6.
Heat exchange
6.1.1 Introduction
The study of heat exchange phenomena can be reduced to
two variables: temperature and heat flux. Temperature is
an indication of the average molecular energy of a system;
the heat flux indicates the thermal energy exchange from
one body to another. On the microscopic scale, thermal
energy is related to molecular kinetic energy; as the
temperature of a material increases, the thermal agitation
of its molecules also increases. As an effect of the second
law of thermodynamics, bodies that have a bigger average
molecular kinetic energy spontaneously transfer this
energy to the bodies that have a smaller average molecular
kinetic energy. Heat fluxes between two regions depend on
several different material properties, among which thermal
conductivity, specific heat, density and surface emissivity.
In the case of heat transfer in a fluid, its flux velocity and
viscosity are very important. Heat transfer mechanisms
can be of three different types: by conduction, convection
and radiation.
Transfer by conduction takes place in solids and fluids at
rest or in laminar motion. It is defined as a process of
molecular character in which part of the energy of molecules
in the hotter portion of a material medium (therefore with a
bigger average molecular energy) gets transferred to the
molecules in the colder portion, without any mass transfer
taking place. Such a process takes place as an effect of direct
molecular collisions, as an effect of the action of vibrations
and collisions. In metals a significant portion of thermal
energy gets transferred thanks to the movement of
conduction electrons.
Convection is defined as the macroscopic process that
takes place only in fluids where the hotter fluid moves (or is
forced to move) in the regions of the fluid at a lower
temperature, where it gets remixed and transfers part of its
thermal energy. If the motion of the fluid is solely generated
by density differences caused by temperature differences,
convection is called natural; if the fluid motion is produced
mechanically, convection is called forced.
Radiation is the mechanism of heat transfer from one
body to another that takes place by radiating energy
emission and absorption. All bodies emit radiating energy in
all directions, in quantities depending on their temperature.
VOLUME V / INSTRUMENTS
In this case, energy is transported by photons of
electromagnetic radiation, mainly in the infrared and visible
region of electromagnetic spectrum. When temperatures are
uniform, radiative flux between objects is in equilibrium
conditions and therefore no net transfer of thermal energy
takes place. When temperatures are not uniform, radiative
flux generates a thermal energy transfer from hotter to
colder surfaces. In many cases, thermal exchange processes
can take place thanks to a combination of all the three above
mentioned processes.
6.1.2 Conduction
The fundamental law regulating heat transfer by conduction
was stated in 1822 by Jean-Baptiste-Joseph Fourier:
dT
[1] q = − kA
dx
where q represents the heat transferred for unit time, A is the
area of the section of the body normal to the axis along
which heat flows, ⫺dTⲐdx is the temperature gradient in the
heat flux direction, and k is a physical parameter, called
thermal conductivity, that depends on the material
characteristics of the body and that usually varies with
temperature.
Table 1 shows conductivity values of some solids and the
value of temperature at which such values have been
measured. In many cases, it is assumed that k varies with
temperature according to a linear law such as:
[2] k = k0 (1+ α T )
It is often convenient to refer to an average value of k in the
temperature interval of interest.
Coefficient k has much smaller values for fluids than
those measured for solids, as can be observed from Table 2.
In the cases where experimental values are missing, thermal
conductivity can be estimated with good approximation
through empirical relationships correlating k with specific
heat, molecular weight and fluid viscosity.
Equation [1] can be used to derive non-stationary
tridimensional equation for heat transmission in solids and
fluids at rest:
303
PROCESS ENGINEERING ASPECTS

Steady conduction
Table 1. Thermal conductivity of some solids When the heat flux is stationary, q in the Fourier
equation is a constant. In the same way, ⭸TⲐ⭸t in [3] is zero.
Conductivity By assuming k constant, [3] becomes:
Material Temperature (°C)
(kJ/h⭈m⭈°C) q⬘
[4] ∇ 2T = −
18 1,053.785 k
Gold With no heat generation (q⬘⫽0), it is possible to directly
100 1,060.07
integrate [1], thereby obtaining:
18 243.02 T1 − T2
Pure iron [5] q = kA
100 228.355 x1 − x2

18 163.41 By suitably varying coordinates, it is possible to derive

Steel (1% C) the solution of heat conduction in different situations of
100 161.315 practical importance.
In order to study the case of radial conduction in a
Magnesium 0-100 574.03 hollow cylinder, the length of the cylinder is indicated by
Cork – 0.155 L, its internal and external radius are indicated by Ri and
Re , whereas Ti and Te are the temperatures of the internal
and external face. Heat flux through an elementary layer at
∂T ∂  ∂T  ∂  ∂T  ∂  ∂T T a distance R from the cylinder axis that has a thickness dR
[3] cp r = k + k + k + q⬘
∂t ∂x  ∂x  ∂y  ∂y  ∂z  ∂z  (Fig. 1) is given by
dT
where cp is the specific heat at constant pressure of the body, [6] q = − k 2pRL
r is its density, x, y and z represent rectangular coordinates dR
and q⬘ indicates the heat generation rate – by unit volume of By separating variables and integrating, it is possible to
the body – due to chemical or nuclear reaction, or caused by obtain:

( )
the passage of an electric current. It is necessary to specify
suitable initial and boundary conditions to obtain the k 2pL Ti − Te
[7] q=
solution of [3], which gives the temperature of the body as a R
function of time and position. Equation [3] can be ln e
Ri
transformed into spherical or cylindrical coordinates to
better adapt to the geometry of the problem in question. It is possible to express the radial flux as if it took place
Notice that if k is considered temperature dependent, [3] through an equivalent plane layer that has a width equal to
becomes non-linear and therefore it cannot be solved (Re⫺Ri ) and an area equal to Aeq:
kAeq (Ti − Te )
analytically anymore, with the exception of a few particular
cases. [8] q=
Re − Ri
being
Table 2. Thermal conductivity of some fluids
[9] Aeq = 2pLRlm
Temperature Conductivity
Fluid
(°C) (kJ/h⭈m⭈°C)
Liquid ammonia ⫺15/⫹30 1.80
30 0.59
Benzene
60 0.54
Glycerine 20 1.02
Ti Te
Ethyl alcohol 20 0.65 R
Methyl alcohol 20 0.78 L
Ri
0 2.14 dR
Water 38 2.26 Re

93 2.45

⫺200 0.18
Hydrogen
0 0.60

Carbon dioxide 0 0.05 Fig. 1. Heat transmission by conduction

in a hollow cylinder.

304 ENCYCLOPAEDIA OF HYDROCARBONS

 HEAT EXCHANGE

where Rlm is average logarithmic radius defined as are in contact is Tn and that of the most external
Re − Ri surface is Te.
Rlm =
( )
[10] For cylinder A it is:
( )
ln Re Ri
k1 2pL Ti − Tn
In order to solve the problem of radial conduction [20] q=
through a spherical layer, it is common to indicate the Rn
ln
internal and external radius of the layer with Ri and Re, and Ri
the temperatures of the corresponding surfaces with Ti and and then
Te. The heat transmitted through an elementary layer at a Rn
distance R from the centre of the sphere that has a thickness q ln
Ri
dR is given by: [21] Ti − Tn =
k1 2pL
dT
[11] q = − k 4pR 2 and for cylinder B:
dR
By separating the variables and integrating, one gets: [22] q=
(
k2 2pL Tn − Te )
[12] q=
(
4pkRi Re Ti − Te ) ln
Re
Rn
Re − Ri and then:
It is also possible in this case to express the heat Re
transmitted through an equivalent plane layer that has a q ln
Rn
width equal to (Re⫺Ri ). The equivalent surface is given by: [23] Ti − Te =
k2 2pL
[13] Aeq = 4pRi Re = Ai Ae
It is therefore possible to derive:
that is, the geometric average of the internal and external
surface of the spherical layer. [24] q=
(
2pL Ti − Te )
1 Rn 1 Re
Another case of great practical interest is that of ln + ln
transmission through a series of conductors that have the k1 Ri k2 Rn
same surface; refer for instance to three consecutive plane However, if a system of two hollow spheres is taken into
walls that have areas A1, A2, A3, widths x1, x2, x3, consideration, it is possible to derive:

( )
conductivities k1, k2, k3, and indicate with DT1, DT2, DT3 the
temperature increase between the surfaces delimiting the 4p Ti − Te
[25] q=
walls. Since the heat flux through each of the three walls Rn − Ri Re − Rn
must be the same, it is +
k1 Ri Rn k2 Rn Re
k1 A1∆T1 k2 A2 ∆T2 k3 A3 ∆T3
[14] q= = = If the temperature of a material is a function of two
x1 x2 x3 spatial coordinates, bidimensional conduction occurs and the
By defining a thermal resistance for each wall: relative equation, having assumed k constant, is:
x ∂ 2T ∂ 2T q⬘
[15] r= [26] + 2 =−
kA ∂x ∂y
2
k
one gets If q⬘ is equal to zero, [26] becomes the Laplace equation,
∆T1 = qr1 ∆T2 = qr2 ∆T3 = qr3 the analytical solution of which is only possible for special
[16]
boundary conditions and geometries (Carslaw and Jaeger,
and summing together the single DTi , it is possible to derive: 1959). When it is not possible to obtain an analytical
[17] q ( r1 + r2 + r3 ) = ∆T1 + ∆T2 + ∆T3 = ∆T solution, graphic or numerical solutions are employed by
using finite differences methods, for instance.
or in other terms:

[18] q=
∑ ∆T = T − T
i i e
Non-stationary conduction
When the temperatures of the material are functions of
rT rT space and time, equations more complex than those
where rT represents the overall resistance given by the sum previously introduced are necessary. The most general
of the single resistances: equation is [3], tridimensional non-stationary. In most cases,
[19] rT = r1 + r2 + ...+ rn the solutions to practical problems involving non-stationary
conduction phenomena need to use numerical solutions to be
Equation [17] is similar to Ohm’s law for a system of developed on the computer. It is possible to find many cases
resistances in series. for problems corresponding to several different geometries
An example of heat transmission through a series and boundary conditions in the literature.
of conductors that have a variable section is given by One-dimensional problems are expressed by the
the system of two concentric cylinders A and B of following equations:
equal length L that have, respectively, conductivities k1
∂T ∂ 2T q⬘
and k2, both considered to be constant. The temperature [27] =α 2 +
of the most internal surface is Ti, whereas the ∂t ∂x cp r
temperature of the surface at which the two cylinders valid for rectangular coordinates;

VOLUME V / INSTRUMENTS 305

PROCESS ENGINEERING ASPECTS

∂T α ∂  ∂T  q⬘ heated from both sides, the width of the slab heated on one
[28] = R +
∂t R ∂R  ∂R  c p r
side and perfectly isolated on the other; x is the distance in the
direction along which heat is transmitted between the surface
for cylindrical coordinates; of a semi-infinite body to the point in question.
Usually, when integrations are performed, cp, hT , k, R,
∂T α ∂  2 ∂T  q⬘
[29] = R + Rm, T⬘, x and r are assumed constant.
∂t R 2 ∂R  ∂R  c p r In Fig. 2 the results in the case of an indefinite slab are
for spherical coordinates. shown; these results are expressed by plotting families of
In the last three equations, parameter a, also called curves, corresponding to different values of m and n, where
thermal diffusivity, was introduced, defined as: Y is plotted in logarithmic scale on the y-axis, and X in linear
k scale on the x-axis; similar results are obtained in the case of
[30] α = a cylinder and a sphere.
cp r The problem of conduction through a slab immersed in a
These equations have been solved in the case of a plane fluid is greatly simplified if the slab is so thin and the
slab, a cylinder, and a sphere immersed in a fluid. Solutions material has an electric conductivity so high that temperature
are obtained as infinite series, and results are usually shown on the entire width of the slab can be considered constant.
in a graph as curves characterized by four adimensional Suppose that the slab has a volume V, total surface A, initial
ratios that, when q⬘⫽0, are (Gurney and Lurie, 1923): temperature T0, and is in contact with cooler air at a uniform
temperature T⬘; at every instant, the quantity of heat dQ
T ⬘− T kt
[31] Y=  X = transferred over time dt is proportional to surface A, to the
T ⬘ − T0 rc p Rm2 difference between the temperature T of the slab and that of
k R x the air through the coefficient h:
[32] m= n= or n =
dQ
= hA (T − T ⬘)
hT Rm Rm Rm
[33]
where T⬘ indicates environmental temperature; T0 is the initial dt
uniform temperature of the body; T is the temperature of the Under the imposed conditions, the numerical value of h
body in a given point at time t, measured from the beginning is relatively small and the corresponding heat transfer
of the cooling or heating process; k is the thermal conductivity velocity per unit area is therefore also small. As an effect of
of the body assumed uniform and constant with varying this, if the value of conductivity k is high and the slab is thin,
temperature; r is the density of the body assumed constant; cp the temperature of the slab can be considered uniform, and
is the specific heat of the body; hT is the global heat transfer by developing a thermal balance, it is possible to obtain:
coefficient between the environment and the surface of the
body, expressed as heat transferred per unit time, unit surface
[34] ( )
dQ = hA T − T ⬘ dt = −V rc p dT
area and unit temperature difference between the surface and By assuming hA/V rcp to be constant, [34] can be easily
the environment; R is the distance in the direction along integrated giving:
which heat is transmitted between the point or the plane in the hA
middle of the body and the point in question; Rm is the radius T − T⬘ −
V rc
t
[35] =e p
of the sphere or the cylinder, half of the width of the slab T0 − T ⬘

Fig. 2. Solution 1.0 m⫽⬁

of the problem 0.7 m⫽6
of non-stationary 0.5
conduction for an undefined
slab that has a thickness 0.3
equal to 2Rm.
0.2 n⫽1
n⫽0.8
m⫽2

n⫽0.6
(T'⫺T)/(T'⫺T0)

0.1 n⫽0.4
n⫽0.2
0.07 n⫽0
0.05 n⫽1
n⫽0.8
n⫽0.6
0.03 n⫽0.4
m⫽1

0.02 n⫽0.2
n⫽1 n⫽0
n⫽0.8
n⫽0.6 n⫽0.8
0.01 n⫽0.4 n⫽0.6
n⫽0.2 n⫽0.4
0.007 n⫽0 n⫽0.2
0.005 n⫽0

0.003 m⫽0 m⫽0.5

0.002
0 1.0 2.0 3.0 4.0 5.0
a⭈t/R2m

306 ENCYCLOPAEDIA OF HYDROCARBONS


6.1.3 Convection
Heat transfer in fluids is helped by the possibility of
remixing in the fluid itself. In the case of natural convection,
such remixing movements derive from density differences in
the fluid mass due to the different temperatures present in
the various points. Movements, however, can also be
produced by stirring, or by the movement of a fluid in a pipe:
in these two cases, heat is said to be transmitted by forced
convection. The latter is the most common case in process
industries where often a hot and cold fluid separated by a
wall are pumped through the heat exchange equipment. In
the study of mechanisms controlling heat exchange in this
kind of equipment, it is necessary to account for the fact that
in the motion of a fluid around a solid, even when the bulk of
the fluid is in turbulent motion, a film in which the fluid is in
laminar motion is created near the surface. In the laminar
film zone, heat gets transferred mainly by molecular
conduction. Resistance of laminar layer to heat flux varies
with its width, and it can be responsible for 95% of total
resistance for some fluids, or only 1% for others. The
turbulent fluid bulk and the transition layer included between
the turbulent zone and the laminar film offer a resistance to
heat transfer that is a function of turbulence and of thermal
properties of the fluid. It is therefore possible to conclude
that in this case the overall heat flux is really the effect of
several different mechanisms, where conduction acts in
conjunction with convective effects.
The study of natural convection problems would
require the simultaneous solution of coupled equations of
momentum and energy balance, but this rigorous approach
is only possible for the geometrically very simple problem
of a vertical plate. Therefore it is common to use
dimensional analysis methods, widely applied when
studying heat exchange problems. They allow the
expressions for the local transfer coefficients to be
determined by using adimensional parameters: geometric
parameters of the problem and fluid properties are grouped
in adimensional groups that represent the significant
variables of the problem, so that the effect of each factor
does not have to be independently determined. This method
only operates on dimensions, and therefore does not
provide numerical values, but it allows the relative
influence of the various parameters to be estimated. By
doing so, the investigation on the experimental correlations
among data is enormously simplified.
Dimensional analysis allows us to derive that the
characteristic parameter for the natural convection problems
is given by the Grashof number, NGr, an adimensional ratio
given by:
l 3 r2 gβ∆T
[36] N Gr =
µ2
where l represents a characteristic dimension of the system,
g is the acceleration of gravity, b is the thermal expansion
coefficient of the fluid, m its dynamic viscosity, and DT the
difference between the temperature at the fluid-wall interface
and that in the fluid bulk. The following correlation, called
Nusselt equation, can be derived:
( )
m
[37] N Nu = a N Gr N Pr
(where a is an adimensional parameter) in which two other
adimensional groups appear, the Nusselt number
VOLUME V / INSTRUMENTS
HEAT EXCHANGE
[38] N Nu =
hl
k
and the Prandtl number
cp µ
[39] N Pr =
k
In [38] and [39] k represents the fluid thermal
conductivity and cp its specific heat at constant pressure.
The properties of the fluid get evaluated at a temperature
Tf that is the arithmetic average between the temperature in the
fluid bulk T⬘ and the temperature of the wall Ts. From equation
[37] it is also possible to derive a dimensional equation, where
the fluid properties are grouped in a single factor b:
h = b ( ∆T ) l 3m−1
m
[40]
Also in the case of forced convection, rigorous
approaches are limited to simple geometries and to laminar
flow situations, whereas the analyses in turbulent flux
situations are usually based on deterministic models, but are
not able to produce correlations that can be applied for
process design. To study complicated geometries, empirical
correlations are used most of the time which have limited
validity, however. In forced convection, transfer coefficients
are strongly influenced by the flow characteristics, which in
turn depend on several factors, such as turbulence intensity,
conditions at the inlet, and conditions at the wall. Also in this
case, therefore, dimensional analysis is used widely.
In order to study the problem of heat exchange towards
or from fluids flowing within cylindrical tubes, three regimes
of motion can be distinguished, depending on the value taken
by the Reynolds number:
rvD
[41] N Re =
µ
where v is the linear velocity within the tube and D is the
diameter of the tube itself. Fluid properties are evaluated in
the bulk of the fluid phase. If NRe⬍2,100, the regime of
motion is laminar: the velocity of fluid particles keeps
parallel to the tube walls and takes on a parabolic profile,
zero at the walls and maximum at the tube axis. When
NRe⬎10,000, the regime of motion is turbulent and
corresponds to a situation where the transversal component
of velocity is also present. In laminar regime, the fluid
motions keeps regular and its particles move without
remixing along stream lines that are easy to identify, whereas
in turbulent regime it is characterized by chaotic motions
(vortexes) that cause remixing. The presence of fluctuations
generated by motion irregularity makes momentum and
energy transfer in the direction normal to the tube walls
stronger, causing an increase of heat transfer coefficients.
For 2,100⬍NRe⬍10,000 the so-called transition regime
occurs where the two motions, laminar and turbulent, coexist
simultaneously in an unstable way and the situation is not
clearly determined.
In the literature several theoretical solutions have been
developed for various geometries and boundary conditions
in relation to the problem of heat transfer in a fluid flowing
in a tube in laminar regime conditions, supposing that it is
mostly due to conduction. However, these solutions mostly
neglect natural convection phenomena, which in practice are
often quite important, and it is therefore advisable to use
experimentally derived empirical relationships. Dimensional
307
PROCESS ENGINEERING ASPECTS

analysis suggests correlating data in terms of the Nusselt It is therefore possible to draw a logarithmic scale
number, NNu, or in terms of the Graetz number: diagram (Fig. 3) from which it is possible to derive the
D transfer coefficients for all motion regions, by plotting
[42] N Gz = N Re N Pr (hⲐcpG)b(cp mⲐk)b2Ⲑ3(mw Ⲑmb)0.14 on the y-axis and (NRe)b on
L the x-axis; G is the mass rate, equal to rv. Equation [49]
whereas the effects of natural convection are collected in the is represented here by a series of curves corresponding
Grashof number, NGr. to different values of the (L/D) ratio and terminating at
For horizontal circular tubes with NGr⬍100, it is NRe⫽2,100. The portions of curves corresponding to the
recommendable to use the Hausen (1943) correlation: transition regime are drawn by hand, so that they can
0.14 start at these terminal points and then tangentially link
0.085 N Gzµ  with the curve corresponding to the turbulent regime.
[43] N Nu = 3.66 + 2/3 
b

1 + 0.047( N Gz )  µ w  This diagram is widely applied, even if it has the
following limitations:
that approximately accounts for the variations of fluid • In laminar regime, it underestimates the transfer
properties through the adimensional factor (mb Ⲑmw), where mb coefficient when DT is quite high and natural convection
is the viscosity at the average fluid temperature (at which streams disturb the velocity profile.
also cp, r and k are evaluated), whereas mw is the viscosity at • In laminar regime, it underestimates the transfer
the wall temperature. coefficient when DT is so high that the film temperature
If NGr⬎100, for small DT and diameters, it is possible to goes slightly above the fluid boiling point, causing
use the Sieder and Tate (1936) equation: vapour bubble formation on the tube surface. If the
0.14
  temperature of the fluid mass is below boiling point,
1 3 µb
[44] N Nu = 1.86( N Gz )   these bubbles condense but their presence causes a
 µw  turbulence that increases the value of the transfer
A more general expression, valid for all diameters and coefficient.
DT, is obtained by adding a factor 0.87[1⫹0.015(NGr)1Ⲑ3] at • In turbulent regime, when 10⬍L/D⬍400, inlet
the second member of [44]. For vertical tubes, however, it is effects become relevant and a more adequate
advisable to use the graphs derived by Pigford (1955). correlation is:
The flux around a body immersed in a fluid is known as −0.054
 L
laminar if the boundary layer is laminar on the entire body, [51] N Nu = 0.036( N Re )0.8 ( N Pr )1 3  
even if the main flow is turbulent. In this case, the general  D
correlation stands: • If the fluid is a liquid metal, NPr takes on low values
( ) (N ) compared to those of other fluids; resistance to heat
m 13
[45] N Nu = Cr N Re Pr
transfer due to the film on the wall becomes low, due to
where the values of Cr and m depend on the body geometry the high thermal conductivity and therefore the exponent
and on its orientation with respect to the main flux. 1/3 for NPr does not give a correct estimate of the heat
In order to predict heat exchange for a fluid flowing with transfer coefficient. On the basis of theoretical
turbulent motion in a tube, several correlations have been calculations and experimental data (Rohsenow and
proposed. The use of dimensional analysis allows the Hartnett, 1973), it is advisable to use the following
following relationship to be derived: correlation:
( ) (N )
e f
N Nu = A N Re
[46]
( )
0.827
Pr
[52] N Nu = 4.82 + 0.0185 N Re N Pr
where e and f are empirical coefficients and must be
experimentally derived. All the equations so far refer to cylindrical pipes but they
The Sieder and Tate equation for turbulent motion is: can easily be extended to pipes that have a non-circular
 µb 
0.14 section, by replacing diameter D with the equivalent
( ) ( )
0.8 0.33
[47] N Nu = 0.023 N Re µ  N Pr diameter Deq, defined by the ratio:
 w
where, similarly to what was done for laminar motion, the
0.14

adimensional coefficient (mb Ⲑmw) was introduced. 0.05

b ⭈(mw/mb)

If the Stanton number, NSt, defined as L/D⫽50

L/D⫽100
N Nu h L/D⫽200
[48] N St = = 0.01 L/D⫽400
N Pr N Re c p v r
(h/cpG)b⭈(cpm/k)2/3

is introduced, the two Sieder and Tate equations can also be

written as
0.14
2 3 µ 
13
 D
N St ( N Pr )  b  = 1.86   ( N Re )
−2 3 0.001
[49]
 µw   L 0.0005
102 103 104 105 106 107
in the case of laminar regime, and
0.14
(NRe)b
2 3 µ 
[50] N St ( N Pr )   = 0.023( N Re )
−0.2
b
Fig. 3. Evaluation of the transfer coefficient
µ
 w for the forced convection problem for a fluid
in the case of turbulent regime. flowing in a tube.

308 ENCYCLOPAEDIA OF HYDROCARBONS

 HEAT EXCHANGE

Table 3. Values of D to be used in the equation [54] as a function of the number of tube rows

Number 10 and
1 2 3 4 5 6 7 8 9
of tubes above
D 0.7 0.72 0.83 0.87 0.92 0.94 0.96 0.97 0.99 1

4A When developing equations to calculate the heat transfer

[53] Deq =
P coefficient for the condensation on a vertical wall, it is
where A indicates the area of the pipe section and P the wet
perimeter, i.e. the perimeter touched by the liquid inside the
pipe.
Sometimes fluids are heated by making them pass
orthogonally to a tube layer and the calculation of h can be
performed by using McAdams equation (1954):
0.33
 cp µ 
0.6
 hDe   Dumax r 
[54]  k  = 0.33∆  µ   
     k 
In this case, fluid properties are estimated at the surface
film temperature, calculated as the arithmetic average of Tb
and Tw; De is the tube diameter and umax is the fluid rate
through the minimum free area between rows of tubes;
finally, the D value depends on the number of layers of
tubes, as shown in Table 3.
There are also graphs that can be used to obtain initial
estimates of transfer coefficients for liquids and gases used
in single phase heating and cooling operations inside tubes,
or externally to tube layers. The values given by these
graphs, however, must then be verified with more detailed
calculations, properly accounting for fluid velocity,
temperature difference and equipment geometry.
6.1.4 Condensation
Exchange phenomena taking place in a fluid undergoing a
phase change require a specific approach. Phase change of
a saturated vapour in its liquid state is called
condensation. Heat transfer associated with condensation
is usually classified as convective but, often, conduction
also provides significant contributions. Heat exchange
from a condensing vapour is a phenomenon that can be
encountered very often in practical applications in the
chemical industry. Vapour heaters and head condensers of
distillation columns are some of the most known
h[ m 2 /k 3 r 2 g] 1/3
assumed that the deposited liquid film flows downwards with
a laminar motion. This liquid film thus represents the biggest
resistance to the flow of the heat freed by the condensing
vapour, and the film width is a function of the flow length, to
be calculated according to appropriate hydrodynamic laws.
By integrating on the entire flow length, Nusselt (1916)
derived the following expression to calculate h:
−1 3
13
 µ2   4Γ 
h 3 2  = 1.47 
 µ 
[55]
 k r g
The 4GⲐm ratio represents the Reynolds number that can
be applied to this kind of flux. For vertical walls, G is
defined as the flow of condensed fluid per unit of wet
perimeter (for tubes) or wet width (for plates).
For tubes:
Γ = WF N t π D
[56]
where Nt represents the number of tubes and WF is the flow in
the film, evaluated at the bottom of the tube, where it takes its
maximum value. However, it was experimentally demonstrated
that [55] underestimates h, and therefore it is preferable to use
a coefficient equal to 1.85 rather than 1.47 (McAdams, 1954).
This discrepancy is probably due to the fact that some
discontinuities at the exchange surface, or some stresses
caused by the vapour at the vapour-film interface, or any other
kind of disturbance can create turbulences in the film.
Reported correlations, however, neglect the influence of
the Prandtl number, which can actually be relevant. This is
the reason why Dukler (1959) proposed an improved version
of Nusselt’s classical theory, obtaining as a result the graph
shown in Fig. 4, which correlates the effects of the Prandtl
1.0
NPr⫽cpm/k 10
5
2

1
examples. It is possible to distinguish two types of
condensations: film and drop condensation. In the first 0.5
case, condensing vapour forms a liquid film on the solid 0.1
surface that flows downwards, wetting it completely. When
the condensing fluid scarcely adheres to the surface, it Dukler 0.1
forms drops that do not wet it completely, thereby leaving a Nusselt
substantial portion in direct contact with the vapour. Drop
condensation gives much higher transfer rates (up to 6 or
0.01
even 18 times) than those obtained with film condensation. 0 1,000 10,000 100,000
Liquid film, in fact, gives higher resistances to heat NRe, 4G/m
transfer. However, since drop condensation is very
unstable, it is seldom used, even though it can be helped in Fig. 4. Evaluation of the transfer coefficient
principle by adding particular substances to the vapour; in for the film condensation of a pure vapour
equipment design calculations, it is often assumed that on a vertical wall, calculated by the Nusselt theory
condensation takes place with a film mechanism. and Dukler theory.

VOLUME V / INSTRUMENTS 309

PROCESS ENGINEERING ASPECTS

For instance, vaporization can take place as an effect of heat

Table 4. Values of factor Fpl according to the variation absorption by radiation and convection at the surface of a
in geometric disposition of the tubes liquid mass, or as an effect of heat absorption by natural
convection from a heated wall placed below the free surface
Tube arrangement Fpl of the liquid; in the latter case, vaporization takes place
when the supersaturated liquid reaches the free surface.
Triangular 2.10 Vaporization can also take place from falling films (and in
this case, the process is the opposite of condensation
Turned traingular 1.21 previously described), or by expansion of superheated
liquids via forced convection under pressure.
Square 1.05
When a liquid mass is heated, not only the heat flux rate,
Turned square 1.49 but also the heat transfer coefficient h vary as the difference
between the fluid-wall interface temperature Tw and that in
the liquid bulk Tb varies. For water under atmospheric
and Reynolds numbers on the transfer coefficient. It is pressure, the coefficient varies slowly as the temperature
interesting to observe that this graph does not clearly show a difference increases, if Tw⫺Tb⫽DT⬍5°C. Much faster is the
well determined transition Reynolds number. Notice also increase for DT⬎5°C, as long as a narrow maximum occurs
that deviations from the results of Nusselt’s theory decrease when the temperature difference is equal to a critical value.
as the terminal Reynolds number decreases. Furthermore, A further temperature difference increase would cause the
when NPr⬍0.4, smaller values of h are derived compared to coefficient to decrease.
those calculated with Nusselt’s theory. This is the explanation: when DT⬍5°C, the phase
However, for condensation on a single horizontal tube, variation occurs by evaporation at the vapour-liquid interface
Nusselt’s theory allows the following correlation to be but when DT⬎5°C, several bubbles form at the heating
derived: surface, from which they subsequently detach. This last heat
13 −1 3 transfer mechanism is called nucleated boiling, and produces
 µ2   4Γ 
[57] h 3 2  = 1.51 high transfer coefficients. However, when DT becomes
 k r g  µ  greater than a critical value DTc, vapour bubbles tend to
In this case, it is: form a continuous layer around the surface, insulating it and
therefore worsening thermal exchange conditions. This
[58] Γ = WF L
mechanism is called film boiling. The effect of DT on the
When the problem involves a set of tubes where the fluid transfer coefficient for water at atmospheric pressure is
drips from the upper tubes to the lower rows thus causing an shown in Fig. 5. In general, the value of DTc depends on the
increase of the width while moving downwards, even though nature of the solid surface and the liquid, but it usually falls
laminar flow is maintained, it is necessary to redefine G. In between 20 and 30°C.
this case, the total condensed flux should be divided by the At the critical temperature difference, the maximum heat
number of vertical rows of tubes Nvtr on which the flux is obtained. In real applications, usually values of DT
condensed liquid drips: lower than DTc are used in order to avoid falling in the
Γ = W N vtr L condition of film boiling, where the heat exchange surface
[59]
If tubes are circular, the value of Nvtr can be estimated as: 60
[60] N vtr = Fp 1 ( N t )0.5
The value of the Fp1 factor depends on the geometrical 40
arrangement of tubes (Table 4). In real conditions,
transfer coefficient (kW/m2⭈K)

condensate dripping from row to row causes the insurgence

of a certain turbulence, and [57] therefore underestimates h. 20
For this reason, at the second member a factor of 1.85 or the
Dukler correlation is often used (see again Fig. 4), even
though this was originally developed for vertical walls. 10
When Reynolds numbers of films fall in the turbulent field, 8
if no specific correlation is available, equation [57] or Fig. 4 6
is used.
4

6.1.5 Boiling
2
Phase change of a saturated liquid to its vapour state is
usually called boiling or evaporation. Boiling is often used to
transfer heat in chemical processes. For instance, the heat 1 2 4 6 8 10 20 40 60
necessary to perform a distillation is generally provided DT (K)
using a boiler. Heat exchange taking place during boiling is Fig. 5. Transfer coefficient for boiling water,
classified as convective, since bubbles cause significant outside a single horizontal tube submerged
turbulence but, generally, several mechanisms play a role. in a mass of water.

310 ENCYCLOPAEDIA OF HYDROCARBONS


works with a low efficiency. Moreover, since the film of
vapour acts as an insulating layer, it is possible to superheat
the metal tube on the hot side. In order to design a boiler, it
is very important to have an estimate of the maximum heat
flux, and this largely depends on the density of generated
vapours. Low pressure and vacuum operations have a bigger
potential compared to the possibility of operating in
conditions of nucleated boiling. The maximum heat flux for
a mass of fluid heated by a single horizontal tube can be
calculated on the basis of a correlation, expressed in terms of
the operating pressure p and of the critical pressure of the
fluid pc, which can be applied to most fluids, including water
and hydrocarbons (Mostinski, 1963):
0.35 0.90
 p  p
[61] ( )
q A max = 367    1 −  pc
 pc   pc 
where pressures are expressed in kPa. According to [61] the
maximum heat flux increases when p increases starting from
0, until it reaches a maximum when it is equal to 28% of pc,
to then decrease back to zero as p approaches pc. Equation
[61] is valid when heating is provided by a single horizontal
tube, but when it is given by a group of horizontal tubes,
vapour release from heating elements is less immediate due
to the proximity of adjacent tubes. In general, by increasing
the number of tubes and decreasing their distance, (qⲐA)max
decreases. In some cases, (qⲐA)max goes down to the 10% of
the value obtained with a single tube.
Heat transfer coefficient in nucleated boiling conditions
can be calculated by using, among others, the following
correlation (Mostinski, 1963):
 q  p  0.17  p  1.2
0.7
 p 
10
[62] h = 0.00441 pc0.69   1.8   + 4 +10
 A  pc  p p 
 c  c
 
This correlation does not account for the nature of the
heating surface and it can be considered acceptable for
commercial tubes, whereas very smooth tubes offer fewer
sites for bubble nucleation, and therefore give lower values
of the transfer coefficient. Turbulence created by boiling in
groups of horizontal tubes can give values of the coefficient
that are at least double those obtained for a single tube.
An expression was also proposed for the heat exchange
in film boiling conditions:
[63] h = 0.620 
( )
 kv3 rl − rv rv g 

 µv Do ∆T 
6.1.6 Heat exchangers
In most industrial equipment for heat transmission, heat gets
transferred from a hot to a cold fluid through a solid surface.
The simplest equipment used for this purpose is the double
tube exchanger (or coaxial fluids exchanger), in which one
of the fluids flows through the external tube, while the other
flows through the internal tube. For double tube exchangers,
two different exchange modes are possible, defined as
countercurrent when the two fluxes proceed in opposite
directions or cocurrent flow when the two fluids proceed in
the same direction.
Let’s suppose Tc and Th are the cold fluid temperature
and the hot fluid temperature, respectively, Tci and Tco the
inlet and outlet temperature of the cold fluid, and Thi and Tho
VOLUME V / INSTRUMENTS
HEAT EXCHANGE
the inlet and outlet temperature of the hot fluid. In cocurrent
flow, the outlet cold fluid temperature can never surpass the
hot fluid outlet temperature (Tho⬎Tco). As the heat exchange
surface gets higher, Tco gets closer to Tho.
In countercurrent it can happen that Tco⬎Tho, since there
is a bigger thermal recovery here. While the temperature
increase in a countercurrent exchanger does not vary
significantly, whereas for cocurrent flow it goes from a
maximum at the inlet section down to a minimum at the exit
section (Fig. 6).
Heat transfer between two fluids separated by a plane
wall is usually described using the equation:
[64] q =UA (Th − Tc )
according to which the heat q transferred in the unit time
from the hot to the cold fluid is equal to the product between
a proportionality factor U, called global transfer coefficient,
the area of the surface A through which heat flows and the
difference between the temperatures of the two fluids, Th⫺Tc.
Equation [64] applies only locally, or when exchange
conditions are kept constant on the whole equipment,
otherwise it should be expressed in differential form:
[65] dq =U (Th − Tc ) dA
The global transfer coefficient accounts for a series of
resistance to heat flow that appear in a system like this,
including the fluids themselves (rh and rc), the deposits due
to the scaling on the hot and cold side of the wall that
separates the fluids (rhs and rcs), and the solid wall itself (rw).
The overall transfer coefficient can be expressed as the sum
of the resistances in series:
 1
[66] U=
rh + rhs + rw + rcs + rc
The inverse of each one of these resistances is called
conductance and is indicated by h. The values of
conductance of the two fluids, he and hc, represent the heat
transfer coefficients. The heat flux (q/A) through each of the
single resistances is the same; temperature increase through
each resistance is inversely proportional to the value of h:
q q 1
[67] ∆Th = rh =
A A hh
The difference between the temperature of the hot fluid
and the cold fluid is given by the sum of DTs for the single
resistances.
T T
Thi Thi
Tho
Tho
Tco
Tco Tc
Tc Tci
Tci
inlet outlet inlet outlet
cocurrent countercurrent
A B
Fig. 6. Temperature trends for the two fluids
in a cocurrent flow exchanger (A) and in a countercurrent
flow exchanger (B).
311
PROCESS ENGINEERING ASPECTS

Usually the solid wall, which has a width equal to y, is However, most equipment has tubular exchange surfaces,
made up of conductor material, and the quantity of heat that where the external area Ao is larger than the internal area Ai.
runs through it is given by: Equation [66] must therefore be modified in order to refer
k all resistances to the same surface, especially the surface
[68] q = A∆Tw used to define A. Usually, in the case of tubular equipment,
y
one refers to the external surface, and the overall transfer
Very often rw is negligible, since the thickness of the coefficient is therefore given by:
wall is very small, but also because the value k of 1
conductivity is very high. [73] U =
Ao Ao A
Scale is a term defining any layer, or deposit, of r + r + o r +r +r
extraneous material on a surface that must exchange heat. Ai i Ai is Alm w os o
These materials usually have low thermal conductivity, and The resistance of the metallic wall of the tube is
therefore they display a high resistance to thermal exchange. multiplied by the ratio between the external surface and an
These layers can be created as an effect of the deposition of average surface, calculated by the average logarithmic value
finely subdivided substances present in the process fluids, or of the radius of the tube defined in [10].
by crystallization of some substance whose solubility at the Finally, it is necessary to take into consideration the case
temperature of the wall is lower than the solubility at the where U is not constant but varies with the temperature of
temperature in the fluid bulk. The formation of rather thick the two fluids. In this case, most of the time it is assumed
and hard layers of scales at the wall is caused by that the dependency is linear, such as:
polymerization reactions and sometimes by corrosion [74] U = α∆T + β
products of the walls themselves.
As noted before, when exchange conditions vary along By integrating [65], the following expression is
the equipment, it is necessary to use equation [65]. An obtained:
expression of heat flux as a function of inlet and outlet
temperatures of the hot fluid (Thi and Tho) and of the cold q= A
( ) (
U c Thi − Tco − U h Tho − Tci )
(T − T )
[75]
fluid (Tci and Tco) can be derived with the following Uc hi co

(T − T )
hypotheses: a) the overall transfer coefficient U is constant ln
Uh ho ci
on all the equipment; b) the exchanger operates in stationary
conditions; c) the specific heat of the two fluids is constant where Uc and Uh are the values of the transfer coefficient at
along the whole length of the exchanger; d ) no phase the coldest point and the hottest point, respectively, of the
variations occur in the system; and e) heat dispersions are exchanger.
negligible. In practice, cocurrent flow heat exchangers are seldom
In particular, if the flows of the fluids take place in used, since they offer a limited exchange area. The most
countercurrent, it is possible to derive: widely used exchangers are those of the shell-and-tube type
that are versatile and can be used also in the presence of
[69] q = UA
(T hi ) (
− Tco − Tho − Tci ) phase changes (for instance, condensation and boiling) with
T −T operating pressures ranging from values close to vacuum
ln hi co conditions to values in the order of 40 MPa. In practice,
Tho − Tci
shell-and-tube exchangers are composed of two plates to
The effective temperature difference, called logarithmic which a certain number of tubes (tube bundle) are connected,
average temperature difference, is then given by: inserted within a shell. With this set-up, it is possible to
compact a great exchange surface into a small space. Fig. 7
[70] ∆Tlm =
(T
hi ) (
− Tco − Tho − Tci ) shows two schematic drawings of exchangers. One of the
T −T two fluids runs through the tube bundle: in the simplest case,
ln hi co it enters at one head and exits at the other; in the more
Tho − Tci
general case, it runs through some of the tubes in one
If, however, the two fluids flow in parallel, it is possible direction, and then inverts the flux direction, and runs
to derive: through other tubes in the inverse way, to then invert the
direction again, and so on, until n runs within the tube
[71] q = UA
(T hi ) (
− Tci − Tho − Tco ) bundle are performed (n stages). The exchanger shown in
Thi − Tci Fig.7A is of the 2-stage type, that shown in 7B is of the
ln 4-stage type. The other fluid runs through the cylindrical
Tho − Tco
shell containing the tubes, hitting the tubes in a rather
In this case, the average logarithmic temperature complex way. Usually, in order to improve heat exchange,
difference is given by: some baffles are inserted within the shell that partially close
the shell section, allowing the tubes to cross them: these
[72] ∆Tlm =
(T
hi ) (
− Tci − Tho − Tco ) baffles are fastened alternatively on opposite sides of the
T −T shell. The fluid therefore zigzags, hitting the tubes in a
ln hi ci direction that is approximately perpendicular in the space
Tho − Tco
between two baffles, whereas it moves more or less parallel
The overall transfer coefficient U can be directly to the tubes when it is close to the baffles’ openings. Section
evaluated by [66] only if the exchange area is constant, as reduction, which causes flow rate increase, and the fact that
happens in the case of transfer through plane surfaces. the fluid partially hits the tube bundle frontally, cause a

312 ENCYCLOPAEDIA OF HYDROCARBONS


A were introduced to derive equation [69]. In general, values of
B
Fig. 7. Shell-and-tube exchangers: A, 2-stage type;
B, 4-stage type.
turbulence increase and therefore an improvement of the
transfer coefficient. Usually, the openings in the baffles
account for about 25% of the total section of the shell.
The calculation of the transfer coefficient for the tube
side is immediate, since it is possible to use the criteria
already seen for the flux in tubular pipes (see again Fig. 3).
Less simple is the calculation of the coefficient for the fluid
flowing within the shell, since the regime of motion is rather
complex in this case. However, a few correlations have been
proposed, such as the one by Kern (1950):
0.14
 Deq Geq   c p µ   µ 
0.55 13
hde
[76] = 0.36 
k  µ   k   µ f 
similar to [54].
Equivalent diameter Deq is calculated by [53]. Geq
indicates an equivalent mass rate. At a baffle opening, where
the flux is parallel to the tubes, a mass rate Gw is estimated
by the ratio between the fluid flow rate and the free section,
Sw, representing the total area of the opening from which it is
necessary to subtract the sum of the sections of the tubes that
cross it. Moreover, in the space included between two
baffles, where the flux is normal to the tubes, the free section
area, Ss, can be approximately calculated using the
expression yLxpDs ⲐyT , where Ds is the shell diameter, yT is the
distance between the axes of two neighboring tubes located
on the same horizontal row, yL is the distance between two
neighbouring tubes on the same horizontal row (yL⫽yT⫺De)
and xp is the distance between subsequent baffles. Notice that
Ds ⲐyT is the approximated value of the number of tubes and
therefore also of the number of free spaces at the shell
diameter, whereas the product yLxp gives the area of the free
space between two neighboring tubes. Therefore in the space
between baffles, it is possible to estimate a mass velocity Gs,
given by the ratio between the fluid mass flow and Ss. The
equivalent mass rate Geq, introduced in [76], can be
calculated by the geometric average of Gw and Gs:
[77] Geq = GwGs
Equations such as [69], derived for cocurrent flow
systems, are also valid for shell-and-tube exchangers, as long
as a corrective factor Y is introduced, since the two flows are
VOLUME V / INSTRUMENTS
HEAT EXCHANGE
neither countercurrent nor cocurrent in this kind of
equipment. The value of Y is always smaller than 1 and it
depends on the type of exchanger, and on the inlet and outlet
temperatures of the two fluids. The values of Y have been
calculated (Kern, 1950) for a series of fluxes of common
interest, based on the assumptions mentioned above, that
Y smaller than 0.8 are considered unacceptable and the
corresponding exchanger configurations is considered
inefficient.
In exchangers the tendency is to maintain the fluid rates
high, so as to obtain transfer coefficients that are as high as
possible, without increasing pressure drops above reasonable
limits.
6.1.7 Radiation
A heated body emits electromagnetic radiations at an
intensity and frequency that depends on its temperature. For
instance, if the wire of an incandescent lamp is electrically
heated, both the energy quantity emitted in unit time and the
portion of visible radiation increase as temperature
increases. At a temperature below 400°C, the radiation
emitted by the lamp is not perceived by the human eye, but
gets felt by human skin as heat.
The quantitative description of this phenomenon is based
on quantum mechanics, but in qualitative terms it can be
explained by observing that, when transferring energy to a
solid body, some of its atoms and molecules jump to excited
states, but they tend to spontaneously go back to lower
energy states. When this occurs, energy gets emitted as
electromagnetic radiations, which depend on the changes
that happen in the electronic, vibrational and rotational state
of atoms and molecules, and are thus distributed on different
wavelengths. The inverse process, called absorption, takes
place when some radiating energy is transferred to a body,
and thus it moves to an upper energy level. This
phenomenon occurs when radiating energy hits a body
thereby.
Radiating energy incident on the surface of a solid
surface is partially absorbed, partially reflected and partially
transmitted. It is therefore possible to write:
a + r +τ =1
[78]
where a indicates absorptance, i.e. the energy fraction
absorbed by the body, r is reflectivity, i.e. the reflected
fraction, and t is transmittivity, i.e. the transmitted fraction.
In most cases, materials of practical interest are
represented by opaque substances with zero transmittivity.
Bodies absorbing all incident energy are called black bodies.
No real surface in practice behaves like a black body. The
best example of a black body is an empty cavity with opaque
walls in communication with the environment through a hole
that has a negligible area compared to the surface of the
cavity. All the radiating energy entering the cavity remains
trapped in it, and the hole behaves like a black body.
Two bodies with areas A1 and A2 contained in an
insulated cavity externally emit radiating energies equal to
A1⭈e1 and A2⭈e2, where e represents the total emissive power,
i.e. the energy emitted per unit time and unit surface in the
hemisphere located above each elementary surface. Thus, the
energy incident on the unit surface of the two bodies coming
313
PROCESS ENGINEERING ASPECTS
from the cavity surface is indicated with eB, and the
absorbing powers of the two bodies, which is the ratio
between the absorbed and the incident radiation, is indicated
with a1 and a2. In thermodynamic equilibrium conditions,
the energy incident on and absorbed by each body must be
equal to the energy emitted, and therefore it is:
[79] eB A1α1 = A1e1; eB A 2α 2 = A 2 e2
or, in other terms:
e1 e2
[80] = =e
α1 α 2 B
This relationship allows to state that in equilibrium
conditions, the ratio between absorbing and emitting power
is equal for all bodies (Kirchhoff ’s law). For a black body,
a⫽1 and therefore e⫽eB. The emitting power of the black
body depends exclusively on temperature.
The ratio between the emissive power of a generic
surface and that of a black body is called surface emissivity
and is indicated by:
ei
[81] ε=
eB
In thermal equilibrium conditions, the emissivity of a
body is equal to its absorbing power. Besides the total
emissive power, it is also possible to define the specific
emissive power el, that represents the energy emitted in the
hemispace for unit surface, unit time, and in a wavelength
interval included between l e l⫹dl. Total emissive power
not only depends on thermodynamic temperature T, but also
on the wavelength l. For the black body, this dependency
was discovered by Planck in 1900 (Planck, 1923):
2π hP c 2 λ −5 c1λ −5
[82] eλ = h c k λT
dλ = c λT
e P B −1 e 2 −1
where c is the speed of light in vacuum, equal to 2.9979⭈108
m/s, kB is the
Boltzmann constant, equal to 1.3807⭈10⫺23 J/K, hP is
Planck’s constant equal to 6.6261⭈10⫺34 J⭈s, whereas c1 and
c2 are the so-called first and second constant of Planck’s law
that are equal to 3.7403⭈10⫺16 J⭈m2/s and 1.4387⭈10⫺2 m⭈K,
respectively.
Fig. 8 shows the dependency of el on the wavelength, at
different values of temperature. Each one of the curves
obtained at a given temperature shows a maximum for a
wavelength that varies as temperature varies, according to
Wien’s law:
0.2898
[83] λ max =
T surface can vary between 0.65 and 0.95, depending on the
with the constant expressed in cm⭈K.
The total emissive power of a black body can be derived
from the specific power through an integration on the whole
wavelength spectrum.
By using the expression of el given by Planck’s law and
calculating the integral, the Stefan-Boltzmann law is
obtained: eB⫽c1T 4Ⲑc24⫽sT 4, where s is equal to 5.67⭈10⫺8
W/(m2⭈K4).
Calculation of radiating energy transferred between
two surfaces
In order to calculate the radiating energy transferred
between two surfaces, it would be necessary to consider the
314
100
el (a.u.)
90
80
2,040°C
70
60
50
40
30
1,540°C
20
10
1,200°C
0
0 1 2 3 4 5
l (mm)
Fig. 8. Dependency of el the specific emissive power
on the wavelength for different values of temperature.
exchange of monochromatic radiations and integrate over the
whole wavelength spectrum. For the sake of simplicity, it is
more convenient to calculate the energy that the two bodies
would exchange if they were black, by using Stefan’s law,
and multiply the result thus obtained by the emissivity of the
surface involved. In general, the emissivities of bodies
depend on temperature and on the nature of surfaces.
Metallic surfaces at low temperature can reach emissivity
values around 0.05, which can increase up to 0.70 under
high temperatures. The emissivity of an oxidized metallic
temperature of the surface itself. Bricks, glass, marble and
paper usually have emissivities larger than 0.90.
The absorptance of a surface not only depends on the
surface nature and its temperature, but also on the spectral
distribution of the incident radiation. If the body receiving
the radiation absorbs all the wavelengths with the same
absorptance, it is called grey and its absorptance is
independent from the energetic distribution of the incident
radiation. In this case, absorptance a is equal to the
emissivity of the black body at its temperature.
At this point, it is possible to calculate the radiating
energy transmission between two surfaces of solids
separated by a non-absorbing medium. If they are two
ENCYCLOPAEDIA OF HYDROCARBONS
 HEAT EXCHANGE

parallel surfaces with an infinite extension, grey, with the cavity can be considered like a single reflecting surface
emissivity e1 and e2 at temperature T1 and T2, body 1 emits AR at uniform temperature TR.
the radiating energy e1⫽e1sT14 for unit time and surface, of If there are two radiating surfaces A1 and A2, by
which the fraction e2 is absorbed by body 2 and the fraction indicating their geometrical factor with F12, the one between
(1⫺e2) is reflected again. This in turn will be partly absorbed the first surface and the reflecting one with F1R, the one
and partly reflected, and so on. between the reflecting surface and the second surface with
It is possible to demonstrate that the energy absorbed by FR2 and, finally, with FRR the reflecting surface with itself (if
surface 2, per unit of emitting surface, is given by: the surface is plane it is FRR⫽0), it is possible to
q1→ 2 σ T14 demonstrate:
[84] = F1R FR 2
A 1 1 F12 = F12 +
+ −1 [91]
ε1 ε 2 1 − FRR
and similarly: in fact, the fraction F12 of the energy emitted by A1 is
q2→1 σ T24 directly absorbed by A2, and the latter also receives the
[85] = fraction FR2 Ⲑ(1⫺FRR) of the radiation initially incident on AR.
A 1 1
+ −1 The considerations developed above are valid provided
ε1 ε 2 the radiating surfaces involved are black. An exact
Therefore the energy transferred in total from 1 to 2 is: calculation of the heat transferred between two non-black
surfaces is, however, too complicated and therefore it is
[86]
q
A1
= σ (T14 − T24 )
1 1
1
simplified by assuming that surfaces are grey. If one
+ −1 considers two radiating surfaces A1 and A2 with emissivities
ε1 ε2 e1 and e2 in a cavity, it is possible to demonstrate that the
If the two surfaces are black, since e1⫽e2⫽1, [86] becomes: exchanged heat, due to the combined mechanism of direct
radiation, reflecting surface and multiple reflection inside
[87]
q
A1
= σ (T14 − T24 ) the cavity, can be expressed by the equation

If, however, the two surfaces are not parallel and have an
[92] ( ) (
q = σ A1Φ12 T14 − T24 = σ A2Φ 21 T14 − T24 )
infinite extension, it is necessary to introduce a geometric where factor F12 is given by:
factor F12, called view factor, defined as the fraction of 1
energy emitted by surface A1 in all directions intercepted by [93] Φ12 =
1 1  A1  1 
+ −1 + − 1
F12  ε1  A2  ε 2
surface A2.
Therefore it is: 

A1F12 = A 2 F21
[88]
In the presence of different surfaces that together receive
all the radiating energy emitted by a single surface A1, the
following relation stands:
[89] F11 + F12 + F13 + F14 + ... = 1
If A1 does not see any part of itself, then F11⫽0. The
values of the geometric factors involved in the solution of
the most common problems have been calculated and shown
in diagrams. The practical calculation of the view factor is,
in any case, often quite complicated. The diagrams derived
by Hottel (1950, 1954) allow to derive the view factors in the
case of direct radiation between two adjacent rectangles
located on perpendicular planes, and in the case of direct
radiation between identical geometrical forms that are
located in opposite positions on parallel planes. When the
two parallel radiating surfaces are connected by a reflecting
surface, the energy does not only transfer directly between
the two surfaces, but in part also by reflection through the
reflecting surface. In this case, the quantity of exchanged
heat can be then expressed by the equation:
[90] ) )
q = σ F12 A 1(T14 − T24 = σ F21 A 2(T14 − T24
_
where the values of the geometric factor F are always bigger
than the values of F since the presence of a reflecting surface
between two parallel surfaces increases the quantity of heat
exchanged in the unit time.
If a closed cavity is taken into consideration, as the
inside of a furnace, it can be subdivided into a certain
number of radiating surfaces A1, A2, …, whereas the rest of
Equation [92], which is the most general among those
presented above to describe heat exchange, can be used to
derive a heat transfer coefficient. In fact, it can be rewritten
in the following form:
T 4 − T24
[94] q = σ A1Φ12 1
T1 − T2 1 2
(
T −T )
from which the following expression for the transfer
coefficient can be derived:
T14 − T24
[95] h = σΦ12
T1 − T2
Gas radiation
When studying solid surfaces radiations, it is possible to
assume with good approximation that bodies under
examination are grey, which means that emissivity or
absorptance are independent from the frequency of the
radiation. In the study of gas radiation, it is necessary to
account for this dependency. In fact, if the radiation of a
black body passes through a mass of gas, absorption takes
place only in some frequency intervals of the infrared
spectrum. Similarly, if the gas gets heated, it irradiates
energy in the same frequency interval. These infrared
frequencies are produced by molecular transitions between
rotational and vibrational quantum levels.
In practice, among the gases contained in combustion
products, carbon monoxide and dioxide, water vapour,
sulphur dioxide, hydrogen chloride, and hydrocarbons have
an appreciable value of emissivity. On the other hand, gases
with symmetric diatomic molecules such as nitrogen, oxygen

VOLUME V / INSTRUMENTS 315

PROCESS ENGINEERING ASPECTS
and hydrogen do not show a significant emissivity.
Emissivity and absorptance of a gaseous mixture depend on
its radiating components, and they are proportional to their
concentration and to the average width L of the gaseous
mass.
Emissivity values of the most important gases depend on
temperature, on their partial pressure, and on the width L of
the gaseous mass, which can be calculated by using the
following equation:
 V
[96] L = 0.85  4 
 S
where V is the gas volume and S is the total surface.
Total emissivity, or absorptance, of a gas mixture is
smaller than the sum of those of the pure components. If, for
instance, one considers a mixture of CO2 and H2O, you get:
[97] ε g = ε CO + ε H O − C
2 2
where eg is the emissivity of the gas at a given temperature,
eCO2 is CO2 emissivity at the same temperature, eH2O is water
emissivity, C is a correction term depending on both eCO2
and eH2O.
Ovens and furnaces
One of the most important industrial applications of heat
transfer by radiation is represented by ovens, which are
largely used in oil refineries, for crude oil atmospheric
pressure and under vacuum distillation, in thermal cracking
and many service operations. Ovens are equipment in which
most of the total heat gets transferred by direct radiation
from a flame towards a surface capable of absorbing heat.
Since radiation essentially moves in a straight line, these
surfaces must be capable of seeing the sources that emits the
radiation. Usually in industrial ovens, the bodies receiving
the heat are tubes containing the fluid that must be heated,
and these tubes in their turn, are contained within walls
made of a high reflectivity material; therefore it is not easy
to theoretically determine the paths covered by radiation and
the thermal transfer rate.
Generally for ovens, heat flux can be expressed through
the following equation:
[98] q ⫽ ℑsA⬘(T14⫺T24)
where T1 is the temperature of the source, T2 is the
temperature of the tubes, A⬘ is an effective surface of the
tubes, and ℑ is an adimensional transfer factor.
In most ovens, tubes are located on one or two rows, right
in front of the walls. Part of the radiation directly hits the
tubes, which absorb it, whereas the remaining portion hits the
walls, which partly reflect it and partly disperse it in the
atmosphere. It is possible to estimate an effective plane
surface, Acp, to which a unit emissivity is assigned, in order to
replace the total tube surface in design calculations,
multiplying the number of tubes by their length exposed to
radiation and by the distance between their centres. Since the
set of tubes does not entirely absorb the energy emitted by the
flame, it is necessary to also calculate an absorption
efficiency factor a. A method to calculate a was developed
by Hottel (1931). Exchange factor value ℑ mostly depends on
the emissivity of the gases produced in the combustion and
on the radiation reflected by the walls. Important parameters,
on the basis of what was discussed above, are therefore the
concentration of CO2 and H2O, gas temperature and the
316
width of the gaseous mass. Since most of the energy incident
on the walls gets reflected in the tubes direction, ovens that
have large parts of walls exposed to radiation are capable of
transferring more heat per unit surface by radiation than those
where walls are greatly shielded by tubes.
Bibliography
Fanaritis J.P. et al. (1980) Heat exchange technologies (heat transfer),
in: Grayson M. (executive editor) Kirk-Othmer encyclopedia of
chemical technology, New York, John Wiley, 1978-1984, 26v.;
v.XII, 129-170.
Graetz L. (1880) «Mathematik für Physiker», 25, 316.
Holman J.P. (2007) Heat transfer, New York, McGraw-Hill.
Incropera F.P., DeWitt D.P. (2001) Fundamentals of heat and mass
transfer, New York, John Wiley.
References
Carslaw H.S., Jaeger J.C. (1959) Conduction of heat in solids,
Oxford, Clarendon.
Dukler A.E. (1959) Dynamics of vertical falling film systems,
«Chemical Engineering Progress», 55, 62-67.
Gurney H.P., Lurie J. (1923) Charts for estimating temperature
distributions in heating or cooling solid shapes, «Journal of
Industrial and Engineering Chemistry», 15, 1170-1172.
Hausen H. (1943) Presentation of heat transfer in tubes by means of
generalized exponential functions, «Zeitschrift des Vereines
Deutscher Ingenieure», 4, 91-98.
Hottel H.C. (1931) Radiant heat transmission between surfaces
separated by non-absorbing media, «American Society of
Mechanical Engineers. Transactions», 53, 265-273.
Hottel H.C. (1950) Radiant heat transmission, in: Chemical engineers’
handbook, New York, McGraw-Hill, 493-498.
Hottel H.C. (1954) Radiant heat transmission, in: McAdams W.H.,
Heat transmission, New York, McGraw-Hill, Chapter 4.
Kern D.Q. (1950) Process heat transfer, New York, McGraw-Hill.
McAdams W.H. (1954) Heat transmission, New York, McGraw-Hill.
Mostinski I.L. (1963) Application of the law of corresponding states
to the calculation of heat transfer and critical heat flux for boiling
liquids, «Teploenergetika», 4, 66-71.
Nusselt W. (1916) Die Oberflachen-Kondensation des Wasserdampfes,
«Zeitschrift des Vereines Deutscher Ingenieure», 60, 541.
Pigford R.L. (1955) Nonisothermal flow and heat transfer inside
vertical tubes, «Chemical Engineering Progress», 51, 79-92.
Planck M. (1923) Vorlesungen über die Theorie der Wärmestrahlung,
Lipsia, Barth.
Rohsenow W.M., Hartnett J.P. (1973) Handbook of heat transfer,
New York, McGraw-Hill.
Sieder E.N., Tate G.E. (1936) Heat transfer and pressure drop of
liquids in tubes, «Journal of Industrial and Engineering Chemistry»,
28, 1429-1435.
List of symbols
A area of the section of the body normal to the heat
flow direction; section of a tube
a absorbivity of a body
Ae external surface of a spheric layer
Aeq equivalent surface for heat exchange problems
through a cave cylinder, or a spherical layer
Ai internal surface of a spherical layer
ENCYCLOPAEDIA OF HYDROCARBONS
 HEAT EXCHANGE

c speed of light in vacuum rw in a heat exchanger, the resistance offered to heat

cp specific heat at constant pressure exchange by the wall
D tube diameter Ss in a heat exchanger, area of the free section in the
Deq equivalent tube diameter space between two baffles.
Ds diameter of the cylindrical shell of an exchanger Sw area of the free section in the shell of an exchanger
e total emissive power of a body T thermodynamic temperature
el specific emissive power of a body t time
Fi,j view factors T⬘ environmental temperature
Fp1 parameter in equation [60] T0 in a non-stationary conduction problem, initial
temperature of a body
g gravity acceleration
Tb temperature in the fluid bulk
G mass velocity of a fluid in an exchanger
Tc in an exchanger, temperature of the cold fluid
h heat transfer coefficient
Tci in an exchanger, inlet temperature of the cold fluid
hT global heat transfer coefficient between the
environment and the surface of a body, Tco in an exchanger, outlet temperature of the cold fluid
in a non-stationary conduction problem Te temperature of the external surface of a hollow
k thermal conductivity cylinder, or of the external surface of a spherical
layer
kB Boltzmann constant
Th in an exchanger, temperature of the hot fluid
L length of a cylinder or a tube; width of a gaseous
mass Thi in an exchanger, inlet temperature of the hot fluid
l characteristic dimension of the system in natural Tho in an exchanger, outlet temperature of the hot fluid
conduction problems Ti temperature of the internal surface of a hollow
NGr Grashof number cylinder, or of the internal surface of a spherical
layer
NGz Graetz number
Tn temperature of the contact surface between two
NNu Nusselt number
concentric cylinders, or spheres
NPr Prandtl number
Ts wall temperature
NRe Reynolds number
U global transfer coefficient
NSt Stanton number
umax fluid flow rate through the minimum free area
Nt number of tubes between tube rows
Nvtr number of vertical rows of tubes on which the V volume of a body
condensed fluid drips
v linear velocity of a fluid within a tube
p pressure
WF flow in a condensing film
pc critical pressure of a fluid
xp distance between two subsequent baffles in an
q heat exchanged per unit time exchanger
q⬘ heat generation rate Y correction factor to use when the fluxes in a heat
r thermal resistance of a wall exchanger are neither parallel nor in countercurrent
R in a non-stationary conduction problem, distance, in Y adimensional ratio (eq. [31])
the direction along which heat is transmitted, yT distance between the axes of two neighbouring
between the point or the plane in the middle of the tubes in an exchanger
body and the point under examination
r reflectivity of a body Greek letters
rc in a heat exchanger, the resistance offered to heat r density
exchange by the cold fluid
a thermal diffusivity
rcs in a heat exchanger, the resistance offered to heat
b coefficient of thermal expansion of a fluid
exchange by the scale on the cold fluid side
m dynamic viscosity of a fluid
Re external radius of a hollow cylinder, or of a
spherical layer mb viscosity of a fluid at its average temperature
rh in a heat exchanger the resistance offered to heat mw viscosity of a fluid at the wall temperature
exchange by the hot fluid D parameter of equation [54]
rhs in a heat exchanger the resistance offered to heat G flux of condensed fluid per perimeter unit
exchange by the scale on the hot fluid side t transmittivity of a body
Ri internal radius of a hollow cylinder, or of a e emissivity of a surface
spherical layer l wavelength
Rlm logarithmic average radius s Stefan-Boltzmann constant
Rm in problems of non-stationary conduction, radius of
the sphere or of the cylinder, or half of the width of Stefano Carrà
the slab heated from both sides, or width of the slab
MAPEI
heated on one side and perfectly insulated on the
other Milano, Italy

VOLUME V / INSTRUMENTS 317