You are on page 1of 147

Heat Transfer

Conduction:
Conduction of heat means transport of thermal energy from a region at high
temperature to a region at lower temperature without any microscopic motion of
the medium. The difference of temperature between these regions causes the flow
of heat and is called thermal driving force. Heat conduction is also called diffusion
of heat.
The mechanism of heat conduction in a medium depends upon the state of the
medium, i.e. whether it is a solid, or a liquid or a gas. Transport of heat in a solid
occurs only by conduction.
The Basic Law of Heat Conduction:
The basic law of heat conduction in a medium was established by J.B.J Fourier in
1822 from his experimental data on the rate of heat flow. The general form of
Fourier’s law or specifically known as Fourier’s first law can be described as
𝐪 = −k𝛁T (1)
where q is the vector form of heat flux (W/m ) and 𝛁T i.e. the temperature
gradient (K/m). Accordingly, the constant of proportionality, k, bear a unit of .
.
The constant of proportionality, k, is known as the “Thermal Conductivity” of the
medium, which may be solid, liquid or gas.
Instead of the vector form, as given in eqn. 1, we may use scalar forms of the
Fourier’s first law, which are
(i) In Cartesian coordinate
q = −k , q = −k , q = −k (2)
(ii) In cylindrical coordinate
q = −k , q = −k , q = −k (3)
(iii) In spherical coordinate (r, θ, ϕ)
q = −k , q = −k , q = −k (4)

Fig. 1

1|
For r, θ, ϕ identification, refer fig. 1.
Anisotropic and Isotropic Continuum:
A continuum is said to be homogeneous if its thermal conductivity does not vary
from point to point.
A continuum is called heterogeneous if there exists, such a variation.
A continuum is termed as isotropic if the conductivity is same in all directions.
A continuum is termed as anisotropic if there exists, directional variation of
conductivity.
In the statement of Fourier’s first law (eqn. 1), we have considered an isotropic
material. In case of anisotropic heat conduction, the component of heat flux in any
direction, for example q in x direction, depends on the temperature gradients in
each of the three coordinate directions. Therefore,
q =− k +k +k (5)
Three conductivity coefficients are also expected to arise in the y and z directions.
Accordingly, the thermal conductivity becomes the following second order tensor
quantity
k k k
𝐤= k k k (6)
k k k
Alternatively, following the summation convention, we may rewrite eqn. 6 as
𝐤 = k 𝐞𝐢 𝐞𝐣 (7)
The heat flux component in the x (i=1 means x coordinate, i=2 means y
coordinate, i=3 means z coordinate according to old convention) direction is then
q = −k (8)
In vector form we may write the same heat flux as 𝐤. ∇T
𝐪𝐱 𝐢 = −𝐤. ∇T = kij 𝐞𝐢 𝐞𝐣 .
∂T
𝐞
∂x 𝐤
k
⟹ 𝐪 𝐱𝐢 = − k +k +k (9)
=k 𝐞𝐢 𝐞𝐣 . 𝐞𝐤
=k 𝐞𝐢 δ
=k 𝐞𝐢

Eqn. 9 represents the general case. Actually we may extract the isotropic
formulation of Fourier’s first law upon substituting k = 0 for i ≠ j and k = k if
i = j. So for isotropic material we have the thermal conductivity tensor as

2|
k 0 0
𝐤 = 0 k 0 = k𝐞𝐢 𝐞𝐢 (10)
0 0 k
So 𝐪𝐱 𝐢 = −𝐤. 𝛁T = −k 𝐞𝐢 = −k𝛁T (same as eqn. 1)
For orthotropic material (example: wood, fibrous materials and numerous
crystalline substances), k = k and k = 0 for i ≠ j. In this case we have
k 0 0
k= 0 k 0 = k 𝐞𝐢 𝐞𝐢 (11)
0 0 k
Therefore,
𝐪𝐱 𝐢 = −𝐤. 𝛁T = −k 𝐞𝐢 𝐞𝐢 . 𝐞𝐤 = −k δ 𝐞𝐢 = −k 𝐞𝐢 (12)

Heat Conduction Equation for Isotropic Materials:


Consider an infinitesimal
volume element of
dimensions x, y, and z
in the Cartesian coordinate
system as shown in fig. 2.
Let the centroid of the
volume element be at point
(x, y, z). According to the
Fourier’s law (or Fourier’s
first law) we have
q = −k , q =
−k , q = −k (13)

Fig. 2
 
So the heat fluxes at the faces located at x − and x + become

q  =q − (14.a)

and q  =q + (14.b)
So the net difference between the outgoing and the incoming heat transfer rate by
conduction in the volume element be

3|
q  −q  yz + q  −q  xz + q  −q  xy =

+ + xyz
(15)
Let Q be the heat generation rate per unit volume (W/m ), then the rate of
generation of heat in that volume element be Q. xyz.
We know the accumulation of thermal energy in the volume would lead to an
increase of temperature. And, the rate of heat accumulation in that volume may be
expressed as ρC xyz , where, C is the specific heat (in J/kg.K) and ρ is the
density of the system (kg/m ).
The general energy balance equation may be expressed as
Rate of energy in – rate of energy Rate of energy
out + rate of energy generation = accumulation
So
∂qx ∂qy ∂qz
− + + xyz + Qxyz = ρC xyz (16)
∂x ∂y ∂z
Combining eqn. 13 (Fourier’s first law) and eqn. 16 we have
k + k + k + Q = ρC
⟹k + + + Q = ρC
⟹ k∇ T + Q = ρC (17)
Introducing the term, ≡ α, i.e. thermal diffusivity (in m /s), we may rewrite
eqn. 17 as
∇ T+ = (18)
It is to be noted that, the Laplacian operator (∇ ) assumes the following forms in
(i) Cylindrical and (ii) Spherical coordinate systems:
(i) Cylindrical: ∇ ≡ r + +
(ii) Spherical: ∇ ≡ r + sin θ + (19.a, b)
(Refer fig. 1)
Actually, the thermal diffusivity is a property of conducting material and basically
signifies the rate at which heat diffuses into the medium during changes in
temperature with time. The larger the value of α, the faster is the propagation of
heat into the medium. A low value of α means that heat is mostly absorbed by the
material and a small amount of heat will be conducted further.

4|
Simple 1-D Steady- state Conduction Problems:
Whenever the geometry of the heat-conducting body is simple and heat transfer is
restricted to one direction only. Moreover, considering steady (i.e. invariant with
time) state, the heat conduction equation can be greatly simplified. Some very
practical heat conduction problems fall into this category, for example, the plane
wall, the hollow cylindrical tube and the hollow sphere.
Plane Wall:
Let us consider a wall of
thickness L and width W
(W ≫ L) as shown in fig. 3.
Considering the height of the
wall to be H, let us assume H
and W≫ L, so that we may
simplify the problem to be a 1-
D heat conduction. We fix our
coordinate system (x)’s origin
on the high temperature
surface, i.e. where, T = T .
Accordingly, the other surface
(T = T ) be at x = L, where L
be the thickness of the wall.
Fig. 3
The general conduction eqn. (eqn. 17) becomes

k + + + Q = ρC
0 0 0 Steady
No heat state
As T = T(x)
generation

⟹k =0
⟹ =0 (20)
Integrating twice, we obtain
T= C x+C (21)
The constants C and C can be evaluated from the following boundary conditions:
(i) At x = 0, T = T and
(ii) At x = L, T = T .
So we obtain, C = T and C = −

5|
Therefore, the temperature profile becomes
T=T − x (22)
Hence T vs. x is a straight line with a negative slope. Let WH = A (cross sectional
area of the wall), so the heat flow through a wall of cross sectional area A can be
obtained as
q = Aq = −kA = (22)
heat flux
It is important to note here the similarity of eqn. 22 to the usual statement of
Ohm’s law. The term is equivalent to the electrical resistance and is
appropriately called the thermal resistance. The thermal circuit is shown in fig. 4.
Now if we have n number of slabs
(of thickness L and thermal
conductivity, k ) leading to a
composite block, as shown in the fig.
5, we have
q = Aq = (23)

We consider, T > T > ⋯ > T >


T > ⋯T .
However, for steady state condition
we should satisfy
Fig. 4
q =q =⋯=q =q (24)
(For the whole wall)

Fig. 6
We can calculate the equivalent
thermal resistance R considering
the connection of individual
Fig. 5 resistance as shown in fig. 6.
6|
It is obviously be the series connection. So
R=∑ R = ∑ (25)
Accordingly, the overall heat transfer rate (q) may be calculated as
q = (26)

In order to calculate the intermediate temperatures, i.e. {T }, i = 1 to n-1, we have


to use the equality of q and, q . So we have
= (27)

For, i = 1, we may calculate T using eqn. 27. Once T is known, upon setting
i = 2 in eqn. 27, we can calculate T and so on.
Hollow Cylinder:
Consider a very long thick-
walled hollow cylinder or tube
(fig. 7) having inside surface
temperature maintained at T (at
r ) and the outer surface
temperature maintained at T (at
r ). Note that, T > T . The heat
flow occurs only in the radial
direction because the tube is very
long (typically, for > 3, the
cylinder may be treated as long
cylinder) and hence axial
conduction effect may be
neglected.
Fig. 7
Furthermore, the inside and the outside temperatures are uniform in the
circumferential direction and therefore, there cannot be any circumferential
variation of temperature in the cylinder wall. This kind of geometry is encountered
in any pipe flow situation. For steady state case (with cylindrical geometry),
considering, T = T(r), we have
r =0 (28)
Integrating twice, we obtain
T = C ln r + C (29)
The appropriate boundary conditions are
(i) At r = r , T = T and (ii) At r = r , T = T .

7|
Applying the boundary conditions to eqn. 29, we obtain
C = , and C = T − ln r (30.a, b)

With eqn. (29) and (30. a, b) we obtain the expression for radial temperature
distribution as
T=T + ln

( )
⟹ = (31)

Fig. 8 represents that the


temperature profile is no
longer linear as in the
plane wall problem. This
is because the area
normal to the heat flux
increases with increase in
the radius. Since, for
steady state, the heat
transfer rate is constant,
must decrease with the
increase of r. The heat
transfer rate may be
calculated as

Fig. 8
q = qA = −k . 2πrL = . (T − T )

Heat flux
⟹q = (32)

From eqn. 32 we can identify the thermal resistance to be


R= ln (33)
It is to be noted that the integrated form of the rate equation in case of plane wall

was found to be q = kA , which is pretty simple. So it may be advantageous to

convert the equation (32) into a mathematical form, same as that of plane wall.

8|
Considering the equivalent area to be A , we have
( ) ( )
q = kA ( )
=

Mathematical form, ⟹ A = 2πL (34)


same as plane wall
Alternatively, we may write, A = 2πLr̅ , where r̅ = = log mean of r and

r . So in the integrated equation of the slab, if we replace x by r − r , and A by


A , [where A = 2πLr̅ and r̅ = log mean (r , r )] we will obtain the heat transfer
rate for a long tube.
We can think of a
composite cylinder to
be a pipe with layers of
various types of
insulation. Let the inner
and outer surfaces are
maintained at temp. T
and, T (T > T ),
shown in fig. 9. The
individual layer
resistance is
R = ln
(35)
For i = 1 to n
Fig. 9
So the overall heat transfer rate may be computed as
q =∑ =
⋯ ⋯

= (36)

In order to calculate the intermediate temperatures {T } i = 1 to n-1, we may use


the equality of heat transfer rate through the individual layer and the overall heat
transfer rate (steady state condition). So
q =q
⟹ = (37)

For i = 1, we may calculate T . Using, T , once again by eqn. 37, we can calculate
T and so on.

9|
Hollow Sphere:
Consider a hollow sphere of inner and outer radii r and r at uniform inner and
outer temperatures T and T respectively (T > T ). If the material of sphere is
homogeneous and the heat transfer is steady, then the temperature distribution in
the shell will be a fraction of r only i.e. T = T(r) and the Fourier’s first law takes
the following form
∇ T= with = 0,
r = 0 for spherical coordinate (38)
with T = T(r)
The boundary conditions are
(i) At r = r , T = T and (ii) At r = r , T = T .
Integrating twice we obtain
T(r) = − + C (39)
Using the boundary conditions, the constants C and C are evaluated as
C = , and C = T +

So we obtain
( )
= (40)

The heat transfer rate through the spherical shell be


( )
q = 4πr q = 4πr −k = 4πkr r ( )
(41)
Therefore the thermal resistance of the spherical shell be, R = −
Similar to cylinder, we may be interested in comparing the form of eqn. 41 with
( )
that obtained for a plane wall, where, q = kA ( )
. Here A is same mean area
to be replaced in the equation of plane wall so as to obtain the expression for heat
( ) ( )
transfer in case of a spherical shell. We have, q = kA ( )
= 4πkr r ( )
⟹ A = 4πr r = 4πr , where, r = r r . So it becomes clear that in the
integrated equation of the slab, if we replace x (thickness) by r − r and A by A
[where A = 4πr and r = r r =geometric mean of r and r ], we will obtain
the heat transfer rate for a hollow spherical shell.
Similar to composite cylinder, we may go for composite sphere assembly (refer
fig. 9). Let the inner and outer surfaces are maintained at temperature, T and T
(T > T ).

We have already obtained the expression of individual resistance as

10 |
R = − (42)
The overall heat transfer rate may be computed as
q =∑ = (43)

In order to calculate the intermediate temperature, {T } i = 1 to n-1, we may use


the equality of heat transfer rate through the individual layer and the overall heat
transfer rate (steady state conduction). So
q =q
⟹ = (44)

For i = 1, we may calculate T . Using, T , once again by eqn. 37, we can calculate
T and so on.

𝐐 𝛛𝐓 𝟏
Boundary Conditions to Solve 𝛁 𝟐 𝐓 + = . , i. e. the General Conduction
𝐤 𝛛𝐭 𝛂
Equation:
The surface boundary conditions usually encountered in the theory of heat
conductions are as follows:

A. Prescribed surface temperature: The surface temperature may be constant or a


function of time or space or both. This is also known as Dirichlet condition and is
the easiest boundary condition to work with. But it must be remembered in
practical that it is often difficult to prescribe the surface temperature.

B. Prescribed heat flux: The heat flux across the boundaries is specified to be a
constant or a function of time or space or both. The mathematical description of
this condition may be given in the light of Kirchhoff’s current law: the algebraic
sum of heat fluxes at the boundary must be equal to zero. The following
conventions are used
 Heat flux to the boundary is positive
 Heat flux from the boundary is negative
The next four cases illustrate the application of Kirchhoff’s current law. In fig. 10,
q is the conductive heat flux and q is the prescribed heat flux at the boundary. n is
the normal to the boundary or surface.

11 |
Fig. 10
Case: a: Referring fig. 10.a, we have q + (−q) = 0
⟹q −q=0
⟹ −k   −q=0 (45)
Case b: Referring fig. 10.b, we have q + q = 0
⟹ −k   +q=0 (46)
Case c: Referring fig. 10.c, we have −q − q = 0
⟹ −k   −q=0 (47)
Case d: Referring fig. 10.d, we have −q + q = 0
⟹ − −k   +q=0 (48)
C: No heat flux across the surface:   = 0 at all points on the surface. Here
denotes the differentiation in the direction of the outward normal to the surface.
The condition B and C are also called Neumann condition.

12 |
D: Interfaces of two media of different thermal conductivities 𝑘 and, 𝑘 :
Let T and T denotes the temperatures in two media. When the two media have a
common boundary, the heat flux across this boundary evaluated from both media,
regardless of the direction of the normal, gives
q −q =0
⟹ −k   − −k   =0
⟹k   =k   (48)
This assumption is valid if the media are solid and in intimate contact, such as a
soldered joint. Examples are composite walls and insulated tubes. The continuity
of heat flux and equality of temperature at the interface of two different media are
also referred to as Compatibility condition.
E: Heat transfer to the surrounding by convection: In order to fix the boundary
condition at solid- fluid interface, we have to calculate the heat transfer rate by
convection.
Film Heat Transfer Coefficient:
Let us consider a solid wall at temperature T (uniform over the entire wall surface)
in contact with a fluid with a bulk temperature, T (T > T ).
Because of temperature
difference, T − T , heat
will be transferred from
wall to the fluid. Let the
temperature profile
(qualitative) at any instant
of time be as shown in fig.
11. The primary interest
here is to determine the rate
of heat transfer. We know
for a flowing fluid over a
solid surface in general,
there exists, (i) a laminar
sublayer (ii) a buffer layer
and (iii) a turbulent core.
Fig. 11
For laminar flow only (i) will present, while (ii), (iii) are missing. It is well known
that in the laminar sublayer the primary mode of heat transfer is conduction.
Accordingly in a general turbulent flow we may infer that the thermal resistance
offered by the sublayer will be maximum, which will also be contributed by (ii)

13 |
and (iii). We know under steady state conduction we have a linear temperature
profile over a slab of finite thickness. Accordingly, we assume a hypothetical film
adjacent to the wall in which laminar flow condition prevails. And the thickness of
the film is such that it offers a thermal resistance same as that of the actual case. So
the temperature profile in the hypothetical film may be contracted by drawing a
tangent to the existing profile at wall and extending the same unless its temperature
is same as T (as shown in fig. 11). This film is known as the “effective film” and
the corresponding thickness (δ) is known as the effective film thickness.
As the prevailing mechanism of heat transfer in that film is conduction, we may
write
q=k . (49)
However, as we are unable to determine δ we combine k and δ into a single
term, h ≡ , known as “effective film heat transfer coefficient” or simply the
“film heat transfer coefficient” (sometimes its referred as simply the “hear transfer
coefficient”).
So
q = h(T − T ) (50)

in SI system, the unit of h be, . The h value depends on space, time, geometry,
orientation of the solid surface, flow conditions and fluid properties. h is either
constant or a function of temperature difference. It is important to note that h being
a function of temperature difference does not challenge the validity of eqn. 50.
Table 1 lists typical values of heat transfer coefficient.

Process h Process h
Free convection Convection with phase
Gas 2- 25 change
Liquid 50- 1000
Boiling 2500-100,000
Forced convection condensation 4000-25,000
Gas 25- 250
liquid 50- 20,000
Table. 1
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

14 |
Referring fig. 12, the boundary
condition for the general
conduction equation with heat
transfer to the surrounding by
convection becomes
q =q
⟹ −k   = h(T − T ) (51)
As h → 0, the boundary condition
tends to condition C and as h → ∞,
it tends to condition A.
The condition E is also known as
Robin’s condition.
Fig. 12
Extended Surface: Fins:
One of the main applications of heat transfer study is to increase the rate of heat
transfer from a heated surface to a cool fluid. A typical example is conventional
heat exchangers in which heat is transferred from one fluid to another through a

Fig. 13
15 |
metal wall. The rate of heat transfer is directly proportional to the surface area of
the wall and temperature difference between the wall and the fluid. In most of the
cases, however, the temperature difference cannot be changed. Therefore, the only
way to increase the rate of heat transfer is, to increase the effective heat transfer
area. The effective heat transfer area on a solid surface can be enhanced by
attaching thin metal strips, called fins or spines (thin cylindrical or tapered rods) to
the surface. Although attaching such extended surfaces effectively increases the
heat transfer area, there is also a price to be paid. That is, these extended surfaces
also act as additional resistances to heat transfer and as a result, there is a
temperature drop in the fins. This means that the average surface temperature of
the fins or spines will not be same as the original surface temperature of the wall.
Some commonly used fins and spines are shown in fig. 13.
In this section, the analysis is limited to one- dimensional extended surface with
the following assumptions:
(i) Heat flow in the extended surface is steady
(ii) Thermal conductivity of fin material is constant
(iii) The thickness of the extended surface is so small compared to its length that
the temperature gradient normal to the surface may be neglected.
(iv) The convective heat transfer coefficient between the fin and the surrounding is
constant.
(v) The base temperature is constant. This is, however, a questionable assumption
but, nevertheless, used for the sake of simplicity.
(vi) The temperature of the surrounding fluid is uniform and constant.
Let us now consider
the heat conduction
in the extended
surface as shown in
fig. 14. Note that the
fin is of variable
cross sectional area.
An energy balance
when applied to the
“system” as

Fig. 14

16 |
shown in fig. 14 yields
q A = q Px + q A + (q A)x (52)
But from Fourier’s law
q = −k (53)
Combining eqn. 52 and 53, we obtain
−kA x = −q Px
⟹ kA =q p (54)
(where p is the perimeter of the system, averaged over the thickness, x)
We know
q = h(T − T ) (55)
Using eqn. 54 and 55, we obtain
kA = hP(T − T ) (56)
Since k is a constant, eqn. 56may be written as
A − (T − T ) = 0 (57)
Let, θ ≡ (T − T ), then eqn. 57 transforms to
A − θ=0 (58)
This is the governing equation for 1D heat transfer from a fin. Since eqn. 58 is of
second order, two boundary conditions are needed in the x direction; one at the
base and the other at the tip of the fin.
Extended Surface with Constant Cross- sections:
For an extended surface with constant cross section, eqn. 58 reduces to
−m θ=0 (59)
Where, m = . The general solution of eqn. 59 can be written as
θ(x) = C e + C e (60)
⟹ θ(x) = C sinh mx + C cosh mx (61)
Where C , C , C , and C are constants of integration, to be determined from the
boundary conditions. Since the base temperature T is constant, the boundary
condition at x = 0 be
BC 1: At x = 0, T = T
Or at x = 0, θ ≡ (T − T ) = θ (let) (62)
The second boundary condition (BC-2) depends on the nature of the problem as
discussed.

17 |
Case A (Infinitely Long Fin):
The extended surface for this case is very long compared to the scale of the
temperature profile. So, in this case, the temperature at the tip is essentially equal
to the temperature of the surrounding fluid. The second boundary condition can
therefore be written as
BC- 2: At x → ∞, T → T (63)
or, At x → ∞, θ → 0
The application of boundary conditions gives, C = 0 and, C = θ , hence the
profile eqn. be
θ=θ e
⟹ =e (64)

The temperature profile is as


shown in fig. 15. The heat
transfer from the fin can now be
calculated by integrating the
local convective heat transfer
over the whole periphery (note
that no heat loss occurs at the fin
tip), so
q = ∫ hP(T − T )dx
= ∫ hPθ(x)dx

Fig. 15
q = hPθ ∫ e dx = = √hPkA. θ (65)
On the other hand, one may infer that the heat transferred from the fin by
convection to the surrounding fluid must be equal to the heat conducted to the fin
at the base. Hence, we may also evaluate the heat transfer from the fin by applying
Fourier’s law at the base:
q = −kA   = −kA   = kAθ m = √hPkAθ (66)
Case B: Fin of Finite Length having Insulated Tip:
For this case, the BC- 2 may be expressed as
BC- 2: At x = L, = = 0 (67)
Applying BCs we obtain C = and, C = 0. Therefore, the temperature
profile becomes

18 |
( )
= (68)
So, heat transfer from the fin is
q = + −kA   = (m sinh m(L − x)x) = 0
= kAθ (tanh mL). m = √hPRAθ tanh mL (69)
Since tanh mL → 1 as, mL → ∞, q approaches that for an infinite fin. This
statement is independent of the boundary condition employed at the tip of the fin,
since the effect of the tip diminishes as, L → ∞.
Case C: Fin of Finite Length with Convective Tip:
In this case Bc-2 becomes
Bc-2: At x = L, −k   = h(T − T ) (70)
Or, at x = L, = − θ (71)
Using the BCs, the temperature profile becomes
( ) ( )
= = (72)
So the heat loss be
q = − +kA   = mkAθ (73)
Evaluation of Fin Performance:
Two yardsticks are used to compare and evaluate extended surfaces in augmenting
the heat transfer from the base area. They are (i) fin efficiency (ii) fin effectiveness.
(i) Fin Efficiency: Fin efficiency (η ) is defined as the ratio of the actual heat
transfer to the heat that would be transformed if the entire fin were at the base
temperature.
Case A:
( )
η = lim → =
⟹η = (74)
Therefore, with the increase of L, η decreases. Actually, if we set mL = 0 (i.e., for
the hypothetical case of L= 0), η becomes 1. Hence, we should not expect to be
able to maximize fin efficiency with respect to the fin length. It is, however,
possible to maximize the efficiency with respect to the quantity of fin material
(mass, volume or cost).
(ii) Fin Effectiveness: Fin effectiveness (ϕ) is defined as the ratio of the actual
heat transfer to the heat that would be transferred from the same base temperature,
T , remaining constant.

19 |
Therefore (for example, for a fin with insulated tip),
∫ ( ) ( )
ϕ= = = ∫ dA (76)

Base area of the fin


Note that
∫ ( ) ( )
η = = ∫ dA (77)
, where, A is the total surface area over which the fin transfers heat to the
surrounding fluid. Comparing eqn. 76 and 77 we obtain
ϕ= η (78)
Unsteady State Heat Conduction:
Unsteady state conduction problems are those where the temperature of the body in
question varies with both space and time (as in a distributed system). Before we
discuss lumped and distributed systems, the concept of Biot number is to be
introduced. We have seen the convective heat transfer to/from the boundaries is
important in the formulation and solution of conduction problems. The Biot
number (Bi) is defined as
Bi = (79)
where h is the convective heat transfer coefficient (fluid property), L is the
characteristic dimension of the solid body and k is the thermal conductivity of the
solid body. Bi can also be written as
Bi = = (80)
When conductive resistance ≪ convective resistance i.e. Bi→ 0, the spatial
temperature distribution inside the solid may be neglected. The aforesaid analysis
is called lumped system analysis. But whenever Bi has a finite value (not close to
zero), we have to consider the spatial variation of temperature. Additionally, when
h → ∞, Bi → ∞ and the boundary temperature approaches ambient temperature.
This and the previous case require a distributed system approach. In short, a
lumped system approach assumes that temperatures at all points in the body are the
same, whereas a distributed system implies that there is a temperature variation
from point to point within the body. Typically, transient conduction in very small
bodies or bodies of high thermal conductivity can be modeled using the lumped
system approach.

20 |
Lumped System transients:
A small object of volume V, surface area, A, density, ρ, specific heat, C , initial
temperature, T , is suddenly exposed to an atmosphere at temperature T .
Considering the lumped system analysis, [T = T(t)], we may write the energy
balance equation as
Rate of thermal energy Rate of heat transfer by convection
loss from the solid object = into the surrounding fluid

So −mC = +hA(T − T )
⟹ + (T − T ) = 0 [as m = Vρ] (81)
Eqn. 81 is the governing equation for all the lumped system analysis. It is a first-
order ordinary differential equation. The IC (initial condition) is as follows:
IC: at t = 0, T = T (82)
The solution of eqn. 81, subject to the initial condition (eqn. 82) is
()
= exp − t (83)
The equation is valid for heating as well as cooling. The applicability of eqn. 83 is
restricted for very small Biot number i.e. typically when Bi< 0.1. So Bi = <
0.1, for eqn. 83 to remain valid. The length scale (L) corresponds to the maximum
spatial temperature difference. While for a plane wall of thickness 2L, the length
scale should be used as L. For very long cylinders or spheres the length scale
should be its radius (R). For complex shapes, however, L may be replaced by the
ratio of volume V to the surface area A.
One Dimensional Transient problems: Distributed System
Consider unsteady state heat
conduction in a plane wall of
large length and breadth but of
finite thickness l. the initial
temperature has been, T ,
throughout the wall. At time
t = 0, both the surfaces are
brought to a temperature T and
maintained at this value for all
subsequent time. The relevant
physical properties are assumed
to remain constant. A section of
the wall is shown in fig. 16.
Fig. 16

21 |
The x axis is taken normal to the wall with origin on a surface of it. In order to
formulate the problem mathematically, we consider a small rectangular volume
element, having area same as the slab. Because the length and breadth of the plate
is large, and the surface temperature is uniform, heat conduction in the wall will be
1D.
So the primary equation to be solved in order to obtain the temperature profile is
=α (84)
Where α is the thermal diffusivity of the, solid. The IC and BCs are
IC: t = 0, 0 < 𝑥 < 𝑙, 𝑇 = T
BC 1: t ≥ 0, x = 0, T = T
BC 2: t ≥ 0, x = l, T = T (85)
Let us define the following dimensionless variables
T= , x = , θ = , in terms of the dimensionless variables, eqn. 84 and 85
becomes
= (86)
and
IC: At θ = 0, 0 < x < 𝑙, T = 1
BC 1: At θ ≥ 0, x = 0, T = 0
BC 2: At θ ≥ 0, x = l, T = 0 (87)
According to the method of separation of variables, we assume a solution of the
dimensionless equation in the following form
T = X(x)τ(θ)
where, X is a function of x only and τ be a function of θ only. So we get
(Xτ) = (Xτ)
⟹X =τ
⟹ = (88)
In eqn. 88, the LHS is a function of θ only, whereas the RHS is a function of x
only, but they are equal. The only possibility is each side equals to the same
constant. Let it be −λ (the constant −λ is deliberately chosen as a negative
quantity in order to obtain non-trivial solution).
So
= = −λ (89)
we may write
τ = Ae θ (A is a constant of integration) (90)

22 |
and
+λ X=0
⟹ X = A cos λx + B sin λx (91)
Hence
T = X(x)τ(θ) = (A′ cos λx + B′ sin λx)Ae
⟹ T = (A cos λx + B sin λx)Ae ,
[where A = AA and B = BB′]
From BC1 (eqn. 87) we know that at x = 0, T = 0, accordingly we may infer,
A = 0 [as cos 0 = 0 ≠ 1], so the solution becomes
T = B sin λx e (92)
From BC2 (eqn. 87) we have at x = 1, T = 0 (for, θ ≥ 0). So we may infer
sin λ = 0
⟹ λ = nπ, where n = 1, 2, 3……
(Eigen values)
Now corresponding to each eigenvalue, we have a solution for T. The general
solution for T will be a linear combination of all such solutions. Thus
T=∑ B sin(nπx)e (93)
where B (n = 1, 2, 3 … . ) are constants, which are yet to be determined. We put
the IC in eqn. 93, which yields
I=∑ B sin nπx (94)
The constants {B } can be determined by using orthogonality property of
sinusoidal function.
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Orthogonal Functions:
A sequence of functions, {ϕ (x)}, i = 1, 2, 3 … … defined in an interval [a, b] is
said to form an orthogonal set if it has the following properties:
∫ ϕ (x)ϕ (x)dx = 0 for m ≠ n
≠ 0 for m = n (95)
Example: If we define {ϕ (x)} = sin , 0 < 𝑥 < 𝑙 and k = 1, 2, 3 … … it can be
verified by direct integration that
∫ sin . sin dx = 0 for m ≠ n
≠ 0 for m = n
Therefore, this set of functions forms an orthogonal set in the given interval.
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––

23 |
Multiplying both sides by (sin mπx) and integrating the equation over, x ⇒ [0,1],
we obtain
∫ sin(mπx)dx = ∫ ∑ B sin(nπx) sin(mπx)dx
⟹ − cos = [1 − (−1) ] = B ∫ sin (mπx)dx
⟹ [1 − (−1) ] =
⟹B = [1 − (−1) ] (96)
If m is an even number, B vanishes. However, when m is an odd number, we
have
B = (for odd m) (97)
Let us put, m = 2p + 1 (p = 0, 1, 2, 3 … . . ), when
B = ( )
(98)
Using the value of B , we have the solution as
( )
T= ∑ ( )
exp[−(2p + 1) π θ] (99)
or in terms of the original variable, we have
( )
= ∑ ( )
exp −(2p + 1) (100)
Eqn. 100 gives the solution for the dimensionless temperature distribution in the
plane wall under unsteady state condition. The dimensionless time, 𝜃 = is called
the “Fourier number”.
Example: One Dimensional Distributed System Approach: Semi- infinite solid
Although all bodies have finite dimensions, a number of cases can be idealized as
semi-infinite solids, in which there will be regions which still remain unaffected by
a change of temperature on one of their surfaces. In other words same parts of the
body may still remain at the initial temperature even after a long time.
Consider a semi-infinite solid rod
that is initially at temperature, T .
Assume that the surface
temperature of the solid at one end
of the rod (let, x = 0) is suddenly
changed to T . We wish to find the
unsteady temperature in the solid.
The schematic of the problem is
shown in fig. 17. Basically we have
Fig. 17 to solve the following equation.

24 |
=α (101)
Subjected to the following IC and BCs
IC: at t = 0 for, x > 0, T = T
BC1: at x = 0 for, t ≥ 0, T = T
BC2: at x → ∞ for, t ≥ 0, T = T (102)
It is interesting to note the features BC2. Basically, it indicates that it will take an
infinity long time for the heat to penetrate to the other end of the rod.
We introduce a dimensionless temperature, θ ≡ . So in terms of θ, the eqn.
101 and the IC and BCs become
= (103)
Subjected to
IC: at t = 0, x > 0, θ = 0
BC1: at x = 0, t ≥ 0, θ = 1
BC2: at x → ∞, t ≥ 0, θ = 0 (104)
In order to solve the PDE we use the similarity solution technique as the
temperature profiles at two different time-instants are similar to each other.
Accordingly, we introduce a combined dimensionless variable (η) as
η≡ (105)

Now
= . = . . − = − η (106)

and,
= . = . ; = . (107)

Using eqn. 106 and 107; eqn. 103 becomes
. = .η .
⟹ + 2η =0 (108)
The corresponding boundary conditions become
IC: at η → ∞, θ = 0; BC1: at η = 0, θ = 1; BC2: at η → ∞, θ = 0. It may be noted
that IC and BC2 are same, so finally it reduces into two equations, consistent for a
second order ODE (eqn. 108).
Let, = q, so in terms of q, eqn. 108 becomes
+ 2ηq = 0
⟹q=C e (109)
where, C is the constant of integration.

25 |
Again, q = . So, =C e . Integrating once again we obtain
θ = C ∫ e dy + C (110)
As at η = 0, θ = 1, we obtain C = 1, so
θ = 1 + C ∫ e dy (111)
Using the second BC (i.e., at η → ∞, θ = 0) we have
0 = 1 + C ∫ e dy (112)
We know, ∫ e dy = (113.a)
Using eqn. 112 and 111 we obtain
θ = 1 − ∫ e dy (113.b)

So in terms of the original variables; the solution becomes
( ,)
=1− ∫√ e dy (114)

By definition, error function of x, [erf(x)] is
erf(x) = ∫ e dy (115)

So the solution be
( ,)
= 1 − erf (116)

The rate of heat transfer at x = 0, may be calculated as
q(0) (heat lux) = −k   = k(T − T ) ∫√ e dy

( ) ( )
⟹ q(0) =   e = (117)
√ √
We define another important quantity, penetration depth at a given time as the
distance up to which the temperature gradient exists and beyond which the body
remains at the initial temperature. Mathematically, this is the distance at which
T − T is 1% of T − T . It is denoted by symbol, δ.
So at x = δ,
= 0.01 = erf (118)

According to the variation of error function we may conclude that
δ = 3.6√αt (119)

Critical Thickness of Insulation:


A special application of thermal resistance formulae developed so far may be
considered in determining the thickness of an annular insulation that should be
applied to the outer surface of a small diameter circular tube wall of a known wall

26 |
temperature. A practical application is the problem of insulating electrical wires
where the objective would be the provision of adequate electrical insulation, at the
same time providing maximum wire cooling.
The reason why this problem has technological importance is as follows:
In the case of small-diameter pipe, the application of insulating material to the
outer surface, may in special instances, increases the heat loss from the surface.
Let us consider the case of a
steam carrying pipe of a fixed
outer radius, r , r denotes the
radius of insulation, as shown in
fig. 18. So the thickness of
insulation be (r − r ), k ≫
k and h ≫h . The
implication of the foregoing first
assumption is that the relative
thermal resistance of the pipe is
so small that there will be
virtually no temperature drop in
the wall of the pipe.
Fig. 18
The assumption of a very high inner heat transfer coefficient, h , implies
negligible inner convective resistance. Therefore, out of four resistances to the heat
flow path from T to T , the first two may be neglected. So the heat flow per unit
length of the pipe is
= (120)

The rate of flow will be maximum when denominator becomes minimum. The
minimum value of the denominator can be calculated by taking the derivative of
the denominator with respect to r, and setting the result equal to zero. Thus
r (Critical insulation = (121)
thickness for pipe)

The heat transfer coefficient, h is considered constant in this calculation. The


conclusion is that tubes whose outer radius (r ) is smaller then r , can have their
heat losses increased by adding insulation upto r (fig. 19.a). Further increase in
the thickness of insulation will cause the heat loss to decrease from this peak value,
but until a certain amount of insulation, denoted by r ∗ is added, the heat loss is still
greater than for the base pipe. Thus, an insulation thickness in excess of (r ∗ − r )
must be added to reduce the heat loss below the un-insulated rate.

27 |
Fig. 19
Fig. 19.b illustrates the case of large pipes, in which r > r and any insulation
will decrease the heat loss.

Problems to be Solved:
1. Draw the analogous electric circuit
for heat conduction through the
composite wall shown in the figure.
State clearly all the assumptions.
2. A large slab concrete, 1m thick, has
both surfaces maintained at 250C.
During the curing process a uniform
heat generation of 60 occurs
throughout the slab.
If the thermal conductivity of concrete is 1.1 W/mK, find the steady state
temperature at the centre of the slab.
3. Develop the expression for the steady state temperature distribution in a sphere
(solid) of radius b, in which heat is generated at a rate of: Q = Q 1 − .
where Q is constant and boundary surface (at r = b) is maintained at T .
4. Consider a shielding wall for a nuclear reactor. The wall receives γ ray flux such
that heat is generated within the wall according to the relation, Q = Q e , where

28 |
Q is the heat generated per unit volume at the inner face of the wall exposed to γ
ray flux and a is a constant.
Derive an expression for the temperature distribution in a wall of thickness L,
where the inside and outside temperatures are maintained at T and T ,
respectively. Also obtain an expression for the maximum temperature in the wall.
5. A thermocouple well wall of
thickness t is in contact with
the gas stream on one side
only and the tube thickness is
small compared to diameter.
The arrangement is shown in
the figure side by. Estimate
the temperature of the gas
stream, if T (temperature
indicated by the thermo-
couple) is 2600C, T (wall
temperature) is 1770C,
h= , t = 2mm and L =
6cm

6. Consider a solid body of volume V and surface area A surrounded by a coolant


of temperature, T . For time t ≥ 0, energy is generated in the solid at an
exponential decay rate per unit volume to Q = Q e , where Q and β are known
constants. Let the heat transfer coefficient be h between the solid body and the
coolant, which remains constant. Obtain an expression for the temperature of the
solid body as a function of time for, t > 0. What will be the maximum solid
temperature and when it will be reached?
7. 1-D steady state heat transfer occurs from a flat vertical wall of length 0.1m into
the adjacent liquid. The heat flux into this fluid is 21 . The wall thermal
conductivity is 1.73 W/mK. If heat transfer coefficient is 30 and Nusselt
number based on wall length is 20, then the magnitude of the temperature gradient
at the wall on the fluid side (in K/m) is
(a) 0.7 (b) 12.14 (c) 120 (d) 140 (2006)
8. A composite wall of an oven consists of three materials A, B and C. Under
steady state operating conditions, the outer surface temperature T is 200C, the
inner surface temperature, T is 6000C and the oven air temperature (T ) is 8000C.
For the following data of thermal conductivities
k = 20W/mK, k = 50W/mK
29 |
Thickness: L = 0.3m
L = 0.15m and L =
0.15m, inner wall heat
transfer coefficient,
h = 25 , the thermal
conductivity of material B,
k in W/mK is calculated
as
(a) 35 (b) 1.53
(c) 0.66 (d) 0.03
(2007)

9. Two plates of equal thickness (t)


and cross sectional area, are joined
together to form a composite as
shown in the figure. If the thermal
conductivities of the plates are k
and 2k then, the effective thermal
conductivity of the composite is
(a) (b)
(c) (d) (2008)

10. A metallic ball (ρ = 2700 kgm and C = 0.9kg/kg℃) of diameter 7.5 cm is


allowed to cool in air at 25℃. When the temperature of the ball is 125℃, it is
found to cool at a rate of 4℃/min. If the thermal gradients inside the ball are
neglected, the heat transfer coefficient (in ) is

(a) 2.034 (b) 20.34 (c) 81.36 (d) 203.4 (2008)

11. For the composite wall shown in below (case 1), the steady state interface
temperature is 180℃. If the thickness of layer P is doubled (case 2), then the rate of
heat transfer (assuming 1-D conduction) is reduced by
(a) 20% (b) 40% (c) 50% (d) 70% (2009)

30 |
Common statement for Q. 12 and 13
A slab of thickness L with one side
(x = 0) insulated and the other side
(x = L) maintained at a constant
temperature, T , as shown in figure. A
uniformly distributed internal heat
source produces heat in the slab at the
rate of, S . Assume steady state 1D
heat transfer.
12. The maximum temperature in the
slab occurs at
(a) x = 0 (b) x =
(c) x = (d) x = L
13. The heat flux at x = L is
(a) 0 (b) (c) (d) SL (2009)
14.

31 |
The fig. above shows steady state temperature profiles for 1D heat conduction
within a solid slab for the following cases:
P: Uniform heat generation with left surface insulated
Q: Uniform heat generation with right surface insulated
R: Uniform heat consumption with left surface insulated
S: Uniform heat consumption with right surface insulated.
Match the profiles with appropriate cases
(a) P-I, Q-III, R-II, S-IV (b) P-II, Q-III, R-I, S-IV
(c) P-I, Q-IV, R-II, S-III (d) P-II, Q-IV, R-I, S-III

Forced Convection:
So far we have discussed problems of heat transfer in conduction. We have
considered convection only in relation to the boundary condition imposed on a
conduction problem. The main purpose of studying convective heat transfer is to
predict the value of the convective heat transfer coefficient, h. The subject of
convection heat transfer requires an energy balance along with an analysis of fluid
dynamics of the problem concerned.
Convection is not a separate mode of heat transfer. However, it is related to a fluid
system in motion. We must allow for the motion of the fluid system in writing an
energy balance, but there is no new basic mechanism of heat transfer involved.
Nevertheless, if the fluid is in motion, heat is transported both by simple
conduction and by the movement of the fluid itself. This complex transport process
is referred as convection. Thus the essential feature of convective heat transfer is
the transport of energy to or from a surface by both molecular conduction process
and gross fluid motion. If a fluid motion involved in the
process is induced by some external
means (pump, blower, wind, vehicle
motion etc.) the process is generally
called forced convection. If the fluid
motion arises due to external force
fields, such as gravity, density
gradient induced by temperature
gradient, the process is usually
referred as free or natural
convection. Consider a fluid having
velocity u and temperature T
flowing over a surface of arbitrary
shape and of area A (fig. 1). The
surface is maintained at temperature,
Fig. 1 T (T > T ).
32 |
The local heat flux q may be expressed as
q = h(T − T ) (1)
where h is the local heat transfer coefficient. The total heat transfer rate (q′) may
be obtained by integrating the local heat flux over the entire surface:
q′ = ∫ qdA = (T − T ) ∫ hdA (2)

Defining h as the average


heat transfer coefficient,
which is expressed as
h = ∫ hdA (3)
eqn. 2 may be written as
q′ = hA(T − T ) (4)
Note that for the special
case of flow over a flat plate
(fig. 2) we have
h = ∫ hdx (5)

Fig. 2
Thermal Boundary Layer:
A thermal boundary layer develops when the free stream and surface temperature
differ. Due to the “no-slip” equivalent thermal condition at the plate surface, the
stationary fluid particles have the same temperature as that of the plate surface
after thermal equilibrium is reached.
The fluid particles in
contact with the plate
exchange energy with
those in the adjoining
layer, and temperature
gradients develop in the
fluid. The region of the
fluid in which
temperature gradients
exist is the thermal
boundary layer (fig. 3),
and its thickness δ is
typically defined as the
value of y for which the
Fig. 3 ratio = 0.99.

33 |
As the distance from the leading edge increases, the effects of heat transfer
penetrate further into the free stream and thermal boundary layer grows. At the
plate surface, since there is no fluid motion and heat transfer can only occur by
conduction, we can apply Fourier’s law to calculate the local surface heat flux as
q = −k   (6)
 
As q = h(T − T ), therefore h= ( )
(7)
Eqn. 7 is the defining relation for the heat transfer coefficient. k is the thermal
conductivity of the fluid. From eqn. 7 it is clear that the conditions in the thermal
boundary layer, which strongly influence the plate surface temperature gradient,
  , will determine the rate of heat transfer across the boundary layer. Note
that, since (T − T ) is constant, independent of x, while δ increases with
increasing x, temperature gradient in the boundary layer must decreases with
increasing x. Accordingly, the magnitude of   decreases with increasing x
(or δ ) and it follows that q and h decrease with increasing x. Since the thermal
boundary layer thickness is zero at the leading edge, the heat transfer coefficient
there, is infinity.
Nusselt number:
Frequently, the heat transfer coefficient is made non-dimensional by using a
characteristic length L and defining Nusselt number:
Nu ≡ (8)
Thus, the local Nusselt number for flat plate is Nu = and the average Nusselt

number for a flat plate of length L is Nu = . Nusselt number is indicative of the


temperature gradient at the wall in the normal direction. We may define the
physical significance of Nusselt number as
 
.
Nu = =  (where T = T − T )
.

So

Nu = ~ (9)

34 |
Prandtl number:
Prandlt number is the one of the most important dimensionless groups in heat
transfer and is defined as
Pr ≡ (10)
where μ, C and k are the viscosity, specific heat and thermal conductivity of the
fluid respectively.
We may define the physical significance of Pr as
( )
Pr =

= = = (11)

Prandtl number also signifies the ratio of the momentum boundary layer thickness
(δ) to the thermal boundary layer thickness, (δ ). Thus
Pr~ (12)
If Pr > 1, it must follow that the momentum boundary layer develops faster than
the thermal boundary layer. If Pr < 1, the opposite holds i.e. the thermal boundary
layer develops more rapidly than the momentum boundary layer.
The Prandtl number spectrum of various fluids is shown in fig. 4.

Fig. 4
Typical Prandtl number are : 0.7 for air, 7 for water at room temperature, oils have
very high Pr and liquid metals have very low Pr. Pr for any particular fluid
generally varies somewhat with temperature as, μ, C and k, all are functions of
system temperature.

35 |
Energy Equation in Thermal Boundary Layer in Laminar Flow over a Flat
Plate:
We are now in a position to derive the energy equation in the thermal boundary
layer in laminar flow over a flat plate using a differential C𝒱 as shown in fig. 5.
The relevant assumptions
are as follows.
(i) The flow is steady,
incompressible, 2D and
laminar.
(ii) The fluid has constant,
μ, ρ, C .
(iii) Negligible heat
conduction in the direction
of flow (this is because
thermal BL is very thin).

Fig. 5
(iv) Viscous dissipation (energy generation due to viscous work) in y direction is
negligible.
(v) and =0
(vi) The KE of the fluid along y direction is negligible, (v ≪ u).
(vii) Pr > 0.5. This implies the present analysis is applicable to most gases and
liquids and δ < 𝛿.

Fig. 6

36 |
Fig. 6 shows the expanded view of the control volume and the energy-in and out
terms. Note that thermal and KE are convected with bulk fluid motion across the
control surface. Moreover, energy is also transferred across the control surface by
conduction as well as by surface forces (pressure, viscous etc.). Only for
supersonic flows or the high- speed motion of lubricating oils, viscous dissipation
may not be neglected.
For the C𝒱 shown in fig. 6, we may write
energy in = energy out
Note that, i is the enthalpy (or thermal energy) per unit mass, C T , is the KE
per unit mass in x direction, q is the conductive heat flux in y direction, and
−uτ is the viscous shear work in the x direction. The negative sign on uτ is
given because viscous work is done on the control surface.
Writing all energy-in terms on the LHS and all energy-out terms on the RHS and
cancelling terms, we have
0= ρv i + + ρu i + + + −uτ (12)

Net energy associated Same as Net heat


with fluid motion transfer by Net rate of work
the first in added to the C𝒱 in
transferred into the C𝒱 in x direction conduction
the y direction, i.e. the net in y y direction due to
flux of enthalpy and KE direction viscous dissipation

Substituting q = −k and τ =μ , we obtain


0 = ρv +u + ρu +u +ρ i+ + −k −
u μ − μ (13)
Since the fluid is incompressible, from the equation of continuity we have
+ =0 (14)
Combining eqn. 13 and 14 we have
0= ρ u +u + ρuv + ρu −k − μu −μ

⟹0=ρ u +v + ρu u +v −ν −k −μ (15)
But from x momentum balance equation of the boundary layer we have
u +v =ν (16.a)

37 |
Combining eqn. 15 and (16.a) we have
0=ρ u +v −k −μ (16.b)
Now as i = C T and, ∂i = C ∂T, we have
0 = ρ uC + vC −k −μ

⟹ ρC u +v =k +μ

⟹u +v =α + (17)

Convection term Conduction Dissipation


term term
We see that the change in KE is balanced exactly by a portion of the work done by
the viscous force. The rest of the viscous work is dissipation.
The term, u + v , represents the net transport of energy by convection into the
C𝒱. The term α is the net heat conducted out of the C𝒱 in y direction. Finally,

the term is the net viscous work done on the element. This is called the
viscous dissipation term.
The relative importance of dissipation term w.r.t the conduction term can be
checked as follows.
We know u is in the order of u , y is in the order of δ (or δ ) and is in the order
( )
of . Therefore
α ~α and ~
Now

~ = Pr ( )
= Pr. Ec (18)

where Ec= Eckert number= ( )


. Therefore, if Pr. Ec ≪ 1, viscous
dissipation can be neglected. So the eqn. 17 takes the following form
u +v =α (19)
To illustrate this concept, consider flow of air at u =5m/s. T =200C, T =600C,
P=1 atm. For this conditions, C =1005J/kg.0C, Pr=0.7. Therefore,
Pr. Ec = Pr. ( )
= 0.7 × ( )
= 0.000435

38 |
Thus for this case we may neglect viscous dissipation.
Governing Equation and Boundary Conditions:
From the foregoing discussion we see the governing differential equations for the
thermal boundary layer over a flat plate at zero incidence, are as follows:
Continuity: + =0 (20)
X momentum: u +v =ν (21)
Energy: u +v =α (22)
For an isothermal flat plate, the boundary conditions are
BC 1: At y = 0, u = 0, v = 0, T = T (23)
BC 2: At y → ∞, u = u , T = T , v = 0 (24)
A close look at eqns. 22 and 21 reveals that they will have exactly the same form
when, α = ν. Thus, we should expect that the relative magnitudes of thermal
diffusivity and kinematic viscosity would have an important influence on
convective heat transfer since these magnitudes relate the velocity distribution to
the temperature distribution. Recall that, Pr = . This explains why Prandtl
number is so important in convective heat transfer studies.
Approximate Analysis: von Karman’s Integral Method:
Instead of the direct solution of eqn. 20 to 22, with BC’s [eqn. 23 and 24], which is
based on similarity analysis [Pohlhausen (1921)], we may also analyze the
properties of a thermal boundary layer by energy integral equation, as proposed by
von Karman. The momentum integral method of analysis of boundary layer flow
can be extended to analyze the phenomenon of convective heat transfer in a
laminar boundary layer. Let us integrate eqn. 23 from the surface of the plate
(y = 0) to the edge of the thermal boundary layer, (y = δ ). Thus we get
∫ u dy + ∫ v dy = α ∫ dy (25)
As, ∫ uTdy = ∫ (uT)dy = ∫ T dy + ∫ u dy
The first term on the LHS of eqn. 25 becomes
∫ u dy = ∫ uTdy − ∫ T dy (26)
Similarly, the second term of eqn. 25 (on LHS) yields
∫ v dy =  vT) − ∫ T dy =  vT) + ∫ Tdy
= −T ∫ dy + ∫ Tdy
=− ∫ uT dy + ∫ Tdy (27)

39 |
In deriving eqn. 27, it is to be noted that at y = 0, v = 0 and =− (from
continuity eqn.), we have
 vT) = T v(δ ) = T ∫ dy = − T ∫ dy

As at y = δt, T = T∞
Combining eqn. 25, 26 and 27 we obtain
∫ uTdy − ∫ uT dy = α   = −α  

⟹ ∫ u(T − T )dy = −α   (28)


Eqn. 28 is known as von Karman energy integral equation for heat transfer in
laminar boundary layer flow.
In order to solve eqn. 28 for boundary layer temperature distribution, we assume
that the dimensionless temperature profile can be expressed as a polynomial in .
Let us consider a third degree polynomial.
=b +b +b +b (29)
The following boundary conditions apply
BC 1: At y = 0, T = T , =0 (30)
BC 2: At y = δ ,T = T , =0 (31)
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
We have, u +v =α
At y = 0 as u = v = 0, so the BC at y = 0 becomes, =0
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Using the BCs we obtain
b = 0, b = , b = 0, b = − (32)
The corresponding expression for the temperature profile becomes
= − (33)
In which the thermal boundary layer thickness δ is yet to be determined.
Similarly, as explained in the note of “boundary layer theory” (in the subject Fluid
Mechanics) we can determine the velocity profile eqn. as
= − (34.a)

40 |
Considering =∑ a and the boundary conditions: at y = 0, u = 0, =
0, and at y = δ, u = u , = 0, = 0.
Using eqn. 33 and 34 in eqn. 28 we obtain
∫ u − (T − T ) −1 + − dy

= −α(T − T )   −

= −α(T − T ) (34.b)
If we substitute χ = and ζ = , the above eqn. reduces to
u ∫ δ χζ − χ ζ 1− ζ+ ζ dζ = (35)
Integrating we have
u δ χ − χ = (36)
An analytical solution of the above equation can be obtained if, χ = < 1. This
happens when, Pr < 1.
Under this condition the term χ in eqn. 36 can be neglected when compared
with χ . With this simplification, the equation reduces to
(δχ ) = (37)
.
If we substitute, δ = , in the above equation, we get

+ χ = . (38)
Putting, η = χ , we have
+ η= . (39)
Integrating we have
η = . + Cx (40)
where C is the constant of integration. From physical reasoning η is finite at x = 0.
This is possible if we put, χ = 0. Therefore
η=χ = .

⟹ = . ≈ Pr (41)
. .
⟹δ = . . = (42)

41 |
The temperature distribution in the boundary layer can be obtained from eqn. 33
after substituting, δ , given by eqn. 42. The local temperature gradient at the plate
surface can be obtained by differentiating eqn. 33. Thus
  = (T − T ) (43)
The local heat transfer coefficient can now be determined as
h = −(   = −( (T − T ) (44)
) )

Substituting δ from eqn. 42, we get


h = Re . Pr = 0.323 . Re Pr (45)
.
The local Nusselt number is obtained as
Nu = = 0.323Re Pr (46)
The above relation for Nusselt number is nearly identical to that obtained by
direction analytical solution, which is
Nu = 0.332Re Pr (47)
This shows that the integral method is a very good technique for analysis of
boundary layer transport.

Importance of Dimensional Analysis:


Exact or approximate mathematical solutions to convective heat transfer problems
in simple situations are often possible especially because of computational
techniques. But many of the practical situations encountered in industrial practice
still defy our analytical and numerical tools and we have to depend upon
“empirical correlations”. A large number of correlations have been developed over
the years to relate the heat transfer coefficient (which itself is an empirical quantity
for all practical purposes) with the relevant properties, parameters and variables of
systems.
Dimensionless Groups in Heat Transfer (Forced Convection):
Before we discuss the important heat transfer correlations available in the
literature, it is pertinent to make a list and also to mention the physical significance
of dimensionless groups which are frequently used in forced convection heat
transfer (table 1).
Name Expression Physical Significance
Nusselt number hl wall temp. gradient
Nu =
k temp. gradient across the luid
Reynolds number lVρ inertial force
Re =
μ viscous force
Prandtl number C μ ν momentum diffusivity
Pr = =
k α thermal diffusivity

42 |
Name Expression Physical Significance

Stanton number h Nu rate of wall heat transfer by convection


St = =
VρC Re. Pr rate of heat transfer by bulk low
hT. l
=
ṁC T
hTl h
= ~
Vl ρC T VρC

ρVlC rate of heat trasfer by bulk low


Peclet number Pe = = Re. Pr
k rate of heat transfer by conduction
ρVC Tl ρVlC
≈ T =
k .l k
l

Graetz number d
Gz = Pe. Similar to Peclet number
L
d (used in the analysis of heat transfer
= Re. Pr. in laminar flow)
L

Table: 1
Here l denotes the characteristic length. A characteristic length can be defined
based on the geometry of the system. For pipe flow we use diameter (d) in place of
l. for Graetz number, L is the pipe length over which the heat transfer occurs.
Turbulent Flow through a Circular Pipe:
For fully developed turbulent flow through a pipe, the following correlation
suggested by Dittus and Boelter (1930) is used widely
Nu = 0.023Re . Pr (48)
where, n = 0.4 for heating [T (wall temp.)> T(bulk temp.)] and n = 0.3 for
cooling (T < T). The conditions for applicability of this correlation are:
(a) 0.7 < 𝑃𝑟 < 160 (b) < 0.1 (c) Re > 10
The above equation is applicable for moderate values of temperature difference
between the wall and the bulk (= T − T). The fluid properties are evaluated at
the arithmetic mean of the bulk temperature (the average of the inlet and the outlet
temperatures of the fluid). The maximum error in the predicted h does not exceed
25%.
With temperature difference between the wall and the bulk substantial, its effect on
the fluid properties, particularly on viscosity, needs to be taken into account. For
such cases, the Sider and Tate equation (1936) is recommended.

43 |
.
. .
Nu = 0.027Re Pr (49)
.
Here is the viscosity correction factor (denoted by ϕ ) that has to be used
when the viscosity at the wall temperature (μ ) is substantially different from that
at bulk temperature (μ). The conditions for applicability are: (a) 0.7 < 𝑃𝑟 <
16,700 (b) Re > 10 (c) < 0.1. The bulk fluid properties must be evaluated at
the mean bulk (arithmetic mean) temperature.
Flow through a Non-Circular Duct:
Ducts of non-circular cross sections, rectangular or square, are often used in the
industrial and other applications. Flow through the annulus of a double pipe heat
exchanger is a common example. The foregoing equations for estimation of the Nu
are generally applicable for heat transfer calculations for flow through non-circular
ducts, but the equivalent diameter (d ) of the duct is to be used in the calculation.
The equivalent diameter (d ) is four times the hydraulic radius, (r ). The
hydraulic radius is defined as
r =
⟹ d = 4r (50)
For example, in case of flow through a rectangular duct of sides l and, l , the
equivalent diameter is
d = ( )
(51)
It should be noted that the
calculation of the wetted perimeter
of a duct in case of heat transfer
calculation may be different from
that used in the case of
determination of pressure drop. To
illustrate this point, we consider
the flow through the annulus
between two concentric tubes in a
double pipe heat exchanger. Let
the outside diameter of the inner
pipe be d and the inside diameter
of the outer pipe be d , as shown
in fig. 7. If our objective is to
calculate the pressure drop for
flow through the annulus, the
Fig. 7 wetted perimeter is π(d + d )
44 |
because both the walls forming the annulus contribute to pressure drop. However,
if we consider heat transfer from a hot fluid flowing through the inner pipe to a
cold fluid flowing through the annulus, wetting of the outer wall of the inner pipe
only becomes relevant to the heat transfer coefficient in the annulus. So, here the
wetted perimeter is simply, πd . The equivalent diameters of the annulus in the
respective cases are
(i) For pressure drop calculation:
d =4 ( )
=d −d (52.a)
(ii) For heat transfer calculation:
d , =4 = d −d (52.b)

Flow over a Flat Plate (external flow):


It has been already established in the foregoing discussion that for heat transfer in
laminar boundary layer flow, the following correlation for the local Nusselt
number can be obtained from the solution of the boundary layer equations. The
relation is
Nu = 0.332Re Pr (53)
(0.5 < 𝑃𝑟 < 50), where Nu = and, Re = , are the local Nusselt number
and Reynolds number respectively. Here x is the distance from the leading edge of
the plate. An average value of heat transfer coefficient over a distance L may be
obtained as
h = ∫ h dx
⟹ = Nu = 0.664Re Pr (54)
For heat transfer in turbulent boundary layer flow, a simple correlation in common
use is
Nu = 0.0296Re Pr (55)
for 0.6 < 𝑃𝑟 < 60.

Flow Past a Sphere: (external Flow)


Whitaker (1972) suggested the following correlation for flow over a sphere:
.
.
Nu = 2 + 0.4 Re + 0.06Re Pr (56)
where Re = , Pr = , μ is the viscosity at wall temperature. The
correlation is applicable for both gas and liquids, and the error in the prediction

45 |
remains within ±30%. The number ‘2’ arises out of the contribution of conduction
in the medium i.e. when, Re = 0.
Laminar Flow through a Circular Pipe:
Laminar flow (Re < 2100) may occur in industrial equipment for heating or
cooling of a considerably viscous material or of a liquid or solution sensitive to
shear stress (for example, the solution of protein, may deteriorate if it is subjected
to high shear stress). Because it is possible to analytically solve the problem of heat
transfer in laminar flow, a large volume of theoretical research work has been
published in this area. However, we have handy correlations for heat transfer
coefficient, similar to the case of turbulent flow. A simple correlation was
suggested by Sieder and Tate (1936).
.
Nu = 1.86 Re. Pr. (57)
.
⟹ Nu = 1.86Gz (58)
The above correlation, eqn. 57 and 58, is applicable when
(i) 0.48 < 𝑃𝑟 < 16,700
(ii) 0.0044 < < 9.75
(iii) Gz > 10
It is to be noted that the correlation 57, 58 includes the pipe length L, including
“thermal entrance length”.
Thermal Entry Length:
If the fluid enters the tube at a uniform temperature that differs from the surface

Fig. 8

46 |
temperature, convective heat transfer occurs and a thermal boundary layer
develops, as shown in fig. 8.
In this region, known as thermal entrance/entry length, 0 < 𝑧 < z , , the shape of
the temperature profile develops i.e. T-profile changes along z. In the thermally
fully developed region z > z , the shape of the temperature profile is preserved.
For laminar flow
,
= 0.05RePr (59)
From the relation of δ and δ (hydrodynamic and thermal boundary layer
thickness) in terms of Pr, we may write
,
= (60)
,
Hydrodynamic
entrance length

So if Pr > 1, the hydrodynamic layer (boundary) develops more rapidly than the
thermal boundary layer, while it is inverse for, Pr < 1.
Moreover, for very high Pr fluids, such as oils, (Pr > 10 ), z , is very much
smaller than z , and it is reasonable to assume a fully developed profile throughout
z , . On the other hand, for very low Pr fluid (liquid metals, 0.006< Pr < 0.03),
z, ≫z, .
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Momentum and Heat Transfer Analogy:
It is well known that the basic laws of transport of momentum and heat are
expressible in similar forms.
Consider the
laminar flow of
fluid through a
circular pipe (fig.
9). The wall of the
pipe is maintained
at a higher
temperature and
the fluid gets
heated as it flows
through the pipe.
We know (for
Newtonian fluid)
Fig. 9

47 |
τ=μ (61)
Shear
stress
Eqn. 61 can be rewritten as
τ = ν (ρu) (62)
The quantity ρu is the volumetric concentration of x momentum. Eqn. 62
physically means

Momentum flux = Momentum × Gradient of concentration (63)


diffusivity of momentum
Now consider the case of heat transfer to the fluid from the wall. The radial heat
flux at the wall is given by Fourier’s law
q=k (64)
Because T increases with r, the negative sign is not used on the RHS. The eqn. 64
can be rewritten as
q= (ρC T) = α (ρC T) (65)
Here α is the thermal diffusivity and ρC T is the volumetric concentration of heat
energy (enthalpy). Therefore

Heat flux = (Thermal (Gradient of the concentration (66)


diffusivity) of heat energy)
The flux equations 62 and 66, and their physical representations given in eqn. 63
and 66, show the similarity of the basic laws of momentum and heat transfer. The
diffusivities of momentum and heat i.e. ν and, α, have identical unite (m /s). In
the case of turbulent flow, a simple way of expressing the momentum and heat flux
was put forward originally by Reynolds, who argued that the laws of diffusional
transport as given by equation 62 and 66, are still applicable in case of turbulent
flow, but the contribution of eddy exchange should be incorporated in terms of
separate parameters. The resulting modified transport ‘laws’ are as follows where
ε and ε stands for the “eddy diffusivities” of momentum and heat respectively.
Turbulent transport of momentum,
τ = +(ν + ε ) (ρu) (67.a)
Turbulent transport of heat,
q = +(α + ε ) (ρC T) (67.b)

48 |
Eqn. (67.a) can be written at r = R, gives the wall shear stress. Thus we have
 (ν + ε ) (ρu) = τ = ρV (68)
where f is the Fanning’s friction factor, τ is the wall shear stress and V be the
mean fluid velocity in the pipe [it is to be noted that by definition, f ≡ ].

⟹ +  = ( )
= ( )
(69)

The RHS of eqn. 68 follows from the Fanning friction factor’s definition.
Similarly, eqn. 67 can be modified as
 (α + ε ) (ρC T) = q = h(T − T ) (70)
where q is the wall heat flux, h is the wall heat transfer coefficient, and T is the
mean fluid temperature. Defining, T = , we can modify equation 70 as
  = (71)
( )
The analogy between momentum and heat transfer in pipe flow was first quantified
by Reynolds on the specific assumption that:
(i) The gradient of dimensionless velocity and dimensionless temperature at wall
are equal and
(ii) (ν + ε ) = (α + ε )
Then from eqn. 69 and 71 we have

( )
= ( )
(72)
Putting ν + ε = α + ε and rearranging, we get
=

⟹ = St = = (73)
.
where St is the Stanton number, eqn. 73 is called the Reynolds analogy and can be
used to determine the heat transfer coefficient if the friction factor f is known.
Prandtl (1910) provided a more realistic picture of turbulent transport by assuming
that the momentum and heat transfer occur through eddy exchange or eddy
transport in a ‘turbulent core’ and through diffusive transport in the ‘laminar
sublayer’ near the wall. With this assumption and by using the ‘universal velocity
profile’ near the wall, Prandtl developed the following equation relating the
Stanton number with the Fanning’s friction factor (f)

49 |
St = (74)
( )

The eqn. 74 is called the Prandtl analogy. It reduces to Reynolds analogy in the
case of Pr = 1.
Chilton and Colburn (1934) observed that experimental heat transfer data could be
better correlated by replacing, 1 + 5 (Pr − 1), by Pr in eqn. 74. That is

St =

⟹ = (75)
.
The LHS of eqn. 75 is the well known Colburn 𝒿 factor, (𝒿 ). That is
=𝒿 = (76)
.
Eqn. 76 is called the Chilton-Colburn analogy. Using the well known correlation
for the friction factor, f = 0.046Re . , for pipe flow, the 𝒿 is given by
𝒿 = 0.023Re . (77)
The Chilton-Colburn analogy as well as the Reynolds analogy can be alternatively
derived from the Dittus- Boelter correlation for turbulent flow, which is
= 0.023Re . Pr . (let it be cooling)
.
⟹ × × = 0.023Re Pr
.
⟹ = 0.023Re Pr
.
⟹ St. Pr = 0.023Re
⟹ . Pr = 𝒿 = 0.023Re . (78)
.
.
As we know for turbulent flow, f = 0.046Re (79)
Therefore,
𝒿 = (80)
which is nothing but Chilton-Colburn analogy. Putting Pr = 1 in the eqn. 80, we
have, 𝒿 = St, so
St = (81)
Eqn. 81 represents the basic eqn. of Reynolds analogy.

50 |
Overall Heat Transfer Coefficient:
In order to understand “overall heat transfer coefficient” we have to have an idea of
double pipe heat exchanger.
A double pipe heat exchanger consists of two concentric pipes with one fluid in the
inner pipe and other fluid in the annulus between the pipes. The heat transfer area
for such heat exchangers is equal to the surface area of the inner tube.
The flow and heat
transfer rates in this type
are moderate because
the equipment is
relatively small. Both
parallel flow and
counter flow
arrangements are
available in these
exchangers. The
schematic diagram of a
double pipe exchanger
is shown in fig. 10.
Fig. 10
Considering the heat transfer at any axial distance from one end of the exchanger
(double pipe), the temperature profile across the wall of the inner pipe and in the
neighborhood may be qualitatively represented as shown in fig. 11.
Here it is to be noted that the
depicted temperature profile
is that under steady state
condition. A closer
investigation reveals that the
magnitude of temperature
gradient always decreases
from wall to the bulk of the
fluid i.e.   and
  are maximum. In
order to answer this
question, we must know the
different regime of flow
existing in a turbulent flow
Fig. 11 system

51 |
As explained previously, we have laminar sublayer, very close to the wall, where
locally laminar flow condition prevails. Accordingly, the only heat transfer
mechanism may be identified as conduction. Next to the laminar sublayer we have
buffer layer (where transition flow condition prevails and the mechanism of heat
transfer is both conduction and convection) and turbulent core (where turbulent
flow condition prevails and the mechanism of heat transfer is only convection).
Under steady state conduction, the rate of heat transfer must remain same in all the
three layers. So in the laminar sublayer, we may write
( )
q = −k (82)
where k ( ) is the effective thermal conductivity in the sublayer. However, as the
( )
mechanism of heat transfer is only conduction, we may conclude, k =k,
where k is the thermal conductivity of the fluid. On the other hand, in the buffer
layer we may write
( )
q = −k (83)
Because of additional convection pathway of heat transfer in the buffer layer we
( ) ( )
have, k > k (= k ). As for steady state, q = q , we must have
> . Following the same logical analysis we may conclude
> > (84)

In the turbulent
Similarly, on the cold fluid side we have core

> > (85)


Thus we can explain the decreasing trend of the magnitude of temperature gradient
from the wall to the bulk fluid in a double pipe heat exchanger.
Now at any section (at a definite axial distance from one end) of the double pipe
heat exchanger, let the temperature of the hot fluid (bulk temperature) be T and
the same of the cold fluid (bulk temperature) be, T . So the overall driving force be
T − T . Considering a small heat transfer area dA, we may define the overall heat
transfer coefficient (U) as
U≡ )
(86)
.(
where dQ is the heat transfer rate over the area dA. Looking into the geometry of
the heat transfer area (as shown in fig. 12), we may consider dA either to be the
inner or the outer surface area of the inner pipe, i.e., we may have
dA = dA = 2πr dL = πD dL (87.a)

52 |
Or
dA = dA = 2πr dL =
πD dL (87.b)
Accordingly, we must
define two different
overall heat transfer
coefficients, one based
on the inner surface area
of the inner pipe, (U ),
whereas the other is
based on the outer
surface area of the of the
inner pipe, (U ). So we
may have either
Fig. 12
dQ = U dA (T − T ) (88.a)
or
dQ = U dA (T − T ) (88.b)
Combining eqn. 87 and 88, we obtain the following relation
= (89)
Next we have to evaluate the value of either U or U (if one is the known, the
other could be obtained by using eqn. 89). The primary question is what sort of
combination of different individual resistances actually leads to overall heat
transfer resistances? Once again, looking into the fig. 11, we may infer that there
exist three individual resistances, namely
(i) Resistance of the cold fluid (may be represented by cold fluid film heat transfer
resistance). (ii) Resistance of the pipe wall. And (iii) Resistance of the hot fluid
(may be represented by the hot fluid film heat transfer resistance). And we may
also conclude that these resistances are combined in series.

Fig. 13

53 |
Considering the analogy of electrical resistance and current flow, we can construct
the above circuit diagram (fig. 13).
In fig. 13 we have considered the hot fluid to be flowing through the inner pipe
(the corresponding resistance be, R ← ) and the cold through the annulus (the
corresponding resistance be, R ← ).
Accordingly, we may identify
R = ,R = , and R =  (90)

where x be the wall thickness of the inner pipe (so 2x = D − D ), k be the
thermal conductivity of the pipe material and dA be the log- mean area over the
axial span from L to L + dL, so dA = πD dL, where
D = (91)

Using eqn. 90 and considering steady state heat transfer, we have


dQ = h dA (T − T ) = k dA = h dA (T − T ) (92.a)

We have
T − T = (T − T ) + (T − T ) + (T − T ) (92.b)
Now using equation (88.a)
= + +


⟹ = + + (93)
Using the expressions of dA , dA and dA in relation to D , D and D
respectively, we have

= + + (94)
So
U =  (95)

Using eqn. 89, we obtain the following expression for U as


U =  (96)

As we have solved for U and U (given by eqn. 95 and 96) without considering
any dirt layer resistances, we may identify the overall heat transfer coefficients (U
and,U ) as “clean overall heat transfer coefficient”. Generally they are represented
by U and,U (‘c’ stands for clean).

54 |
Dirt Resistance or Fouling factor:
Prolonged operation of a heat exchanger may result in the heat transfer surfaces
being coated with various deposits in the flow system. Furthermore, the surfaces
may become corroded as a result of the interactions between the fluids and the
material used for the construction of the heat exchanger. In either case, this coating
gives rise to an additional resistance to the heat flow, and thus results in low
performance. The overall effect is usually represented by a quantity called “dirt
factor” or “fouling factor”. The fouling factor effect must be included in the overall
heat transfer coefficient, after considering the fouling effects (U is alternatively
known as “design overall heat transfer coefficient” because design calculations
should always be performed with U and not with, U ). We may define fouling
resistance (R ) as
Based on inside area R = − (97.a)
Or
Based on outside area R = − (97.b)
Alternatively, let h and h be the inner and outer wall (of the inner pipe) heat
transfer coefficients of the dirt layer, which (in the form of inverse) in terms of
thermal resistance acts in series with the resistances shown in fig. 13. The modified
thermal circuit including the dirt resistances is shown in fig. 14

Fig. 14
where R = and R = . Following the same procedure, we may
show
U = (98.a)

Or
U = (98.b)

55 |
Combining eqn. 95, 96, 97 and 98 we may identify
R = + . (99.a)
and
R = + . (99.b)
Heat Transfer with a Variable Driving Force: Co-current and Counter-
current Operations:
Let us consider the schematic of a countercurrent flow system shown in fig. 15.
The following notations will be used.
m : Mass flow rate of the fluid stream (kg s )
C : The specific heat (J/ kg℃)
T , T : Temperature of the fluid streams at the ends 1 and 2 of the device
respectively.
T: (= T − T ), the driving force at any section.

Fig. 15
The subscript ‘h’ and ‘c’ refer to the hot and the cold fluids respectively, and
T and T are the local temperatures of the respective streams. We arbitrarily
assume that the hot fluid flows through the inner tube. The temperatures of both
the streams vary with the position along the device.
Now consider a ‘thin’ section of the device having a heat transfer area dA. The
local temperature of the fluids are T and, T , whereas U be the overall heat
transfer coefficient. The rate of heat transfer, dQ, at steady state through a small
area dA may be written as
dQ = UdA(T − T ) = UdAT (100)

56 |
Also, a local heat balance over the section gives
dQ = −m C dT = m C dT (101)
Here dT and dT are the changes in the temperatures of the hot and the cold
streams over the thin section because of heat exchange. Obviously dT < 0 and,
dT > 0.
Next we put, T = T + T , so we obtain (from eqn. 101)
dQ = −m C d(T + T ) = −m C dT−m C dT
= −m C dT − dQ
Therefore
.
dQ = d(T) (102)
Substituting for dQ from eqn. 102 in eqn. 100 and rearranging, we get

=− UdA (103)
 .
Integrating eqn. 103 from (1) [A = 0, T = T ] to (2) [A = A, T = T ] we
obtain
 
− ∫ = . U ∫ dA
 .

⟹ ln = . UA (104)
 .
Let Q be the total rate of heat transfer, for the entire device, then we have
Q=m C T −T =m C T −T
⟹ + =T −T +T −T
= T −T − T −T = T − T
Therefore
.
Q= . (T − T ) (105)
If we define T as an appropriate mean driving force over the entire length of the
device, we have
Q = UAT (106)
From eqn. (106) and (105), we get
.
UAT = . (T − T ) (107)
Comparing eqn. 104 and 107, we get
 
T =  (108)

Eqn. 108 is one of the most important eqn. is heat transfer calculation. It shows
when the temperature driving force varies from one point to the other, the “log

57 |
mean temperature difference”,T , given above is the appreciable mean driving
force.
Next we will consider the case of heat exchange between a hot and a cold fluid in
countercurrent flow as shown in fig. 16.

Fig. 16
Here
dQ = UdA(T − T ) = UdAT (109)
Once again, dQ = −m C dT = m C dT
The capacity eqn.(Q = UA(LMTD)) for counter-current flow situation are similar
to those for counter current flow. So we have
Q = UA(LMTD) (110)

Problems to be Solved :
1. Experiments conducted with sparingly dissolving cylinder wall in a flowing
liquid yields the following correlation for Sherwood number,
Sh = 0.023Re . Sc
Assuming applicability of Chilton- Colburn analogy, the corresponding correlation
for heat transfer is
. .
(a) St = 0.023Gr Pr (b) Nu = 0.023Re Pr
(c) 𝒿 = 0.023Re . Pr (d) Nu = 0.069We . Pr (2006)
2. Water enters a thin walled tube (L=1m, D=3mm) at an inlet temperature of
970C and mass flow rate, 0.015kg s . The tube wall is maintained at a constant
temperature of 270C. Given the following data for water
ρ = 10 kgm , μ = 489 × 10 Pa. s

58 |
C = 4184 . Inside heat transfer coefficient, h = 12978 . The outlet
.
temperature of water in 0C is
(a) 28 (b) 37 (c) 62 (d) 96 (2007)
3. A hot fluid entering a well- stirred vessel is cooled by feeding cold water
through a jacket around the vessel. Assume the jacket is well- mixed. For the
following data
ṁ = 0.25 , ṁ = 0.4 , C = 6000 ,C = 4184 . The inlet
. .
0 0
and outlet temperature of hot fluid is 150 C and 100 C respectively. Inlet
temperature of cold water=200C, U = 500 , the heat transfer area (in m ) is
(a) 1.82 (b) 2.1 (c) 3 (d) 4.26
4. Hot liquid is flowing at a velocity of 2m/s through a metallic pipe having an
inner diameter of 3.5cm and length 20m. The temperature at the inlet of the pipe is
900C. Following data are given
ρ = 950kgm , C = 4.23 , μ = 2.55 × 10 s, k = 0.685 . The heat
.℃ . ℃
transfer coefficient (in ) inside the tube is

(a) 222.22 (b) 111.11 (c) 22.22 (d) 11.11 (2008)
5. The temperature profile for heat transfer from one fluid to another separated by a
solid wall as

(2008)
59 |
6. A double- pipe heat exchanger is to be designed to heat 4kg/s of a cold feed
from 200C to 400C, using a hot stream available at 1600C and a flow rate 1kg/s.
The two streams have equal specific heats and, U = 640 . Then the ratio of the
heat transfer areas required for the co-current to counter current modes of
operation is
(a) 0.73 (b) 0.92 (c) 1.085 (d) 1.25 (2009)
Common statement for Q. 7 and 8 (2010)
0 0
Hot oil at 150 C is used to preheat a cold fluid at 30 C in a shell and tube heat
exchanger. The exit temperature of the hot oil is 1100C. Heat capacities (product of
mass flow rate and specific heat capacity) of both the streams are equal. The heat
duty is 2kW.
7. Under co-current flow condition, the overall heat transfer resistance is
℃ ℃ ℃ ℃
(a) 0.4 (b) 0.04 (c) 0.36 (d) 0.036
8. Under counter-current flow condition, the overall heat transfer resistance is
℃ ℃ ℃ ℃
(a) 0.4 (b) 0.04 (c) 0.36 (d) 0.036
9. Heat is generated at a steady rate of 100W due to resistance heating in a long
wire, (L = 5m, D = 2mm). This wire is wrapped with an insulation of thickness
1mm that has a thermal conductivity of 0.1 . The insulated wire is exposed to air
at 300C. The convective heat transfer between the wire and surrounding air is
characterized by a heat transfer coefficient of 10 . The temperature (in 0C) at
the interface between the wire and the insulation is
(a) 211.2 (b) 242.1 (c) 311.2 (d) 484.2 (2012)
10. In a counter-flow double pipe heat exchanger, oil ṁ = 2 kg s, C = 2.1
.℃
0 0
is cooled from 90 C to 40 C by water, ṁ = 1 kg s, C = 4.2 , which enters
.℃
the inner tube at 100C. The radius of the inner tube is 3cm and its length is 5m.
Neglecting the wall resistance, the overall heat transfer coefficient based on the
inner radius (in ) is
(i) 0.743 (b) 7.43 (c) 74.3 (d) 2475
11. Consider steady state heat transfer (by conduction) from a sphere of diameter d,
having a constant surface temperature, T , to a large volume of stagnant fluid. The
temperature of the fluid is T at a point far away from the sphere. Show that
Nu = 2.
12. Air at 200C flow over a block of ice, 1m × 1m free surface area, at a velocity
of 8m/s. If the ice is at its melting point, calculate the rate of melting of ice.

60 |
13. Air at 700C and essentially atmospheric pressure is flowing through a 15cm ×
10cm rectangular duct at a velocity of 12m/s. If the wall temperature of the duct is
320C, and the length of the duct is 50m, calculate the drop in air temperature.

61 |
Radiation Heat Transfer:
Heat transfer by conduction or convection from one point in the medium to another
occurs in the presence of a temperature difference. Heat transfer from a body by
radiation, to the contrary, does not need a temperature driving force or a medium.
However, radiation heat exchange between two bodies at different temperatures
always results in a net transfer of heat energy from the body at a higher
temperature to the other at a lower temperature.
Thermal radiation is nothing but an electromagnetic wave. Radiation propagates at
the speed of light, c = 2.998 × 10 m⁄s. A radiation or an electromagnetic wave
is characterized by its wavelength λ or its frequency ν related by
c = λν (1)
Emission of radiation from a body is not continuous, but occurs only in the form of
discrete quanta. Each quanta has an energy of
E = hν (Plank’s law) (2)
where h is the Plank’s constant, h = 6.6256 × 10 J. s.
The electromagnetic radiation spectrum comprises extremely short wavelength (or
high, ν) gamma rays and cosmic rays on one end to long wavelength microwaves
and radio waves on the other. The names of different bands of wavelengths of the
spectrum are shown in table–1.
Type Wavelength band (m) Type Wavelength band (m)
Cosmic rays up to 4 × 10 Infrared rays 7.8 × 10 – 1 × 10
γ rays 4 × 10 – 1.4 × 10 Thermal 1 × 10 – 1 × 10
X rays 1 × 10 – 2 × 10 radiation
UV rays 1 × 10 – 3.9 × 10
Visible light Microwave, 1 × 10 – 2 × 10
3.9 × 10 – 7.8 × 10
radar, TV, radio
waves
Table-1
Basic Concepts of Radiation from a Surface:
Thermal radiation incident on a body tends to increase its temperature. Generally,
the incident radiation may be absorbed, reflected or transmitted, partly or fully.
Absorption of radiation cause, a rise of temperature, primarily localized over the
surface exposed to radiation. Conduction of heat from the surface to the interior of
the body, raise the temperature of the entire body. If the body is permeable to the
incident radiation, a part of it is transmitted through the body. Additionally, for
polished surface we may have a part of incident radiation being reflected back by
the body.
The fraction of incident radiation absorbed by a body is called absorptivity (α), the
fraction reflected is reflectivity (ρ) and the fraction transmitted through the body is
the transmissivity (τ). The factor should add up to unity i.e.

62 |
α+ρ+τ= 1 (3)
Absorption and reflection of light on a surface are responsible for our perception of
color. The color of the body is determined by the wavelength of radiation it
reflects. For example, a ‘red’ object appears so because its surface reflects only the
red component of light and absorbs the rest. A surface, which absorbs light of all
wavelengths in the visible range, is black, whereas a surface that reflects light of
all wavelengths in this range and does not absorb any preferentially, is called
white. Thus for an opaque black surface, ρ = 0 and, τ = 0, i.e., α = 1. For an
opaque white surface, α = 0, τ = 0, i.e., ρ = 1.
Blackbody Radiation:
The fundamental laws and relations for radiation have been developed for surfaces
on the basis of the concept of a blackbody. A blackbody is a surface that has the
following properties
(i) A blackbody completely absorbs the incident radiation irrespective of its
wavelength. A blackbody is ‘black’ because it does not reflect any radiation.
(ii) A black body is a perfect emitter. No other surfaces can emit more radiation
than a blackbody provided that they are at the same temperature.
(iii) Emission of a blackbody occurs in all possible direction.
Now we order to quantify the energy emitted by a blackbody, we must have certain
standardized terms related to the radiation energy of a body in general. The two
most commonly used terms are
(i) Monochromatic Emissive Power(𝐸 ):
The amount of energy emitted by a body (by a surface) per unit area, per unit time
and per unit wavelength (in the wavelength range λ to λ + dλ) is known as
monochromatic emissive power. Its unit is μm
(ii) Total Emissive Power (E):
The amount of energy emitted by a body (or by a surface) per unit time, per unit
area, over the entire wavelength range, i.e., 𝜆 ⇒ [0, ∞[ is known as total emissive
power.
So we may obtain E by integrating E over the entire wavelength range of,
𝜆 ⇒ [0, ∞[ i.e.
E = ∫ E dλ (4)

Plank’s Law:
In 1900, Max Plank, with the help of his quantum theory, derived the following
equation for monochromatic emissive power (or spectral emissive power), E of a
blackbody as a function of λ and the temperature of the body (T). Here ‘b; in the
subscript of ‘E ’ represents ‘black’ body. The equation as derived by Plank is

63 |
E =

⟹E = (5)
where h is the Plank’s constant, k is the Boltzman constant, c is the velocity of
light, K = 2πhc and K = . Equation 5 is known as Plank’s distribution law.
This is a one parameter
distribution function; the
parameter is the absolute
temperature T. A plot of
Plank’s distribution of E
with λ is shown in fig. 1. The
plot says quite a few things.
At any temperature, a
blackbody emits radiation
covering a range of
wavelengths. Any of the
curves corresponding to a
particular temperature has a
maximum (at, λ = λ ) and
the energy of the emitted is
rather concentrated in a
region of wavelength (λ ).
Fig. 1
The peak of the curve is shifted towards the shorter wavelength side as temperature
increases. The value of λ decreases as the temperature of the blackbody
increases. This means that a blackbody of high temperature emits more radiation of
short wavelength.
Wein’s Displacement Law:
It can be derived from eqn. 5 that the wavelength, λ corresponding to the peak
of λ vs. E plot (fig. 1), is inversely proportional to the temperature of the
blackbody. In other words
λ T = constant = 2898μm. K (6)
Eqn. 6 is called Wein’s displacement law.
Taking the derivative of E as in eqn. 5 wrt λ and setting it to zero ( = 0 for,
λ = λ ) we get
( )
= − × exp + =0

64 |
⟹ 5 1 − exp − − =0
The transcendental eqn. can be solved for to give, = 4.965. Therefore
. ×
λ T= = = μm. K
. . .
⟹ λ T = 2898 μm. K (Wein’s displacement law)
Wein actually had propounded his law in 1893, before Plank’s quantum theory was
proposed. In this context, it is interesting to note how the gradual change of ‘color’
of the body occurs as it is heated. At low temperature, the body predominantly
emits infrared radiation and cannot be seen/detected by the eye. As temperature
increases it emits more and more, short wavelength radiation. While being heated,
the color of the body, at some point of time appears dark red. This indicates that
the body is emitting radiation in considerable amount of wavelength equal to that
of red light. At gradually increasing temperatures, the color appears bright red,
then bright yellow and then white. In the “white- hot” stage, the body emits
radiation covering the entire visible range. However, never does the color of the
body appears to change from red to orange to yellow, etc. because at no stage the
body emits radiation of a single wavelength. The emitted radiation always covers a
band or range of wavelength that theoretically varies from zero to infinity.

Stefan-Boltzman Law:
Plank distribution law gives the monochromatic emissive power of a blackbody. It
is rather simple to determine the total emissive power of a blackbody over the
entire spectrum by integrating, E .
If the total emissive power of a blackbody is denoted by E , then
E = ∫ E dλ = σT (7)

Evaluation of the above integral results the constant, σ = , which is known as


Stefan-Boltzman constant. Putting K and K from Plank’s law we get, σ =
5.669 × 10 . Eqn. 7 is called the Stefan-Boltzman law, which allows us to
calculate the total amount of radiation of all wavelength emitted in all possible
directions by a blackbody. The value of σ has been determined by a number of
researchers. The accepted experimental value of σ = 5.729 × 10 , is
generally used in radiation calculation.

Kirchoff’s Law:
Let us consider a large enclosure of surface temperature, T . The enclosure
virtually behaves like a blackbody. It is assumed that the radiation heat flux

65 |
incident on any surface in the enclosure is q. There are several bodies in the
enclosure, which are sufficiently small so that, the radiation effect on the surface of
the enclosure is small. Let the surface area of one of the bodies (call it the first
body) is A . If E denotes the emissive power of the body, we have
The rate at which body-1 emits radiant energy= E A
The rate at which body-1 receives radiant energy= α qA
where α is the absorptivity of the body. Therefore, if body-1 is in thermal
equilibrium with the enclosure, then the rate of emission of radiation must be equal
to the rate of absorption.
Therefore, E A = α qA
⟹E =α q (8)
Now, if body-1 is replaced by a blackbody, we should put E = E and α = α =
1 (because the absorptivity of a blackbody is unity). Therefore
E = q and = =α (9)
Let us now define emissivity (ε) of a body or surface as the ratio of emissive
power to that of a blackbody. So for body-1, we have
ε = (10)
Therefore, from eqn. 9 we obtain
ε =α (11)
Similar relation is also valid for other bodies in the enclosure. So removing the
subscript in eqn. 11 we obtain
ε=α (12)
Eqn. 12 states that the emissivity of a body which is in thermal equilibrium with its
surroundings is equal to its, absorptivity. This is known as Kirchoff’s law.
However, it should be noted that the stated law is valid only when the source
temperature of irradiation is equal to the temperature of the irradiated surface. But
as a matter of fact, the absorptivity of most real surfaces is relatively insensitive to
temperature and wavelength. So, for practical purposes, it is customary to assume
that the emissivity and the absorptivity of a surface are equal even when it is not in
thermal equilibrium with its surroundings.
Gray Body:
In the previous discussion, we have assumed that for all practical purposes. α = ε.
In this context we define gray body whose α and ε are independent of wavelength,
(λ). Thus a gray body is also an ideal body with α and ε < 1.

66 |
The fig. 2 shows
(qualitatively) the nature of
dependence of
monochromatic emissive
power of a blackbody, a
gray body and a real surface
upon the wavelength of
radiation. For a real surface,
in general, the emissivity
depends upon the
wavelength of radiation.
Correspondingly, we define
spectral emissivity: ε (λ) as
the ratio of E to E . So
ε ≡ (13)

Fig. 2
where as
ε= (14)
Radiation Intensity of a Blackbody:
The rate of emission of radiation by a surface is conveniently expressed in terms of
radiation intensity. Two types of radiation intensity are defined––the spectral
intensity and the total intensity. The spectral intensity is defined as the rate of
emission of radiant energy in a particular direction per unit area of the emitting
surface normal to the direction (i.e., projected area), per unit wavelength, 𝜆, per
unit solid angle around this direction.
We denote it by I , where
subscript ‘b’ implies black
body. It has the unit of
.μm. Sr. Here ‘Sr’ is
sterdian, the unit of solid
angle. The representation of
solid angle (ω) can be
explained by fig. 3.
The quantity, I , may be
Fig. 3 explained with the help of
fig. 3 and fig. 4
67 |
Consider an elementary
‘black’ area dA that emits
radiation and is surrounded
by a hemisphere of arbitrary
radius r and surface area,
2πr . The hemisphere is
constructed in such a way, so
that the elementary area (dA)
is coplanar with the base of
the hemisphere and the
midpoint of dA coincides
with the centre of the
hemisphere. The hemisphere
is assumed to behave like a
blackbody.
Fig. 4
Consider another elementary area, dA or the surface of the hemisphere located at
the cone angle θ and the planar angle ϕ in the spherical coordinate system. Now
the projection of dA normal to the r direction is dA cos θ. Therefore, the
differential rate of radiant energy emission from the surface around a wavelength λ
should be proportional to:
(i) The projected area of the elementary surface normal to r direction i.e. dA cos θ.
(ii) The solid angle dω subtended by the elementary receiving surface dA of the
hemisphere, (dω) and
(iii) The small interval of wavelength d λ.
Hence
d Q ∝ dA cos θ . dω. dλ (15)
where d Q is the rate of radiant energy from the small area dA in the direction r
covering a solid angle dω for the interval of wavelength dλ. Note that, d (. )
implies the third order differential term, proportional to the product of three first
order terms, dA, dω and, dλ. Introducing the proportionality constant I and
substituting dω = , we get
d Q = (16)
The constant of proportionality, I , is called the “spectral intensity of radiation”
and has the unit of . μm. Sr, as stated before. The total intensity (I ) can be
determined by integrating I over the entire range of wavelength. Thus, we get

68 |
I = ∫ I dλ (17)
Spectral Emissive Power of a Blackbody over a Hemisphere:
We shall now determine the monochromatic or spectral emissive power of the
blackbody over the hemisphere shown in fig. 4. We have
d Q = I dA cos θ dωdλ (18)
In the spectral polar coordinate system, the elementary or differential area dA on
the surface of the sphere is given by
dA = r sin θdθdϕ (19)
i.e. dω ≡ = sin θdθdϕ (20)
The hemispherical spectral emissive power E can be obtained as
E =∫ ∫ I cos θ sin θ dθdϕ

⟹E =I ∫ − cos 2θ dϕ = πI
⟹ E = πI (21)
Eqn. 21 embodies a very important result, i.e., the hemispherical spectral emissive
power of a blackbody is π times the spectral intensity of radiation. Integrating over
the entire wavelength we get
∫ E dλ = π ∫ I dλ
⟹ E = πI (22)

Radiative Heat Exchange between Surfaces: The View Factor:


Calculating the heat exchange between two or more surfaces is of great practical
importance. Heat transfer by radiation between neighboring surfaces is pretty
common in industrial practice. For example, a furnace wall or door loses heat by
radiation, in addition to other modes of heat loss, to the surrounding. A natural
question is:
What are the factors that determine the rate of heat exchange between two bodies?
Two factors are: the temperature of the individual surfaces, their emissivities (1 for
black surfaces) and how well one surface can see the other. Let us attempt to
quantify the last factor in terms of quantity known as “view factor” assuming the
intervening medium between the surfaces does not absorb any radiation.
The fraction of the total energy radiated by a surface i, which is intercepted by the
surface j is called the view factor/configuration factor/shape factor. We denote
view factor by F where i refere to emitting surface and j refer to intercepting
surface. Thus view factor can be expressed as

69 |
the fraction of the total radiation
view factor F = emitted by surface i and intercepted (23)
by the surface j
In order to develop a mathematical expression for view factor, we consider two
surfaces A and A of arbitrary geometry and orientation. Two small surface
elements of A and A are denoted by dA and dA and are located at a distance r
apart.
The distance segment r subtends
an angle θ with the normal to
the surface element, dA , and an
angle θ with the normal to the
surface element dA (fig. 5). If
I is the local intensity of
radiation emitted by the surface,
A , then the amount of radiation
emitted by the small surface dA
and intercepted by another small
surface dA (both of which are
assumed be blackbodies) is
Fig. 5 given by
d Q = I dA cos θ dω (24)
But the solid angle, dω = . Therefore we have
d Q = (25)
The notation, d Q , indicates that it is a second order differential term involving
the product of two first order differential terms, dA and dA . Integrating eqn. 25
over the surface A and A we obtain
Q =I ∫ ∫ dA dA (26)
The total amount of radiation Q emitted by the surface A obtained from its
hemispherical emissive power as
Q = A E = A πI (27)
The view factor may be determined as the ratio of Q and, Q . That is
F = = ∫ ∫ dA dA (28)
We can obtain F simply by interchanging the subscripts. That is
F = = ∫ ∫ dA dA (29)

70 |
From eqn. 28 and 29 we obtain
AF =AF (30)
Eqn. 30 is known as the reciprocity relation between the two view factors.
Example (Calculation of View Factor from the First Principle):
Consider a circular disc of diameter D and area A above a plane surface of area,
A ≪ A . The surfaces are parallel to each other and A is located at a distance L
from the centre of, A . Obtain an expression for the view factor, F .
Solution: Recall
F = ∫ ∫ dA dA
As A is pretty small, θ , θ and r
are approximately independent
of A . Thus
F =∫ dA
From fig. 6, we have
θ = θ = θ (let)
and r +L =r
Whereas dA = 2πr dr and,
cos θ = .

Fig. 6
So we have
F =∫ . . 2πr dr

⟹F =L ∫

⟹F =L − = − L

⟹F = (31)
View Factor Algebra:
Determination of view factor by using equation 28 and 29 often becomes difficult
for bodies having complex geometries. Sometimes the problem can be tackled by
making use of the definition of view factor and the reciprocity relation (eqn. 30).
View factor algebra is a methodology to determine the view factor for a collection
of surfaces with the help of the following two relations.

71 |
In case of a blackbody enclosure (fig. 7) having N number of surfaces or walls, the
radiation emitted by any surface, for example, surface–1, may be incident on any
of the surfaces including itself (because if a surface is concave, it can see “itself”).

So
F + F + ⋯ + F = 1 (32)
For any arbitrary surface i of the
enclosure, we have
∑ F =1 (33)
This is known as summability
relation, to be used in view factor
algebra.
The other important equation is
the reciprocity relation, which has
been already derived as
Fig. 7 F A =F A (34)
Now we shall show that for an N-walled enclosure, some of the view factor may be
determined from the a priori knowledge of the rest. There is a total N view factor
which may be represented in the matrix form given below:
F F …..F
𝐅= F F …….F (35)
F F ……..F

Now, from summability relation (eqn. 33) we have N number of equations (for
surfaces i=1, 2, 3……, N). Now considering the reciprocity relation (eqn. 34), we
( )
have equations of view factors [for a given i, we have (N − 1) values of j,
now i may be varied from 1 to N, so there are N(N − 1) equations, but each eqn.
( )
has occurred twice. Therefore the actual number of equations is ]. Thus we
have the total number of equations available be
( ) ( )
N+ = (36)
which is the number of the unknown view factors. That can be determined from the
view factor algebra.
So the number of view factors to be determined a priori becomes
( ) ( )
N − = (37)
For example, if we have a 4 surface enclosure:
The total number of view factor involved= N = 4 = 16.

72 |
( ) ×
The number of view factors that should be known = = =6
The number of view factors that can be determined from view factor algebra
( ) ×
= = = 10
Example 1:
Consider an enclosure consisting of a
hemisphere of diameter D and a flat
surface of the same diameter (shown in
figure). Determine the relevant view
factors.
Solution:
Here the number of surfaces, N = 2.
So the total number of view factors=
N = 2 = 4. i.e. F , F , F , F .
Now the flat surface cannot see itself, so, F = 0. We know the number of view
( ) ×
factors to be known a priori, = = = 1 and that is F . The rest of the
F , we can determine by view factor algebra.
From summability we have, F + F = 1, so F = 1, as F = 0.
By reciprocity we have, F A =F A
⟹ 1. D = F . D

So, F = = 0.5. Once again by applying summability for body 2 we have


F = 1 − F = 1 − 0.5 = 0.5
So finally we have the view factor matrix as
0 1
𝐅=
0.5 0.5
Example 2:
An enclosure of triangular cross
section is made up of three
plane plates, each of finite
width and infinite length (thus
forming an infinitely long 
prism). Derive an expression
for the configuration/ view
factors between any two of the
plates of widths, L , L and L
(see fig.)

73 |
Solution: For plate 1, summability gives the following relation
F +F +F =1
as F = 0, we have
F +F =1 (i)
Using similar relation for other planes we have
F +F =1 (ii)
and F +F =1 (iii)
Multiplying eqn. (i), (ii) and (iii) by, A , A and A respectively we have
A F +A F =A (iv)
A F +A F =A (v)
A F +A F =A (vi)
Applying reciprocating relations we have
A F +A F =A (vii)
A F =A F
A F +A F =A (viii)
A F =A F A F +A F =A (ix)

Thus eqn. (vii) to (ix) gives three equations for three unknown view factors.
Solving the eqn. simultaneously [(vii)+ (viii) –(ix)] we obtain
F = ⟹F = (x)
Similarly we can determine F = = etc.
Hottel’s Crossed- Strings Method:
The view factor relation, developed in example-2, may be used to determine all
view factors with long enclosure having constant cross section. The method was
first proposed by hotel (1954) and is called cross-strings method since the view
factors can be determined in a laboratory by a person. All he/she needs are four
pins, a roll of strings, and a measuring tape.
Consider fig. 8. Our objective here is to determine F . The surfaces shown are
rather irregular and the view between them may be obstructed. Once can imagine
how difficult it can be to obtain the view factor by integration. In the crossed
strings method, four pins are placed at the two ends of each surface, as indicated
by a, b, c and d, in fig. 8. Now all the points are connected by tight strings.

74 |
Fig. 8
Now, assuming the strings to be to the imaginary surfaces, A , A and A for
the triangle abc, we can write
F = (38)
It would be noted here that since a tightened string always forms a convex surface,
the rule of the  developed in example-2 can be used. Similarly, for  abd,
F = (39)
From the summation rule,
F +F +F =1 (40)
Thus, using eqn. 38– 40 we get
( ) ( )
F = (41)
It is also observed from fig. 8 that all radiation leaving A travelling to A will be
received by the surface, A . At the same time all radiation from A going to A
must pass through, A . Therefore,
F =F (42)
Using the reciprocity relation and repeating the argument for surface A and, A ,
we see
F =F = .F = .F (43)
Using eqn. 41 and 43 we have
( ) ( )
F = (44)

75 |
or

F = )
(45)
×(
Example 3:
Two infinitely long semi cylindrical
surfaces of radius R are separated by a
minimum distance D as shown in the
fig. derive the view factor, F .
Solution:
The length of the crossed string abcde
is denoted as, L , and of uncrossed
string as L . From the symmetry of the
problem,
F = =
=
The length L is given by
L = D + 2R
The length, L = 2 × length of cde
The segment L from c to d is found from the right- angled triangle ocd to be

L , = −R = D +R

And the segment of length L from d to e is L , = Rθ = sin . Therefore


, ,
F = =

⟹F = (Answer)

Rate of Radiation Exchange between Blackbodies:


The definition of view factor can be used to calculate the rate of exchange of
radiation between two blackbodies of any geometry and orientation. Consider
again two blackbodies 1 and 2 of surface areas A and A , temperatures T and T ,
and the total emissive powers E and E , respectively. Then, the rate at which
radiation emitted by body 1 is intercepted and absorbed by body- 2
=Q =A E F (45)

76 |
The rate at which radiation emitted by body– 2 is intercepted and absorbed by
body– 1
=Q =A E F (46)
Therefore, the net rate of radiation exchange between bodies 1 and 2
= [Q ] = A E F − A E F (47)
Putting, E = σT , E = σT and using the reciprocity relation, A F = A F ,
we get
[Q ] = A F σ(T − T ) (48)
[Q ] denotes the net rate of gain of radiation by body 2 because of its
Radiative interaction with body 1.
The result of eqn. 48 can be easily extended to Radiative exchange among the N
surfaces of an enclosure. The net rate of radiation energy loss by the surface i
because of its Radiative interactions with all the surfaces including itself is given
by
[Q ] = ∑ A F σ T − T (49)

It is to be noted that if the surfaces of the enclosure are at thermal equilibrium, we


get
T = T , and [Q ] = 0

Exchange of Radiation between Diffuse Gray Surfaces:


In the preceding section, we discussed the method of calculation of the rate of
exchange of radiation in a block enclosure, that is, an enclosure having walls that
behave like a black bodies. A real surface does not behave like a blackbody,
although it may sometimes be close approximation to a blackbody. For a non-
blackbody we cannot write 𝜌 = 0, so the total quantity of radiation that leaves a
non- black surface is the combination of emitted radiation and reflected radiation.
So more information is needed for the calculation of radiation exchange in a non-
black environment and the equations for such calculations have to be developed.
Before we proceed to analyze radiation exchange between non-black bodies, we
shall discuss a little more about their characteristics and learn a few more terms.
We have already defined a gray body or surface (which is also an idealization). A
gray body or surface is sometimes called a diffuse gray. A surface is called diffuse
gray if its spectral emissivity or absorptivity are independent of the angle of
incident and the angle of emission. In fact, a surface is called diffuse if it emits (or
absorbs) radiation equally in all directions without any directional preference.
As a matter of fact, for a real surface, the emissivity and absorptivity may depend
upon the direction of emitted radiation or the incident radiation as the case may be.
Also, reflection from a real surface may be secular or diffuse. However, the

77 |
analysis will be quite complicated if all these characteristics are taken into
consideration. Because many real surfaces have characteristics close to those of
diffuse gray surfaces, the equation for radiation exchange that will be presented
here can be used in many practical situations with satisfactory accuracy.
We will introduce here two new terms: irradiation, denoted by 𝐺, and radiosity
denoted by J. They are defined as under.
Irradiation (G): The total radiation that hits the surface per unit area per unit time,
.
Radiosity (J): Total radiation that leaves a surface per unit area per unit time, .
Since a blackbody does not reflect any radiation, we have J = E (total emissive
power).
Radiation Exchange in a Grey Enclosure:
Radiosity of a diffuse gray body can be related to the irradiation it receives in the
following way. If E is the total emissive power of a blackbody at the same
temperature, then we have
Emissive power of gray body= εE
Radiation reflected by gray body= ρG
where, ε is the emissivity and ρ is the reflectivity. Therefore
J = εE + ρG (50)
If the transmissivity of the body is zero, (τ = 0), we have α + ρ = 1. Also from
Kirchoff’s law we have, α = ε.
So ρ=1−α=1−ε
Putting the result in eqn. 50, we get
J = εE + (1 − ε)G (51)
If A is the surface area of the body, the net rate of loss of radiant energy from unit
area of the surface is
= J − G = εE + (1 − ε)G − G
= εE − εG (52.a)
( )
From eqn. 51, we have, G = ( )
. Therefore
( )
= εE − ( )

Q = ( ) (52.b)
We may consider the RHS of eqn. (52.b) as the ratio of a potential difference and a
surface resistance of radiation heat transfer. The radiant heat loss can, therefore, be
simulated by an electric current flow network diagram as shown in fig. 9.

78 |
Now let us consider radiation heat
exchange in an enclosure, assuming
the surfaces to be diffuse gray. It is
easy to understand that the view
factor and the emissivities of the
surfaces will be involved in the
analysis. Radiation from each of the
N surfaces will, in general, be
incident on all the surfaces. If we
consider the ith surface, the
irradiation G of the surface will be
Fig. 9
the sum of the radiosities of N surfaces multiplied by the respective areas and view
factors. That is
AG =∑ AF J (53)
Using the reciprocity relation, A F = A F , we get
AG =∑ AF J (54)

The net loss (rate) of radiant energy by the ith surface is


Q , = A (J − G ) = A J − ∑ A F J (55)
From eqn. of summability relation we have
∑ F =1
⟹J =J ∑ F (56)
Using eqn. 55 and 56 we have
Q, =A J ∑ F −∑ A F J
=∑ A F J −J =∑ Q (57)
Here Q denotes the net rate of radiant energy loss from the ith surface
because of radiation exchange with the jth surface. Writing eqn. 52 for the ith
surface, we have
Q, = (58)

From eqn. 57 and 58, we get


Q, = =∑ (59)

Also, the net rate of exchange of radiation between the surfaces i and j is

79 |
Q, = (60)

Eqn. 59 is very important for the calculation of radiation exchange in a grey


enclosure. It may be noted that J values should be known in order to calculate,
Q , . Actually for N- surface enclosure, eqn. 59 provides N simultaneous linear
equations that may be solved to provide N number of J values. Let us illustrate the
application considering, N = 2.

Radiation Exchange in a Two-surface Grey Enclosure:


Eqn. 59, written for i = 1 and, i = 2, gives
i = 1, = A F (J − J ) (61)

i = 2, = A F (J − J ) (62)

Subtracting eqn. 62 from eqn. 61 and putting, F = . F , we get


E −E = (J − J ) 1 + F . +F .
The above equation can be solved for, (J − J ). Putting this value in place of,
(J − J ) in eqn. 60 and simplifying, we get
Q , = ( ) = (63)

Here Q , is the net rate of loss of radiant energy from surface 1. The RHS of
eqn. 63 can be visualized as the ratio of radiation potential difference E − E
and the radiation resistance. The resistance term (i.e. the denominator) is the sum
of three individual resistances in series.
: Radiation surface resistance of surface 1
: Radiation surface resistance of surface 2
: Space resistance to radiation
It is to be noted that the above resistances are notional and should not be confused
with the heat transfer resistances in conduction or convection. The network
diagram for radiation exchange in a two surface enclosure is shown in fig. 10.

80 |
Fig. 10
If we put the expressions of E and E in eqn. 63, we get
Q , = (64)
where T and T are the absolute temperatures of the surface 1 and surface 2.
Now, if the surfaces are flat (so that, F = 0), like two very large parallel flat
surfaces, we have F = F = 0. So, F = F = 1, so = 1. So under this
condition, the net flux of radiation energy loss from surface 1 is
,
= (65)

Emissivity Factor:
This is a term sometimes used in connection with the calculation of radiation
exchange among diffuse gray surfaces. It is also called the emissivity correlation.
For an enclosure made up of two surfaces 1 and 2, it is defined as follows.
Q , = F F A σ(T − T ) (66)
where Q , is the net rate of radiation loss by surface-1. F is the emissivity
factor of surface– 1.
Comparing eqn. 66 with eqn. 64, we have
F = (67)

The emissivity factor for the other surface can be defined similarly.

A Gray Enclosure with Re-radiating Surfaces:


A re- radiating surface is one which radiates the entire amount of radiation that it
receives from other surfaces. At steady state there is no net radiation absorption or
net radiation emission of the re-radiating surface. In other words J = G and,
Q , = 0. A re-radiating surface is functionally adiabatic.

81 |
Let us consider an enclosure consisting of N gray active surfaces and M re-
radiating surfaces. Because the ith active surface exchanges radiation with all
(N + M) surfaces, we can rewrite eqn. 59 as
Q, = =∑ (68)

where j = 1,2,3, … , N, r , r … , r and i = 1,2,3, … , N


For M re-radiating surfaces we have
,
0= =∑ (69)

where k = 1,2,3, … , M and j = 1,2,3, … , N, r , r … , r


The subscript r denotes re- radiating surface. Equation (68) and (69) together
constitute a set of (N + M) simultaneous algebraic equations which can be solved
to determine the radiosities J , J … , J , … , J , J , … , J of the surfaces. Hence, the
net rate of heat transfer Q , can be calculated.
It is to be noted that for re- radiating surface
E , −J =0 (70)
⟹ E , = σT = J (71)
Once the radiosity of a re-radiating surface is known its temperature can be easily
calculated from eqn. 71.

Use of the Network Diagram to Calculate radiation Exchange:


The radiation network diagram for a three- surface enclosure consisting two active
and one re- radiating surfaces is shown in the fig. 11.
The nodes in the diagram are denoted by J , J , J (a node is a point where two or
more heat currents meet). The algebraic sum of the ‘heat currents’ at each of the
nodes is zero as represented by the following equations.

Fig. 11

82 |
= + + (72)

= + + (73)

0= + + (74)
For example, from eqn. 72, we have
+ + (75)
The LHS of eqn. 75 is the sum of all radiation currents reaching the node, J . The
total resistance between J and J can be obtained from fig. 11 in which the
resistances are in series- parallel arrangement, and is given by
=

The total resistance between the points E and E is


+ + (76)

Hence the rate of heat flow from surface 1 to surface 2 is given by


Q , = (77)
.

Radiation shield:
A radiation shield is a barrier wall of low emissivity placed between two walls in
order to reduce the exchange of radiation between them. Radiation shields do not
add or remove any energy from the system. They essentially put additional
resistances to the transfer of radiation energy as shown in fig. 12.

Fig. 12

83 |
Let us determine the benefits in terms of reduction in the rate of radiation exchange
that may be reduced by using a radiation shield. We assume two walls are ‘large’
and have temperatures T and T and emissivities ε and ε respectively. If T >
T , it is expected that there will be a net loss of radiation by wall-1 and net gain by
wall-2. The radiation shield will have an intermediate at steady state. Neglecting
conduction and convection, the net rate of radiation energy loss per unit area of
wall–1, in the absence of shield is
,
= (78)
In the presence of the radiation shield, at steady state, the net rate of transport of
radiant energy from wall-1 to the shield will be equal to that from the shield
surface to wall-2. We may write these rates per unit area as follows (let the shield
be surface 3).
= = = (79)
Eqn. 79 may be solved for T and the net rate of radiation energy loss from wall-1
may be determined. Then the % reduction in radiation exchange in the presence of
shield may be calculated.
As a special case, if we assume, ε = ε = ε = ε, we have
=

⟹T = (80)
Again from eqn. 79 we have (using eqn. 80)

= = (81)
Dividing eqn. 81 by eqn. 78 (with, ε = ε = ε) we have
( )
( )
= (82)
Therefore, there will be 50% reduction in the radiation energy transfer by putting a
radiation shield between the walls, if the emissivities of the walls and the shield are
all equal. The result does not depend upon the exact position of the shields between
the walls.
Following the logic of induction it can be shown for N identical shields between
wall 1 and 2 (so that, ε = ε = ε = ⋯ = ε ). We will have
= (83)

84 |
Problems to be solved:
1. An insulated cylindrical pipe of 0.2m diameter has a surface temperature of
450C. It is exposed to black body surrounding at 250C. The emissivity and
absorptivity of the insulation surface are 0.96 and 0.93 respectively. The
convective heat transfer coefficient outside the insulation surface is 3.25 . The
Stefan-Boltzman constant is 5.67× 10 . The surrounding fluid may be
assumed to be transparent. The % contribution from radiation to the total heat
transfer rate to the surrounding will be
(a) 30.9 (b) 50.0 (c) 57.6 (d) 68.4 (2006)
2. A fluid flows through a cylindrical pipe under fully developed, steady state
laminar flow conditions. The tube wall is maintained at constant temperature.
Assuming constant physical properties and negligible viscous heat dissipation,
governing eqn. for the temperature profile is (z, axial direction; r, radial direction)
(a) u 1− = r +
(b) u 1− = r +
(c) 2u 1− = r +
(d) u 1− = z + (2006)
3.
A well insulated hemispherical
furnace (radius =1m) is shown in
fig. The self view factor of
radiation for the curved surface 2
is
(a) (b)
(c) (d) (2009)

4. The view factor matrix for two


infinitely long co-axial cylinders,
shown in the fig. is
0 1 0 1
(a) (b)
0.5 0.5 1 0
1 0 0.5 0.5
(c) (d)
0 1 0 1
(2010)

85 |
Common statement for Q. 5 and 6:
Hot oil at 1500C is used to preheat a cold fluid at 300C in a 1:1 heat exchanger. The
exit temperature of the hot fluid is 1100C. Heat capacities (= ṁC ) of both streams
are equal. The heat duty is 2kW. (2010)
5. Under co-current flow conditions, the overall heat transfer resistance is (in,

)
(a) 0.4 (b) 0.04 (c) 0.36 (d) 0.036
6. Under counter-current flow condition, the overall transfer resistance is (in,

)
(a) 0.4 (b) 0.04 (c) 0.36 (d) 0.036
7.
For the enclosure formed between
two concentric spheres as shown
below, (R = 2R ), the fraction
of radiation leaving the surface
area A that strikes itself is
(a) (b)
(c) (d) (2012)

8. Show the view factor F between two


infinite parallel plate just one above
other (see fig. I) is

F = 1+ −

Fig. I

86 |
When one parallel plate is
shifted by a distance L (fig.
II) with respect to the other,
show that

F = 1+ − −

1+

Fig. II

9.

Fig. III
For the configuration shown in fig. III, calculate, F .

10. A hemispherical furnace of 1m radius has the inner surface (ε = 1) of its roof
maintained at 800K, while its floor (ε = 0.5) is kept at 600K. Calculate the net
Radiative heat transfer from the roof to the floor.

87 |
Natural/ Free Convection:
Natural or free convection flow arises when a heated object is placed in a quiescent
fluid (a fluid at rest), the density of which varies with temperature. The density in
the neighborhood of the surface of the object decreases, which in a normal fluid, is
associated with temperature increase. This causes the layers near the surface to rise
causing the surrounding cold fluid to take its place. This, in turn, causes a free
convection flow, which transports the heat away from the surface. The reason of
such a flow is a body force. Buoyancy force is the primary driving force for free
convection. And it is well known that buoyancy force is originated due to the
combined presence of a density gradient within the fluid and a body force which is
proportional to the fluid density.
Typical applications of natural/ free convection are as follows:
(i) Heat transfer from pipes and transmission lines as well as from various
electronic devices.
(ii) Dissipation of heat from the coil of a refrigeration unit to the surrounding air.
(iii) Heat transfer from a heater to the ambient air.
(iv) Heated and cold enclosure.
(v) Quenching, wire drawing etc.
(vi) Atmospheric and oceanic circulation.
Free convection flows at or over a heated surface–– for example, a wall, a wire or a
cylinder –– occurs in the form of a plume. The plume width varies with position as
shown in fig. 1.

Fig. 1

88 |
However, when a fluid is enclosed between two horizontal surface (at a distance L
apart), the lower surface being hot and the upper surface cold, heat transfer occurs
practically by pure conduction when the Grashof number, Gr (a dimensionless
group characterizing free convection, will be discussed later) is less than about
1700. Above this value of Gr , free convection begins and a pattern of hexagonal
prismatic cells is formed as shown in fig. 2.
At the middle of each prism, the
warmer fluid ascends; the colder
fluid descends along the ‘walls’
of the hexagonal prism. The top
surface looks like the cross
section of a honeycomb. This
phenomenon was first observed
by Benard in 1901 while
heating a thin layer of liquid
from below. These free
convection cells are known as
Benard cells. Other common
situations of free convection
heat transfer are from a
Fig. 2 horizontal (or inclined) plate,
hot or cold, facing upwards or downwards. Free convection in a vertical or inclined
enclosure made up of two parallel, cylindrical or spherical surfaces is also
important from practical angle. The natural flow patterns in a few common
geometries are qualitatively shown in fig. 3

Fig. 3
89 |
Free Convection from a Vertical Plate:
Let us now consider the natural convection heat transfer from a heated vertical
plate placed in an extensive quiescent medium (fig. 4). Since, T > T , the fluid
adjacent to the vertical surface gets heated, becomes lighter and rises. The fluid
from the neighborhood rushes in to take place of this rising fluid.
Eventually, a flow of the form
shown in fig. 4, known as boundary
layer flow, develops adjacent to the
vertical surface. The analysis and
study of such a flow gives the
desired information on heat transfer
rates, flow field, temperature field
etc. The x-component velocity is
positive and increases upwards.
The y-component velocity is
negative and decreases upward.
The u- velocity is zero at the plate,
and at the edge of the boundary
layer, (contrast this with forced
convection, where the velocity is
maximum at the edge of the
Fig. 4 boundary layer).

The u-velocity reaches the maxima, in the boundary layer. The temperature
decreases monotonically from the wall towards the bulk fluid in the surrounding.

Analysis:
To analyze the boundary layer shown in fig. 4, we make the following assumptions
(i) The flow is steady, laminar and 2D.
(ii) The temperature difference between the plate and the fluid is small to
moderate, in which case the fluid may be treated as having constant properties.
(iii) We consider Bossinesq approximation to be valid.
(iv) Boundary layer approximations are valid.

Governing Equations:
The X momentum equation reduces to
u +v =− +ν + (1)
The body force per unit volume is,
X = −ρg (2)

90 |
Therefore, eqn. 1 becomes
u +v =− −g+ν (3)
Invoking boundary layer approximation, we can write
=0 (4)
Furthermore, the x-pressure gradient at any point in the boundary layer must be
equal to the pressure gradient in the quiescent region outside the boundary layer.
However, in this region, u = v = 0. Therefore,
= −ρ g (5)
Hence
u +v = − (−ρ g) − g + ν
⟹u +v = (ρ − ρ) + ν (6)
Eqn. 6 must apply at every point in the momentum boundary layer. The first term
on the RHS of eqn. 6 is the buoyancy force, and the flow originates because the
density ρ is a variable. This is related to the temperature difference by using the
definition of volumetric thermal expansion coefficient, β, which is
β=−   (7)
Now β may be expressed in the following approximate form
β=− (8)
P is the reference pressure level, i.e. , pressure at the bottom of the plate. For most
natural convection flows, the pressure correction is not required. Using eqn. 6 and
8 we obtain
u +v = gβ(T − T ) + ν (9)

Inertia Buoyancy Viscous


The presence of temperature in the buoyancy term of the momentum equation
(eqn. 9) compels the flow to the temperature field and vice versa. The overall mass
and energy conservation equation remains unchanged. Finally, the set of governing
equations may be expressed as follows:
Continuity: + =0 (10)
x– momentum: u +v = gβ(T − T ) + ν (11)
Energy: u +v =α (12)

91 |
Note that the viscous dissipation term has been neglected in the energy equation
because of small velocities associated with free convection. Free convection effects
obviously depends on the expansion coefficient, β. The manner in which β is
obtained depends on the nature of the fluid. For a perfect gas β = (ideal gas) and
T is taken as .

Non-dimensionalization of The Momentum and Energy Equation:


Using the following dimensionless parameters, let us now non-dimensionalize the
momentum and energy equations:
x ∗ = , y ∗ = , u∗ = , v ∗ = and T ∗ = (13)
L is the characteristic length (in this case, the length of the plate) and u is an
arbitrary reference velocity.
∗ ∗ ( ) ∗
x momentum: u∗ ∗
+ v∗ ∗
= T∗ + . ∗
(14)
∗ ∗ ∗
Energy: u∗ ∗
+ v∗ ∗
= . ∗
(15)
In eqn. 14, the coefficient of T ∗ can be expressed as
( )
=
where Re = is the Reynolds number based on arbitrary reference velocity and
( )
Gr = is the Grashof number.
×
( ) ( ) ×
It is to be noted that, Gr = = = ( )
.
.
As in natural convection inertial force is in the order of viscous force, we may
write, Gr = .
So basically represents the ratio of buoyancy force to inertia force. Therefore,
in a flow where both forced and free convection effects are significant (mixed
convection)
Nu = f(Re , Gr , Pr), so if, = 1 (mixed convection), for, ≪ 1 (forced
convection) and if, ≫ 1, (free convection).
It is to be noted that larger be the temperature difference between the wall and the
fluid, greater be the induced flow. When the flow velocity is large enough the free
convection boundary layer changes from, laminar to turbulent. This transition has a
strong influence on the rate of heat transfer. The condition under which the

92 |
transition takes place is given by a critical value of a dimensionless number called
Rayleigh number, Ra .
Rayleigh number is defined as (where L is the characteristic length)
( )
Ra = Gr . Pr = (16)
For the case of a vertical surface critical Rayleigh number is taken as, Ra ≈ 10 .
Above this value of Rayleigh number the flow changes from laminar to turbulent.
However, the flow may remain laminar over a part of the surface, and turbulent
flow over the rest. The distance from the bottom edge (if the surface is hot) where
transition occurs is given by local Rayleigh number, Ra = 10 . A few important
correlations for free convection heat transfer from surface of common geometries
are discussed below. These correlations are usually of the form, Nu = f(Ra, Pr),
and the relevant physical properties are taken at the mean film temperature.
Free convection from a flat surface:
(i) For a vertical plate, hot or cold, the following eqn. may be used for the entire
range of Rayleigh number
⎡ ⎤
⎢ . ⎥
Nu = = ⎢0.825 + ⎥ (17)
⎢ . ⎥
⎣ ⎦
The equation is applicable for an inclined plate if the angle of inclination with the
vertical, θ < 60°. Here h denotes the average heat transfer coefficient and Nu is
the average Nusselt number.
(ii) Different flow situations arise when heat transfer by natural convection occurs
from a horizontal plate depending upon whether heat transfer surfaces faces up or
down. The nature of flow in such cases has been shown qualitatively. A
characteristic length in all the cases is defined as: L = .
A few useful correlations are given below
(a) Heat transfer from the upper surface of a flat plate
Nu = = 0.54Ra (18)
for Ra ⇒ [10 , 10 ]
(b) Heat transfer to the lower surface of a cold plate
Nu = = 0.15Ra ; 10 ≤ Ra ≤ 10 (19)
(c) Heat transfer from the lower surface of a hot plate or to the upper surface of a
cold plate
Nu = = 0.27Ra ; 10 ≤ Ra ≤ 10 (20)

93 |
Free convection from a cylinder:
(i) For heat transfer from or to a horizontal cylinder, Morgan (1975) recommended
a correlation of the following form
Nu = = C. Ra (21)
where d is the diameter of the cylinder. The constant C and exponent n vary with,
Ra .
(ii) In the case of a vertical cylinder, the eqns. for the vertical plates are applicable
if the thickness of the free convection boundary layer is much smaller than the
cylinder diameter. The cylinder may be considered to be equivalent to a vertical
plate of breadth equal to the circumference of the cylinder. The criterion for this is
quantitatively given by
≥ (22)

Here L is the height of the vertical cylinder. For smaller cylinder, we have to solve
the problem numerically.
Free convection from a sphere:
The following correlations suggested by Churchill may be used
.
Nu = = 2+ ; (23)
.

Pr ≥ 0.7 , and, Ra ≤ 10

Combined Free and Forced Convection:


There are practical situations in which contributions of both free and forced
convection should be taken into account for heat transfer calculations, ~1 . If
heat transfer occurs in the mixed convection regime, the following eqn. may be
used to calculate the Nusselt number.
Nu = nu ± Nu (24)
where Nu and Nu are Nusselt number for forced and free convection. A
value of m=3 is usually recommended. Positive or negative sign is taken depending
upon whether the free convection flow occurs in the same or in the opposite
direction of the forced convection flow.

94 |
BOILING:
The course of heating of a liquid first leading to evaporation and then to boiling
can be followed quantitatively if the temperature of the hot surface and
consequently the heat flux is varied systematically over a wide range. There are
quite a few classical works on the boiling phenomenon (e.g. Nukiyama, 1934;
Farber and Scorah, 1948 etc.). An electrically heated surface (wire) was used by
many workers because of the ease with which the rate of heat input and the
temperature can be controlled.
An experimental arrangement is described here briefly. A schematic of the set-up
is shown in fig. 5.
An electrically heated wire
immersed in the liquid supplies
the heat required for boiling.
The rate of heat input is
calculated from the measured
values of the applied voltage
and the current. The
measurement also gives the
resistance of the wire and hence
the wire temp. T .
Fig. 5
The surface heat flux q can be calculated from the rate of electrical power input
and the area of the wire. A set of temperature- flux data can be generated by
changing the applied voltage over a reasonably wide range.
In a boiling liquid, the temperature of the hot surface must be higher than the
boiling point of the liquid, i.e. T > T , where T is the temperature of the
saturated liquid or the boiling point of the liquid. The temperature driving force for
the boiling liquid which is also called the excess temperature is defined as follows
excess temperature, T = T − T (25.a)
A plot of the logarithm of the heat transfer coefficient at the wire surface (h ) vs.
the excess temperature (T ) typically looks like fig. 6. The modes or regimes of
boiling over different ranges of excess temperature are identified in this figure. It is
to be noted that these results are characteristic of pool boiling. Pool boiling refers
to boiling of a quiescent liquid in which the motion is caused by free convection
and by the formation, growth, detachment and rise of the bubbles. Pool boiling of
the liquid occurs in the experimental arrangement shown in fig. 5. There is another
kind of boiling, called forced boiling or forced convection boiling in which motion
in the boiling medium is caused by an external means (like a pump) in addition to
the factors involved in pool boiling.

95 |
Fig. 6
Over the section A–B (regime I) in fig. 6, the wire temperature is slightly above the
saturation temperature of the liquid, (T ≤ 2℃). The rate of vaporization of the
liquid or the rate of formation of vapor bubble is pretty small. Motion in the liquid
medium is caused principally by free convection. The hot liquid vaporizes only at
the free top surface. This regime is called the interfacial evaporation regime.
The section BD is characterized by nucleate boiling. Here the excess temperature
varies from T = 2℃ at B to T = 35℃ at D. The nucleate boiling regime may be
subdivided into two smaller regimes. Over the section B–C (T ≈ 2 to 60C),
representing regime II, isolated bubbles are formed at the surface of the wire, but
most of them collapse before reaching free surface of the liquid. Because bubble
formation at a reasonable rate starts at B, the point indicates the onset of nucleate
boiling. Over the remaining section, C–D, which is regime III, a large number of
vapor bubbles are generated and vigorously rise through the liquid and escape at
the free surface. Near point D, the vapor bubbles rise as jets or columns and then
from bigger bubbles slugs. Breakage and coalescence of vapor bubbles also occur
because of intense motion generated in the liquid.
When the excess temperature exceeds that at point D, the mode of boiling
gradually changes from nucleate to film boiling. Here the bubbles are formed in so
large numbers that they coalesce right on the heating surface to form a film of
vapor. This gradually reduced the heat transfer coefficient. The film boiling regime
96 |
may be subdivided into regimes IV, V, and VI. In regime IV (section D–E) the
formation of the vapor film on the wire surface is not complete, and the film is
discontinuous. The film may even disappear momentarily and then reappear. This
regime represents partial nucleate boiling or transition boiling. Beyond the point
E, start the regime V (section E–F) where stable film boiling occurs and a
continuous vapour film always blankets the heating surface. The surface heat flux
decreases significantly although the excess temperature may be as high as a few
hundred degrees. After the point F, the wire temperature increases so much that
Radiative heat transfer becomes important. This is characteristic of regime VI
(section F–G). If the excess temperature is too high, the heating wire may even
melt. The melting or burn out of heat transfer surface is sometimes called boiling
crisis. In some cases a direct shift from point D to point G may occur as
temperature increases. The excess temperature and heat flux in some electrical
heating devices or in a nuclear reactor may be so large that it may lead to boiling
crisis. This possibility should be taken into account at the design stage. The boiling
phenomenon in the different regimes is also shown in fig. 7.

Fig. 7
It is pertinent at this point to mention the so called Leidenfrost phenomenon. When
water droplets fall on a sufficiently hot plate, the droplet bounces up and down few
times before they disappear by vaporization. The above phenomenon was observed
by Leidenfrost in 1756. The mechanism is related to film boiling. When a water
droplet lands on a hot surface, a film of steam immediately forms between the
droplet and the hot surface. Because the thickness of this vapour film increases
very quickly by further vaporization of liquid, the droplet experiences an upward
thrust and it bounces. It falls back on the surface and bounces again by the same
97 |
mechanism and reduces in size at every touch with the surface. The droplet
disappears eventually. So the Leidenfrost phenomenon is caused by film boiling
and occurs for evaporating droplets in a hot surface corresponding to regime V of
the boiling curve.

Hysteresis in the Boiling Curve:


This phenomenon was first reported by Nukiyama (1934) in his experiments on
pool boiling of water using a hot nichrome wire. Nukiyama reported a boiling
curve shown in fig. 8 in which surface heat flux is plotted vs. the excess
temperature on the logarithmic scale. Although he did not identify the different
regimes, he observed a hysteresis phenomenon in the boiling curve.
As the excess temperature T
increased, the surface heat
flux q followed the path A–
B–C reaching a maximum at
C. on further increase in the
power input to the wire, the
excess temperature suddenly
jumped to a very high value
at E following the path C–E.
When the power input to the
wire was reduced, the drop
in T and the flux q
followed the path E–D–B–A.
Had there been no jump in
Fig. 8
the temperature after the point C during heating, the boiling curve should have
been like A–B–C–D–E, although C–D represents an unstable regime. The point C
in fig. 8 corresponds to the maximum heat flux or the critical heat flux. The
Leidenfrost phenomenon of an evaporating droplet also occurs at the point C.

The Mechanism of Nucleate Boiling:


Nucleate boiling proceeds through the formation of bubble nuclei in a superheated
liquid, growth of the bubbles and their escape from the liquid. There are quite a
few questions associated with this phenomenon. Does bubble form in the bulk of
the superheated liquid or on the hot surface with which the liquid is in contact?
How to determine the rate of bubble growth? The answers are as follows.
Formation of tiny bubble in a liquid is called nucleation. A superheated liquid is
said to be in a metastable condition i.e. it cannot exist in this condition indefinitely
and any disturbance in superheated liquid may be considered as a source of
98 |
disturbance. If a bubble nucleus is formed in the bulk of a superheated liquid, it is
called homogeneous nucleation. On the other hand, if a nucleus is formed on a hot
surface or on a solid particle suspended in the superheated liquid, we call it
heterogeneous nucleation.
The condition of mechanical equilibrium at a liquid- vapour interface is given by
Gibb’s equation
P −P =σ + (25.b)
where P and P are the pressures exerted by vapour and liquid phases at the
interface, σ is the interfacial tension, and r and r are the principal radii of
curvature of the interface. If the interface is spherical, i.e. r = r = r, then eqn.
(25.b) reduced to
P −P = (26)
It is clear that, P > P . The pressure difference can be related to the liquid
superheat by using Clausius- Clapeyron equation. Thus we have
= (27)
where L is the molar heat of vaporization and R is the gas constant. Let us assume
that T is the temperature of the superheated liquid, P is the pressure of the
superheated liquid and T is the corresponding boiling point of the liquid. The
vapour in the bubble is in thermal equilibrium with the liquid, i.e. the vapour
temperature is also, T . As the vapour is saturated, the pressure in the eqn. 27 is
within the appropriate limits. Thus we have
∫ d ln P = ∫
Or ln = −
Because ln = ln 1 + ≈
Therefore, =
Substituting for P − P from eqn. 26, we get
T −T = .

⟹T −T ≈ . (28)
The radius of a vapour bubble in mechanical equilibrium in a superheated liquid
can be determined from the above equation. For example, if we consider a pool of
water with 100C superheat at atmospheric pressure, we have
T = 110℃, T = 100℃, P = 1.013 bar, σ = 0.063 , L = 4.07 × 10
R = 8.314
.

99 |
Therefore
× . . ×
T −T = 10℃ = ( . )
×
× . ×
⟹ r = 3.55 × 10 m
If a bubble nucleus of a radius smaller than above, say, 10 m, is generated in the
liquid by some way it will collapse. Because the nucleus cannot remain in
mechanical equilibrium, if the superheat is only 100C, eqn. 28, thus explains why
some tiny bubble collapse in liquid if the degree of superheat is insufficient.
Eqn. 28 also tells us that a high degree of superheat is necessary for the generation
of a tiny nucleus in the bulk liquid. However, our experience shows that a liquid
boils vigorously even at a superheat of several degrees. The reason is that, bubble
nuclei are formed by heterogeneous nucleation at the minute cavities and crevices
present at the heating surface. This is more favored on a rough surface than it is on
a smooth one.

Condensation:
Condensation means the change of phase from the vapour to liquid. If the
temperature of a vapour is reduced below its saturation temperature, condensation
occurs.
If a mixture of a vapour and a gas is cooled, the vapour condenses to form minute
droplets suspended in the carrier gas. This is called homogeneous condensation. In
contrast, if a vapour or a gas-vapour mixture comes in with a cool surface,
condensation occurs on the surface. This is called surface condensation. If the
condensate wets the surface, it flows down as a continuous film. This is called film
condensation. However, if the surface is not wetted by the condensate, drops
(instead of film) appear on the surface, which grow in size and then tickle down
the surface. This is called drop wise condensation. This occurs when the surface is
contaminated or the condensate liquid does not have any ‘affinity’ for the surface.
These two modes of condensation is shown in fig. 9.

Fig. 9

100 |
In film condensation, the latent heat is transferred through the liquid film and then
conducted through the wall to the cooling fluid on the other side of the wall. The
condensate film offers considerable heat transfer resistance. In drop-wise
condensation, on the other hand, a part of the surface is covered by liquid drops
and the rest is directly exposed to the vapour. As a result, heat transfer coefficient
for film condensation is considerably lower than that for drop-wise condensation.

Film Condensation on a Vertical Plate:


A theoretical analysis of laminar film condensation of a vapour on a vertical plate
was given by Nusselt in 1916. This analysis is based in the following major
assumptions.
(i) The condensate film is in ‘locally’ fully developed laminar flow with zero
interfacial shear and constant liquid properties.
(ii) The vapour is saturated.
(iii) Heat transfer through the condensate film occurs by conduction only, and the
temperature profile in the film is linear.
Considering the well-known
Navier-Stokes equation along x
direction, we have
(steady) 0 0(fully developed)
ρ +u +v +w =
ρ = ρg − +μ + +
0 0 0
0
⟹ 0 = ρg + μ (29)
[as g = g = 9.8 ]
Integrating the equation we have
=− y+C (30)
Fig. 10
At y = δ i.e., at the gas-liquid interface, the shear stress, τ =μ is zero,
hence, = 0 at, y = δ, hence C = . δ. Incorporating the value of C in eqn.
30 we have
=− y+ δ

101 |
Integrating again, we have
u(y) = − y + δy + C (31)
Now by no-slip condition at y = 0, u = 0, so
u(y) = − y + δy (32)
eqn. 32 gives the velocity profile in the freely flowing film. The profile is half
parabolic. The flow rate of the condensate (per unit breadth of the film) at any
location x is obtained by integrating eqn. 32 as
ṁ = ∫ u(dy. 1)ρ = ∫ ρ − y + δy dy
⟹ ṁ = (33)
As x increases downwards, the film thickness also increases because of
condensation of more and more vapour. The rate of condensation on an
‘elementary surface’ of size dx × 1 exposed to the vapour can be obtained from the
rate of heat transfer through this area. The temperature profile in the condensate
film is assumed to be linear and we may, therefore, write
The rate of heat transfer = k dx. 1
The change in the condensate film flow-rate over the length dx be
Latent heat of dṁ λ = k dx (34)
condensation
( )
= , let T= T − T
.
( ) 
⟹ . 3δ = =
.

⟹ = (35)
Integrating eqn. 35 and using the boundary condition: at x = 0, δ = 0, we have

= .x

⟹δ= (36)
If h is the heat transfer coefficient for the condensate film, heat flux through the
film at any location x is
h= (37)
Combining eqn. 36 and 37 we have

h(x) = . (38)

102 |
It is to be noted that δ ∝ x and h ∝ x i.e., film thickness increases and local
heat transfer coefficient decreases in the downward direction.
The average heat transfer coefficient over a length L is

h = ∫ hdx = 0.943 (39)



While using eqn. 39, the liquid properties should be taken at the mean film
( )
temperature defined as . If condensation occurs on an inclined surface, g
in the above equation should be replaced by g cos θ, where θ is the angle of
inclination of the surface to the vertical.

Condensation outside a Horizontal Tube or a Tube Bank:


Condensation outside a vertical tier of horizontal tube occurs in many condensation
units. The possible physical picture of the condensation phenomenon over a tube
bank is shown in fig. 11. If there is a single horizontal tube, condensate flows as a
film along the cylindrical surface (fig. 11a). If there is another tube below, the
condensate film flows down from the bottom edge of the upper tube to the upper
edge of the lower tube (fig. 11b).

Fig. 11

This goes on from an upper tube to the lower of the tube tier, and the thickness of
the condensate film grows. It may also happen that the condensate from the bottom
edge of the upper tube drips down to the lower tube (fig. 11c).

103 |
In order to determine the average heat transfer coefficient for laminar film
condensation on a horizontal tube, let us consider fig. 12, which shows a
condensate film over a horizontal tube.
An energy (heat) balance on the
condensate film between θ and
θ + dθ per unit length of the
cylinder gives
( )
λ dṁ = dx (40)
Here in deriving eqn. 40, we
have neglected the curvature
effect.
Noting that, θ = , we get the
following form as
̇ 
= (41)
The component of gravity
acting tangentially to the tube is
g sin θ.
Fig. 12
Thus, with the parabolic velocity distribution used for a vertical plate, replacing g
by g sin θ, the condensate mass flow rate is given by
ṁ = ρ ∫ u. 1. dy (42)
u is to be substituted in the above expression from eqn. 32 (replacing g by g sin θ).
Therefore
ṁ = (43)
Also

δ=
Putting, x = Rθ , we obtain

δ= (44)
Combining eqn. 41 and 43 we get

ṁ dṁ = . sin θdθ (45)

104 |
Integrating eqn. 45 from θ = 0 to θ = π, we get the condensate production from
one side as

ṁ = 1.924 (46)
An energy balance on the entire tape yields
2λṁ = h 2πr. 1(T − T ) (47)
Average heat transfer
Coefficient for a single tube

Therefore, from eqn. 46 and 47, we obtain

h = 0.728 (48)

where D (=2R) is the outer diameter of the tube.
Now condensers have banks of horizontal tubes in a vertical row as depicted in fig.
11b. For n horizontal tubes in a vertical row, the average heat transfer coefficient
for a single tube array is
h =h n (49)

Drop-wise Condensation:
In drop-wise condensation, the vapour condenses on tiny condensate nuclei on the
surface. These nuclei grow in size to form drops. The bigger drops run down the
surface and further condensation occurs on the nuclei. The nuclei originate in
cracks and pits on the surface. The condensate drops occupy 90% of the total
surface area.
As stated before, the condensation heat transfer coefficient is usually very high for
drop-wise condensation, and contributed very little towards the total heat transfer
resistance. In most practical situations this contribution can be safely neglected.
Drop-wise condensation of steam is most common and the presence of non-
condensable gases drastically reduces the rate of condensation.

Problems to be Solved:
1. Water is boiling in a pan at atmospheric pressure. If the water has a 60C
superheat, calculate the pressure inside a vapour bubble of 5mm diameter.
2. If condensation of a saturated vapour occurs on the outside of a tube under ‘zero
gravity’ condition, a condensate film will form on the surface but will not run off.
Show that in such a condition if a pseudo steady state approximation is done, the
heat transfer coefficient is given by
h= ( )

105 |
where R is the radius of the tube and R is the outer radius of the film at time t and
is given by
+ ln − = t
Common statement for q.3 and 4: (2006)
In a film condensation on a vertical plate,
the x directional velocity distribution is
given by
( )
u(y) = δy − y
where δ is the film thickness at any x.
3. The mass flow rate of the condensate
m(x) through any axial position x per unit
width of the plate is given by
( ) ( )
(a) m(x) = (b) m(x) =

(c) m(x) = (d) m(x) =


4. Differentiate m(x) with respect to δ to get the differential increase in condensate
mass dm with film thickness i.e., . Then obtain assuming heat flux through
the film to be due to conduction based on linear temperature profile between the
vapour and the wall. Hence determine, . Here μ is the liquid viscosity, k is
thermal conductivity and λ is latent heat of condensation. T is the vapour
temperature and T is the wall temperature.
( ) ( )
(a) = ( )
(b) = ( )
( ) ( )
(c) = ( )
(d) = ( )
5. Consider a liquid stored in a container exposed to its saturated vapour at
constant temperature, T .
The bottom surface of the container is
maintained at constant temperature,
T < T , while its side walls are
insulated. The thermal conductivity, k ,
of the liquid, its latent heat of
vaporization λ and density ρ are known.
Assuming a linear temperature
distribution in the liquid, the expression
for the growth of the liquid layer δ as a
function of time is given by
106 |
( ) ( )
(a) δ(t) = t (b) δ(t) = t
( ) ( )
(c) δ(t) = t (d) δ(t) = t
6. A 4cm diameter shell ball ρ = 7800 , C = 473 at 900C is superheated

0
in still air at 25 C. If the ball cools down by free convection heat loss only,
calculate the time required for the temperature of the ball to drop down to 35 0C.
7. A laboratory water bath has an immersed horizontal heating element, 2.54cm in
diameter and 30cm in length, with a power input of 500W. If the bulk water
temperature is 380C and heat transfer occurs by free convection only, calculate the
surface temperature of the heater.
8. Saturated steam condensed at 1 atm pressure on a vertical wall, 1m high,
maintained at a uniform temperature of 780C. If film condensation occurs,
calculate the following at a location 0.5m below the top of the plate:
(a) The film thickness
(b) The average velocity of the condensate, and
(c) The local heat transfer coefficient.
Also, calculate the average heat transfer coefficient over the total length of the
plate.
9. Calculate the rate of condensation of saturated acetone at 2bar total pressure on a
tier of six horizontal tubes, given, T = 40℃.

107 |
Evaporation and Evaporators:
Evaporation ordinarily means vaporization of a liquid or that of a solvent from a
solution. In chemical engineering terminology, evaporation means removal of a
part of the solvent from a solution of a non-volatile solute by vaporization. The
objective of vaporization is to concentrate the solution. It is one of the most
fundamental operations in the process industries. Typical example includes
concentration of a cane sugar juice in a sugar industry, concentration of an aqueous
solution of (NH4)2SO4 in a fertilizer plant, concentration of a dilute recycled NaOH
in an alumina plant and many others.

Types of Evaporators:
Evaporators used in process industry mostly have tubular heating surface. An
adequate number of tubes are provided through which the solution circulates or
may pass through the tube just once. Following tree diagram represents the general
classification of evaporators.
Evaporator

Once through Circulation-evaporator


evaporator

Natural circulation Forced circulation


Falling Rising Agitated evaporator
evaporator
film film thin-film (liquid circulation
(avg. velocity of liquid
evaporator evaporator evaporator velocity = 16–20 ft/s)
in the tube= 1–4 ft/s)

Horizontal tube (only Calandria/ or Basket type Long tube


evaporator with steam short tube vertical vertical
in tube side, now-a- vertical evaporator evaporator
days obsolete) evaporator (LTV)
(STV)

With outside With inside With outside


vertical heating vertical heating horizontal heating
element element element

108 |
We will now describe one of evaporators in detail construction from each group.

(1) Natural Circulation: STV or Calandria Evaporator:


A short tube vertical evaporator
(STV) has a short tube bundle
enclosed in a shell. This is called
a Calandria (fig. 1). The
Calandria is of annular
construction. The feed is supplied
to the through a nozzle above the
upper tube sheet and saturated
steam (generally,𝑃 < 4𝑎𝑡𝑚)
is supplied to the shell side of the
Calandria. The solution is heated
and partly vaporized in the tube.
The liquid flows up because of
the prevailing density gradient.
The liquid flows down through
the central opening of the
Calandria, known as downtake or
downcomer (generally of cross
sectional area=40% of the total
tube area). Thus a continuous
recirculation is established.
Fig. 1
Thick product liquor is withdrawn from the bottom of the evaporator as shown in
fig. 1.
Tubes of 50 mm to 76 mm (2” to 3”) diameter and 1.2 to 2 m length are commonly
used. We may vary the cross sectional area of the downtake from 40% to even
100% of the total tube flow area. The steam condensate leaves through a drain
nozzle connected to a steam trap. A bleed or vent line is provided in the shell for
the release of the non-condensable in the steam. There is a vapour outlet at the top
of the evaporator body. An entrainment separator (will be discussed later) in this
line, separates out the liquid droplets entrained in the vapour stream.
Short tube vertical evaporators are used to concentrate a variety of solutions. A
common example is concentration of sugar solution. However, these evaporators
are not suitable for solutions in which precipitation or salting out of solid may
occur. The problem may be overcome by installing a propeller in the downtake
pipe to increase the circulation rate. This construction is called a propeller
Calandria.
109 |
Basket type vertical evaporator has operational features similar to those of STV
type. However, the heating element is a tube bundle with fixed tube sheets welded
to the shell pieces, thus forming a basket. The baskets are welded to the evaporator
body and the tube bundle, in turn; is bolted to the brackets, thus allowing removal
of the bundle by opening the flanged top of the evaporator for the purpose of
maintenance and cleaning, which is the main advantage of basket type evaporators
over STV. Other features are similar to those of the STV evaporator.
There is another widely used natural circulation evaporator, known as Long Tube
vertical (LTV) evaporator. It has a long vertical tube bundle fitted with the shell
(25 to 50 mm diameter, 5–10 m long tubes are generally used). The shell is
projected into a large diameter vapour head at the top. The feed enters the tube
bundle at the bottom, flows through the tube, while undergoing vigorous boiling
and discharges into the vapour head and impinges on a deflector plate above the
top end of the tube bundle. This removes most of the entrained liquid droplets from
the vapour. Generally, with the above stated construction LTV seems to be an once
through evaporator, with concentrated liquor being withdrawn through a pipe
fitted in a vapour chamber. Now if a part of the liquid leaving the vapour head is
re-circulated via the feed line it becomes a long- tube re-circulation evaporator.
The principal advantages of the LTV evaporators are: high heat transfer
coefficient, low cost, low liquid hold- up and less floor space requirement. The
disadvantages are: high headroom requirement, unsuitable for viscous and scale-
forming materials.

(2) Falling-Film Evaporator (Once-Through Evaporator Group):


In a falling-film evaporator (fig. 2a), the liquid flows down the inner walls of the
tubes in a vertical tube bundle. The tubes are heated by condensing steam or any
other hot liquid on the shell side. As the feed liquor flows down the tube wall, the
water vaporizes and the liquid concentration gradually increases. The thick liquor
is withdrawn from the bottom and the vapour goes to separator where the entrained
liquid droplets are separated.
The most important factor for satisfactory operation of a falling film unit is liquid
distribution. In one arrangement, each of the tubes has a number of notches,
usually rectangular, cut at the periphery at the top end. The notched part of a tube
projects above the tube sheet (fig. 2b). The feed enters at a point above the upper
tube sheet and forms a liquid pool on it. An impingement plate may be provided at
the feed inlet. The feed liquor enters the tube through the notches or weirs and
forms a falling film in each of the tubes. Other types of distributors are also used.
The number of tubes in the bundle is determined by the liquid rate. A minimum
liquid rate should be maintained in order that a continuous film is formed instead

110 |
of rivulets. The film Reynolds number [Re = , where Γ is the liquid rate per unit
circumference of a tube and μ is the liquid viscosity] should be above 2000. At
higher Re ripples appear on the free surface of the film. The liquid rate must be
high enough to ensure complete wetting of the tube walls so that no dry spot is
formed. The minimum liquid rate for complete wetting may be calculated as
Γ = 0.128(μsσ) . (1)
where Γ is in kg/m of circumference s, μ is in kg/m.s and s is the specific gravity of
the liquid and σ is the surface tension in N/m.
The heat transfer rate in a falling film evaporator is pretty high. The contact time
remains low. For this reason, the device is suitable for heat sensitive materials.
However, the solution viscosity must not be high (otherwise the film will be too
thick). The maximum allowable viscosity is 500 cP (and 100 cP< μ < 500 𝑐𝑃).
The device is widely used for concentration of fruit juice and many other solutions
(ammonium nitrate solution). It is not suitable if salting out of the solution occurs.
The vapour passes through a wire mesh entrainment separator and leaves through
the top of the evaporator.
In addition to the falling film evaporator in the ‘once through group’ we have
climbing/rising film evaporator and agitated thin film evaporator.

(a)

(b)

Fig. 2 (2a & 2b)

111 |
A climbing film evaporator looks similar to the long tube vertical evaporator. It has
a vertical tube bundle heater and a vapour chamber. The operation can be easily
understood by referring to the discussion on boiling of a liquid in up-flow through
a vertical tube. After gaining sensible heat at the lower part of the tube, the liquid
starts boiling. If the heat transfer rate is sufficiently high and the tube is long, the
volume fraction of the vapour in the two phase mixture increases and bubbles
forms slugs rising through the tube at high velocity. Thin liquid slugs remain
sandwiched between vapour and slugs. A thin liquid film is dragged along the tube
wall by rising slugs, thus generating a climbing or rising film, as shown in fig. 3.
The heat transfer
coefficient in such a
device is very high.
On the other hand if
the viscosity of the
liquid is very high or
viscosity increases
sharply with
concentration of the
feed, even a forced
circulation evaporator
may not offer an
adequate rate of heat
transfer. A falling
film evaporator is not
suitable either,
Fig. 3 because the film
thickness of a highly viscous liquid becomes high. A solution to the problem is
offered by an agitated thin-film evaporator. It consists of a vertical steam-jacketed
cylinder along the inner surface of which the feed liquid flows as a film. Four
vertical blades mounted on a central shaft continuously agitate the film. The blades
maintain a small clearance from the wall. The action of the agitator greatly
increases the heat transfer coefficient, reduces fouling of the surface and keeps the
film thickness small. Highly viscous, heat sensitive and fouling liquids may be
evaporated in this equipment. Typical application includes concentration of tomato
paste, candies, beer malt, meat extract, tannin extract, gelatin, water soluble
polymers etc. The main disadvantages are high capital cost, moving internals and
high maintenance cost.

112 |
(3) Forced Circulation Evaporator with Outside Horizontal Heating Element:
In a natural-circulation evaporator the liquid enters the tubes at 1 to 4 ft/s velocity.
The linear velocity increases greatly as vapour is formed in the tubes, so that in
general the rates of heat transfer are satisfactory. With viscous liquids, however,
the overall coefficient in a natural circulation unit may be uneconomically low.
Higher coefficients are obtained in forced circulation evaporators, an example of
which is shown in fig. 4.

Fig. 4
Here a centrifugal pump forces liquid through the tubes at an entering velocity of 6
to 18 ft/s. the tubes are under sufficient static head to ensure that there is no boiling
in the tubes; the liquid becomes superheated as the static head is reduced during
flow from the heater to the vapour space, and it flashes into a mixture of vapour
and spray in the outlet line from the exchanger just before entering the body of the
evaporator. The mixture of liquid and vapour impinges on a deflector plate in the
vapor space. Liquid returns to the pump inlet, where it meets incoming feed;
vapour leaves the top of the evaporator body to a condenser or to the next effect.
Part of the liquid leaving the separator is continuously withdrawn as concentrate.
In the design, shown in fig. 4, the exchanger has horizontal tubes and is two pass

113 |
on both tube and shell sides. In others vertical single pass exchangers are used. In
both types heat transfer coefficients are high, especially in horizontal orientation,
as h ∝ and h ∝ (from Nusselt equation), where d and L are

the O.D and length of a tube (with d ≪ L for standard tubes). It is to be noted that
with thin liquids (low viscosity) the improvement with forced circulation does not
warrant the added pumping costs over natural-circulation, but with viscous
material the added costs are justified, especially when expensive metals must be
used. An example is caustic soda concentration, which must be done in nickel
equipment. In multiple-effect evaporators producing a viscous final concentrate the
first effect may be natural- circulation units and the later ones, handling viscous
liquid are forced circulation units. Because of high-velocities in a forced
circulation evaporator, the residence time of the liquid in the tubes is short–– about
1 to 3s–– so that moderately heat sensitive liquids can be concentrated in them.
They are also effective in evaporating salting liquids or those that tend to foam.

Evaporator Auxiliaries:
There are quite a few common accessories required for the operation of an
evaporator. If an evaporator is operated under vacuum to reduce the boiling
temperature of the liquid, there must be a vacuum producing device, for example a
vacuum pump or more commonly a steam jet ejector coupled with a barometric
condenser. Surface condensers are also used for vapour condensation. The steam
condensate in the steam chest or shell is drained through a steam trap. An
entrainment separator is used in the vapour outlet line in order to remove the
entrained liquid droplets from the vapour. The construction of some of these
accessories is described below.

(1) Vacuum Devices:


Evaporators, particularly the last one is a multiple effect evaporator, are frequently
operated under varying degree of vacuum. The vapour from the final evaporator is
drawn by putting vacuum. The vapour, which is generally water vapour, is
condensed and the non-condensables, which include dissolve gas in the feed liquor
as well as the air that leaks into the vacuum evaporators, are removed. Vacuum
devices are therefore, the common auxiliaries of evaporators. They are broadly
classified into two categories–– the vacuum pumps and the steam jet ejectors.

(I) Vacuum Pumps:


There are few types of common vacuum pumps. A ‘reciprocating vacuum pump’
works on the principle of one or more pistons in to- and- fro motion in a cylinder

114 |
(s). Both ‘dry’ and ‘wet’ machines are used. In a dry pump, the vapour enters the
pump directly. But a wet pump is provided with a ‘surface condenser’.
A reciprocating vacuum pump can produce reasonably high vacuum and has a
good efficiency.

(II) Steam Jet Ejector and Barometric Condenser:


Installation of a vacuum pump to maintain the desired low pressure by
withdrawing the vapour often proves prohibitive in cost. A steam-jet-ejector
combined with a condenser of the vapour is a simple and cheap method of
maintaining vacuum.

Fig. 5
115 |
Condensation may be done by using a surface condenser or barometric condenser.
The later is more common and cheap device used for maintaining vacuum in air
evaporator.
A barometric condenser coupled with a steam jet ejector is shown in fig. 5. Steam
(HP steam) is issued through the nozzle into the suction chamber at a high velocity.
The pressure in the chamber remains very low as the pressure head of the steam is
converted to velocity head accordingly to Bernoulli’s equation. Steam and non-
condensable are drawn into the suction chamber. The gas-vapour mixture moves
through a Venturi shaped converging-diverging section. The diverging section is
called the diffuser in which the velocity head is converted to pressure head so that
the mixture can be discharged into the atmosphere.
Water vapour with non-condensables, if any, from the evaporator first enters the
barometric condenser as shown in fig. 5. It is a contact condenser, i.e. the vapour
condenses in direct contact with water fed to the condenser. A number of splash
plates are provided in the condenser to enhance condensation. Water leaves the
condenser through a tail pipe or barometric leg of a height about 34 ft. it
discharges into a sump called hot well. The hot well is called because the
temperature of water may be as high as 50 to 55 0C. In an ideal condition the
pressure inside the barometric condenser should be equal to the vapour pressure of
water at the condenser temperature. However, there may be some non-
condensables in the inlet vapour which accumulate in the condenser chamber and
tend to raise the pressure gradually. So, the barometric condenser is connected to
the suction of the steam jet ejector to withdraw the non- condensables and water
vapour continuously. In this way, the vacuum in the condenser and in the
evaporator is stabilized. The vapour and non- condensables sucked from the
condenser are ultimately discharged into the atmosphere.
Steam ejectors may be classified into single stage and multi stage types. Multi-
stage ejectors may be further classified into condensing and non- condensing types.
The single stage ejector is generally recommended for suction pressures from
atmospheric to about 80 mm of Hg pressure (up to 680 mm of Hg vacuum). Multi-
stage ejectors are available in two to six stages. Inter-condensers between stages
condense steam from preceding stage, reducing ‘the load’ to be compressed in the
succeeding stages. Four, five or six stage ejectors are used to achieve very low
pressure.

Steam Trap:
In all the types of steam-heated equipment like evaporators, re-boilers, steam-
jacketed vessels etc., steam condenses after giving away its latent heat. The
condensate must be removed continuously without allowing the steam to blow up.
A steam trap is an automatic valve that allows the condensate to discharge but
116 |
traps (prevents the flow up) steam. It distinguishes between the condensate and
steam on the basis of (i) density difference (ii) difference in temperature (iii)
difference in flow characteristics.
Mechanical steam traps work on the principle of density differences whereas
thermoplastic traps work based on temperature difference.

Entrainment Separator:
In a equipment like an evaporator, vapourizer or reboiler, vapour bubbles burst
vigorously on reaching the liquid surface. In that process, a lot of liquid droplets
are thrown up in the vapour space above the boiling liquid. The droplets vary
widely in size. The bigger one immediately falls back in the liquid. Most of the
small droplets, however, are carried away or entrained by the outgoing vapour.
This phenomenon is called entrainment. The entrained liquid particles must be
removed before the vapour finally leaves the system such that any loss of liquid or
solution is prevented. An entrainment separator is a device that retains the liquid
droplets when the vapour passes through it. The capture of the tiny droplets may
occur through three mechanisms–––
(i) diffusion, (ii) interception, and (iii) internal compaction.
Different types of entrainment separators are in use; two of them are shown in fig.
6. The helical buffle or catchall (fig. 6a) has two turns of helical baffles filled to
the central vapour inlet pipe. While the vapour flows through the baffled annular
space, it attains a swirling motion and strikes the wall. The liquid particles are
mostly retained on the wall. The liquid flows down the wall, accumulates at the
bottom and leaves through a drain.

Fig. 6

117 |
A simpler type of entrainment separator is made of several layers of wire mesh to
form a pad (fig. 6b). This is supported on brackets below the vapour exit pipe. As
the vapour flows through the pad, the liquid droplet impinges on the wires of the
pad, and get separated from the vapour by impaction mechanism. The retained
liquid drips back to the liquid (boiling). The device is also known as demister
(because it removes ‘mist’ or fine liquid droplets). A typical wire-mesh is made
from 0.011 inch wire. Such a mist pad can remove droplets larger than, 6μm in size
at 99% efficiency.

Principles of Evaporation and Evaporators:


A solution boils in an evaporator to give off the vapour, and thereby becomes
concentrated. However, the physical phenomenon and principles involved in the
evaporation operation are more varied than those discussed under section ‘boiling’.
For example, chemical evaporators are often built and operated as multiple-effect
unit for achieving better steam utilization pattern. Solutions often exhibit boiling
point elevation, thus reducing the driving force for heat transfer. It is necessary to
consider these before we go for actual calculations.

(i) Capacity and Economy:


The capacity of an evaporator means its vapourization capacity, i.e. , kg of water
vapourized per hour.
The economy (or steam economy) is the kg of water evaporated from all the effects
taken together per kg of steam fed to the first effect.
For a single effect unit, it is less than unity. For an n-effect unit, the capacity is
roughly n times that of a single effect unit.

(ii) Boiling Point Elevation: (BPE)


Most evaporators produce concentrated solutions having a boiling point
substantially higher than that of water (or solvent) at the pressure prevailing in the
vapour space. This difference in b.ps is called boiling point elevation or boiling
point rise. Consider a strong solution of caustic soda that boils at 1150C at
atmospheric pressure. If the solution is obtained from an evaporator operating
under atmospheric pressure, the BPE of the solution be 1150C– 1000C=150C. So
the BPE of a thick-liquor must be evaluated for evaporator design, because BPE
always reduces the driving force for heat transfer.
BPE calculation via theoretical means is complicated for different concentrated
solutions, particularly for the solution of non-electrolytes. So, experimental BP
data for solutions of different concentrations and pressures are, therefore,
necessary.
118 |
However, in many practical situations, only a limited volume of BP data is
available. The best use of these data can be made by taking help of an empirical
rule, called the “Dühring’s rule”. The rule states that, the BP of a solution (let it be
𝑇 ) of given concentration is a linear function of BP of the solvent, (let 𝑇 ), which
is mostly water.
So
T = aT + b (1)
Thus if we plot T at several pressures against the corresponding, T , we get a
straight line.
A set of straight lines will be obtained if such plots are made for solutions at
different concentrations. However, the lines are not exactly parallel, the slope
being larger at higher concentration. The rule works very good for moderate ranges
of pressures, but does not give satisfactory correlation at a higher pressure.
Nevertheless it is practically useful rule. The main advantage is that if the BP of a
solution at two different pressures is known and the corresponding BPs of water is
read from the steam tables, a ‘Dühring’s line’ through the points can be drawn, the
BP of any solution at any pressure can be estimated by linear interpolation. During
plots for NaOH solution is shown in fig. 7.
(iii) Change of Latent Heat
of Evaporation:
Similar to the change of BP
from the pure solvent (water)
to the solution, latent heat of
vaporization also changes
from pure water (let λ ) to the
corresponding solution at a
definite concentration and
pressure (let it be λ ).
Knowing λ form the steam
table, one can readily calculate
λ using an empirical eqn.
known as Dühring’s equation,
provided.
Fig. 7
We know, T , T (= T + BPE) and the slope of the particular Duhring line

= from which T was estimated using Dühring’s plot knowing T .

119 |
The equation is described as

= . (2)

Inverse of Ratio of T
the slope of and T ′ in
Duhring line absolute scale
λ can also be evaluated by using another equation, known as Othmer equation
(not being described here).

(iv) Correction for Hydrostatic Head (𝐭 𝐡 ):


Using Dühring’s plot one can evaluate the Bp from the solution (T ) knowing pure
water BP (T ) and the concentration of the solution. However, it is to be noted
that this corrected BP (= T ) is with reference to the free surface of the solution.
Actually, the real boiling takes place beneath the free surface, which is specifically
true for circulation evaporators, i.e., inside the tubes of the evaporator. Therefore,
we may infer that vaporization is actually taking place under an increased
hydrostatic head, which leads to a further elevation of BP (= t ). So the actual
BP of water from the solution be
t = t + t (3)
Hydrostatic head
True BP = T + BPE correction

t can be calculated using the following empirical equation


t = 0.06 .z (4)
where t is the BP elevation due to additional hydrostatic head, T : t expressed
in Rankine [T = t (in 0F) +460], v: molar volume of the generated vapour
(ft ⁄lb = mole), λ : latent heat of vapourization (of water) from the solution (in
Btu/lb) and z is the hydrostatic head (additional) due to the depth of boiling spot
wrt the free surface (in ft).
So
t = t + BPE + t (5)

(v) Heat Transfer Coefficient:


The h value (in the shell side) for saturated steam is pretty high. As a result, the
liquid side heat transfer coefficient virtually controls the rate of heat transfer.
However, prediction of liquid side heat transfer coefficient is difficult. The
entering liquid velocity is generally low, and therefore, the liquid side heat transfer
coefficient is generally low. The coefficient increases greatly as the liquid starts

120 |
boiling after reaching a certain height in the tubes. If the evaporator operates at a
very high liquid velocity such that no boiling takes place inside the tubes, we may
write
Inner = 0.0278Re . Pr . (6)
diameter
of the tube
Thermal Reynolds Prandtl
conductivity number number
of the liquid
There is no practically useful method of estimation of tube side coefficient if there
is boiling inside the tubes of the evaporator. Design calculations are mostly based
on data obtained from an operating evaporator. Some typical values of overall heat
transfer coefficients are: Calandria 750– 2500 kcal⁄hr. m ℃ , LTV (natural)
evaporators, 1000– 3000 kcal⁄hr. m ℃ , falling film evaporator,
500– 2500 kcal⁄hr. m ℃ etc.

(vi) Single and Multiple-Effect Evaporators:


The simplest method of evaporation is to feed the solution to the evaporator which
is provided with sufficient heat transfer area. The vapour generated is condensed
using a ‘surface condenser’ or ‘direct contact condenser’. The concentrated
product is drawn from the bottom. This is called a single-effect evaporator.
Although simple in operator, a single effect does not utilize steam efficiently. More
than a kg of steam (1.1 to 1.3 kg is common) is needed in order to vapourize 1 kg
of water from the solution. In order to increase the steam economy, the vapour
generated in one effect (evaporator) is fed to the steam chest of the next
evaporator. As a result, steam economy improves and it becomes greater than one.
The details of multiple effect evaporators’ feeding arrangement will be discussed
later.

Single Effect Evaporator calculation:


The schematic of a single effect evaporator is shown in fig. 8. In order to develop
the governing equations of the system, we assume
(i) Steady state
(ii) No heat loss in the surrounding
(iii) No entrainment of the liquid droplets (y = 0)
(iv) Only latent heat is transferred from the steam side.

121 |
Fig. 8

With the stated assumptions, the governing balance equations are


Overall material balance: F=V+L (7)
Component balance (solute): Fz = Lx (8)
Energy balance equation: Q = ṁ λ = FC (T − T ) + Vλ (9)

Sub-cooled
heat load
Total heat load
Latent heat
load

Capacity equation: Q = U A(t − t ) (10)


In a design problem, we have to determine the area (A) for a given duty (F, z , x
and T are given). From eqn. 7, 8, and 9, we can solve for ṁ , L and V. Next, by
using the capacity equation (eqn. 10) one can solve for A, provided all the
parameters, namely, C , λ , λ , T , U are known. We have already discussed the
process except, C i.e. the specific heat of the feed solution at concentration of z .
If the C value of the solution is known at any concentration, its value can be
determined at any other concentration by linear interpretation as
C =C − C −C (11)
where C is the specific heat of the solution having mass fraction x , C is the
same at mass fraction x and C is the specific heat of the solvent.

122 |
If the required data for an aqueous solution are not available, the following eqn.
may be used
C = C (1 − x) for x < 0.2 (12)
Sp. heat of the solvent

If, on the other hand, the sp. heat of the anhydrous solute is known, (C ), the
following eqn. may be used
C = C (1 − x) + C x (13)

Classification Based on the Mode of Feed Supply:


Depending upon the directions of flow of the heating medium and of the feed or
the liquor, multiple effect evaporators can be classified into four categories:
(i) Forward feed, (ii) Backward feed, (iii) Mixed feed, and (iv) Parallel feed.

(i) Forward Feed:

Fig. 9
A forward feed unit has already been shown in fig. 9. The feed is introduced to the
first effect. Partly concentrated liquor flows to the second effect, and then from the
second effect to the third effect. Thick liquor is withdrawn from the last effect.
Steam is fed to the shell of the first effect, and the vapour generated there in flows
to the shell of the second effect and acts as a heating medium there. Vapour

123 |
generated in the second effect supplies heat for boiling the liquor in the third effect.
The vapour from the third effect is condensed in the barometric condenser
connected to a steam jet ejector. The flow of the liquid solution from one effect to
the other occurs spontaneously because of the difference of pressure maintained in
the successive effects. In a standard forward feeding system we should install only
two pumps i.e., the feed pump and the product withdrawal pump.

(ii) Backward Feed:

Fig. 10
The backward feed arrangement has been shown in fig. 10. Here the feed is
introduced to the last effect. Partly concentrated liquor flows to the second and
then to the first from which the thick liquor is withdrawn. Pumps are used to
maintain the flow of liquor against a positive pressure. For a backward feed
evaporator with n effects, we need to have (n-1) additional pumps compared to the
forward feeding system.

(iii) Mixed feed:


In a mixed feed arrangement, the dilute feed enters an intermediate effect and
flows to the next higher effect till it reaches the last effect. On this section, the
liquid flow occurs in the forward feed mode. Partly liquor is then pumped to the
effect before the one to which the feed is introduced. It then flows towards the first
effect in the backward feed mode. Thick liquor is withdrawn from the first effect.
124 |
This feeding arrangement for the case of a triple effect evaporator is shown in fig.
11.

Fig. 11
(iv) Parallel Feed:

Fig. 12
In crystallization operation, the evaporator is fed with recycled mother liquor
mixed with fresh feed. The liquid from the evaporator flows to the crystallizer.
There is no question of withdrawal of thick liquor as in the case of other
125 |
evaporators when concentration of the feed is the only objective. In such an
application (fig. 12), the feed is divided into a number of streams and each is fed to
an effect in the unit. Concentrated liquor is withdrawn from each effect separately.
The successive effects however, operate under gradual decreasing pressure so that
the vapour generated in one can act as the heating medium in the next.

Comparison between the Forward and Backward Feed Modes:


Of the two mentioned modes of liquid flow in an evaporator, the former one is
operationally simple. It does not need intermediate pumps for the transfer of the
liquid from one effect to the next; the liquid flows automatically because the
pressure in the next effect is lower. The backward feed technique offers some
advantages over the forward feed arrangement, although at higher capital and
operating costs because of intermediate pumping. Three other important situations
in connection with these two feeding modes are discussed below.
(i) Consider the case when the thick liquor or the final product is highly viscous. In
a forward feed evaporator, we are withdrawing the most concentrated liquor from
the coldest effect. So the viscosity of the liquor in the last effect will be very high
and heat remains pretty low. Conversely, if the backward feeding is used, the
liquor of highest concentration will remain in the first i.e. , the hottest effect.
Therefore, the viscosity of the liquor will remain low, leading to a high heat
transfer coefficient and enhances the capacity of the system. This is a definite
advantage of using backward feed over the forward feed if the liquor viscosity
increases sharply with concentration and the liquor is not heat sensitive.
(ii) Now let us consider a sub-cooled feed. If this feed is admitted to the first effect,
a considerable amount of external steam supplied to this effect is consumed to raise
the temperature of the feed to the boiling point. Because this amount of steam does
not contribute to the generation of vapour, multiple reuse of the steam in the later
effect becomes impossible. On the other hand, if backward feed is used, the cold
feed is introduced in the last effect, the feed absorbs heat from the steam which
does not have the potential of reuse. At the same time because of decreasing
pressure from one effect to another, the BP will remain minimum, in the last effect.
So we may conclude that backward feed is advantageous if the feed is sub-cooled.
(iii) If the feed is hot, the forward feed arrangement is likely to offer a better steam
economy. It dies not absorb any sensible heat when it is introduced to the
evaporator. It rather produces some ‘flash steam’ when it flows to the effect
operating at a lower pressure. In fact, by vapour flashing some evaporation in a
forward feed multiple effect units occur in each effect starting from the second. In
the backward feeding arrangement the flashed steam will be lost to the condenser,
lowering the economy of the system. Hence, for a superheated feed one should
always recommend forward feeding arrangements.
126 |
Multiple Effect Evaporator Calculation:
Let us consider a triple effect, forward feed evaporator, as shown in fig. 13.

Fig. 13
In a standard design problem, the parameters given are
(i) Feed flow rate, composition and temperature (F, z , t )
(ii) Condition of utility steam (P or t )
(iii) Number of effects (n, which can also be determined by economic
optimization).
(iv) Thick liquor concentration (x )
(v) Temperature of utility water, (t ), to be used in direct contact condenser.
The process design steps are as follows:
(Note that if we know the feed temperature, we can also determine the feeding
system/ pattern via economic optimization and with thumb rules).

Design Based on Equidistribution of Pressure:


Step 1: Considering at least an approach of 200C, fix the temperature of the vapour
in the last effect, let it be t , so t ≥ t + 20℃. And from steam table
determine the corresponding pressure P (in general, P ) in the last effect.
Calculate P = P − P

127 |

Step 2: Equidistribute the pressure in each effects. Let P = (in general n)
and set the pressures in each of the effects as
P = P − P
P = P − P
P = P − P
………………….and so on.
Step 3: determine the BPs of pure water at pressures, P , P , P . Let they are, t ,
t , t , which can be obtained from steam table. Also determine the
corresponding latent heats, λ , λ and λ .
Step 4: Assume, x and x , we may consider equal contribution of all of the effects
as an initial guess. Thus if x = x − z , then
 
x =z + ,x =z + etc.
th
So the mean concentration in the i effect may be calculated as
x = ,x = ,x =
Step 5: Knowing T and a tentative set of {x }, determine t for i=1 to n using
Duhring’s plot, such that
t = T + BPE(x )
Similarly, knowing, λ , determine λ using Duhring’s equation (eqn. 2).
Step 6: Knowing the type of evaporator, i.e., knowing an average length of the
tubes, determine the hydrostatic head correction of BP, according to eqn. 4. Let it
be, {T }. Therefore, the corrected BPs becomes
t = t + T
⟹ t = T + BPE + T
for i=1 to n.
Step 7: Determine, C , C , C , etc. by using either of eqn. 11––13.
Step 8: Solve the following equations (specific for triple effect, forward feed
evaporator system) in order to obtain, ṁ , V , L , x ,V ,L ,x V ,L .
F=V +L (14.a) Material
Fz = L x (14.b) balance First
effect
FC t − t + V λ = ṁ λ (14.c) Energy
balance
L =V +L (15.a) Material
L x =L x (15.b) balance Second
L C t −t +V λ =V λ (15.c) Energy effect
balance

128 |
L =V +L (16.a) Material
L x =L x (16.b) balance
Third
L C t −t +V λ =V λ (16.c) Energy effect
balance
Step 9: If |x − x | and |x − x | are less than ε (a small number may be in
the order of 0.01), then go to step 10, else set, x = x and, x = x . Hence
calculate
x = ,x = , and x = , and go back to step 5.
Step 10: Determine the heat transfer rate in each of the effects as
Q = ṁ λ (17.a), Q = V λ (17.b) Q = V λ (17.c)
Step 11: Knowing, U , U and U either from old data or from some suitable
thumb rules, determine
A = , A = ,A = (18 a, b, c)
Step 12: Evaluate A= max (A , A , A ). Next, knowing the standard tube size
evaluate the number of tubes (n ) using the following equation
A = (πD L)n (19)
OD of tubes Length of tubes
The step 12 marks the end of process design. We can revise the procedure for
backward, mixed or parallel feed as well.
Alternative of Equi-distribution of overall pressure drop, we may calculate the
overall temperature drop, t = t − t . Next by considering, Q = Q = Q
and, A = A = A , we may distribute the temperature drop in each effects (t )
as Q =Q =Q , and
 A =A =A
= So (20)
 ∑
U t = U t = U t

129 |
Problems to be Solved:

1. A solid sphere with an initial temp. T is immersed in a large thermal reservoir of


temp. T . The sphere reaches a steady temperature after a certain time, t . If the
radius of the sphere is doubled, the time needed to reach steady-state will be
(a) (b) (c) 2t (d) 4t (2012)

2. If the Nu for heat transfer in a pipe varies with Re as, Nu ∝ Re . , then for
constant average velocity in the pipe, the heat transfer coefficient varies with pipe
diameter D as
(a) D . (b) D . (c) D . (d) D . (2012)

3. An aqueous solution of a high molecular weight solute is concentrated from 5%


to 40% at a rate of 100 . The feed temperature is 250C and concentrated product
leaves at its BP. Calculate the rate at which heat must be supplied if evaporation
occurs at (i) 1 atm (ii) a vacuum of 650 mm of Hg.
Given: ρ = 1020 , C = 4.1 ,C = 3.9
℃ ℃
(Use steam table for other information)
Solve the problem by enthalpy balance.

4. A dilute solution is concentrated from 10% solids to 50% solid at the rate of 500
kg solid/ hr. The feed is available at 700C and steam is supplied at 2 bar (gauge).
The evaporator operates at a pressure of 100 mm of Hg.
(i) Calculate the heat load of the evaporator.
(ii) Determine the steam consumption rate and required surface area.
(iii) If the evaporator pressure is raised to 300 mm of Hg, by what % would the
capacity of the evaporator and the steam rate changes?
Data: C = 0.883 , C = 0.75 , U = 500 , BPE may be
℃ ℃ ℃
neglected (for additional data use steam table).

5. A 200 W heater has a spherical case of diameter 0.2 m. The temperature of the
ambient air is 300C and, k = 0.02 .
(i) Find the heat flux from the surface under steady state (ii) Find the steady-state
surface temperature of the casing.
Combined statement for Q.6 and 7:
Air at 300C flows over a flat plate at a velocity of 8 m/s. The temperature of the
surface is 2000C.
130 |
Given: γ = 2.5 × 10 , Pr = 0.69, k = 0.036 .

6. The thermal boundary layer thickness at x = 0.75m (from the leading edge) in
mm is
(a) 6.6 (b) 7.6 (c) 8.6 (d) 9.6

7. The local heat transfer coefficient at the same point (in, ) is



(a) 3.9 (b) 6.9 (c) 4.9 (d) 5.9

131 |
Heat Exchanger (Shell and Tube):
A shell and tube heat exchanger is the most widely used heat exchanger
equipment. We shall first describe the constructional features of this type of heat
exchanger equipment. This shall be followed by discussions of few other types.
A simplified diagram of a typical shell and tube, floating head type, ‘1–2 pass’
equipment is given in fig. 1. It essentially consists of a shell to which tube sheets
are welded, one at each end. The heat exchanger tubes are fixed to the tube sheets.
There are two heads, the bonnet and the channel, through which flows the tube
fluid.

Fig. 1
There are nozzles on both tubes and shell sides for inlet and outlet of the two
fluids. Fig. 1 shows 1–2 pass heat exchanger i.e., there is one shell pass and two
tube passes. The shell side fluid enters at one end of the shell and leaves at the
opposite end. The tube side fluid as shown in figure, enters at the right end flows
through the tubes, reaches the bonnet on the left, takes a U turn and returned to the
other half of the partitioned channel and then leaves the exchanger. The channel
here is divided into two compartments by a ‘pass partition plate’. The number of
the tube and shell side passes can be increased by using more pass partition plates
for both sides. If no plate is used, we get a unipass system.

132 |
The basic construction and important components of a shell and tube heat
exchanger are described below:

The Shell:
The shell is the enclosure and passage of the shell- side fluid. It has a circular cross
section and is made by rolling a metal plate of suitable dimension into a cylinder
and welding along the length. However, a section of a pipe of suitable thickness
can be used as the shell if the required diameter is not large, (≤ 60cm). The
selection of material for the shell depends upon the corrosiveness of the fluid and
the working temperature and pressure. Carbon still is a common material for the
shell under moderate working condition.

The Tubes:
The tubes provide the heat transfer area in a shell and tube heat exchanger. One of
the fluids flows through the tubes and the other flows through the shell along the
outside of the tubes. Both ‘seamless’ and welded tubes are used, and a wide variety
of materials including low carbon steel, stainless steel, cupronickel, copper, brass,
aluminium etc. are used. The choice of the material depends on the nature of
application.
" "
Tubes of and 1" diameter are more commonly used than larger tubes of 1 and
"
occasionally 1 size. Narrower tubes may be used in smaller heat exchangers
working with clean fluid. Tubes of 8, 16, 12, 20, 24 ft sections are regarded as
standard tube lengths. It is preferred not to go beyond 16 ft, otherwise tube bending
due to the pressure of shell side fluid may becomes probable.
Tubes are generally arranged in a regular manner over the tube sheet and the
distance of two adjacent tubes, should not be less than 1.25 times the tube
diameter. Details of tube arrangements will be discussed later.

Tube Sheet:
The tube sheets are circular, thick metal plate which holds the tube at the ends. In
the type of exchanger shown in fig. 1 (called fixed- head or fixed tube sheet
exchanger), the tube sheets are welded to the shells at the ends. The tubes are
inserted into the holes of the two tube sheets. Perfect alignment of the holes is
required; this is achieved by drilling tube holes through the sheet fastened together
for the purpose, particularly for small thickness tube sheets. The diameter of the
tube holes are made slightly larger than the diameter. The arrangement of tubes on
a tube sheet in a suitable pitch is called ‘tube- sheet layout’.

133 |
Two common techniques of fixing the ends of a tube to the tube sheets are (i)
expanded joints, and (ii) welded joints. Expanded joints may be made with or
without grooves in the tube sheet holes.
Groove joints are made by cutting
small grooves in the tube sheet along
the periphery of each hole (fig. 2) and
expanding the tube into the grooves. A
very strong and durable joint can be
made in this way.
If a welded joint is used, the tubes and
the tube sheets should be of the same or
very similar material so that they are
‘corrosion compatible’. In fact, if the
tube sheets are exposed to a corrosive
environment, they must also be
electrochemically compatible to each
other even if a welded joint is avoided.
Fig. 2
Section of a welded joint is shown in
fig. 3.

Bonnet and Channel:


Both the tube sheets of a shell and tube
heat exchanger should have a closure or
heads. The space inside the closure is
occupied by tube- side fluid. A closure
is called bonnet or channel depending
upon its shape and construction. A
bonnet has an integral cover, a channel
closure has a removable cover.

Fig. 3
A bonnet closure is shown in fig. 1, consists of a short cylindrical section with a
bonnet welded at one end and a flange welded at the other end. The flange is bolted
to the tube sheet after inserting a suitable gasket material to make the joint leak
proof. A bonnet- type closure is used when it does not require to be fitted with any
nozzle.
A channel type- closure is also shown in fig. 1. It is also made from a piece of
cylindrical barrel with flanges welded to both ends. One of the flanges is bolted to
a tube sheet using a gasket for sealing.

134 |
The other flange of the channel is bolted to a flat channel cover plate. One may
have access to the tubes after unbolting the channel cover plate.

The Pass Partition Plate:


A ‘pass partition plate’ or ‘pass divider’ is shown in the fig. 1. The plate causes
two passes for the tube side fluid.
In order to have two passes of the shell- side fluid, a ‘longitudinal shell pass baffle’
is used. This is a flat plate, usually 6–13 mm thick, which runs axially along the
shell to divide it into two semi- cylindrical halves. The plate is either welded to the
shell or fitted into guide channels that are welded longitudinally along the shell
wall.

Nozzles:
Small sections of pipes welded to the shell or to the channel which act as the inlet
or outlet of the fluids are called nozzles. The end of the nozzle is flanged to
connect it to the pipe carrying a fluid.
The shell- side inlet nozzle is often provided with an ‘impingement plate’ (fig. 4).
The impingement plate prevents impact of the high velocity inlet fluid stream on
the tube bundle. Such impact can cause erosion and Cavitation of the tubes just in
front of the nozzle, and can also cause vibration of the tubes. The erosion problem
is aggravated (i) if the inlet liquid has suspended solid particles or (ii) if the inlet
gas stream has suspended dust or solid particles or even liquid droplets. An
impingement plate is also called the ‘impingement baffle’.
Baffles:
A baffle (or a shell-side
baffle) is a metal plate usually
in the form of the segment of
a circle having holes to
accommodate tubes. Shell-
side baffles have two
functions––(i) to cause
changes in the flow pattern of
the shell side fluid creating
parallel or cross flow to the
tube bundle (thus increasing
turbulence and, therefore, the
heat transfer coefficient), and
Fig. 4 (ii) to support the tubes.

135 |
Segmental Baffle:
This is the most popular type of baffle. A segmental baffle may have horizontal or
vertical cuts as shown in fig. 5.
The cut- out portion (called
baffle window) provides the
area for flow of the shell fluid.
This area may vary from 15%
to about 50% (but definitely
less than 50%). However, the
Horizontal cut baffles net flow area of the shell fluid
is calculated by subtracting the
cross- sectional area of the
tubes passing through the cut-
out portion from the area of
the cut- out portion itself. A
small notch or segment is cut
out at the lowest part of a
baffle in a horizontal
Vertical cut baffles
exchanger.
Fig. 5 (a, b)
This provides the passage to the liquid when the shell is required to be drained.

Other Types of Baffles:


Among the other types of shell baffles, the disk- doughnut baffle and orifice
baffles are shown in fig. 6 and 7.

Fig. 6 & 7
136 |
The figures also explain their construction and the fluid flow patterns.

Baffles as Tube support:


Segmental baffles also act as support to the baffles. Vertically- cut segmental
baffles provides better tube support than that provided by horizontally cut type. At
least two baffles in between the tube sheets are required to support all the tubes.
The baffle prevents the tubes from sagging and also from vibrations caused by the
flow of the fluids. A hole in a baffle is slightly larger than the outer diameter of the
tube. If the clearance is larger, the edge of a baffle hole may cut the tube because
of tube vibration. In fact, ‘flow- induced tube vibration’ in a heat exchanger can
turn out to be a serious problem and should be properly taken care of in the design.
Large baffle spacing reduces the shell-side pressure drop, but simultaneously
reduces both the turbulence and shell side heat transfer coefficient. A smaller
baffle spacing increases both. Baffle spacing is selected in consideration of the
allowable shell-side pressure drop and the heat transfer coefficient desired. The
minimum spacing of segmental baffles is one-fifth of the shell diameter or 5 cm,
whichever is larger, (B = D or 5 cm), whereas the maximum spacing
should be equal to the shell diameter itself (B = D ). So for the segmental
baffles we have baffle spacing (B) as, ≤𝐵≤𝐷 .

Tie Rods and Baffle Spacers:


A few tie rods having
threaded ends are used to
hold the baffles in position.
One end of a tie rod is
screwed to a tube sheet. A
section of a tube or pipe
(does not mean heat
exchanger tube) of suitable
diameter (the ID of the piece
of tube should be slightly
larger than tie rods) is slid
over it. A baffle is then
inserted followed by another
section of the tube.
Fig. 8
The tube sections are called spacers because they maintain the spacing or distance
between successive baffles. After the last baffle is inserted, locknuts are tightened

137 |
at the threaded free end of the tie rod. This arrangement fixes all the baffle
families, and prevent their movement.
The tie rod and spacers are also shown in fig. 8. In this figure the tie rod is shown
by a dotted line and the spacer tubes by firm lines.

Fouling of a Heat Exchanger–– the Dirt Factor or Fouling Factor:


Process fluid streams may contain suspended matters or dissolved solids. When
such a fluid flows through a heat exchanger over a long period of time, deposition
on the tube surfaces occurs. The surfaces may also be corroded by the fluids slowly
and the resulting corrosion products also get deposited on the surface. In either
case, a coating of deposits forms whose thickness increases with time. The coating
(called scale) has a thermal conductivity much less than that of tube wall and
therefore offers significant resistance to heat transfer. Formation of a scale or
deposit on a heat transfer surface is called fouling, and the heat transfer resistance
offered by the deposit is called fouling factor or dirt factor; commonly denoted by
R .
The fouling factor is zero for a new heat exchanger. It increases over the period of
operation, and the tubes have to be cleaned after certain time. While designing a
heat exchanger, a value of the fouling factor has to be included as an additional
resistance to heat transfer. The fouling factor or the dirt factor cannot be estimated;
it can only be determined from experimental data on heat transfer coefficient of a
‘fouled’ exchanger and a clean exchanger of similar design operated at identical
conditions.

Log Mean Temperature Difference Correction Factor:


Temperature driving force varies with position in a countercurrent or concurrent
double pipe heat exchanger. It has also been shown that a log mean temperature
difference (LMTD) has to be used as the average driving force. A countercurrent
device is more efficient for exchange of heat. But in many situations, a multi-pass
exchanger gives a highest heat transfer coefficient than a single-pass one because
of larger fluid velocity. However, in a multi-pass exchanger like 1-2, 1-4, 2-4 etc.
the fluids are not always in countercurrent flow. If one tube pass is countercurrent
to the shell fluid, the other tube pass will be concurrent. This deviation from truly
countercurrent flow causes a change in the average driving force. A correction
factor (F ) is used to get the true mean temperature difference or the effective
driving force. The heat transfer rate becomes
Q = U AF (LMTD) (1)
where U is the design overall heat transfer coefficient that takes into account the
fouling or the dirt factor, R . F (LMTD) is the true temperature difference.

138 |
If U is the clean overall coefficient, then we have
= +R (2)
The LMTD correction factor F can be directly obtained from the available charts.
These charts have been prepared from the results obtained theoretically by solving
for the temperature distribution in a multi-pass exchanger.
Temperature distribution of hot and cold streams, in a 1–2 pass countercurrent
exchanger, is shown in fig. 9.

Fig. 9
For complex flow models, the temperature distribution will predictably be
different. Consider a 1–2 co-current exchanger, the qualitative temperature profile
is as shown in fig. 10.

Fig. 10
139 |
If the fluid streams enters at the opposite end (fig. 9), then a temperature cross may
occur sometime. Temperature cross means that the difference between the cold
and the hot fluid temperatures is positive when these streams leave the exchanger,
= T − T . If it occurs, the cold fluid temperature reaches the maximum at a
point inside the exchanger and not at its exit. This point also coincides with the
point of intersection of the temperature profiles of the hot fluid and the concurrent
zone of the cold fluid (fig. 9). The quantity, T − T , is called the temperature
cross of the exchanger. However, if temperature cross does not appear (if T <
T ), then T − T is called the approach.
Typical charts for LMTD correction factor (F ) is given in fig.(11.a) and (11.b).
One of the factors governing the choice of numbers of shell and tube passes is the
value of the correction factor, F . F < 0.8 is generally not acceptable.

Fig. 11.a, b

140 |
Here τ is known as temperature ratio and R is known as capacity ratio.
Actually, lengthy algebraic exercise is necessary in order to express the correction
factor F in terms of τ and R. Instead of expression based evaluation one should
use the F chart (plot of F vs. τ at different R) for ready evaluation of F .

Individual and Overall Heat transfer Coefficient:


In order to calculate the heat transfer area of a shell and tube heat exchanger by
using the eqn.
̇  ̇ 
A= ( )
(3)
We have to know U in addition to ṁ , C , t or (ṁ , C , t ), F and LMTD.
Once again from eqn. 2 we know, = + R . So, the objective reduces to the
calculation of ‘clean overall heat transfer coefficient’. In terms of inside (or
outside) diameter we may write
U =  (4)
.

So finally, we simplify the problem of calculating U in the determination of h


and h .

Inside Film Coefficient:


A number of correlations for the estimation of h has been cited previously. For a
shell and tube exchanger, the Dittus-Boelter or Sider-Tate equation may be used
depending on whether the viscosity ratio is significantly different from unity
or not. A chart of Colburn factor may be used over a wide range of Reynolds
number covering both laminar and turbulent flow.

Outside Film Coefficient:


For a double pipe heat exchanger, the outside film coefficient can be calculated by
using the Dittus-Boelter or the Sider-Tate equation on the basis of equivalent
diameter of the annulus. A similar approach may be adopted for shell- and- tube
exchanger if there are no baffles on the shell side. However, baffles are invariably
used to enhance turbulence and thereby to increase the shell side heat transfer
coefficient, although the presence of baffles increases the pressure drop on the
shell side. Given below are the methods of determination of shell side heat transfer
coefficient in a baffled heat exchanger. Kern (1950) suggested the following eqn.
for this purpose
. . .
= 0.36 (5)

141 |
where
D : Hydraulic/ equivalent diameter of the shell side
k : Thermal conductivity of shell side fluid
G : Mass flow rate of shell side.
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Tube Layout:
We may arrange the tubes on the tube sheet either in (i) square pitch or (ii) in
triangular pitch layouts as shown in fig. (12. a) and (12. b).

Fig. 12 (a, b)
For tubes arranged in square pitch we have
D = (6)
[As wetted area = P − d . 4 and wetted perimeter, = ]
On the other hand for tubes arranged in triangular pitch we have

D = (7)

[As for equilateral triangle, the area is P and out of which tubes occupy,
° ° ° ° ° °
= × , the wetted perimeter: πd × ].
° °
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
In order to calculate, G we have to use the following eqn.
̇
G = Mass flow rate of the shell fluid (8)

Flow area

142 |
a i.e., the flow area may be calculated based on fig. 13 (determine the hatched
area of fig. 13).
As
a = .c B (9)

Area of
No. of individual
rectangles rectangle
where
c : Clearance between two
adjacent tubes.
B: baffle spacing
D : Inside diameter of the shell.
The quantities in eqn. 6, 7 and 8,
9 should be taken in consistent
units. The baffle spacing (for
25% cut baffles) is usually
chosen to be within D to D .
Fig. 13
Pressure Drop calculation:
There are quite a few correlations and charts available for the calculation of
pressure drops over the tube and the shell sides of a heat exchanger. One method is
described below.

The Tube Side Pressure Drop (kern, 1950):


Use the following equation
P = (10)
where f is the Fanning’s friction factor, G is the mass velocity of the tube fluid, L
is the tube length, n is the number of tube passes, ρ is the density of the tube fluid,
d is the inside diameter of the tube, ϕ is the viscosity ratio, P is the tube side
pressure drop.
Here, ϕ = , m=0.14 for Re>2100
and, m=0.25 for Re<2100.
In addition to the friction loss for flow through the tubes, the head loss owing to
the change of direction of the fluid in a multi-pass exchanger has to be taken into
account. This is sometimes called the return loss.

143 |
The pressure drop owing to the return loss (P ) is given by
P = 4n ρ (11)
where V is the linear velocity of the tube fluid. So the total tube- side pressure drop
is
P = P + P (12)

Shell Side Pressure Drop:


For an unbaffled shell the following eqn. may be used
P = (13)
where L is the shell length, N is the number of shell basses, ρ is the shell fluid
density, G is the shell side mass velocity, D is the hydraulic diameter of the shell,
ϕ is the viscosity correction factor for the shell side fluid.
Now
.
D = (14)
where d is the outer diameter of the tubes, D is the inside diameter of the shell,
.
N is the number of tubes in the shell and ϕ = .
For a shell with segmental baffles, a modified eqn. may be used
( )
P = (15)
where N is the number of baffles, D is the hydraulic diameter of the shell. The
shell side Reynolds number may be obtained as, Re = .
The friction factor for the shell side flow (f ) may be obtained from Moody chart
or by using a suitable correlation.

144 |
Problems to be Solved:
1. A process fluid has to be cooled from 220C to 20C using brine in a 2–4 shell and
tube heat exchanger shown below. The brine enters at -3 0C and leaves at 70C. The
overall heat transfer coefficient is, 500 . The design heat load is 30 kW. The
brine flows through tube side and process fluid on shell side. The heat transfer area
in (m ) is

(a) 1.1 (b) 5.77 (c) 6.59 (d) 7.53 (2006)

2. The list of options P, Q, R and S are some important consideration in the design
of shell and tube heat exchanger.
P. Square pitch permits the use of more tubes in a given shell diameter
Q. The tube side clearance should not be less than one fourth of the tube diameter
R. Baffle spacing is not greater than the diameter of the shell or less than one-fifth
of the shell diameter
S. The pressure drop on the tube side is less than 10 psi.
Pick the correct combination of TRUE statements:
(a) P, Q & R (b) Q, R & S (c) R, S & P (d) P, Q, R & S (2007)

3. In a shell and tube H.E, if the shell length is L , the baffle spacing is L and the
thickness of the baffle is t , the number of baffles in the shell side be
(a) (b) − 1 (c) +1 (d) +2 (2008)

4. In a 1–1 pass floating head type shell and tube H.E, the tubes (od=25 mm, id=
21 mm) are arranged in a square pitch. The tube pitch is 32 mm. The thermal

145 |
conductivity of the shell side fluid is 0.19 , and the Nusselt number is 200. The
shell side heat transfer coefficient (in ) rounded off to the next integer is
(a) 1100 (b) 1400 (c) 1800 (d) 2100 (2011)

5. Steam is to be condensed in a shell and tube H.E, 5 m long and of shell diameter
1 m. Cooling-water is to be used for removing heat. Heat transfer for cooling
water, whether on shell or tube is same. The best arrangement is
(a) Vertical H.E with steam in tube side
(b) Vertical H.E with steam in shell side
(c) Horizontal H.E with steam in tube side
(d) Horizontal H.E with steam in tube side.

Combined statement for Q. 6 and 7


Two ends of a 5 cm diameter and 50 cm long Al rod, with curved surface perfectly
insulated are maintained at 300C and 3000C, respectively. The temperature
dependent k of the metal can be given as, k = 202 + 0.0545T .

0
where T is in C.


6. Calculate the temperature gradient at the end of the rod (in )
(a) 660 (b) 360 (c) 560 (d) 760

7. The temperature midway in the rod (in 0C) is


(a) 267 (b) 67 (c) 467 (d) 766

8. If the constants {A } are suitably chosen so as to satisfy the initial conditions


and n is an integer, which one of the following are a valid solution for the unsteady
∗ ∗
state, one dimensional conduction equation: ∗
=α ∗
in the domain 0 ≤ x ∗ ≤ 1
(where, x ∗ = and T ∗ = ) with BC: T ∗ = 0 at x = 0 and T ∗ = 0 at x = L.
(a) ∑ A sin exp (b) ∑ A cos exp
(c) ∑ A sin exp − (d) ∑ A cos exp −

9. A heat exchanger tube of D =20 mm diameter coveys 0.0983 kg/s of water


inside (Pr =4.3, k = 0.632 , ρ = 10 and μ = 0.651 × 10 ) and is
.
used to cool a stream of air at the outer side, where the external heat transfer

146 |
coefficient (h ) is 100 . Ignoring the resistance of tube wall evaluate the value
of U assuming a suitable correlation for h
(a) 95.2 (b) 59.2 m (c) 29.5 (d) 25.9

147 |

You might also like