative deviation less than 7.1%.

6. Conclusions
Several techniques to estimate dynamic viscosity and thermal conductivity of fluids [28, 29, 30] have
been published including the theoretical and semi-theoretical investigations concerned with the atomic
and molecular structure of the different compounds in the gaseous or the liquid phase. In this paper
separate sections are presented to describe the more important methods able to estimate with
acceptable accuracy dynamic viscosity and thermal conductivity of gases and liquids. These methods,
very useful for engineering purposes, are essentially based on group contributions, or on the
corresponding stetes principle, or on empirical calculations obtained from the experimental available
data. Sometimes it is difficult to realize the difference among these categories, and the distinction is
not clear (many methods present both corresponding-states and empirical features). Equations based
on group contributions and empirical correlations often lead to big tabulations of constants for
individual compounds. The methods adopted for mixtures are often valid at ambient temperature
and the lack of experimental data make difficult to known their degree of accuracy. Summarizing,
empirical estimation techniques find their greatest applications for low-pressure gas and gaseous
mixtures. The dense fluid region is generally described by corresponding-states methods.

7. Nomenclature

 acentric factor
T absolute temperature, K
Tc critical temperature, K
Tb normal boiling point, K
Tf Normal freezing point, K
Pc critical pressure, bar
Vc critical volume, cm3mol-1
Tr reduced temperature = T/Tc
M molecular weight, g/gmol or kg/kmol
 dynamic viscosity, poise (P)
 Kinematic viscosity, stoke (St)

8. References

[1] van der Waals J D 1873 Over de continuiteit van den Gas – en Vloeistoftoestand University of Leiden
[2] Joback K G 1984„„A Unified Approach to Physical Property Estimation Using Multivariate Statistical
Techniques,‟‟ S.M. Thesis, Department of Chemical Engineering, Massachusetts Institute of Technology,
Cambridge, MA
[3] Joback K G and Reid RC 1987 Chem. Eng. Comm. 57: 233

11
35th UIT Heat Transfer Conference (UIT2017) IOP Publishing
IOP Conf. Series: Journal of Physics: Conf. Series 1234567890
923 (2017) 012001 doi:10.1088/1742-6596/923/1/012001

[4] Poling BE, Prausnitz JM and O‟Connell JP 2001 ”The Properties of Gases and Liquids”, 5th ed
McGrawHill
[5] Constantinou L and Gani R 1994 AIChE J. 40: 1697
[6] Chapman S and Cowling TG 1939 The Mathematical Theory of Nonuniform Gases, Cambridge, New York
[7] Chung T, Ajlan M, Lee LL and Starling KE 1988 Ind. Eng. Chem. Res. 27: 671
[8] Chung TH, Lee LL and Starling KE 1984 Ind. Eng. Chem. Fundam. 23: 8
[9] Lucas K, VDI-Warmeatlas, Abschnitt DA, 1984 „„Berechnungsmethoden für Stoffeigenschaften,’’Verein
Deutscher Ingenieure, Düsseldorf,
[10] Reichenberg D, August 1971 DCS report 11, National Physical Laboratory, Teddington, England
[11] Reichenberg D, 1979 Symp. Transp. Prop. Fluids and Fluid Mixtures Natl. Eng. Lab., East Kilbride,
Glasgow, Scotland
[12] Lucas K 1981 Chem. Ing. Tech. 53: 959
[13] Andrade E N da C, 1930 Nature 125: 309
[14] Andrade E. N. da C 1934 Phil. Mag. 17: 497, 698
[15] Grunberg L and Nissan AH 1949 Nature 164: 799
[16] Baroncini C, Di Filippo P, Latini G and Pacetti M 1981 Intern. J. Thermophys. 2(1): 21
[17] Lenoir J M 1957 Petrol. Refiner 36(8): 162
[18] Missenard A 1970 Rev. Gen. Thermodyn. 101(5): 649
[19] Latini G and Baroncini C 1983 High Temp.-High Press. 15: 407
[20] Filippov L 1955 Vest. Mosk. Univ., Ser. Fiz. Mat. Estestv. Nauk (8)10(5): 67–69
[21] Filippov L P and Novoselova NS 1955 Vestn. Mosk. Univ., Ser. Fiz. Mat. Estestv Nauk (3)10(2):37–40
[22] Jamieson D T, Irving J B and Tudhope JS 1975 Liquid Thermal Conductivity. A Data Survey to 1973, H.
M. Stationary Office, Edinburgh
[23] Baroncini C, Di Filippo P, Latini G and Pacetti M : 1981 (pub. 1983) Thermal Cond. 17th Plenum Pub. Co.,
p. 285
[24] Baroncini C, Di Filippo P and Latini G March 1983 Workshop on Thermal Conductivity Measurement,
IMEKO, Budapest
[25] Baroncini C, Latini G and Pierpaoli P 1984 Intern. J. Thermophys. 5(4): 387
[26] DIPPR-801, Project 801 2006 Evaluated Process Design Data, Public Release Documentation, Design
Institute for Physical Properties (DIPPR),American Institute of Chemical Engineers, AIChE
[27] Vargaftik NB, Filippov LP, Tarzimanov AA and Totskii EE 1994 “Handbook of Thermal Conductivity of
Liquids and Gases”, Boca Raton: CRC Press
[28] Assael MJ ,Trusler JPM and Tsolakis TF 1996 “Thermophysical Properties of Fluids”, Imperial College
Press
[29] Millat J, Dymond JH and Nieto de Castro CA 1996 “Transport of Fluids”, Cambridge University Press
[30] Horvath AL 1992 “Molecular Design, Chemical Structure Generation from the Properties of Pure Organic
Compounds”, Elsevier
[31] Gharagheizi F, Ilani-Kashkouli P, Sattari M, Mohammadi AH, Ramjugernath D and Richon D 2013 AIChE
J. 59:1702

12