Indira Gandhi MRW-002

National Open University

School of Engineering & Technology HEAT TRANSFER

BLOCK 1

Introduction to Heat Transfer

 Indira Gandhi
National Open University
MRW – 002
School of Engineering & Technology Heat Transfer

Block

1
INTRODUCTION TO HEAT TRANSFER
UNIT 1
Concepts of Heat Transfer 5

UNIT 2
Different Modes of Heat Transfer 27

UNIT 3
Mass Transfer 63
GUIDANCE
Prof. Nageshwar Rao Prof. Satyakam Prof. Ashish Agarwal
Vice-Chancellor, IGNOU PVC, IGNOU Director, SOET, IGNOU
COURSE CURRICULUM DESIGN AND DEVELOPMENT COMMITTEE
Prof. K.A. Subramanian, Prof. Ashish Agarwal,
Head, Dept. of Energy Science and Director, School of Engg. & Tech.
Engineering, IIT Delhi IGNOU, New Delhi

Prof. J Ram Kumar, Prof. Ajit Kumar

Mechanical Engineering, IIT Kanpur Prof. in Civil Engineering, School of Engg. & Tech.
IGNOU, New Delhi
Prof. G.N. Tiwari Prof. P S Kumar
Retired Professor Professor in Civil Engineering, School of Engg. & Tech.,
Centre for Energy Studies, IIT. Delhi IGNOU, New Delhi

Prof. Ramchandra Prof. K.T. Mannan

Prof. of Physics Professor in Mechanical Engineering, School of Engg. &
Ex V.C., Monad University, Hapur Tech., IGNOU, New Delhi
Dr. Anand Kumar Tewari, Prof. N. Venkateshwarlu
(Retired Executive Director) Professor in Mechanical Engineering
IOCL School of Engg. & Tech., IGNOU, New Delhi
Prof. R.P. Saini, Dr. Shashank Srivastava
Dept. of Hydro and Renewable Assistant Professor in Mechanical Engineering
Energy, IIT Roorkee School of Engg. & Tech., IGNOU, New Delhi

Dr. Dibakar Rakshit, Prof. Sanjay Agrawal

Associate Professor, Professor in Electrical Engineering, School of Engg. &
Department of Energy Science and Tech., IGNOU, New Delhi
Engineering, IIT Delhi

Dr. Rhythm Singh, Prof. Rakhi Sharma

Assistant Professor Professor in Electrical Engineering, School of Engg. &
Dept. of Hydro and Renewable Technology, IGNOU, New Delhi
Energy, IIT Roorkee
Dr. M.K. Bhardwaj
Dr. S.K. Vyas Associate Professor in Civil Engineering, School of
Associate Professor in Civil Engineering & Technology, IGNOU, New Delhi
Engineering, School of Engg. &
Tech., IGNOU, New Delhi Dr. Shweta Tripathi
Assistant Professor in Mechanical Engg., School of Engg. &
Dr. Anuj Purwar Technology, IGNOU, New Delhi
Assistant Professor in Civil
Engineering, School of Engg. &
Tech., IGNOU New Delhi

COURSE PREPARATION TEAM

Programme Coordinator Course Coordinator
Dr. Shweta Tripathi Dr. Shashank Srivastava
School of Engg. & Tech., IGNOU School of Engg. & Tech., IGNOU

BLOCK PREPARATION TEAM

Units Written by Edited by Compiled and Collated by
Dr. Pinakeswar Mahanta Prof. Subhasis Maji Prof. Ashish Agarwal
Dept. of Mechanical Engineering SOET, IGNOU SOET, IGNOU
IIT, Guwahati
Dr. Shweta Tripathi
Prof. N. Venkateshwarlu
SOET, IGNOU
SOET, IGNOU
Dr. Shashank Srivastava
SOET, IGNOU
PRODUCTION
XXXXXXXXXXXX XXXXXXXXXX
Section Officer (Publication) Proof Reader
©Indira Gandhi National Open University,

All rights reserved No part of this work may be reproduced in any from, by mimeograph or any other means, without
permission is writing from the Indira Gandhi National Open University.

Further Information on the Indira Gandhi National Open University courses may be obtained from the University’s
office at Maidan Garhi, New Delhi-110068.

Printed and publish on behalf of the Indira Gandhi National Open University by Prof. Subhasis Maji, Director SOET.

Laser Typesetting by School of Engineering and Technology, IGNOU, New Delhi-110 068. Phone 29532863

Printed at XXXXXXXXXXXXX
Paper Used: -------------------------
HEAT TRANSFER
This course consists of 5 blocks.
Block 1, Introduction to Heat Transfer, deals with the basic concepts of transfer
and gives a bird’s eye view of the mechanisms. It builds up the necessary
background for the understanding of the physical significances of the conduction,
convection and radiation heat transfer. The phenomena of mass transfer,
mechanism of mass transfer and various types of mass transfer are discussed.

Block 2, Conduction, deals with the governing equations or conduction heat

transfer. Application of Fourier’s law to steady and unsteady state conduction has
been analysed.The governing equation of heat conduction is presented in
rectangular, cylindrical as well as spherical coordinates. Formulation and solution
are done for extended surfaces. Performance parameters such as fin efficiency and
effectiveness are discussed in details.

Block 3, Convection, deals with heat transfer by convection and discuses the
physical and mathematical basis for the understanding of convective transport and
to reveal various heat transfer correlations. The analysis of convention is
complicated because the fluid motion is affected by pressure drop, drag force and
heat transfer. The literature of convective heat transfer is overwhelming and ever
growing.

Block 4, Radiation, deals with the third mode of heat transfer, i.e. radiation has
been considered. Thermal radiation is that electromagnetic radiation emitted by a
body as a result of its temperature. The principles of radiation, radiation intensity
and emissive power are discussed. Concept of blackbody and laws of radiation are
discussed.

Block 5, Heat Transfer Applications, deals with applications of heat transfer in

practical engineering fields. Different types of heat exchangers utilized in
industries are discussed. The very purpose of the entire course will remain
incomplete if we do not discuss about heat transfer associated with chemical
processes. Chemical processes such as evaporation, liquid-liquid extraction,
adsorption and membrane separation are in use in industry where lot of mass
transfer is involved. This block mainly devoted for applications of heat and mass
transfer in chemical processes.
INTRODUCTION TO HEAT TRANSFER

Unit 1 introduces the basic concepts of heat transfer and gives a bird’s eye view
of the mechanisms. A discussion of units and dimension is also presented. Unit 2
builds up the necessary background for the understanding of the physical
significance of the conduction, convection and radiation heat transfer. Emphasis is
placed on the mathematical formulation of practical heat conduction problems.
This matter is illustrated with numerous representative examples. Interaction of
conduction, convection and radiation is also discussed. Unit 3 deals with the
phenomena of mass transfer. Mechanism of mass transfer and various types of
mass transfer are discussed. The unit is supported with solved problems for better
insight of the mass transfer mechanism. All understanding the basic heat transfer
phenomena.
 MRW – 002
HEAT TRANSFER

BLOCK 1 : INTRODUCTION TO HEAT TRANSFER

Unit 1 : Concepts of Heat Transfer
Unit 2 : Different Modes of Heat Transfer
Unit 3 : Mass Transfer

BLOCK 2 : CONDUCTION
Unit 4 : Governing Equations of Heat Conduction
Unit 5 : Numerical Methods to Solve Conduction Problems
Unit 6 : Heat Transfer from Extended Surfaces

BLOCK 3 : CONVECTION
Unit 7 : Convection Heat Transfer
Unit 8 : Boundary Layer Formation

BLOCK 4 : RADIATION
Unit 9 : Radiation Principles
Unit 10: Radiation Exchange

BLOCK 5 : HEAT TRANSFER APPLICATIONS

Unit 11: Heat Exchangers
Unit 12: Boilers

MPDD-IGNOU/P.O.5H/___________2020 (Print)

ISBN-
 Concepts of Heat and
UNIT 1 CONCEPTS OF HEAT AND MASS Mass Transfer

TRANSFER
Structure
1.1 Introduction
Objectives
1.2 Heat Transfer in Engineering
1.2.1 Aims of Studying Heat Transfer
1.2.2 Applications of Heat Transfer
1.3 Modes of Heat Transfer
1.4 Mechanism of Heat Transfer
1.4.1 Conduction Heat Transfer
1.4.2 Convection Heat Transfer
1.4.3 Radiation Heat Transfer
1.5 Mass Transfer
1.6 Temperature Field
1.7 Relationship to Thermodynamics
1.8 Units and Dimension
1.9 Summary
1.10 Key Words
1.11 Answers to SAQs

1.1 INTRODUCTION
The study of transfer phenomena has been considered as a unified discipline of
fundamental importance. Such phenomena include wide range of transfer processes such
as, heat, mass, momentum and energy. Usual practice to study these phenomena is on the
basis of generalized fluxes and forces. All type of transfer processes involves a flux (or
force) and its effect is called current. For example,
(a) Heat flows due to the temperature gradient from higher to lower temperature.
(b) Mass transfer occurs due to a concentration gradient from higher to lower
concentration.
(c) Momentum transfer occurs due to a velocity gradient from higher to lower
velocity.
(d) Electric current flows due to a potential gradient.
(e) Chemical process occurs due to the flux called chemical affinity.
All such processes mentioned above are spontaneous processes and terminates once the
force is withdrawn or the same diminishes to zero. It is the law of nature, which states
that driving force causes the respective flux from a higher to lower potential. The transfer
process indicates the tendency of a system to proceed towards the equilibrium.
Interestingly, in all such processes, driving force is linearly proportional to the gradient.
Objectives
After studying this unit, you should be able to
 define conduction, convection and radiation,
 differentiate between heat and mass transfer,
 know the various types of heat exchangers, and 5
Introduction to Heat  understand the various applications of heat transfer.
and Mass Transfer

1.2 HEAT TRANSFER IN ENGINEERING

1.2.1 Aims of Studying Heat Transfer
Heat transfer is the energy interaction due to a temperature difference in a medium or
between media. Heat is not a storable quantity and is defined as energy in transit due to a
temperature difference. The primary aims of studying the science of heat transfer are :
(a) to understand the mechanism of heat transfer processes, and
(b) to predict the rate at which heat transfer takes place.
Whenever we refer to heat transfer we actually imply the heat transfer rate. It is this
“rate” that differentiates the field of heat transfer from thermodynamics.
1.2.2 Applications of Heat Transfer
The applications of heat transfer are diverse, both in nature and in industry. Climatic
changes, formation of rain and snow, heating and cooling of the earth’s surface, the
origin of dew drops and fog, spreading of forest fires are some of the natural phenomena
wherein heat transfer plays a dominant role. The existence of living beings is possible
due to the existence of the sun.
The importance of heat transfer in industry, including medical applications, can be seen
by focusing on the following classes of problems.
Designing of Energy Conversion Devices
One of the most important applications of heat transfer is in the designing of steam
generators, turbine, internal combustion engines, gas turbine, jet and rocket
propulsion, etc. Another important aspect is designing of waste heat recovery
devices.
Heat Exchangers
There are various types of heat ex-changers with wide application in industries,
such as heat wheel, shell and tube type heat ex-changers, plate type, plate-fin type,
spiral type, stationary re-generator, metallic recuperators, run around coil, heat
pipe, etc. (Figure 1.1). Sizing and rating problems of heat ex-changers require
expertise in the subject.
Hot fluid

Cold fluid

Heating zone Cooling zone

Empty

(a) Ljungstorm Heat Exchanger

6
 Concepts of Heat and
In In
Mass Transfer

A B

C D

(b) Storage type of Heat Exchanger

Hot air out

Hot gas in

Rotating hood

Stationary heat exchanger

Cooled out gas

Cool air in
(c) Ruthemule Heat Exchanger

From Elevator

Ports
A

Pressure
B Hot Air Outlet
Control
Connections

Cold Air Inlet

To Elevator

(d) Pebble Bed Heat Exchanger

7
Introduction to Heat Header
and Mass Transfer

Hot fluid

Cold fluid out Cold fluid in

Hot fluid out

(e) Plate type of Heat Exchanger

Waste gas

Hot air to process

Cold air in

Flue Gas
(f) Metallic Recuperator

Waste gas tf, out

tf, in

Cold air
fa, out fa, in

(g) Run Around Coil

Hot out

Cold in (h) Spiral Heat Exchanger

Figure 1.1 : Different Type of Heat Exchangers

8
Industrial Processes Concepts of Heat and
Mass Transfer
Designing of industrial furnaces, incinerators, autoclave, etc.
Thermal Insulations
In many of the heat transfer devices thermal insulators are used to reduce the heat
loss from the devices. Typical examples are: Thermos flask, ice box, hot box,
building walls, steam pipes, and cryogenics. In this classes, the maximum and
minimum temperatures (Tmax and Tmin) experienced by a heat transfer medium are
usually fixed. The main objective is to reduce the “heat loss” or “heat leak”. The
thermal design involves judicious changes in the constitution of insulation (that is,
its size, material, shape, structure, flow pattern) so that the heat transfer indeed
decreases while Tmax and Tmin remain fixed.
Heat Transfer Enhancement (Augmentation)
The main application is in the design of heat exchangers where the total heat
transfer rate (q) between the hot and cold fluid streams separated by solid surface
is usually a prescribed quantity. The objective is to transfer q across a minimum
temperature difference. This can be done by changing the flow patterns of the two
streams with breaking of the boundary layer (causing more turbulence). There are
three distinct methods for heat transfer enhancement, i.e.
(a) passive method (Figure 1.2(a)),
(b) active method (Figure 1.2(b)), and
(c) combined method (i.e. combination of both the active and passive
method).
Passive method does not require any external energy whereas active method
involves external energy input to implement. Twisted tape, wire coil, indentation
of flow surface, surface roughening, etc. are some of the passive methods of heat
transfer enhancement. Picture of some twisted tapes are given in Figure 1.2(c).

H Tape
A

Direction of
Flow

A Tube
D
D0

Section A-A
Figure 1.2(a) : Twisted Tape Inserted in a Pipe for Heat Transfer Enhancement

Tube

Direction of Flow

Reciprocating Plunger

Figure 1.2(b) : Active Method with a Reciprocating Plunger

9
Introduction to Heat
and Mass Transfer

Figure 1.2(c) : Picture of Some Twisted Tapes

Temperature Control
In many areas, overheating of a heat-generating body is not permissible. Examples
of temperature control applications include cooling of electronic equipments such
as personal computers and super computers, cooling of nuclear reactor cores, and
the cooling of the outer surface of space vehicles during re-entry. Cooling of high
heat flux surfaces such as electronic chips in a tightly packaged set of electronic
circuits is quite challenging because of the size limitations. The temperature of the
electronic chips cannot rise much above the ambient temperature, because high
temperatures drastically reduce their performance. Another important application
of temperature control is the film cooling of gas turbine blades by routing air
through channels within the blades.
Bio-Heat Transfer
Heat transfer plays a very important role in living systems as it affects the
temperature and its spatial distribution in tissues. The primary role of temperature
is the regulation of a plethors of rate processes that govern all aspects of the life
process. These thermally driven rate processes define the difference between
sickness and health, injury and successful therapy, comfort and pain, and accurate
and limited physiological diagnosis. Typical applications of bio-heat transfer
include human thermoregulation, thermal surgical procedures such as microwave,
ultrasound, radio frequency and laser, cryo-preservation of living cells, and
thermal burn injury.
Materials Processing
Recent years have seen surging interest among researchers to understand heat
transfer aspects in various material processing systems such as solidification and
melting, metal cutting, welding, rolling, extrusion, plastic and food processing, and
laser cutting materials. This has led to improved designs of material processing
systems.
Other areas of heat transfer applications are in power production, chemical and
metallurgical industries, heating and air conditioning of buildings, design of
internal combustion engines, design of electrical machinery, weather prediction
and environmental pollution, oil exploration, drying, and processing of solid and
liquid waste. The list is endless.
It is no wonder that J. B. Joseph Fourier, the father of the theory of heat diffusion,
made this remark in 1824 : “Heat like gravity, penetrates every substance of the
universe; its rays occupy all parts of space. The theory of heat will hereafter form
one of the most important branches of general physics.”
Chemical Processes
Chemical processes involve heating and cooling of materials or fluid streams in a
number of steps. For example, consider the process of manufacturing nitric acid by
catalytic air oxidation of ammonia (Figure 1.3). Liquid is the raw material.

10
 Concepts of Heat and
To Stack Mass Transfer
Water

Ammonia 6
From the
Storage Tank
5

11
4

1 2 3
Air
7 8 9 10

Product Nitric
Acid
Figure 1.3 : Heat Transfer in a Chemical Process of Making Nitric Acid

It is a common practice to store ammonia at atmospheric pressure and at a

temperature of – 33oC (approx). A refrigeration unit is used to keep the liquid cool
(in other words, to continuously remove the heat that enters the storage tank from
the ambient air). The storage tank is properly insulated to substantially reduce the
transfer of heat from the outside air to the liquid ammonia. Liquid ammonia from
the tank is pumped to a vaporizer (5) where heat supplied by steam (or any other
hot fluid as available) converts the liquid to a vapour.
Compressed air is used for the oxidation of ammonia. Air is usually compressed in
a two stage centrifugal compressor (2) provided with a interstage cooling. When
compressed, the temperature of the gas increases. Hot air leaving the first stage of
the compressor is cooled in a heat exchanger by cold water. The cooled gas moves
to the second stage of the compressor and exchange heat with the tail gas (4) from
the absorption tower (11). Ammonia vapour is mixed with this compressed air in
an appropriate proportion and fed to the reactor (6). The reaction products attain a
temperature of 800oC (approx). The hot product stream, therefore, contains a huge
amount of energy. This energy is recovered in a waste heat boiler, which is a part
of the reactor unit, in which transfer of heat from hot gas to the boiling water
occurs, and thus a large quantity of steam is generated. Even after the gas leaves
the waste heat boiler, its temperature remains at about 275oC and has to be further
reduced before the gas reaches the absorption tower (the gas containing the oxides
of nitrogen is absorbed by water in a number of absorption towers in succession to
produce nitric acid). Cooling to this gas from 275oC is carried out in a series of
heat exchangers (7-10). In the first heat exchanger (7), heat transfer from the hot
gas occurs to the tail gas leaving the final absorption tower. The heated tail gas is
then fed to a turbine (3) that drives the air compressor. The second heat exchanger
(8) acts as an economizer, and the third one (9) is used to heat boiler feed water. In
this way, most of the heat energy is recovered. The gas flows through a series of
absorption towers (11) after passing through a cooler condenser (10). Absorption
of the gas by water gives nitric acid. The absorption process is exothermic. As a
result, each absorption tower is required to be provided with a liquid cooler (i.e.
heat exchanger).

1.3 MODES OF HEAT TRANSFER

Heat transfer or simply heat is the thermal energy in transit due to a temperature
difference. Whenever there exists a temperature difference in a medium or between two
media, heat transfer must occur. Different types of heat transfer processes may be
classified in 3 (three) different modes :
(a) Conduction
11
Introduction to Heat (b) Convection
and Mass Transfer
(c) Radiation
In reality, heat transfer in a system occurs in combination of these modes. However,
analysis can be done for each mode of heat transfer separately. Figure 1.4 presents the
3 (three) modes of heat transfer.
Ty > T
T1 > T2
T1 T2

Moving Fluid, T
Surface, T1

q”
q”1 Surface, T2

q” q”2
T,

Conduction through a Solid Convection from a Surface Net Radiation Heat Exchange
or a Stationary Fluid to a Moving Fluid between Two Surfaces
Figure 1.4 : Different Modes of Heat Transfer
When a temperature gradient exists in a stationary medium, which may be a solid or fluid,
we use the term conduction to refer to the heat transfer that will occur across the medium.
For example,
(a) A solid rod is insulated along its periphery and kept the two ends open to
atmosphere. If heat is added to one end of the rod, you can feel that the other
end is hot after sometime. It is a case of pure conduction.
(b) Spoon used to stir a hot cup of coffee got heated at the other end is an
example of a heat conduction.
If the heat transfer occurs between a surface and a moving fluid (obviously with a
temperature difference between the two), the mode of heat transfer is called convection.
For example,
(a) Water heated in a pan is an example of convection.
(b) Steam flowing from boiler to turbine is an example of convection.
The third mode of heat transfer is the radiation. All surfaces at finite temperature emit
energy in the form of electromagnetic wave. Hence, in absence of an intervening medium,
there is net heat transfer by radiation between two surfaces at different temperatures.
Some examples of radiation are,
(a) Heat received by us from sun is a case of radiation. Similar is the case when
one feels hot from a fire away from him.
(b) Outside body of the air plane gets heated if flying over high. It is basically
radiation heat transfer from sun. In many situations, heat transfer occur as
mixed mode, i.e. conduction, convection and radiation can be prevalent in
reality.

1.4 MECHANISM OF HEAT TRANSFER

1.4.1 Conduction Heat Transfer
Conduction refers to the transfer of heat between two bodies or two parts of the same
body through molecules, which are, more or less, stationary, as in the case of solids.
The physical mechanism of conduction is most easily explained by considering a gas and
using ideas similar to thermodynamics.
Consider a gas in which there exists a temperature gradient and assume that there is no
bulk motion. The gas may occupy the space between the two surfaces that are maintained
12
at different temperatures as shown in Figure 1.5. We associate the temperature at any Concepts of Heat and
Mass Transfer
point with the energy of gas molecule in proximity to the point. This energy is related to
the random translational motion, as well as to the internal rotational and vibrational
motions of molecules.
T T1 >T2

q”x
X0 q”x

x T0

Figure 1.5 : Mechanism of Conduction Heat Transfer

Higher temperatures are associated with higher molecular energies, and when
neighbouring molecules collide, as they are constantly doing, a transfer of energy from
the more energetic to the less energetic molecules must occur. In the presence of a
temperature gradient, energy transfer by conduction must then be occur in the direction
of decreasing temperature. This transfer is evident from Figure 1.5. The hypothetical
plane at x0 is constantly being crossed by molecules from above and below due to their
random motion. However, molecules from above are associated with a larger temperature
than those from below, in which case there must be a net transfer of energy in the
positive x direction. We may speak of the net transfer of energy by random molecular
motion as a diffusion of energy.
The situation is much the same in liquids, although molecules are more closely spaced
and molecular interactions are stronger and more frequent. Similarly, in case of a solid,
conduction may be attributed to atomic activity in the form of lattice vibrations. The
modern view is to ascribe the energy transfer to lattice waves induced by atomic motion.
In a non-conductor, the energy transfer is exclusively via these lattice waves, in a
conductor it is also due to the translational motion of free electrons.
It is possible to quantify heat transfer processes in terms of appropriate rate equations.
These equations may be used to compute the amount of energy being transferred per unit
time. For heat conduction, the rate equation is known as Fourier’s law (J. B. J. Fourier,
French scientist, 1822). For the one dimensional or uni-directional plane wall
(Figure 1.6), having a temperature distribution T (x), the rate equation is expressed as :

T1

T(X)
q”x

T2

x
L
Figure 1.6 : One Dimensional Conduction Heat Transfer

dT
q x  . . . (1.1)
dx
13
Introduction to Heat dT
and Mass Transfer or, q x =  k . . . (1.2)
dx
dT
where q x is the rate of heat flux (a vector) in W/m2, is the temperature gradient in
dx
the direction of heat flow x and k is the constant of proportionality, which is a property of
the material through which heat propagates. This property of the material is called
thermal conductivity (W/MK). The negative sign is used because heat flows from a high
dT
to low temperature and the slope is negative.
dx
SAQ 1
(a) What are the various modes of heat transfer? Explain their differences.
(b) State the Fourier’s law of heat conduction.

Example 1.1
The wall of an industrial furnace is constructed from 0.20 m thick fireclay brick
having a thermal conductivity of 1.7 W/m.K. Measurements made during steady
state operation reveal temperatures of 1500 K and 1020 K at the inner and the
outer surfaces, respectively. What is the rate of heat loss through a wall that is
0.6 m by 1.2 m on a side?
Solution
Known : steady state conditions with prescribed wall thickness, area, thermal
conductivity and surface temperatures.
Find : Wall heat loss.
Schematic :
w
k

T1 T2 H

qx
q”x

Wall Area, A
x
L
x L
Assumptions :
(a) Steady state conditions.
(b) One dimensional conduction through the wall.
(c) Constant thermal conductivity.
Analysis :
Since heat transfer through the wall is by conduction, the heat flux may be
determined from Fourier’s law of heat conduction. Using Eq. (1.2), we have
T (1500  1020) K 480
qx  k  1.7 W/m.K   1.7   4080 W/m 2
L 0.2 m 0.2
The heat flux represents the rate of heat transfer through a section of unit
area, and it is uniform (invariant) across the surface of the wall. The heat
loss through the wall of area A = H  W is then
q x  ( H  W )  q x  (0.6  1.2)  4080  2937.6 W
14
Example 1.2 Concepts of Heat and
Mass Transfer
The heat flow rate through a wood board with a 5 cm thickness is 500 W/m2.
Temperature difference along the direction of flow of heat between the faces of the
wood board is 55oC. Calculate the thermal conductivity of the wood.
Solution
We know that conduction heat transfer across the wood board is given by
T
qx  k
L
qx . L
Hence, k
T
Here, T = 55oC, L = 5 cm and qx = 500 W/m2

Hence, thermal conductivity of the wood is

500 W/m 2  0.05 m

k  0.45 W/m.oC
55o C
1.4.2 Convection Heat Transfer
The convection heat transfer mode is comprised of two mechanisms. In addition to
energy transfer due to random molecular motion or diffusion, energy is also transferred in
bulk or macroscopic, motion of the fluid. This fluid motion is associated with the fact
that, at any instant, large numbers of molecules are moving collectively or as aggregates.
Such motion, in the presence of a temperature gradient, contributes to the heat transfer.
Because the molecules in the aggregate retain their random motion, the total heat transfer
is then due to a superposition of energy transport by random motion of the molecules and
by bulk motion of the fluid. It is customary to use the term convection when referring to
the cumulative transport and the term advection when referring to transport due to bulk
fluid motion.
We are specially interested in convection heat transfer, which occurs between a fluid in
motion and a bounding surface when the two are at different temperatures. Consider fluid
flow over the heated surface (Figure 1.7).

Fluid
y y
u T

Velocity Temperature
Distribution Distribution
u(y) T(y)
q”
Ts

u(y) Heated T(y)

Surface

Figure 1.7 : Development of Boundary Layer in Convection Heat Transfer

Consequence of the fluid surface interaction is the development of a region in the fluid
through which the velocity varies from zero at the surface to a finite value u associated
with the flow. This region of fluid is known as the hydrodynamic or velocity boundary
layer. Moreover, if the surface and flow temperatures differ, there will be a region of the
fluid through which the temperature varies from Ts at y = 0 to T in the outer flow. This
layer is called the thermal boundary layer, may be smaller, larger or of the same size as
that through which the velocity varies. In any case, if Ts > T, convection heat transfer
will occur between the surface and the outer flow.
15
Introduction to Heat The convection heat transfer mode is sustained both by random molecular motion and by
and Mass Transfer the bulk motion of fluid within the boundary layer. The contribution due to random
molecular motion (diffusion) dominates near the surface where the fluid velocity is low.
In fact, at the interface between the surface and the fluid (y = 0), the fluid velocity is zero
and heat is transferred by this mechanism only. The contribution of bulk fluid motion
originates from the fact that the boundary layer grows as the flow progresses in the
x-direction. In effect, the heat that is conducted into the layer is swept downstream and is
eventually transferred to the fluid outside the boundary layer. Appreciation of the
boundary layer phenomena is essential to understanding convection heat transfer. It is for
this reason that the discipline of fluid mechanics will play a vital role in our later analysis.
Convection heat transfer may be classified according to the nature of the flow. We speak
of force convection when the flow is caused by external means, such as by a fan, a pump,
or atmospheric winds. As an example, consider the use of a fan to provide forced
convection air-cooling of hot electrical components on a stack of printed circuit boards
(Figure 1.8). In contrast, for free (or natural) convection the flow is induced by buoyancy
forces, which arises from density differences caused by temperature variations in the
fluid. An example is the free convection heat transfer that occurs from hot components
on a vertical array of circuit boards in still air (Figure 1.8(b)). Air that makes contact with
the components experiences an increase in temperature and hence a reduction in density.
Since, it is now lighter than the surrounding air, buoyancy forces induce a vertical motion
for which warm air ascending from the boards is replaced by an inflow of cooled ambient
air.
Buoyancy-driven
Flow

Forced Air
Flow
q” Hot Components
On Printed q”
Circuit Boards

Air

(a) (b)

Water
Moist air Droplets

Vapor Cold
Bubbles Water Water

Hot Plate

(c) (d)
Figure 1.8 : Convection Heat Transfer Processes (a) Forced Convection and (b) Natural Convection

While we have presumed pure forced convection in Figure 1.8(a) and pure natural
convection in Figures 1.8(b)-(d), conditions corresponding to mixed convection, i.e.
combined forced and natural convection may exist.
For example, if the velocities associated with the flow in Figure 1.8(b) are small and/or
buoyancy forces are large, a secondary flow comparable to the imposed forced flow
could be imposed. The buoyancy induced flow will be normal to the forced flow and
could have a significant effect on convection heat transfer from the components. In
Figure 1.8(b) mixed convection would result if a fan were used to force air upward, there
by opposing the buoyancy force.
16
We have described the convection heat transfer mode as energy transfer occurring within Concepts of Heat and
Mass Transfer
fluid due to the combined effects of conduction and bulk fluid motion. Typically, the
energy that is being transferred is the sensible, or internal thermal energy of the fluid.
However, there are convection processes for which there is, in addition, latent heat
exchange. This latent heat exchange is due to the phase change between the liquid and
vapour state of the substance. These cases are dealt in Unit 15.
Regardless of the particular nature of convection heat transfer process, the appropriate
rate equation is of the form
q  h (Ts  T ) . . . (1.3)
where q = Convective heat flux (W/m2),
Ts = Surface temperature (K),
T = Fluid temperature (K), and
h = Convective heat transfer coefficient (W/m2.K)
Eq. (1.3) is also known as the Newton’s law of cooling.
Example 1.3
A hot plate of area 0.5 m2 is maintained at a temperature of 60oC by a 100 W
electric heater when room temperature is 30oC. The appropriate convection
coefficient is 2.15 (T)1/3 W/m2K. What fraction of the heat supplied is lost by
natural convection? What happens to the rest of the heat supplied?
Solution
The convective heat transfer coefficient
1 1 1
h  2.15 T 3  2.15  (60  30) 3  2.15  30 3  6.605
Q  h A T  h  0.5  30  6.605  0.5  30  99.075

Q
Fraction of supplied heat lost by convection is  0.99075  99.075%
100
The remaining 0.00925 or 0.925% is lost by radiation.
Example 1.4
Air at 30oC flows over a hot plate (50 cm  75 cm) maintained at 200oC with the
help of an electric heater. The convection coefficient is 20 W/m2. Calculate the
heat transfer.
Solution
From Newton’s law of cooling
q  hA (Tw  T )
 20  (0.5  0.75) (200  30)
= 1275 W
SAQ 2
(a) What is convection heat transfer? Why is it regarded as a mode of heat
transfer?
(b) Give two examples each of natural and forced circulation.

17
Introduction to Heat 1.4.3 Radiation Heat Transfer
and Mass Transfer
Consider a solid initially at a high temperature Ts than that of its surroundings Tsur as
shown in Figure 1.9. The entire medium between the solid and the surrounding is
vacuum. Consideration vacuum precludes other modes of heat transfer (i.e. conduction
and convection) between the solid and the surroundings. Our intuition tells us that the
solid will cool and eventually achieve thermal equilibrium with its surroundings. This
cooling of solid is associated with a reduction in the internal energy stored by the solid
and is a direct consequence of the emission of thermal radiation from the surface. In turn,
the surface of the solid will intercept and absorb radiation originating from the
surroundings. If, Ts > Tsur, the net heat transfer by radiation qrad, net is from the solid
surface and the same will cool until Ts reaches Tsur.

Gas
T, h Gas
T, h
G E Surroundings
q”con At Tsur q”rad q”con
v v

Surface of Emissivity Surface of Emissivity

, Absorptivity , and ,=  Area A, and
Temperature T, Temperature T,
(a) (b)
Figure 1.9 : Radiative Heat Transfer

Radiation emitted by the surface originates from the thermal energy of matter bounded
by the surface, and the rate at which energy is released per unit area (W/m2) is termed the
surface emissive power E. There is an upper limit to the emissive power, which is
prescribed by the Stefan-Boltzmann law

Eb   Ts4 . . . (1.4)
where Ts = absolute temperature of the surface (K), and
 = Stefan-Boltzmann constant = 5.67  10– 8 W/m2K4.
In the Eq. (1.4) we have assumed the surface of the solid to be a perfect emitter. Details
regarding other type of emitters are discussed in appropriate units.
Now, heat transfer between the surface and surrounding is given by

Es  sur   (TS4  Tsur

4
) . . . (1.5)

We associate thermal radiation with the rate at which energy is emitted by matter as a
result of its finite temperature. At this moment thermal radiation is being emitted by all
the matters that surrounds you, by the furnitures, walls of the room, etc. if you are indoor,
or by the ground, buildings, atmosphere and sun, if you are outdoors. The mechanism of
emission is related to the energy released as a result of oscillations or transitions of the
many electrons that constitute the matter. These oscillations are again sustained by
internal energy, and therefore the temperature of the matter. Hence we associate the
emission of thermal radiation with thermally excited conditions within the matter.
We know that radiation originates due to emission by matter and that its subsequent
transport does not require the presence of any matter. At this point we should know the
nature of transport of radiation. One theory views radiation as the propagation of a
collection of particles termed photon or quanta. Alternatively, radiation may be viewed as
the propagation of electromagnetic waves. In any case we wish to attribute to radiation
the standard wave properties of frequency  and the wavelength . For radiation
propagating in a particular medium, the two properties are related by
c
 . . . (1.6)

18
where c is the speed of light in the medium. For propagation in a vacuum, Concepts of Heat and
Mass Transfer
c0 = 2.998  108 m/s. The unit of wavelength is commonly the micrometer (m), where
1 m = 10– 6 m.
Example 1.5
The quantity of radiation received by earth from the sun is 1.4 kW/m2 (solar
constant). Assuming that sun is an ideal radiator, calculate the surface temperature
of the sun. The ratio of the radius of earth’s orbit to the radius of the sun is 216.
Solution
Total radiation from the sun

Qr  1.4  4 R 2

where R  radius of the earth’s orbit.

Total radiation emitted by the sun

Qr    4 r 2  T 4
where r = radius of the sun, and
T = surface temperature of the sun.

Now,   4 r 2  T 4  1.4  4 R 2
2
1.4  R  1.4  10 3
or, T4      (216)2  0.1152  1016 K 4
 r 5.67  10  8
Hence, T = 5826 K.
SAQ 3
(a) What is the mode of heat transfer in vacuum?
(b) State the Stefan-Boltzmann law of radiation.

1.5 MASS TRANSFER

Diffusional mass transfer occurs at a microscopic or molecular level which deals with the
transport of one constituent of a fluid solution or gas mixture from a region of higher
concentration to a region of lower concentration. Heat is transferred in a direction which
reduces an existing temperature gradient, and mass is transferred in a direction which
reduces an existing concentration gradient.
The rate of molecular diffusion is proportional to the concentration gradient
NA dC A
 . . . (1.7)
A dy
NA
where  diffusion rate per unit area (kg mol/m2s) of the diffusion species A,
A
dC A
 concentration gradient (kg mol/m4) in the direction of diffusion Y, and
dy
CA = concentration of species A in kg mol/m3.
Now Eq. (1.7) can be written as
NA dC A
=D . . . (1.8)
A dy 19
Introduction to Heat where D  constant of proportionality, called the diffusivity (m2/s).
and Mass Transfer
Eq. (1.8) is called the Fick’s law of diffuision, which is similar to the Fourier’s law of
heat conduction.
Whenever there is concentration gradient there will be mass transfer, till the
concentration of the particular constituent becomes uniform.
When mass transfer occurs in a fluid at rest, the mass is transferred by purely molecular
diffusion resulting from concentration gradient. The process is similar to heat diffusion
resulting from temperature gradient.
When the fluid is in motion, mass transfer takes place by both molecular diffusion and
convective motion of the bulk fluid. In such a case knowledge of velocity field is needed
to solve the mass transfer problem. It may be noted that bulk velocities are insignificant
in case of molecular diffusion. However, convective mass transfer involves bulk motion
of the fluid. The convective mass transfer is given by the relation
m A
 hM (Cw  C  ) . . . (1.9)
A
where Cw and C are the concentrations of the species A at the wall and free stream; hM is
the convective mass transfer coefficient (ms– 1).
For low concentration of the mass in the fluid and low mass transfer rates, the convective
heat and mass transfer processes are analogous and many of the results derived in
connection with convective heat transfer are applicable to convective mass transfer.
Therefore, mass transfer equations discussed in Units 3, 11-13 are obtained by analogy
directly from the corresponding heat transfer equations. However, under high mass flux
condition and with chemical reactions there are significant differences between the heat
and mass transfer processes. Such high level situations are beyond the scope of this
course.
Mass transfer processes occur in variety of applications in mechanical, chemical and
aerospace engineering, physics, chemistry and biology. Typical examples are :
(a) Transpiration cooling of rocket motors and jet engines.
(b) Ablative cooling of space vehicles during reentry into the atmosphere.
(c) Mass transfer from laminar and turbulent streams onto the surface of a
conduit.
(d) Evaporation and condensation on the surface of a tube or plate.
(e) Processes such as absorption, desorption, distillation, solvent extraction,
drying, humidification, sublimation, etc.
(f) Biological applications – oxygenation of blood, food and drug assimilation,
respiration, etc.
Example 1.6
Gaseous hydrogen is stored at elevated pressure in a rectangular steel container of
10 mm wall thickness. The molar concentration of hydrogen in steel at the outer
surface is 2 kg mol/m3, while the concentration of hydrogen in steel at the outer
surface is 0.5 kg mol/m3. The binary diffusion coefficient for hydrogen in steel is
0.26  10– 12 m2/s. What is the mass flux of hydrogen through the steel?
Solution
Assumptions
(a) Steady state, one-dimensional diffusion of hydrogen.
(b) No chemical reaction involved.
(c) Molar concentration of hydrogen is much less than that of steel
(CA < < CB). So C = CA + CB is uniform

20
 C A2  C A1 Concepts of Heat and
NA
Now,   DAB Mass Transfer
A L
NA (2  0.5) m 2 (kg mol/m 3 )
or,  0.26 10 12  .  0.39 10 10 kg mol/m 2 .s
A 0.01 s m
NA
Mass flux of hydrogen =  M  0.39  10 10  2  7.8  10 11 kg/m 2 .s
A
SAQ 4
(a) State and explain the Fick’s law of diffusion.
(b) What is mass diffusivity? What is its dimension?
(c) How does diffusional mass transfer differ from convection mass transfer?

1.6 TEMPERATURE FIELD

As mentioned in the previous section, heat transfer occurs due to the temperature
difference or a temperature gradient. As long as the temperature difference exists
between two locations, heat will always transfer from higher to the lower temperature. A
majority heat transfer problem involves calculation of temperature at a location in the
domain under consideration provided the heat flux and geometry are known. A
temperature field in the domain of study can indicate the direction of heat flow.
In some cases, boundary condition may appear as temperature boundary condition. In
some cases, temperature gradient may be used as a boundary condition. In a domain, loci
of all the points having same temperature is known as isothermal line. No heat can flow
along the isothermal line. However, heat will always flow from one isotherm to the
another (both isotherms being at two different temperature).
In the heat transfer study, temperature may be unsteady or steady. Unsteady temperature
involves variation of temperature with time. For example, if we heat one end of a
metallic rod and hold the other end with hand, we can feel the variation in warmth with
time. This variation will continue till the heat transfer attains a steady state (invariant
with time). Transient temperature being time variant, it can be plotted for each location
with time as a variable. In case of steady state temperature, plot of temperature can be
done with spatial variables (x, y and z).

1.7 RELATIONSHIP TO THERMODYNAMICS

Thermodynamics essentially deals with systems in equilibrium state to another. But,
thermodynamics may not be used to calculate the rate at which this change takes place,
since the system is not in equilibrium during the process. Heat transfer utilizes the first
and second laws of thermodynamics in addition to the rate laws such as Fourier’s law of
heat conduction, Newton’s law of cooling and the Stefan-Boltzmann law of radiation.
To show the difference between heat transfer and thermodynamics, let us take a practical
application such as quenching of a hot steel bar in an oil bath. Thermodynamics will help
us calculate the final equilibrium temperature of the steel bar-oil combination.
However, it will not be helpful in predicting the time that will be taken by the steel
bar-oil combination to reach the steady state or the rate at which the temperatures of the
steel bar and oil will change with time. On the other hand, heat transfer will help us in
finding out the time-temperature history of both the bar and oil.
Thermodynamics is concerned with interaction of heat and work at the boundary of the
system and is concerned with the equilibrium states of the matter. Equilibrium state
necessarily precludes the existence of a temperature gradient. Thermodynamics is
governed by the laws of thermodynamics. Amount of heat and/or work and directions of
flow during a process can be estimated by using the laws of thermodynamics. 21
Introduction to Heat Heat transfer is inherently a non-equilibrium process and it occurs only because of a
and Mass Transfer temperature gradient. Heat transfer quantifies the rate at which the same occurs in terms
of degree of thermal non-equilibrium. This is done through the rate equations for the
three modes as discussed by Eqs. (1.2, 1.3, 1.4, 1.8 and 1.9).

1.8 UNITS AND DIMENSIONS

The physical quantities of heat transfer are specified in terms of dimensions, which are
measured in terms of units. Four basic dimensions are required for the development of
heat transfer; they are length (L), mass (M), time (t) and temperature (T). All other
physical quantities of interest may be related to these four basic dimensions. In this
course we will be using SI system of units throughout.
Table 1.1 : Dimensions and Units in SI System
Dimension Unit
Length (L) meter (m)
Mass (M) Kilogram (kg)
Time (t) second (s)
Temperature (T) kelvin (K)

SI base and supplementary units are given in Table 1.2

Table 1.2 : Supplementary Dimensions and Units in SI System
Dimension Unit
Length (L) meter (m)
Mass (M) kilogram (kg)
Concentration (C) mole (mol)
Time (t) second (s)
Electric current (I) ampere (A)
Thermodynamic temperature (T) kelvin (K)
Plane angle () radian (rad)
Solid angle () steradian (sr)
Luminous intensity Candela (d)

With regard to these units note that 1 mol is the amount of substance that has as many
atoms or molecules as there are atoms in 12 g of carbon-12 (12C); this is the
gramsmole (mol). Although mole has been recommended as the unit quantity of matter
for the SI system, it is more consistent to work with kilogram – mol (kmol, kg-mol).
One kmol is simply the amount of substance that has as many atoms or molecules as
there are atoms in 12 kg of 12C. As long as the use is consistent within a given problem,
no difficulties arise in using either mol or kilogram-mole. The molecular weight of a
substance is the mass associated with a mole or a kilogram-mole. For oxygen, as an
example, the molecular weight M is 16 g/mol or 16 kg/kmol.
Although the SI unit of temperature is the Kelvin, use of the Celsius temperature scale
remains widespread. Zero on the Celsius scale (0oC) is equivalent to 273.15 K on the
thermodynamic scale in which case

T ( K )  T ( o C)  273.15 . . . (1.10)
However, temperature differences are equivalent for the two scales and may be denoted
as oC or K. Also, although the SI unit of time is the second, other units of time (minute,
hour, and day) are so common that their use with the SI system is generally accepted.
The SI units comprise a coherent form of the metric system. That is, all remaining units
may be derived from the base units using formulas that do not involve any numerical
factors. Derived units for selected quantities are given in Table 1.3.
22
 Table 1.3 : Derived Units Under SI System Concepts of Heat and
Mass Transfer
Dimension Name and Symbol Formula SI Base Unit
Force Newton (N) m.kg/s2 m.kg/s2
Pressure/stress Pascal (Pa) N/m2 kg/m.s2
Energy Joule (J) N.m m2.kg/s2
Power Watt (W) J/s m2.kg/s3

Note that force is measured in newtons, where a 1-N force will accelerate a 1-kg mass at
1 m/s2. Hence, 1 N = 1 kg.m/s2. The unit of pressure (N/m2) is often referred to as the
pascal. In the SI system there is one unit of energy (thermal, mechanical, or electrical),
called the joule (J), and 1 J = 1 N.m. The unit for energy rate, or power is then J/s, where
one joule per second is equivalent to one watt (1 J/s = 1 W). Since it is frequently
required to work with extremely large or small numbers, a set of standard prefixes has
been introduced to simplify matters (Table 1.4). For example, 1 megawatt (MW) = 106 W
and 1 micrometer (m) = 10– 6 m.
Table 1.4 : Multiplying Prefixes
Prefix Abbreviation Multiplier
pico p 10– 12
nano n 10– 9
micro  10– 6
milli m 10– 3
centi c 10– 2
hecto h 102
kilo k 103
mega M 106
giga G 109
tera T 1012

SAQ 5
(a) Describe the relationship between heat transfer and thermodynamics.
(b) What are the basic units? How do you obtain force, pressure, energy and
power from the basic units?
Exercise 1.1
(a) A temperature difference of 510oC is maintained across a fireclay brick
25 cm thick with thermal conductivity 3.5 W/moC. Determine heat transfer
rate per square meter area.
(b) A brick wall 20 cm thick with thermal conductivity 2.45 W/moC is
maintained at 32oC on one face and 220oC at the other face. Determine the
heat transfer rate across 4.2 m2 surface area of the wall.
(c) The inside and outside surface temperatures of a window glass are 22oC and
– 15oC, respectively. If the dimension of the glass is 82 cm  43 cm and the
thickness of the glass is 1.75 cm, determine the heat loss through the glass
over 4 hours. Assume the thermal conductivity of the glass
kglass = 0.778 W/moC.
(d) 250 W is transferred by conduction heat transfer through a 0.58 m2 section
of a 4.2 cm thick insulating material. Determine the temperature difference
across the insulating layer if the thermal conductivity is 0.12 W/moC.
23
Introduction to Heat Exercise 1.2
and Mass Transfer
(a) Water at mean temperature of 22oC flows over a flat plate maintained at
84oC. Determine the heat transfer per square meter of the plate over 5 hours.
Consider heat transfer coefficient to be 220 W/(m2.oC).
(b) Heat is supplied to a plate from its back surface at a rate of 1100 W/m2 and
is removed from its front surface by air flow at 27oC. If the heat transfer
coefficient between the air and the plate surface is 28 W/(m2.oC). What is
the temperature of the front surface of the plate?
(c) The inside surface of an insulating layer is at 266oC, and the outside surface
is dissipating heat by convection into air at 33oC. The insulating layer is
12 cm thick and has thermal conductivity k = 0.22 W/m.oC. What is the
minimum value of the heat transfer coefficient at the outside surface if the
outside surface temperature should not exceed 55oC?
(d) A thin metallic plate is insulated at the back surface and is exposed to the
sun at the front surface. The front surface absorbs the solar radiation of 1200
W/m2 and dissipates it mainly by convection to the ambient air at 25oC. If
the heat transfer coefficient between the plate and the air is
25 W/m2.oC, what is the temperature of the plate?
Exercise 1.3
(a) A thin metal plate 0.15 m by 0.1 m is placed in a large evacuated container
whose walls are kept at 320 K. The bottom surface of the plate is insulated,
and the top surface is maintained at 550 K as a result of electrical heating. If
the emissivity of the surface of the plate is  = 0.75, what is the rate of heat
exchange between the plate and the walls of the container?
Take   5.67  10  8 W/(m 2 .K 4 ) .

(b) Two very large, perfectly black parallel plates, one maintained at 1250 K
and the other at 650 K, exchange heat by radiation. Determine the heat
transfer rate per square meter of the surface.
(c) One surface of a thin plate is exposed to a uniform flux of 600 W/m2, and
the other side dissipates heat by radiation to an environment at – 8oC.
Determine the temperature of the plate assuming blackbody conditions for
radiation.
(d) A pure gas is stored at elevated pressure in a rectangular stainless steel
container of 20 mm wall thickness. The molar concentration of the gas in
stainless steel at the outer surface is 4.5 kg mol/m3, while the concentration
of the gas in stainless steel at the outer surface is 1.5 kg mol/m3. The binary
diffusion coefficient for the gas in stainless steel is 0.88  10–10 m2/s. What
is the mass flux of the gas through the steel?

1.9 SUMMARY
It is understood that heat and mass transfer are transport processes and they occur due to
the presence of finite gradients (temperature gradient for heat transfer and concentration
gradient for the mass transfer). In the present unit, importance of heat transfer is
discussed and application to many practical devices like heat exchangers are cited. A
short description is provided about various modes of heat transfer. Basics of mass
transfer is highlighted. Further, relationship between the heat transfer and
thermodynamics is discussed. The idea of units and dimensions are incorporated which
will help in problem solving. Theory is supported with solved problems for better
understanding of the unit. At the end, some unsolved problems are given which will help
you to understand the topic.
24
 Concepts of Heat and
1.10 KEY WORDS Mass Transfer

Heat Transfer : Heat transfer is the process of transfering heat

from a hot body to a cold body.
Conduction : It is the mode of heat transfer due to molecular
diffusion.
Convection : It is the mode of heat transfer due to the motion of
fluid molecules.
Radiation : It is the mode of heat transfer due to the
propagation electromagnatic wave or photon.
Mass Transfer : Mass transfer is the process of transfering mass
through a concentration gradient.

1.11 ANSWER TO SAQS

Refer the preceding text for all the Answers to SAQs.

25
 Different Modes of
UNIT 2 DIFFERENT MODES OF HEAT Heat Transfer

TRANSFER
Structure
2.1 Introduction
Objectives
2.2 Conduction
2.2.1 Heat Conduction through a Plane Wall
2.2.2 Resistance Concept
2.2.3 Composite Walls
2.2.4 Contact Resistance
2.2.5 Thermal Conductivity
2.2.6 Heat Conduction through a Cylinder
2.2.7 Heat Conduction through a Sphere
2.2.8 Overall Heat Transfer Coefficient
2.2.9 Thermal Insulation
2.2.10 Critical Radius of Insulation
2.3 Convection
2.4 Radiation
2.5 Combined Heat Transfer Mechanism
2.6 Comparison of Conduction, Convection and Radiation
2.7 Summary
2.8 Key Words
2.9 Answers to SAQs

2.1 INTRODUCTION
Present unit is devoted to the three modes of heat transfer. Application of Fourier’s law
of heat conduction for calculation of heat flow in some simple one dimensional systems
such as plane wall, cylinder and sphere are discussed in details. Some important aspects
associated with conduction heat transfer, such as, thermal insulation, contact resistance,
etc. are described in this unit. More exposure is given to convection and radiation heat
transfer.
Objectives
After studying this unit, you should be able to
 find heat transfer rate from a plane wall, cylinder or a sphere,
 recognize the importance of thermal conductivity,
 identify insulating materials and thickness of insulation calculation,
 estimate convective and radiative heat transfer and combined heat transfer
due to the presence of all mode of heat transfer, and
 appreciate the difference between the different modes of heat transfer.

2.2 CONDUCTION
An introduction to the conduction heat transfer and the mechanism involved was
introduced in Unit 1 (Section 1.4.1). Fourier’s law associated with conduction heat
transfer is phenomenological. This implies that it is developed from observed phenomena
27
Introduction to Heat rather than being derived from first principles. Hence, we view the rate equation as a
and Mass Transfer generalization based on much experimental evidence.
Consider the steady-state conduction through a cylindrical rod as shown in Figure 2.1.
The rod is insulated on its lateral surface, while its end faces are maintained at different
temperatures, with T1 > T2. We assume that the material properties of the rod are known.
The temperature difference causes conduction heat transfer in the positive x direction.

A, T1 T=T1 - T2 T2

qx

x
x

Figure 2.1 : Steady State Heat Conduction through a Cylindrical Rod

We are able to measure the heat transfer rate qx, and we seek to determine how qx
depends on the following variables : T, the temperature difference; x, the rod length;
and A, the cross-sectional area.
We might imagine the following :

(a) Consider that both the T and x are constant and A is a variable. If we do
so, we find that qx is directly proportional to A.

(b) Similarly, holding T and A constant, we observe that qx varies inversely

with x.

(c) We can also assume that both the A and x are constant, we find that qx is
directly proportional to T.
The collective effect is then

T
qx  A . . . (2.1)
x
In changing the material (e.g. from a metal to a plastic), we would find that the above
proportionality remains valid. However, we would also find that, for equal values of A,
x, and T, the value of qx would be smaller for the plastic than for metal. This suggests
that the proportionality may be converted to an equality by introducing a coefficient that
is a measure of the material behavior. Hence, we write

T
q x   kA . . . (2.2)
x
where k, the thermal conductivity (W/m.K), is an important property of the material.
Evaluating this expression in the limit as x  0, we obtain for the heat rate

dT
q x   kA . . . (2.3)
dx
or for the heat flux

qx dT
qx  k . . . (2.4)
A dx

28
As mentioned in Unit 1, the minus sign is necessary because heat is always transferred in Different Modes of
Heat Transfer
the direction of decreasing temperature.
Fourier’s law, as written in Eq. (2.2), implies that the heat flux is a directional quantity.
In particular, the direction of qx is normal to the cross-sectional area A. or, more
generally, the direction of heat flow will always be normal to a surface of constant
temperature, called an isothermal surface. Figure 2.2 illustrates the direction of heat flow
dT
qx in a plane wall for which the temperature gradient is negative. From
dx
Eq. (2.2), it follows that qx is positive. Note that the isothermal surfaces are planes
normal to the x direction.

T(x)

T1

q”x

T2

Figure 2.2 : Heat Transfer through a Plane Wall

Recognizing that the heat flux is a vector quantity, we can write a more general statement
of the conduction rate equation (Fourier’s law) as follows :
 T ˆ T T 
q   k T   k  iˆ  j  kˆ  . . . (2.5)
 x y z 
where  is the three-dimensional del operator and T (x, y, z) is the scalar temperature
field. It is implicit in Eq. (2.3) that the heat flux vector is in a direction perpendicular to
the isothermal surfaces. An alternative form of Fourier’s law is therefore,
T
qn   k . . . (2.6)
n
where qx is the heat flux in a direction n, which is normal to an isothermal, as shown for
the two-dimensional case in Figure 2.3.
qy” qn”

qx”
y

n x
Isotherm

Figure 2.3 : Heat Flux Vector Normal to an Isotherm in a 2-D Coordinate System

The heat transfer is sustained by a temperature gradient along n. Note also that the heat
flux vector can be resolved into components such that, in Cartesian coordinates, the
general expression for q is
q  i qx  j qy  k qz . . . (2.7)

where, from Eq. (2.5), it follows that 29

Introduction to Heat T
and Mass Transfer qx   k . . . (2.8)
x
T
qy   k
y
T
qz   k
z
Each of these expressions relates the heat flux across a surface to the temperature
gradient in a direction perpendicular to the surface. It is also implicit in Eq. (2.5) that the
medium in which the conduction occurs is isotropic. For such a medium the value of the
thermal conductivity is independent of the coordinate direction.
Fourier’s law is the cornerstone of conduction heat transfer, and its key features are
summarized as follows :
(a) It is not an expression that may be derived from first principles; it is instead
a generalization based on experimental evidence.
(b) It is an expression that defines an important material property, the thermal
conductivity.
(c) In addition, Fourier’s law is a vector expression indicating that the heat flux
is normal to an isotherm and in the direction of decreasing temperature.
(d) Finally, note that Fourier’s law applies for all matter, regardless of its state
(solid, liquid, or gas).
2.2.1 Heat Conduction through a Plane Wall
For the simple case of steady-state one-dimensional heat flow through plane wall
(Figure 2.4), the temperature gradient and the heat flow do not vary with time, so from
Eq. (2.4),
L T2
dT
qx  dx   k  dx
. . . (2.9)
0 T1

Physical System

T1 = Thot
Tx
qk

T2 = Tcold

Figure 2.4 : Steady State Conduction through a Plane Wall

where the temperature at the left face (x = 0) is uniform at T1 and the temperature at the
right face (x = L) is uniform at T2.
If k is independent of T, we obtain after integration of Eq. (2.9)
T1  T2
qx  k . . . (2.10)
L
T
or, qx  k . . . (2.11)
L
If A is the surface area normal to heat flow, then the rate of heat transfer in Watt is
30
 T Different Modes of
Q x = kA . . . (2.12) Heat Transfer
L
dT q dT
Since, =  x , for the same qx, if k is low (i.e. for an insulator), will be large, i.e.
dx k dx
there will be large temperature difference across the wall, and if k is high (i.e. for a
dT
conductor), will be small, or there will be a small temperature difference across the
dx
wall (Figure 2.5).
L

T1

T2

L
R= Q
kA

Q
T1 T2
Figure 2.5 : Thermal Resistance Offered by a Plane Wall
2.2.2 Resistance Concept
Heat flow has an analogy to flow of electricity. Ohm’s law states that the current i
flowing through a wire (Figure 2.6) equal to the voltage potential E1 – E2, divided by the
electrical resistance Re
Thermal Circuit
qk

T1 T2

L
Rk =
Ak

Electrical Circuit
i

E1 E2

Re
Figure 2.6 : Analogy Between Thermal and Electrical Circuits Corresponding to Figure 2.4
E1  E2
or I = . . . (2.13)
Re
Since the temperature difference and heat flux in conduction are similar to the potential
difference and electric current respectively, the rate of heat conduction through the wall
Eq. (2.12) can be written as
T  T2 T1  T2
Q= 1  . . . (2.14)
L Rc
kA
L
where Rc  is the conductive thermal resistance to heat flow offered by the wall.
kA 31
Introduction to Heat Again electrical resistance is related to the specific resistance as
and Mass Transfer
l
Re   . . . (2.15)
A
where  is the specific resistance (.m), l is the length of the conductor and A is the
cross-sectional area of the conductor. Eq. (2.13) can be written as
E1  E2 E  E2 dE
I =  A 1  A . . . (2.16)
l l dl

A
I dE
or, i = . . . (2.17)
A dl
1 dE
where   is the electrical conductivity and is the potential gradient. The
 dl
reciprocal of the thermal resistance is referred to as thermal conductance, KC defined by
kA
KC  . . . (2.18)
L
k
The ratio is the thermal conductance per unit area.
L
Concept of resistance is very useful for analysis of conduction heat transfer problems,
particularly for composite systems.
2.2.3 Composite Walls
In industrial heat transfer problems one is often concerned with conduction through walls
made up of layers of various materials, each with its own thermal conductivity. We can
establish how various resistances to heat transfer are combined into a total resistance.
The composite wall, as shown in Figure 2.7, has three materials of different thicknesses
L1, L2 and L3 with different thermal conductivity k1, k2 and k3, respectively.
Physical System

Material 1 Material 2 Material 3

K1 K2 K3

qk qk
qk

L1 L2 L3

Thermal Circuit

T1 T2 T3 T4

L1 L2 L3
R1  R2  R3 
K1A K2A K3A

Figure 2.7 : Heat Conduction Through Composite Walls

Wall 1 is in contact with a fluid at temperature T1. There are three resistances in series.
The rate of heat conduction is the same throughout the sections. The slope of the
temperature profile in each depends on the thermal conductivity k, the more will be the
32
slope and the higher is the temperature difference. The higher the k, the less will be the Different Modes of
Heat Transfer
slope and lower is the temperature difference.
The total thermal resistance is
L1 L L
R  R1  R2  R3   2  3 . . . (2.19)
k1 A k2 A k3 A

Now, the rate of heat conduction is

T1  T4 T1  T4
QC   . . . (2.20)
R L L L
1
 2  3
k1 A k2 A k3 A

 kA   kA   kA 
Also QC    (T1  T2 ) =   (T2  T3 )    (T3  T4 ) . . . (2.21)
 L 1  L 2  L 3
where T2 and T3 are the interface temperatures. The walls are assumed to be in good
thermal contact, with no contact resistance.
Conduction can occur in a wall with two different materials in parallel (Figure 2.8).
T1 T2

A1 k1

A qk

A2 k2

Physical System

Figure 2.8 : Conduction through Two Resistances in Parallel

If the temperature over the left and right faces are uniform at T1 and T2, the equivalent
thermal circuit is shown in Figure 2.9.
R1 = L
k1A1

T1 T2

L
R2 =
k2A2

Thermal Circuit

Figure 2.9 : Thermal Circuit for Figure 2.8

The total resistance R is given by

1 1 1
  . . . (2.22)
R R1 R2
33
Introduction to Heat R1 R2
and Mass Transfer or R . . . (2.23)
R1  R2
L L
where R1  and R2 
k1 A1 k2 A2
The rate of heat conduction is
T1  T2 (T  T2 )
QC   (k1 A1  k 2 A2 ) 1 . . . (2.24)
R L
Since heat is conducted through two separate parallel paths between the same
temperature difference,
T1  T2 T1  T2
QC  Q1  Q2   . . . (2.25)
 L   L 
   
kA
 1  kA  2
Example 2.1
Consider a conduction heat transfer problem through a complex composite
structure as shown in Figure 2.10. Here thermal resistances are both in series and
parallel (Figure 2.10). Total resistance of the system as a whole is
RT  R1  R2  R3  R4  R5
= R1  R3  R5  ( R2  R4 ) . . . (2.26)

Section1 Section 2 Section 3

T1 Material 1 Material 2 Material 4

K1 K2 K4
qk

Material 3
K3 If K2 > K3
More Heat Flows Thru B:

2
L1 L2 = L3 L4

L
R2 = 2 1 4
k A
2 2
3
T1 qk Tx Ty qk T2

L1 L4
R1 = R4 =
Actual heat flow assumed as if heat flow is 1-D
k1A1 k4A4 and isothermal interfaces to yield the said result
L3
R3 =
k3 A3

Figure 2.10 : Heat Conduction Through Resistances in Series and Parallel

SAQ 1
(a) How does conduction occur in a composite wall with different materials put
in (i) series and (ii) parallel?
(b) How are Fourier’s law and Ohm’s law similar?
(c) Explain the resistance concept to illustrate the analogy of heat flow and the
flow of electricity.
2.2.4 Contact Resistance
Thermal contact resistance develops when two conducting surfaces do not fit tightly
together and a thin layer of fluid is trapped between them (Figure 2.11). This resistance is
primarily a function of surface roughness, the pressure holding the two surfaces in
34
contact. The interface fluid resistance and interface temperature. The direct contact Different Modes of
Heat Transfer
between the two solid surfaces, as shown in the expanded view (Figure 2.11), takes place
at a limited number of spots, and the voids between them are usually filled with air or
surrounding fluid. Heat transfer through the fluid filling the voids is mainly by
conduction, since there is no convection in such a thin layer of fluid and the radiation
effects are negligible at normal temperatures.

Interface Fluid

2
qk

1 2

Expanded View
of Interface

Contact Interface

T
Temperature Drop
Through Contact
Ts1
Resistance =  T1

T1 contact

T2 contact Ts2
x

Figure 2.11 : Contact Resistance between Two Solid Bodies

If the heat flux through the two solid surfaces in contact is q and the temperature
difference across the gap is  Ti (  Ts1  Ts 2 ) , the interface resistance Ri is defined by
Ti
Ri  . . . (2.27)
q
The effect of contact pressure on the thermal contact resistance between metal surfaces
under vacuum conditions is presented in Table 2.1. An increase in contact pressure can
reduce the contact resistance significantly.
Table 2.1 : Thermal Contact Resistance at different Contact Pressures
under Vacuum Conditions [1]
Resistance Ri Contact Resistance Ri (m2K/W)
Interface Material
Pressure (1 Bar) Contact Pressure (100 Bar)
Stainless steel 6-25 0.7-4.0
Copper 1-10 0.1-0.5
Magnesium 1.5-3.5 0.2-0.4
Aluminium 1.5-5.0 0.2-0.4

The interfacial fluid also affects the thermal resistance, as shown in Table 2.2. Putting a
viscous liquid like glycerin on the interface reduces the contact resistance 10 times with
respect to air at a given pressure. A thermally conducting liquid called a thermal grease
such as silicone oil is applied between the contact surfaces before they are pressed
against each other. This is commonly done when attaching electronic components such as
power transistors to heat sinks.

35
Introduction to Heat Table 2.2 : Thermal Contact Resistance for Aluminium-Aluminium Interface
and Mass Transfer with different Interfacial Fluids having 1 m Surface Roughness
under 1 Bar Contact Pressure [1]

Interfacial Fluid Resistance Ri (m2 K/W)

Air 2.75  10– 4
Helium 1.05  10– 4
Hydrogen 0.72  10– 4
Silicon oil 0.525  10– 4
Glycerin 0.265  10– 4

SAQ 2
(a) What do you understand by thermal contact resistance?
(b) What parameters does the contact resistance depend?
(c) Explain the effect of contact pressure on thermal contact resistance.
2.2.5 Thermal Conductivity
As defined by Fourier law (Eq. 2.3), the thermal conductivity is
q
k . . . (2.28)
dT
dx
This equation can be used to determine the thermal conductivity of a material. A layer of
solid material of known thickness and area can be heated from one side by an electric
resistance heater of known output. If the other surface of the heater is perfectly insulated,
all the heat generated by the resistance heater will be transferred through the material
whose conductivity is to be determined. Then measuring the two surface temperatures T1
and T2 of the layers of material when steady state has been reached, the thermal
conductivity can be estimated, as shown in Figure 2.12.

Insulation

k
T1 Sample
Material
Electric
Heater

T2
Insulation

Q = We

We

Insulation

L
k Q
A (T1  T2 )

Figure 2.12 : Schematic Diagram of an Experimental Set Up to Determine

the Thermal Conductivity of a Material

36
For engineering purposes, the experimentally measured values of thermal conductivity Different Modes of
Heat Transfer
are generally used. These values can be predicted fairly well for gases with the help of
kinetic theory of gases. But in case of liquids and solids, theories are not adequate to
predict the thermal conductivity with sufficient accuracy.
Table 2.3 gives values of thermal conductivity for several materials. It may be noted that
pure metals are the best conductors and gases are the poorest one.
Table 2.3 : Thermal Conductivity of Some Materials

Thermal Conductivity
Sl. No. Material at 300 K
(W/m.K)

1. Copper 396.00

2. Aluminium 238.00

Carbon steel,
3. 42.0
1% C

4. Glass 0.81

5. Plastics 0.2-0.3

6. Water 0.6

7. Ethylene Glycol 0.26

8. Engine oil 0.15

9. Freon (liquid) 0.07

10. Hydrogen 0.18

11. Air 0.026

The mechanism of thermal conduction in a gas can be explained on a molecular level

from basic concepts of the kinetic theory of gases. The kinetic energy (KE) of a molecule
is a function of temperature. Molecules in a high temperature region have higher KE and
hence higher velocities than those in a lower-temperature region. Since molecules are in
continuous random motion, as they collide with one another they exchange energy as
well as momentum. When a molecule moves from higher temperature region to a lower
temperature region, it transports KE from the
higher-to-lower temperature part of the system. Upon collision with slower molecules,
the faster molecules gives up some of its energy. In this manner thermal energy is
transferred from higher to lower temperature regions in gas by molecular motion.
The faster the molecules move, the faster they will transport energy. Thus, the transport
property called thermal conductivity depends on the temperature of the gas. At moderate
pressures the space between molecules is large compared to the size of a molecule.
Thermal conductivity of gases is therefore essentially independent of pressure (or
density). Figure 2.13 shows how the thermal conductivities of some typical gases vary
with temperature.
The basic mechanism of heat conduction in liquids is qualitatively similar to that in gases.
However, molecular conditions in liquids are more difficult to describe and details of the
conduction mechanisms in liquids are not well understood. The curves in Figure 2.14
show the thermal conductivity of some non-metallic liquids as a function of temperature.

37
Introduction to Heat
and Mass Transfer
0
F
0 200 400 600 800 1000
0.5
0.3

0.4
0.2

Thermal Conductivity K
0.3 H2

(W/mK)

(Btu/h.ft,oF)
He
0.2

0.1

0.1
O2
Air
0 CO2
0 100 200 300 400 500 (oC)
273 373 473 573 673 773 (K)
Temperature

Figure 2.13 : Variation of Thermal Conductivity with Temperature for Gases

0
F
0 100 200 300 400 500

0.4
Water (Saturated Liquid)

0.6 0.3
Thermal Conductivity K
(W/mK)

0.4 0.2
Glycerin

0.2 Benzene 0.1

Light Oil

Freon 12

0 50 100 150 200 250 (oC)

273 323 373 423 473 523 (K)
Temperature
Figure 2.14 : Variation of Thermal Conductivity with Temperature for Liquids

For most liquids, the thermal conductivity decreases with temperature, but water is a
notable exception. Generally thermal conductivity of liquids decreases with increasing
molecular weight.
Solid materials consist of free electrons and atoms in a periodic lattice arrangement. Heat
can be conducted in a solid by two mechanisms :
(a) Migration of free electrons, and
(b) Lattice vibration.
38
These two effects are additive. But in general, the transport due to electrons is more Different Modes of
Heat Transfer
effective than the transport due to vibrational energy in the lattice structure. Since
electrons transport electric charge in a manner similar to the way in which they carry
thermal energy from higher to a lower temperature region, good electrical conductors are
also good heat conductors, whereas good electrical insulators are poor heat conductors. In
non-metallic solids there is little or no electronic transport and conductivity is therefore
determined primarily by lattice vibration. Thus, these materials have lower thermal
conductivities than metals. Thermal conductivities of some typical metals and alloys are
shown in Figure 2.15.
500
1

2
3
200
1 Copper 6 Inconel 600
2 Gold 7 SS 304
3 Aluminum 8 SS 316
Thermal Conductivity

100
4 Iron 9 Incoloy 800
5 Titanium 10 Haynes 230
(W/mok)

50 4

7
5
20
6
8
10
9
10
0 200 400 600 800 1000 1200
Temperature ( C)
o

Figure 2.15 : Variation of Thermal Conductivity with Temperature for Metals and Alloys

SAQ 3
(a) Define thermal conductivity.
(b) How does thermal conductivity vary with temperature for metals and alloys?
(c) How does thermal conductivity can be measured experimentally?
2.2.6 Heat Conduction through a Cylinder
Let us assume that the inside and outside surfaces of the cylinder (Figure 2.16) are
maintained at temperature T1 and T2, respectively (T1 > T2). Heat will be assumed to be
flowing under steady state condition only in the radial direction and there is no heat
conduction along the length or periphery of the cylinder. The rate of heat transfer through
the thin cylinder of thickness dr is given by
dT dT
Qk   kA   k (2  r L ) . . . (2.29)
dr dr
where L is the length of the cylinder.
T2 r2
Qk dr
Now,  dT    2kL r
. . . (2.30)
T1 r1

Qk r
or T2  T1   ln 2 . . . (2.31)
2 k L r1

39
Introduction to Heat 2 k L (T1  T2 )
and Mass Transfer Hence, Qk  . . . (2.32)
r 
ln  2 
 r1 

r L
r2
r1 dr

q
T2
T1
In (r2 / r1 )
Rth 
2 kL
Figure 2.16 : One Dimensional Heat Flow through a Hollow Cylinder with Electrical Analogy

Alternatively,
2 ( r2  r1 ) kL (T1  T2 ) ( A  A1 ) (T1  T2 ) T  T1
Qk  k 2   k Alm 2 . . . (2.33)
 2 r2 L   A  r2  r1 xw
( r2  r1 ) ln   ln  2 
 2 r1 L   A1 
A2  A1
where Alm  = log-mean area,
A2
ln
A1

A2  2 r2 L  outside surface area,

A1  2 r1 L  inside surface area, and

xw  r2  r1  wall thickness of the cylinder.

The thermal resistance offered by the cylinder wall to radial heat conduction is
T2  T1 x
Rk   w . . . (2.34)
Qk kAlm

From Eq. (2.29),

Qk dr dr
dT    C1 . . . (2.35)
2 k L r r

T  C1 ln r  C2 . . . (2.36)

where C1 and C2 are constants to be evaluated from the conditions :

(a) when r = r1, T = T1
(b) r = r2, T = T2
Thus the temperature distribution along the radial direction of the cylinder is
40
   Different Modes of
Heat Transfer
T  T2  T  T2 
T  1 ln r  T1  1 . ln r1  . . . (2.37)
r  r 
ln 1 ln 1
r2 
 r2 

The temperature across the wall of the cylinder varies logarithmically with the radius.
For two concentric cylinders, the fitted one over the other, resistances are in series
(Figure 2.17).
q

q
r1 r2
T2 R1 T2 R2 T3 R3 T4
r3 T1
T1 T3
r4
1 T4 In (r2 / r1 ) In (r3 / r2 ) In (r4 / r3 )
2
3
2k1 L 2k2 L 2k3 L

(a) (b)
Figure 2.17 : One Dimensional Heat Flow through Multiple Cylindrical Sections
and Electrical Analogy

xw1 xw 2
R  R1  R2   . . . (2.38)
k1 Alm1 k2 Alm 2

where xw1  r2  r1 , xw 2  r3  r2

A2  A1 2 (r2  r1 ) L
Alm1   . . . (2.39)
A r
ln 2 ln 2
A1 r1

A3  A2 2 (r3  r2 ) L
Alm 2   . . . (2.40)
A3 r
ln ln 3
A2 r2

The rate of heat transfer will be

T1  T3 T1  T2 T2  T3
Qk    . . . (2.41)
R R1 R2

From the above equation, the interface temperature T2 can be evaluated.

2.2.7 Heat Conduction through a Sphere
Consider a sphere as shown in Figure 2.18.
Heat flowing through a thin spherical strip of the sphere at radius r of thickness dr is
dT
Qk   kA . . . (2.42)
dr

41
Introduction to Heat
and Mass Transfer
T2

dr
r1
r

T1 r2
T1

Figure 2.18 : Heat Conduction through a Sphere

where A is the spherical surface at radius r normal to heat flow
dT
Qk   k (4 r 2 ) . . . (2.43)
dr
T2 r2
Qk dr
or  dT   
4 k r 2
. . . (2.44)
T1 r1

Qk  1 1 
T2  T1      . . . (2.45)
4 k  r1 r2 
4 k (T1  T2 ) r1 r2
or Qk  . . . (2.46)
( r2  r1 )
T2  T1
Qk   k Agm . . . (2.47)
xw
1 1
where Agm = geometric mean area  ( A1 . A2 ) 2  (4 r12 . 4 r22 ) 2  4 r1 r2
and xw  r2  r1  wall thickness of the sphere.
Here, thermal resistance offered by the wall to heat conduction is
xw
Rk  . . . (2.48)
k Agm
It is observed that thermal resistance for plane wall, cylinder and a sphere are similar
(Table 2.4).
Table 2.4 : Thermal Resistances for Different Geometry
Geometry Thermal Resistance
x
Flat plate Rk  w
kA
x
Cylinder Rk  w
kAlm

x
Sphere Rk  w
kAgm

42
2.2.8 Overall Heat Transfer Coefficient Different Modes of
Heat Transfer
The problem largely encountered in engineering practice is heat being transferred
between two fluids of specified temperatures separated by walls (Figure 2.19). In such a
situation the surface temperatures are not known, but they can be calculated if the
convection heat transfer coefficients on both sides of the wall are known.

That, hc, hot hc, cold, Tcold

x Tcold
Thot

1 x 1
R1  
R2  R3  
(h c A) hot kA (h c A) cold

Figure 2.19 : Conduction Heat Transfer through a Slab with Convection at Boundaries

From the Figure 2.19 it is clear that there are three resistances in series.
1 x 1
R  R1  R2  R3    . . . (2.49)
hc ,1 A kA hc ,2 A

Th  Tc Th  Tc
Now, Qc    UA (Th  Tc ) . . . (2.50)
R  1   x   1 
       
 hc ,1 A   kA   hc ,2 A 
where U is known as the overall heat transfer coefficient (W/m2K) and is given by
1 1 x 1
   . . . (2.51)
UA hc ,1 A kA hc ,2 A

1 1 x 1
or    . . . (2.52)
U hc ,1 k hc ,2

For a composite wall with three different layers in series

1 1 L L L 1
  1  2  3  . . . (2.53)
UA hc , hot A k1 A k 2 A k3 A hc , cold A

and Q  UA (Th  Tc ) . . . (2.54)

Similarly, heat transfer from a hot fluid inside a cylinder to the cold fluid outside
(Figure 2.20).
43
Introduction to Heat
and Mass Transfer
T0
h0
T1

kw
Th ro
r1

h1
Tc

Q
Q
T1 T0 Tc
Th
1 Xw 1
hi Ai k w Alm h0 A0
Figure 2.20 : Heat Flow through a Cylindrical Pipe with Cold Fluid Inside
and Hot Fluid Outside the Cylinder

Th  Tc Th  Tc
Qc    U o Ao (Th  Tc ) . . . (2.55)
R1  R2  R3 1 xw 1
 
hi Ai k w Alm ho Ao

1 1 xw 1
where    . . . (2.56)
U o Ao hi Ai k w Alm ho Ao

Uo being the overall heat transfer coefficient based on the outside surface area Ao, hi the
inside heat transfer coefficient and ho the outside heat transfer coefficient.
1
Now, Th  Tc  Qc R1  Qc . . . (2.57)
hi Ai

xw
T1  T2  Qc R2  Qc . . . (2.58)
k w Alm

1
T2  Tc  Qc R3  Qc . . . (2.59)
ho Ao

From which the interface temperatures T1 and T2 can be estimated. When the wall
thickness xw is small
Ao  Alm  Ai . . . (2.60)

1 1 x 1 1
Then   w   . . . (2.61)
U o hi k w ho U i

where Ui is the overall heat transfer coefficient based on the inside surface area Ai. It may
be noted that U o Ao  U i Ai .

If more resistances are put in series, these are to be added up and the same procedure will
follow.

44
SAQ 4 Different Modes of
Heat Transfer
Derive an expression for overall heat transfer coefficient for a infinitely long
cylinder through which hot water is conveying at T1. The cylinder is exposed to
atmosphere at temperature T2.
2.2.9 Thermal Insulation
In most of the engineering applications, such as heat exchangers, building, steam pipes in
power plants, etc. it is essential to reduce the heat loss from the devices. This is
accompliced by providing layers of low thermal conductivity materials over the device.
Thermal insulation materials must have a low thermal conductivity. In most of the cases
it is achieved by trapping air or some other gases inside small cavities in a solid. It uses
the low thermal conductivity of a gas to inhibit heat flow. Heat can be transferred by
natural convection inside the gas pockets and by radiation between the solid enclosures.
The overall thermal conductivity of the insulating material is the result of a combination
of heat transfer mechanism as shown in Figure 2.21.
Conduction
Convection Radiation

Air Pockets

Insulating Material with

Voids

Figure 2.21 : Insulating Material with Voids

The insulating materials can be classified as :

Fibrous Material
High porosity material consists of small diameter particles or filaments of low
density that can be poured into the gap as “loose-fill” or formed into boards or
blankets. Mineral wool, fibre glass, alumina and silica are some of the fibrous
insulators.
Cellular
These are closed or open cell materials that are usually in the form of flexible or
rigid boards. They can also be formed or sprayed to achieve desired geometrical
shapes. Low density, low heat capacity and good compressive strength are their
advantages. Polyurethane and expanded polystyrene foam are some examples of
cellular insulators.
Granular
Granular insulation consists of small flakes or particles of inorganic materials
bonded into shapes or used as powders. Examples are polyurethane and expanded
polystyrene foam.
Both the fibrous and granular insulations can be evacuated to eliminate convection and
conduction, thus decreasing the effective thermal conductivity effectively. Sometimes
45
Introduction to Heat reflective sheets are used to provide the insulation. An example is the thermos bottle, in
and Mass Transfer which the space between the reflective surfaces is evacuated to suppress convection and
conduction.
SAQ 5
(a) Explain the characteristics of thermal insulating materials.
(b) What are the different types of insulating materials? Give a comparative
estimate of fibrous, cellular and granular materials in providing insulation.
(c) What are the insulating materials used in high temperature applications?
2.2.10 Critical Radius of Insulation
Let us consider a small-diameter tube, cable or wire the outside surface of which has a
constant temperature and dissipates heat by convection into the surrounding air. Let the
surface be covered with a layer of insulation. It is desired to examine the variation in heat
loss from the tube surface as the thickness of insulation is changed. As insulation is
added to the tube, the outer exposed surface temperature will decrease because of higher
conduction resistance but at the same time the surface area available for convective heat
dissipation will increase causing more heat loss. These two opposing effects lead to an
optimum insulation thickness.
Let the tube of radius rt and temperature Tt be covered with insulation. At the outer radius
of the insulation ro, a surface coefficient ha is assumed for heat transfer to the atmosphere
at temperature Ta (Figure 2.22). As ro increases, xw (  ro  r1 ) increases and Q decreases.
Again, as ro increases, Ao increases and Q increases. There is thus an optimum ro at which
Q is the maximum.
Insulation

r0

Rconv
rt Rins
Ki Ta
Tt
ha

Tt Ta
L

Figure 2.22 : Thermal Resistance with Insulating Materials Over a Cylinder

  ro  
 ln   
r 1
Tt  Ta  Q (R1  R2 )  Q   i    . . . (2.62)
 2 ki L 2 ro L ha 
 
 
2 L [Tt  Ta ]
Q . . . (2.63)
1 r 1
ln o 
ki rt ha ro
Since Tt, Ta, ki, ri and ha are all constant, the heat loss Q depends only on ro.
1 r  1
As ro increases, ln  o  increases, but decreases. Differentiating Q with ro,
ki  rt  ( ha ro )

dQ  1 rt 1 1 
  2 L (Tt  Ta )    . . . (2.64)
dro  ki ro rt ha ro2 
46
 ki Different Modes of
( ro )cr  . . . (2.65) Heat Transfer
ha

If rt  (ro )cr , as ro increases, Q increases till ro  (ro )cr (Figure 2.23).

Qmax

Cable or Pipe
wire

r0  (r0)cr r0  (r0)cr

(r0)cr r0

Figure 2.23 : Variation of Insulation Radius Influence Heat Loss to the Outside

If rt  (ro )cr , as increase ro, Q decreases. If rt  (ro )cr , any increase of insulation will
decrease the rate of heat transfer. If ro  (ro )cr , the increase of insulation will increase Q
till Q = Qmax.
For pipes, r1, is taken higher than (ro)cr, so that any insulation added will only decrease
the heat loss from the pipe. For wires and cables, rt is kept lower than (ro)cr so that added
increases the heat loss from the wire or cable. An insulated small diameter wire has a
higher current carrying capacity than an uninsulated one. If the current flowing through
an uninsulated wire increases, I2R increases, and if heat dissipation from the wire is not
equal to I2R, the temperature of the wire goes on increasing till it exceeds the melting
point and the wire snaps. If the wire is insulated, it can dissipate more heat (till rt = (ro)cr)
and the wire temperature remains below the melting point.
In the case of a sphere, by following a similar procedure, it can be shown that the critical
radius of insulation is given by
2k
( ro )cr  . . . (2.66)
ha

SAQ 6
(a) What do you mean by critical radius of insulation?
(b) A pipe is insulated to reduce the heat loss from it. However, measurements
indicate that the rate of heat loss has increased instead of decreasing. Can
the measurements be right?

2.3 CONVECTION
Consider the flow condition of Figure 2.24. A fluid of velocity U and temperature T
flows over a surface of arbitrary shape and of area As. The surface is presumed to be at a
uniform temperature Tw, and if Tw  T, we know that convection heat transfer will occur.

47
Introduction to Heat y Velocity Flow
Temperature
and Mass Transfer Profile Profile

U T

qc
T
U(y)
Y y 0
T(y)

Heated
y=0
Surface
Tw

Figure 2.24 : Velocity and Temperature Profiles for a Convection Heat Transfer
( Forced Convection)

The local heat flux q may be expressed as

q  h (Tw  T ) . . . (2.67)

where q = Convective heat flux (W/m2),

Tw = Wall surface temperature (K),

T = Fluid temperature (K), and

h = Local convection heat transfer coefficient (W/m2.K).
Eq. (2.67) is also known as the Newton’s law of cooling.

In Eq. (2.67), the heat flux is assumed positive if Tw > T, i.e. heat is transferred from
surface to the fluid and same is considered negative if heat flows to the surface
(T > Tw). In case of the heat transfer from fluid to the surface, we can rewrite the
equation in the form

q  h (T  Tw ) . . . (2.68)

where T > Tw.

It may be worth mentioning here that the proportionality constant known as convective
heat transfer constant h depends on
(a) Condition of the boundary layer which itself depends on the surface
geometry.
(b) Nature of the fluid motion.
Any study of convection ultimately reduces to a study of the means by which h may be
determined. Convection heat transfer coefficient will frequently appear q = h (T  Tw)
as a in Eq. (2.67), because the flow conditions vary from point to point on the surface,
both q and h also vary along the surface.

The total heat transfer rate q may be obtained by integrating the local flux over the entire
surface.
That is q  q  d As . . . (2.69)
As

or, from Eq. (2.67),

48
 Different Modes of
q  (Tw  T )  h dAs . . . (2.70) Heat Transfer
As

It follows that the average and local convection coefficients are related by an expression
of the form
1
As A
h  h dAs . . . (2.71)
s

Defining an average convection heat transfer coefficient h for the entire surface, the
total heat transfer rate may be expressed as
q  h As (Tw  T ) . . . (2.72)
Convection heat transfer coefficient will frequently appear as a boundary condition in the
solution of conduction problems. In the solution of such problems we presume h to be
known. Some of the typical values of convective heat transfer coefficient h are given in
Table 2.5.
Table 2.5 : Typical Values of the Convection Heat Transfer Coefficient
Convective Heat Transfer
Sl. No. Process
Coefficient h (W/m2.K)
Free convection
1. Gases 2-25
Liquids 50-1000
Forced convection
2. Gases 25-250
Liquids 100-20,000
Convection with phase change
3.
Boiling or condensation 2500-100,000

SAQ 7
(a) State the Newton’s law of cooling.
(b) Define heat transfer coefficient. On what factor does it depend?

2.4 RADIATION
Radiative heat transfer is one of the most fundamental and pervasive process interacting
with every natural and man made system on earth. Every emerging technology such as
global warming to optical computing, energy conversion devices, industrial heating and
drying, rocket nozzles, space vehicles reentry, nuclear fisson, fusion, plasma involves the
study of radiative heat transfer. Recent increase in interest for study of radiation is due to
the development of high temperature applications in furnaces, engines, MHD generators,
Circulating fluidized bed boilers, development of high temperature ceramics, etc.
Radiation is energy emitted by matter at finite temperature. All bodies emit radiation
whenever the same are at a temperature more than the ambient temperature/reference
temperature. Although we will focus on radiation from solid surfaces, emission may also
occur from liquid and gases. Regardless of the form of matter, emission may be attributed
to the changes in the electron configurations of the constituent atoms or molecules. The
energy of the radiation field is transported by electromagnetic waves (Photons).
Thermal radiation is a part of radiation emitted by particles of matter as they undergo
internal energy state transition. Generally the internal energy state transition is in
equilibrium and hence the phenomenon of thermal radiation is associated with the
temperature of the matter. Radiation is emitted and absorbed by electromagnetic waves
or photons by lowering or raising the molecular energy level of any material. The
strength of emission or absorption covers a wide range of wave lengths starting from
gamma rays
( 10– 4 m), X-rays (10– 4 – 10– 2 m)), ultraviolet rays (10– 2 – 0.4  m), visible
(0.4 – 0.7  m), infrared (0.1 0 100  m) to microwaves (> 100 m) (Figure 2.25). 49
Introduction to Heat
Visible
and Mass Transfer

Green

Yellow
Violet
Blue

Red
Infrared
X-rays
Ultraviolet Micro ways

Gamma rays
Thermal Radiation

0.4 0.7

10-5 10-4 103 10-2 10-1 1 10 102 103 104

(m)

Figure 2.25 : Spectrum of Electromagnetic Radiation

Thermal radiation is in the wave length rage of 0.1 – 100 m. That gives mankind heat,
light, photosynthesis and all their attendant benefits. At higher and higher temperature,
radiation is restricted to near visible or infrared region of the spectrum.
2.4.1 Distinction between Surface and Volumetric Radiation
Thermal radiation (as shown in Figure 2.26) may be categorically divided into two parts
(a) surface radiation, and
(b) volumetric radiation.
Radiation Emission

Radiation Emission

Gas or Vacuum

Solid or Liquid

High Temperature Gas or

Semitransparent Medium

(a) (b)
Figure 2.26 : The Emission Process (a) Surface Radiation and (b) Volumetric Radiation

In most solids and liquids, radiation emitted from interior molecules is strongly absorbed
by adjoining molecules. Accordingly, radiation emitted from a solid or liquid originates
from the molecule, those are within a distance of 1 m from the exposed surface. It is for
this reason that emission from a solid or liquid into an adjoining gas or vacuum is viewed
as a surface phenomenon. Both these radiation phenomena will be discussed in details in
Units 9 and 10.
Consider radiative heat transfer processes for ideal surface in Figure 2.27.

50
 Different Modes of
Heat Transfer

Radiation from
Surroundings qrad, net

Solid
Surface
Radiation
Emission

Ts

Tsur Vacuum

Surroundings

Figure 2.27 : Radiation Heat Transfer from an Ideal Surface Bounded by an Enclosure

Radiation emitted by the surface originates from the thermal energy of matter bounded
by the surface, and the rate at which energy is released per unit area (W/m2) is termed the
surface emissive power E. There is an upper limit to the emissive power, which is
prescribed by the Stefan-Boltzmann law :

Eb   Ts4 . . . (2.73)

where Ts = absolute temperature of the surface (K), and

 = Stefan-Boltzmann constant = 5.67  10– 8 W/m2 K4.
The surface, as mentioned earlier is an ideal radiator or a blackbody. The heat flux
emitted by a real surface is complex with involvement of different spatial, directional,
spectral properties, surface properties as well as enclosure properties. Hence the heat flux
for real surface is less than that of an ideal surface at the same temperature and is given
by

Eb    Ts4 . . . (2.74)

where  = radiative property of the surface termed as emissivity. With values in the range
0    1, this property provides a measure of how efficiently a surface emits energy
relative to a blackbody. It depends strongly on the surface and finish.
A portion or all, of the irradiation may be absorbed by the surface, thereby increasing the
thermal energy of the material. The rate at which the radiant energy is absorbed per unit
surface area may be evaluated from the knowledge of a surface radiative property termed
the absoptivity . That is
Gabs   G . . . (2.75)

where 0    1. If   1 and the surface is opaque, portions of the irradiation are

reflected. If the surface is semitransparent. Portion of the irradiation also be transmitted.
However, while absorbed and emitted radiation increase and reduce, respectively, the
thermal energy of the matter, reflected and transmitted radiation have no effect on this
energy. Note that the value of  depends on the nature of the irradiation, as well as on the
surface itself. For example, absorptivity of a surface to solar radiation may differ from
the absorptivity to radiation emitted by the wall of a furnace.
A special case that occurs frequently involves radiation exchange between a small
surface at Ts and a much larger, isothermal surface that completely surrounds the smaller
one (Figure 2.28).

51
Introduction to Heat
and Mass Transfer

Gas
Thot,h
Gas
Surroundings T,h
G E
at Tsur
qconv qrad qconv

Surface of emissivity,
Surface of emissivity, 
E = , area A, and
absorptivity 1 and
temperature Ts
temperature Ts

(a) (b)
Figure 2.28 : Radiation Exchange between a Surface and Large Surroundings

The surroundings could, for example, be the walls of a room or a furnace whose
temperature Tsw differs from that of an enclosed surface (Ts  Tsur). For such a condition
the irradiation may be approximated by emission from a black body at Tsur in which case
4
Gabs   Tsur . If the surface is assumed to be one for which  =  (a gray surface), the
net rate of radiative heat transfer from the surface, expressed per unit area of the surface,
is
q
 
qrad   Eb (Ts )   G    (Ts4  Tsur
4
) . . . (2.76)
A
Eq. (2.76) gives the difference between the thermal energy that is released due to
radiation emission and that which is gained due to radiation absorption.
There are many applications for which it is convenient to express the net radiation heat
exchange in the form
  hr A (Ts  Tsur )
qrad . . . (2.77)

where in Eq. (2.77), the radiation heat transfer coefficient hr is

hr    (Ts  Tsur ) (Ts2  Tsur

2
) . . . (2.78)

Here we have modeled radiation mode in a manner similar to convection. In this sense
we have linearized the radiation rate equation, making the heat rate proportional to a
temperature difference rather than to the difference between the temperatures to the
fourth power. Note, however, that hr depends strongly upon temperature, while the
temperature dependence of the convection heat transfer coefficient h is generally weak.
SAQ 8
(a) Distinguish between the surface and volumetric radiation.
(b) What is the wavelength range for infra red and visible radiation?
(c) Define radiative heat transfer coefficient.

52
 Different Modes of
2.5 COMBINED HEAT TRANSFER MECHANISM Heat Transfer

In the earlier sections we have discussed the heat transfer mechanism, conduction,
convection and radiation separately. In many practical situations heat transfer from a
surface takes place simultaneously by convection to the ambient air and by radiation to
the surroundings. For example, consider a small plate of surface area A and emissivity 
maintained at temperature Ts (Figure 2.22). This plate exchanges heat by (a) convection
with a fluid at temperature T with a heat transfer coefficient h and (b) radiation with the
surroundings at Tsur.
For the conditions of Figure 2.22, the total rate of heat transfer from the surface is then

q  q conv  q rad  h A (Ts  T )   A  (Ts4  Tsur

4
) . . . (2.79)

If | Ts – Tsur | < < Ts, the second term in Eq. (2.79) can be linearized. We then obtain
q  q conv  q rad  h A (Ts  T )  hr (Ts  Tsur ) . . . (2.80)

where hr  4  Ts3 . . . (2.81)

2.6 COMPARISON OF CONDUCTION, CONVECTION

AND RADIATION
(a) Conduction and convection occur due to the direct exchange of kinetic
energy between particles of matter, whereas radiation is the energy in transit
propagated by the electromagnetic wave or photons.
(b) No medium is essential for the transport of energy by radiation whereas both
the conduction and convection mode of heat transfer can occur only if there
is a material medium. Radiation is the dominating mode of heat transfer in
case of vacuum.
(c) Heat transfer by conduction or convection between two locations depends
on the temperature differences of the locations to the first power. Transfer of
heat by radiation between two bodies depends on the differences of
individual absolute temperature of each body raised to the power of four.
Thus, importance of radiation will be significant at high temperature.
Example 2.2
A wall of a furnace consists of 120 mm thick refractory bricks and 150 mm thick
insulating firebricks separated by an air gap. A 15 mm thick plaster covers the
outer surface wall. The inner surface of the wall is at 1200oC and the ambient
temperature is 25oC. The heat transfer coefficient on the outside wall to the air is
27 W/m2K, and the resistance to heat flow of the air gap is 0.14 K/W. The thermal
conductivities of refractory brick, insulating firebrick and plasters are 2.6, 0.8 and
0.44 W/mK, respectively. Calculate :
(a) the rate of loss of heat per unit area of the wall,
(b) the interface temperatures throughout the wall, and
(c) the temperature at the outside surface of the wall.
Solution
Consider 1 m2 of surface area of the wall.
x1 0.12
R1 = Resistance of the refractory brick    0.0461 K/W
k1 A 2.6  1

x2 0.150
R2 = Resistance of the insulating firebrick    0.1875 K/W
k2 A 0.8  1
53
Introduction to Heat x3 0.015
and Mass Transfer R3 = Resistance of the plaster    0.0341 K/W
k3 A 0.44  1

1 1
R4 = Resistance of the air film on outside surface    0.037 K/W
h0 A 27  1
R5 = Resistance of the air gap = 0.14 K/W.
Insulated Fire Brick
(K = 0.8 w/mk)

ho = 27 w/m2k
1200 C
o

Refractory Brick RthT= 0.14 k/w

(k = 2.6 w/mk) To = 25oC

Air Gap

120 mm 150 mm 15 mm
Plaster
(k = 0.44 w/mk)
Figure 2.29
Total resistance
RT  R1  R2  R3  R4  R5
 0.0461  0.1875  0.0341  0.037  0.14  0.442 K/W
Rate of heat loss per unit area
T1  T2 1200  25
Q   658.37 W  2.658 kW
RT 0.442
The interface temperatures are T3, T4 and T5 and the outside surface is at T6.
Applying electrical analogy to each layer,
T1  T3 1200  T3
Qk  2658.37 = 
R1 0.0461

T3  1077.45o C
T3  T4 1077.45  T4
Qk  2658.37 = 
R5 0.14

T4  705.277 o C
T4  T5 705.277  T5
Qk  2658.37 = 
R2 0.1875

T5  206.83o C
T5  T6 206.83  T6
Qk  2658.37 = 
R3 0.0341
Temperature at the outside surface of the wall is
T6  116.18o C
54
Example 2.3 Different Modes of
Heat Transfer
Steam at 380o is flowing in a pipe (k = 100 W/mK) of 8 cm inner diameter and
8.5 cm outer diameter is covered with 10 cm thick insulation of thermal
conductivity k = 0.15 W/mK. Heat is lost to the surroundings at 8oC by natural
convection and radiation, the combined h being 40 W/m2K. Taking the heat
transfer coefficient inside the pipe as 40 W/m2K, determine
(a) the rate of heat loss from the steam per unit length of the pipe, and
(b) the temperature drop across the pipe and the insulation.
Solution
For steady and one dimensional heat transfer through the pipe, the thermal
resistances in series are given in the Figure 2.30.

T2
h2

Insulation Q
r1 r2

Steam r3
T1
T1 T2
h1
T3
T1 T1 T2 T3 T2

R1 R2 R3 R0
Figure 2.30
D1 8
r1    4 cm
2 2
D2 8.5
r2    4.25 cm
2 2
r3  r2  t  4.25  10  14.25 cm

A1  2 r1 L  2    0.04  1  0.251 m 2

A3  2 r3 L  2    0.1425  1  0.895 m 2
1 1
Ri    0.0996 K/W
h1 A1 40  0.251

r   4.25 
ln  2  ln  
r  4.0   9.65  10  5 K/W
R1   1  
2 k1 L 2    100  1

r   14.25 
ln  3  ln  
r  4.25   1.283 K/W
R2   2  
2 k2 L 2    0.15  1
1 1
R0    0.0279 K/W
h2 A3 40  0.895
55
Introduction to Heat Rtotal  Ri  R1  R2  R0
and Mass Transfer
 0.0996  9.65  10  5  1.283  0.0279
 1.41 K/W
(a) Rate of heat transfer
T  T2 380  8
Q  1   263.83 W
Rtotal 1.41

(b) Tpipe  Q  R1  263.83  9.65  10  5  0.025 o C

Tinsulation  Q  R2  263.83  1.283  338.494 o C

Example 2.4
A tube with outer diameter 1.5 cm is maintained at uniform temperature and is
covered with an insulative tube having thermal conductivity k = 0.18 W/mK in
order to reduce the heat loss. Heat is dissipated from the outer surface of the cover
by natural circulation with h0 = 22 W/m2K into the ambient air at constant
temperature. Determine the critical thickness of the insulation of insulation.
Calculate the ratio of the heat loss from the tube with insulation to that without any
insulation for
(a) the thickness of insulation equal to that at the critical thickness, and
(b) the thickness of insulation 4.0 cm thicker than the critical thickness.
Solution
The critical radius roc of insulation is determined by
k 0.18
roc    0.0082 m  0.82 cm
h0 22
Then the critical thickness of insulation is = 0.82 – 0.75 = 0.07 cm
The heat losses from the tube with and without insulation are respectively,
1
Qwith  2 ro H ho  T
r h   ro 
1  o o  ln  r 
 k   i 
Qwithout  2 ri H ho  T
where H = tube length, and
ri = tube radius.
Heat loss ratio becomes
1
Qwith r  r 
 o  1  ln oc 
Qwithout ri  ri 
For ro = roc, this ratio reduces to
1
Qwith r  r 
 oc  1  ln oc 
Qwithout ri  ri 
ho
Since, roc 1
k
For the critical radius roc = 0.82 cm, we get
1
Qwith 0.82  0.82 
  1  ln   1.004
Qwithout 0.75  0.75 

56
 This shows that the heat loss is increased about 0.4% despite the fact that there is Different Modes of
Heat Transfer
an insulation of thickness 0.07 cm. If another layer of 4.0 cm insulation is added,
we have ro = 0.82 + 4.0 = 4.82 cm, and the heat loss ratio is
1
Qwith 4.82  0.0482  22 4.82 
 1  ln   0.537
Qwithout 0.75  0.18 0.75 
This implies that with a 4.82 cm thick insulation layer, the heat loss is reduced by
about 46.3%.
Example 2.5
Two large aluminium plates (k = 250 W/mK) each 3 cm thick, with 8 m surface
roughness are placed in contact under 105 N/m2 pressure in air as shown in
Figure 2.31 given below.

8 m
Surface
Roughness

1.5 1.5
cm cm
Figure 2.31

The temperature at the outside surfaces are 420oC and 450oC. Calculate :
(a) the heat flux,
(b) the temperature drop due to the contact resistance, and
(c) the contact temperatures.
Thermal contact resistance with air as the interface fluid for 8 m roughness is
2.65  10– 4 m2 K/W.
Solution
(a) The rate of heat flow per unit area
T1  T2 T
q 
R1  R2  R3  L  L
   Ri   
 k 1  k 2
where Ri = 2.65  10– 4 m2 K/W and each of the other two resistances is
equal to
L 0.03
  1.2  10  4 m 2K/W
k 250
Heat flux, q 57
Introduction to Heat 450  420
and Mass Transfer q 4
 5.94  10 4 W/m 2
1.2  10  2.65  10  4  1.2  10  4

(b) The temperature drop in each section is proportional to the resistance. The
fraction of the contact resistance is

Ri 2.65  10  4
  0.525
R (1.2  2.65  1.2)  10  4

Temperature drop = 0.525  (450 – 420) = 15.74oC.

(c) The temperature drop in each aluminium plate is
30  15.74
T   7.13o C
2

TC1  450  7.13  442.87 o C

TC 2  420  7.13  412.87 o C

These are the contact surface temperatures.

Example 2.6
A steam pipe ( = 0.85) of 0.5 m diameter has a surface temperature of 550 K. The
pipe is located in a room at 30oC, and the convection heat transfer coefficient is
28 W/m2K. Calculate the combined heat transfer coefficient and the rate of heat
transfer per unit length of the pipe.
Solution
Rate of radiant heat transfer

Qr   A1  (T14  T24 )  hr A1 (T1  T2 )

hr    (T1  T2 ) (T12  T22 )

 5.67  10  8  0.85  (550  303) (550 2  3032 )

 16.21 W/m 2K

Example 2.7
A small hot surface at temperature T1 = 650 K having an emissivity 1 = 0.8
dissipates heat by radiation into surrounding area at T2 = 610 K. If this radiation
transfer process is characterized by a radiation heat transfer coefficient hr, calculate
the value of hr.
Solution
Refer to Eqs. (2.66) and (2.67)
Here, T1  Ts  650 K

T2  Tsur  610 K

As T1  T2  Ts  Tsur   T1

We can apply Eq. (2.67)

hr  4  Ts3  4 [0.8  5.67  10 8  (650) 3 ] W/m 2 .K  49.83 W/m 2 .K

Exercise 2.1
58
 (a) A steam pipe is covered with two layers of insulation. The inner layer Different Modes of
Heat Transfer
(k = 0.19 W/m.K) is 36 mm thick and the outer layer (k = 0.048 W/m.K) is
60 mm thick. The pipe is made of steel (k = 60 W/m.K) and has the inner
diameter of 170 mm and thickness of 10 mm. The temperature of saturated
steam is 439oC and ambient air is at 28oC. If the inside and outside heat
transfer coefficients are 28 and 6 W/m2.K, respectively, calculate the rate of
heat loss per unit length of the pipe.
(b) A 60 W lamp is buried in soil (k = 0.92 W/m.K) at 23oC and switched on.
Find the temperature 0.25 m and 0.50 m away from the lamp, when steady
state is reached.
(c) A 1.2 m high and 2 m wide double-pane window consists of two 3 mm thick
layers of glass (k = 0.78 W/m.K) separated by a 12 mm wide stagnant air
gap (k = 0.026 W/m.K). Determine the steady rate of heat transfer through
this double paned window and temperature of its inner surface for a day
during which the room is maintained at 24oC while the temperature of the
outdoors is – 5oC. Take the convection heat transfer coefficients on the inner
and outer surfaces of the window to be hi = 10 W/m2.K and
ho = 25 W/m2.K, respectively. Neglect heat transfer by radiation.
(d) A composite wall consisting of four different materials is shown in
Figure 2.32. Since the upper and the lower surfaces are insulated, the heat
flow can be considered to be one-dimensional. The dimensions and thermal
conductivity of each layer are indicated in the figure. Using the thermal
resistance concept, determine the heat flow rate per square meter of the
exposed surface for a temperature difference of T = 300oC between the two
outer surfaces.
Insulated

A B C

KB = 70
HB = 1 m W/ (moC)

H=2m
EC = 100 EC = 70
W/ (moC) W/ (moC)

HD = 1 m KD = 20
W/ (moC)
D

Insulated

L1 = 4 cm L2 = 10 cm L3 = 5 cm

Figure 2.32

Exercise 2.2
(a) Derive an expression for the critical radius appropriate for the insulation of a
sphere. An electrically heated sphere with diameter D = 6 cm is exposed to
an ambient at T = 25oC with a convection heat transfer coefficient
h = 20 W/m2.oC. The surface of the sphere is to be maintained at Ti = 125oC.
Calculate the rate of heat loss from the sphere for (i) the un-insulated sphere,
(ii) the sphere covered with an insulation (k = 1.0 W/m.oC) with the radius
corresponding to the critical radius of the insulator.
(b) A flat surface has one surface insulated and the other surface exposed to the
sun. The exposed surface absorbs solar radiation at a rate 800 W/m2 and
dissipates it by both convection and radiation into ambient air at 300 K. If
the emissivity of the surface is  = 0.9 and the convection heat transfer 59
Introduction to Heat coefficient between the plate and air is12 W/m2.oC, determine the
and Mass Transfer temperature of the plate.
(c) An overhead 25 m long, uninsulated industrial steam pipe of 100 mm
diameter is routed through a building whose walls and air are at 25oC.
Pressurized steam maintains a pipe surface temperature of 150oC, and
coefficient associated with natural convection is h = 10 W/m2.K. The
surface emissivity is  = 0.8.
(i) What is the rate of heat loss from the steam line?
(ii) If the steam is generated in a gas fired boiler operating at an
efficiency of f = 0.90 and natural gas is priced at Vg = Rs 2/MJ, what
is the annual cost of heat loss from the line?
(d) A spherical interplanetary probe of 0.5 m diameter contains electronics that
dissipate 150 W. If the probe surface has an emissivity of 0.8 and the probe
does not receive radiation from other surfaces, what is its surface
temperature?
Exercise 2.3
(a) A surface of area 0.5 m2, emissivity 0.8, and temperature 150oC is placed in
a large, evacuated chamber whose walls are maintained at 25oC. What is the
rate at which radiation is emitted by the surface? What is the net rate at
which radiation is exchanged between the surface and the chamber walls?
(b) Air at 40oC flows over a long, 25 mm diameter cylinder with embedded
electrical heater. In a series of tests, measurements were made of the power
per unit length, P, require to maintain the cylinder surface temperature at
300oC for different free stream velocities V of air. The results are as
follows :

Air Velocity V (m/s) 1 2 4 8 12

Power, P (W/m) 450 658 983 1507 1963

(i) Determine the convection coefficient for each velocity and display the
results graphically.
(ii) Assuming the dependence of the convection coefficient on the
velocity to be of the form h = CVn, determine the parameters C and n
from the results of part (a).
(c) An electric resistance heater is embedded in a long cylinder of diameter
30 mm. When water with a temperature of 25oC and velocity of 1 m/s flows
crosswise over the cylinder, the power per unit length required to maintain
the surface at uniform temperature of 90oC is 28 kW/m. When air, also, at
25oC, but with a velocity of 10 m/s is flowing, the power per unit length
required to maintain the same surface temperature is 400 W/m. Calculate
and compare the convection heat transfer coefficients for the flows of water
and air.
(d) Find the heat transfer rate for the following composite layers. Assume a
temperature difference between the two extremes, if not provided in the
Figures 2.33(a) to (e).

60
 Different Modes of
Heat Transfer
K2 K3
K1

T2 T3
T1
T0

L2 L3
L1

(a)

B
A Convection into an
ambient at Tb with
Convection into an a heat transfer
ambient at Ta with hc
Kb coefficient hb
a heat transfer Ka
coefficient ha

Lb
La

(b)

Insulated

A
HB B F

kb HF
K
H
kf Maintained
HC C E
at T2
Maintained kc
at T1
Ke G
HG
W/(moC) Ke
HD W/(moC) kk
D
kd
Kg

Insulated
L1 L2 L3 L4 L5

(c)

K3 = 100
K2 = 50 W/(m0C)
K1 = 20 W/(m0C) T2 = 50oC
W/(m0C)

5 cm 10 cm 15 cm

T1 = 400 Co

(d)

61
Introduction to Heat hc = 3 W/(m2.oC)
and Mass Transfer

Kb = 0.05
Ka = 0.1
W/(moC) Convection into an
Convection into an ambient at
ambient at Tb = 50oC with
Ta = 200oC with hb = 25 W/(m2.oC)
ha = 15 W/(m2.oC)

Lb = 2 cm Lb = 4 cm

(e)
Figure 2.33

2.7 SUMMARY
Fourier’s law of heat conduction can be applied to simple surfaces such as plane wall,
cylinder and sphere to evaluate heat transfer rate. Resistance concept is a simple method
to solve such problems. In case of composite systems, application resistance method
gives easy solution. Physical significance of thermal conductivity is discussed in this unit.
Importance of thermal insulation, contact thermal resistance are discussed. Discussion on
heat transfer by convection and radiation are also presented. Heat transfer may occur in
real situation.

2.8 KEY WORDS

Thermal Conductivity : Property of a material medium arises from
Fourier’s law of conduction.
Thermal Insulation : Materials with poor thermal conductivity used to
reduce heat loss from a system.
Thermal Contact Resistance : Resistance developed when two conducting
surfaces do not fit tightly together and a thin layer
of fluid is trapped between them.
Spectral Property : Property dependent upon the wavelength.
Emissive Power : Indicates rate of radiation emitted from a surface.
Semitransparent Medium : Portion of the irradiation is transmitted through
the medium.
Opaque Surface : Surface does not allow radiation to penetrate.

2.9 ANSWERS TO SAQs

Refer the preceding text in this unit for all the Answers to SAQs.

62
 Mass Transfer
UNIT 3 MASS TRANSFER
Structure
3.1 Introduction
Objectives
3.2 Concentration
3.3 Examples of Mass Transfer Process
3.4 General Molecular Transport Equation
3.5 Physical Origins and Rate Equations
3.6 Mixture Composition
3.7 Fick’s Law of Diffusion (Vector Form)
3.8 Restrictive Conditions
3.9 Convective Mass Transfer
3.10 Types of Mass-Transfer Coefficients
3.10.1 Definition of Mass-Transfer Coefficient
3.10.2 Mass-Transfer Coefficient for Equimolar Counter Diffusion
3.10.3 Mass-Transfer Coefficient for A Diffusing through Stagnant, Non-diffusing B
3.11 Summary
3.12 Key Words
3.13 Answers to SAQs

3.1 INTRODUCTION
We have learned that heat is transferred if there is temperature difference in a medium.
Similarly, if there is a difference in the concentration of some chemical species in a
mixture, mass transfer must occur. Mass transfer is mass in transit as the result of a
species concentration difference in a mixture.
Just as a temperature gradient constitutes the driving potential for heat transfer, a species
concentration gradient in a mixture provides the driving potential for transport of the
species.
It is important to understand clearly the context in which the term mass transfer is used.
Although mass is certainly transferred whenever there is bulk fluid motion, this is not
what we have in mind. For example, we do not use the term mass transfer to describe the
motion of air that is induced by a fan or the motion of water being forced through a pipe.
In both cases, there is gross or bulk fluid motion due to mechanical work. We do,
however, use the term to describe the relative motion of species in a mixture due to the
presence of concentration gradients. One example is the dispersion of oxides of sulfur
released from a power plant smoke stake into the environment. Another example is the
transfer of water vapour into dry air, as in a home humidifier.
There are modes of mass transfer that are similar to the conduction and convection modes
of heat transfer. Mass transfer may occur by diffusion or by convection. Mass transfer by
diffusion is analogous to conduction heat transfer.
Mass transfer occurs in distillation, absorption, drying, liquid-liquid extraction,
adsorption and membrane processes. When mass is being transferred from one distinct
phase to another or through a single phase, the basic mechanisms are same whether the
phase is a gas, liquid or solid. Mass, heat and momentum transfer processes are similar.
Objectives
After studying this unit, you should be able to
understand the principle of mass transfer, 63

Introduction to Heat  recognise the physical origin and rate equation describing mass transfer,
and Mass Transfer
 distinguish different modes of mass transfer (molecular diffusion and
convective mass transfer),
 recognise the applications of mass transfer processes,
 determine mass transfer in various specific cases,
 evaluate mass transfer coefficient, and
 apply theory in solving simple mass transfer problems.

3.2 CONCENTRATION
The concentration of a chemical solution refers to the amount of solute that is dissolved
in a solvent. We normally think of a solute as a solid that is added to a solvent (e.g.
adding table salt to water), but the solute could just as easily exist in another phase. For
example, if we add a small amount of ethanol to water, then the ethanol is the solute and
the water is the solvent. If we add a smaller amount of water to a larger amount of
ethanol, then the water could be the solute.
3.2.1 Units of Concentration
Once we have identified the solute and solvent in a solution, we are ready to determine
its concentration. Concentration may be expressed several different ways, using percent
composition by mass, mole fraction, molarity, molality, or normality.
Percent Composition by Mass (%)
This is the mass of the solute divided by the mass of the solution (mass of solute
plus mass of solvent), multiplied by 100.
Example
Determine the percent composition by mass of a 100 g salt solution which contains
20 g salt.
Solution
20 g NaCl/100 g solution  100 = 20% NaCl solution.
Mole Fraction (X)
This is the number of moles of a compound divided by the total number of moles
of all chemical species in the solution. Keep in mind, the sum of all mole fractions
in a solution always equals 1.
Example
What are the mole fractions of the components of the solution formed when 92 g
glycerol is mixed with 90 g water? (Molecular weight water = 18; molecular
weight of glycerol = 92).
Solution
90 g water = 90 g  1 mol/18 g = 5 mol water
92 g glycerol = 92 g  1 mol/92 g = 1 mol glycerol
Total mol = 5 + 1 = 6 mol
xwater = 5 mol/6 mol = 0.833
xglycerol = 1 mol/6 mol = 0.167
It’s a good idea to check your math by making sure the mole fractions add up to
1 : xwater + xglycerol = 0.833 + 0.167 = 1.000

64
Molarity (M) Mass Transfer

Molarity is probably the most commonly used unit of concentration. It is the

number of moles of solute per liter of solution (not necessarily the same as the
volume of solvent!).
Example
What is the molarity of a solution made when water is added to 11 g CaCl2 to
make 100 mL of solution?
Solution
11 g CaCl2/(110 g CaCl2/mol CaCl2) = 0.10 mol CaCl2
100 mL  1 L/1000 mL = 0.10 L
Molarity = 0.10 mol/0.10 L
Molarity = 1.0 M
Molality (m)
Molality is the number of moles of solute per kilogram of solvent. Because the
density of water at 25°C is about 1 kilogram per liter, molality is approximately
equal to molarity for dilute aqueous solutions at this temperature. This is a useful
approximation, but it should be remembered that it is only an approximation and
doesn’t apply when the solution is at a different temperature, is not dilute, or uses a
solvent other than water.
Example
What is the molality of a solution of 10 g NaOH in 500 g water?
Solution
10 g NaOH/(4 g NaOH/1 mol NaOH) = 0.25 mol NaOH
500 g water  1 kg/1000 g = 0.50 kg water
Molality = 0.25 mol/0.50 kg
Molality = 0.05 M/kg
Molality = 0.50 m
Normality (N)
Normality is equal to the gram equivalent weight of a solute per liter of solution. A
gram equivalent weight or equivalent is a measure of the reactive capacity of a
given molecule. Normality is the only concentration unit that is reaction dependent.
Example
1 M sulfuric acid (H2SO4) is 2 N for acid-base reactions because each mole of
sulfuric acid provides 2 moles of H+ ions. On the other hand, 1 M sulfuric acid is
1 N for sulfate precipitation, since 1 mole of sulfuric acid provides 1 mole of
sulfate ions.
Dilutions
We dilute a solution whenever we add solvent to a solution. Adding solvent results
in a solution of lower concentration. We can calculate the concentration of a
solution following a dilution by applying this equation :
Mi Vi = Mf Vf
where M is molarity, V is volume, and the subscripts i and f refer to the initial and
final values.
Example
How many millilieters of 5.5 M NaOH are needed to prepare 300 mL of 1.2 M
NaOH?
65
Introduction to Heat Solution
and Mass Transfer
5.5 M  V1 = 1.2 M  0.3 L
V1 = 1.2 M  0.3 L/5.5 M
V1 = 0.065 L
V1 = 65 mL
So, to prepare the 1.2 M NaOH solution, you pour 65 mL of 5.5 M NaOH into
your container and add water to get 300 mL final volume.
SAQ 1
Define the following :
(i) Mole fraction
(ii) Dilution
(iii) Normality
(iv) Molality
(v) Molarity

3.3 EXAMPLES OF MASS TRANSFER PROCESS

Mass transfer is important in many areas of science and engineering. Mass transfer
occurs when a component in a mixture migrates in the same phase or from phase to phase
because of a difference in concentration between two points. Many familiar phenomena
involve mass transfer. Some examples are :
(a) Liquid in an open pail of water evaporates into still air because of the
difference in concentration of water vapour at water surface and the
surrounding air. There is driving force from surface to the air.
(b) A cube of sugar added to a cup of coffee eventually dissolves by itself and
diffuses to the surrounding solution.
(c) When a newly cut moist green timber is exposed to atmosphere, the wood
will dry partially when moisture in the timber diffuses through the wood, to
the surface, and then to the atmosphere.
(d) In a fermentation process nutrients and oxygen dissolved in the solution
diffuse to the microorganisms.
(e) In a catalytic reaction the reactants diffuse from the surrounding medium to
the catalyst surface where the reaction occurs.
(f) Mass transfer process occurs in the uranium processing. In such a process,
uranium salt in solution is extracted by an organic solvent.
We can treat mass transfer in a manner some what similar to that used in heat transfer
with Fourier law of conduction. However, an important difference between the two is
that in molecular mass transfer one or more of the components of the medium is moving.
In heat transfer by conduction the medium is usually stationary and the only energy in the
form of heat is being transported.

3.4 GENERAL MOLECULAR TRANSPORT EQUATION

All three molecular transport processes (heat, momentum and mass transfer) are
characterized by the same general type of equation as indicated in unit 1.
driving force
Rate of a transport process = . . . (3.1)
resistance
66
This can be written as follows for molecular diffusion of the property momentum, heat, Mass Transfer
and mass
d
z    . . . (3.2)
dz
Molecular diffusion equation for momentum, heat and mass transfer.
Newton’s equation for momentum transfer for constant density can be written as follows
in a manner similar to Eq. (3.2).
 d
 zx   (vx  ) . . . (3.3)
 dz

where zx is momentum transferred/s.m2,  kinematic viscosity in m2/s, z = distance

in m, and vx  = momentum per m3, where the momentum has units of kg.m/s.
Fourier’s law for heat conduction can be written for constant  and Cp as
qz d
 ( C p T ) . . . (3.4)
A dz
qz
where is the heat flux in W/m2,  is the thermal diffusivity in m2/s and C pT
A
in J/m3.
The equation for molecular diffusion of mass is called the Fick’s law and is similar to Eq.
(3.2). It is written for constant total concentration in a fluid as
d CA
J *Az   DAB . . . (3.5)
dz
where J*Az is the molar flux component A in the z direction due to molecular diffusion in
kg mol A/s.m2, DAB is the molecular diffusivity of the molecule A in B in m2/s, CA is the
concentration of A in kg mol/m3, and z is the distance of diffusion in m.
The similarity of Eqs. (3.3)-(3.5) for momentum, heat and mass transfer is obvious. All
the fluxes on the left hand side of the three equations have as units transfer of a quantity
of momentum, heat or mass per unit time per unit area.
The transport properties  ,  and DAB all have units of m2/s, and the concentrations are

represented as momentum/m3, J/m3 or kg mol/m3.

3.5 PHYSICAL ORIGINS AND RATE EQUATIONS

From the standpoint of physical origins and the governing rate equations, strong
analogies exist between heat and mass transfer by diffusion.
3.5.1 Physical Origins
Both conduction heat transfer and mass diffusion are transport processes that originate
from molecular activity. Consider a chamber in which two different gas species at the
same temperature and pressure are initially separated by a partition. If the partition is
removed, both species will be transferred by diffusion. Figure 3.1 shows the situation as
it might exist shortly after removal of the partition.

67
Introduction to Heat
and Mass Transfer CA
Concentration of CB Concentration of
Species A Species B

ZD
Z

Figure 3.1 : Mass Transfer by Diffusion in a Binary Gas Mixture

A higher concentration means more molecules per unit volume, and the concentration of
species A (light dots) decreases with increasing z, while the concentration of B increases
with z. Since mass diffusion is in the direction of decreasing concentration, there is net
transport of species A to the right and of species B to the left. The physical mechanism
may be explained by considering the imaginary plane show as a dashed line at z0. Since
molecular motion is random, there is equal probability of any molecule moving to the left
or the right. Accordingly, more molecules of species A cross the plane from the left
(since this is the side of higher A concentration) than from the right. Similarly, the
concentration of B molecules is higher to the right of the plane than to the left, and
random motion provides for net transfer of species B to the left. Of course, after a
sufficient time, uniform concentrations of A and B are achieved, and there no net
transport of species A and B across the imaginary plane.
Mass diffusion occurs in liquids and solids, as well as in gases. However, since mass
transfer is strongly influenced by molecular spacing, diffusion occurs more readily in
gases than in liquids and more readily in liquids than in solids. Examples of diffusion
gases, liquids, and solids, respectively, include nitrous oxide from an automobile exhaust
in air, dissolved oxygen in water, and helium in Pyrex.

3.6 MIXTURE COMPOSITION

A mixture consists of two or more chemical constituents (species), and the amount of any
species i may be quantified in terms of its mass density i (kg/m3) and molar
concentration Ci (k mol/m3). The mass density and molar concentration are related
through the species molecular weight, Mi (kg/k mol), such that
i  M i Ci . . . (3.6)
With i representing the mass of species i per unit volume of the mixture, the mixture
mass density is

68
    i . . . (3.7) Mass Transfer
i

Similarly, the total number of moles per unit volume of the mixture is
C   Ci . . . (3.8)
i

The amount of species i in a mixture may also be quantified in terms of its mass fraction
i
mi  . . . (3.9)

or its mole fraction
Ci
xi  . . . (3.10)
C
From Eqs. (3.7) and (3.8), it follows that
 mi  1 . . . (3.11)
i

and  xi  1 . . . (3.12)
i

For a mixture of ideal gases, the mass density and molar concentration of any constituent
are related to the partial pressure of the constituent through the ideal gas law. That is,
Pi
i  . . . (3.13)
Ri T
Pi
and Ci  . . . (3.14)
RT
where Ri is the gas constant for species i and R is the universal gas constant. Using
Eqs. (3.10) and (3.14) with Delton’s law of partial pressures,
P   Pi . . . (3.15)
i

It follows that
Ci Pi
xi   . . . (3.16)
C P

3.7 FICK’S LAW OF DIFFUSION (VECTOR FORM)

Since the same physical mechanism is associated with heat and mass transfer by
diffusion, it is not surprising that the corresponding rate equations are of the same form.
The rate equation for mass diffusion is known as Fick’s law, and for the transfer of
species A in a binary mixture of A and B, it may be expressed in vector form as
j A    DAB  m A . . . (3.17)

or J A   CDAB  x A . . . (3.18)
These expressions are analogous to Fourier’s law, Eq. (2.3). Moreover, just as Fourier’s
law serves to define one important transport property, the thermal conductivity, Fick’s
law defines a second important transport property, namely, the binary diffusion
coefficient or mass diffusivity, DAB.
The quantity jA (kg/s.m2) is defined as the mass flux of species A. It is the amount of A
that is transferred per unit time and per unit area perpendicular to the direction of transfer,
and it is proportional to the mixture of mass density,    A   B (kg/m 3 ) , and to the

gradient in the species mass friction, m A  A . The species flux may also be evaluated

on a molar basis, where J A (k mol/s.m ) is the molar flux of the species A. It is
* 2 69
Introduction to Heat proportional to the total molar concentration of the mixture, C = CA + CB (k mol/m3), and
and Mass Transfer C
to the gradient in the species mole fraction, x A  A1 . The foregoing forms of Fick’s law
C
may be simplified with the total mass density  or the total molar concentration C is a
constant.
SAQ 2
(a) Define Fick’ s law of diffusion.
(b) Explain how does diffusion take place?

3.8 RESTRICTIVE CONDITIONS

Despite its analogous behaviour, mass diffusion, a good deal, more complicated than
conduction. Complications are associated with two restrictive conditions inherent in
Eqs. (3.17) and (3.18). First although mass diffusion may result from a temperature
gradient, a pressure gradient, or an external force, as well as from a concentration
gradient, we are assuming that these additional effects are not present or are negligible. In
most problems, this is in fact the case, and the dominant driving potential in the species
concentration gradient. This condition is referred to as ordinary diffusion. Treatment of
the other (higher order) effects is presented by Bird, et al. [1-3].
The second restrictive condition is that the fluxes are measured relative to coordinates
that move with the average velocity of the mixture. If the mass or molar flux of a species
is expressed relative to a fixed set of coordinates. Eqs. (3.17) and (3.18) are not generally
valid.
To obtain an expression for the mass flux relative to a fixed coordinates system, consider
species A in binary mixture of A and B. The mass flux n nA relative to a fixed coordinate
system is related to the species absolute velocity VA by

n nA   A VA . . . (3.19)
A value of VA may be associated with any point in the mixture, and it is interpreted as the
average velocity of all the A. Particulars in a small volume element about the point. An
average or an aggregate, velocity may also be associated with the particulars of species B
in which case

nBn   B VB . . . (3.20)
A mass-average velocity for the mixture may than be obtained from the requirement that

V  n n  n nA  n Bn   A V A   B VB . . . (3.21)
giving U  m A V A  m B VB . . . (3.22)
It is important to note that we have defined the velocities (VA, VB, V) and the fluxes
nA , nB , n as absolute quantities. That is, they are referred to axes that are fixed in space.
The mass average velocity V is a useful parameter of the binary mixture, since it need to
be multiplied by the total mass density to obtain the total mass flux with respect to fixed
axes. Moreover, since the velocities VA, VB and V are averages associated with aggregates
of particles, the fluxes nA , nB and n may be associated with transport due to bulk or
gross motion.
We may now define the mass flux of species A relative to the mixture mass-average
velocity as
j A   A (VA  V ) . . . (3.23)

70
Whereas nA is the absolute flux of species A, jA.is the relative or diffusive flux of the Mass Transfer

species. It represents the motion of the species relative to the average motion of the
mixture. It follows from Eq. (3.19) that
nA  j A   A V . . . (3.24)
This expression indicates that there are two contributions to the absolute flux of species
A : a contribution due to diffusion (i.e. due to the motion of a relative to the
mass-average motion of the mixture) and a contribution due to motion of A with the
mass-average motion of the mixture. Substituting from Eqs. (3.17) and (3.21), we obtain
nA    DAB m A  m A (nA  nB ) . . . (3.25)
At this point it is useful to recap what we have done by noting the alternative
formulations for the mass flux of species A. The form given by Eq. (3.17) determines the
transport of A relative to the mixture mass-average velocity, whereas the form given by
Eq. (3.25) determines the absolute transport of A. If the second term on the right hand
side of the Eq. (3.25) is not zero, the expression for absolute flux of species A.
Eq. (3.25), is not analogous to that for heat flux, Eq. (2.3).
The foregoing considerations may be extended to species B. The mass flux of B relative
to the mixture mass-average velocity (the diffusive flux) is
jB   B (VB  V ) . . . (3.26)
where jB    DBA mB . . . (3.27)
It follows from Eqs. (3.17), (3.23), and (3.26) that the diffusive fluxes in a binary mixture
are related by
j A  jB  0 . . . (3.28)
If Eqs. (3.17) and (3.27) are substituted into Eq. (3.18), and it is recognized that
m A   mB , since mA + mB = 1 for a binary mixture, it follows that

DBA  DAB . . . (3.29)

Hence, as in Eq. (3.25), the absolute flux of species B may be expressed as :
nB    DAB mB  mB (nA  nB ) . . . (3.30)
Although the foregoing expressions pertain to mass fluxes, the same procedures can be
used to obtain results on a molar basis. The absolute molar fluxes of species A and B may
be expressed as
N A  C A V A . . . (3.31)
and N B  C B VB . . . (3.32)
and a molar-average velocity for the mixture, V*, is obtained from the requirement that

N   N A  N B  CV *  C A V A  C B VB . . . (3.33)

Giving V *  x A V A  x B VB . . . (3.34)

The significance of the molar average velocity is that, when multiplied by the total molar
concentration C, it provides the total molar flux N with respect to a fixed coordinate
system. Eqs. (3.31) and (3.32) provide the absolute molar flux of species A and B. In
contrast, the molar flux of A relative to the mixture molar average velocity J*A, termed
the diffusive flux, may be obtained from Eq. (3.17) or from the expression

J *A  C A (VA  V * ) . . . (3.35)

To determine an expression similar in form to Eq. (3.25), we combine Eqs. (3.31), (3.32)
and (3.35) to obtain
71
Introduction to Heat
and Mass Transfer
N A  J A*  C A V * . . . (3.36)

or, from Eqs. (3.18) and (3.33),

N A   CDAB x A  x A (N A  N B ) . . . (3.37)

Compare the molar fluxes given by Eqs. (3.18) and (3.37). In the first case the molar flux
is relative to the mixture molar-average velocity, and in the second case it is the absolute
molar flux. Note also that Eq. (3.37) represents the absolute molar flux as the sum of a
diffusive flux and a flux due to the bulk motion of the mixture. For the binary mixture, it
also follows that

J *A  J B*  0 . . . (3.38)

A special case for which the absolute flux of a species is equal to the diffusive flux
pertains to what is termed a stationary medium. In terms of mass units, it is a medium for
which V = 0, in which case j A  nA . In terms of molar units, it is a medium for which
V* = 0 and hence J A  N A . For this special case, the analogy between heat and mass
transfer is complete, since the rate equations have the same physical form regardless of
the reference frame.

3.9 CONVECTIVE MASS TRANSFER

In the previous sections of this unit we have emphasized molecular diffusion in stagnant
fluids or fluids in laminar flow. In many cases the rate of diffusion is slow, and more
rapid transfer is desired. To do this, the fluid velocity is increased until turbulent mass
transfer occurs.
To have a fluid in convective flow usually requires the fluid to be flowing by another
immiscible fluid or by a solid surface. An example is a fluid flowing in a pipe, where part
of the wall pipe is made by a slightly dissolving solid material such as benzoic acid. The
benzoic acid dissolves and is transported perpendicular to the main stream from the wall.
When a fluid is in turbulent flow and is flowing past a surface, the actual velocity of
small particles of fluid cannot be described clearly as in laminar flow. In laminar flow the
fluid flows in streamlines and its behaviour can usually be described mathematically.
However, in turbulent motion there are no streamlines, but there are large eddies or
“chunks” of fluid moving rapidly in seemingly random fashion.
When a solute A is dissolving from a solid surface there is a high concentration of this
solute in the fluid at the surface, and its concentration, in general, decreases as the
distance from the wall increases. However, minute samples of fluid adjacent to each
other do not always have concentrations close to each other. This occurs because eddies
having solute in them move rapidly from one part of the fluid to another, transferring
relatively large amounts of solute. This turbulent diffusion or eddy transfer is quite fast in
comparison to molecular transfer.
When a fluid is flowing outside a solid surface in forced convection motion, we can
express the rate of convective mass transfer from the surface to the fluid, or vice-versa,
by the following equation
N A  k c ( C L1  C L 2 ) . . . (3.39)
where kc is the convective mass transfer coefficient in m/s, CL1 is the bulk fluid
concentration in kg mol A/m3, CL2 is the concentration in the fluid next to the surface of
the solid. This mass transfer coefficient is very similar to the heat transfer coefficient h
and is a function of the system geometry, fluid properties, and flow velocity.
Three regions of mass transfer can be visualized. In the first, which is adjacent to the
surface, a thin viscous sub layer film is present. Most of the mass transfer occurs by
molecular diffusion, since few or no eddies are present. A large concentration drop
occurs across this film as a result of the slow diffusion rate.
72
The transition or buffer region is adjacent to the first region. Some eddies are present and Mass Transfer
the mass transfer is the sum of turbulent and molecular diffusion. There is a gradual
transition in this region from the transfer by mainly molecular diffusion at the one end to
mainly turbulent at the other end.
In the turbulent region adjacent to the buffer region, most of the transfer is by turbulent
diffusion, with a small amount by molecular diffusion. The concentration decreases is
very small here since the eddies tend to keep the fluid concentration uniform. A typical
plot for the mass transfer of a dissolving solid from a surface to a turbulent fluid in a
conduit is given in Figure. 3.2.

CA 1

CA

CA 2

O z
Distance from Surface
Figure 3.2 : Concentration Profile in Turbulent Mass Transfer from a Surface to a Fluid

The concentration drop from cA1 adjacent to the surface is very abrupt close to the surface
and then levels off. This curve is very similar to the shapes found for heat and
momentum transfer. The average or mixed concentration c A is shown and is slightly
greater than the minimum cA2.
SAQ 3
(a) Distinguish between the molecular diffusion and convective mass transfer.
(b) What is turbulent mass transfer?
(c) How does the concentration profile changes in case of turbulent mass
transfer?
3.9.1 Mass Diffusion Coefficient
Considerable attention has been given to predicting the mass diffusion coefficient DAB for
the binary mixture of two gasses, A and B. Assuming ideal gas behaviour, kinetic theory
may be used to show that
3
1
DAB ~ p T 2 . . . (3.40)
This relation applies for restricted pressure and temperature ranges and is useful for
estimating values of the diffusion coefficient at condition other than those for which data
are available. Bird, et al. [1 – 3] provide detail discussion of available theoretical
treatments and compressions with experiment.
For binary liquid solution, it is necessary to rely exclusively on experimental
measurements. For small concentration of A (the solute) in B (the solvent), DAB is known
to increase with increasing temperature. The mechanism of diffusion of gases, liquids and
solids in solids is extremely complicated and generalized theories are not available.
Furthermore, only limited experimental results are available in literature.
Data for binary diffusion in selected mixtures are presented in Table 3.1.

73
Introduction to Heat Table 3.1 : Binary Diffusion Coefficients at One Atmosphere
and Mass Transfer
Substance A Substance B T DAB
(K) (m2/s)
Gases
NH3 Air 2982 0.28  10– 4
H2O Air 298 0.26  10– 4
CO2 Air 298 0.16  10– 4
H2 Air 298 0.41  10– 4
O2 Air 298 0.21  10– 4
Acetone Air 273 0.11  10– 4
Benzene Air 287 0.88  10– 5
Naphthalene Air 300 0.62  10– 5
Ar N2 293 0.19  10– 4
H2 O2 273 0.70  10– 4
H2 N2 273 0.68  10– 4
H2 CO2 273 0.55  10– 4
CO2 N2 293 0.16  10– 4
CO2 O2 273 0.14  10– 4
O2 N2 273 0.18  10– 4
Dilute Solutions
Caffeine H2O 298 0.63  10– 9
Ethanol H2O 298 0.12  10– 8
Glucose H2O 298 0.69  10– 9
Glycerol H2O 298 0.94  10– 9
Acetone H2O 298 0.13  10– 8
CO2 H2O 298 0.20  10– 8
O2 H2O 298 0.24  10– 8
H2 H2O 298 0.63  10– 8
H2 H2O 298 0.26  10– 8
Solids
O2 Rubber 298 0.21  10– 9
N2 Rubber 298 0.15  10– 9
CO2 Rubber 298 0.11  10– 9
He SiO2 293 0.4  10– 13
H2 Fe 293 0.26  10– 12
Cd Cu 293 0.27  10– 18
Al Cu 293 0.13  10– 33

SAQ 4
Show that for a equimolar counter diffusion of two species A and B
DAB  DBA

3.10 TYPES OF MASS-TRANSFER COEFFICIENTS

3.10.1 Definition of Mass-Transfer Coefficient
Since our understanding of turbulent flow is incomplete, we attempt to write the
equations for turbulent diffusion in a manner similar to that for molecular diffusion. For
turbulent mass transfer for constant c,
74
 dc A Mass Transfer
J A   ( DAB   M ) . . . (3.41)
dz
where DAB is the molecular diffusivity in m2/s and M is the mass eddy diffusivity in m2/s.
The value of M is a variable and is near zero at the interface or surface and increases as
the distance from the wall increases. We then use an average value M since the
variation of M is not generally known. Integrating Eq. (3.41) between points 1 and 2,
DAB  M
J A1  (c A1  c A2 ) . . . (3.42)
z2  z1

The flux J A1 is based on the surface area A1 since the cross-sectional area may vary. The
value of z2 – z1, the distance of the path, is often not known. Hence, Eq. (3.42) is
simplified and is written using a convective mass-transfer coefficient kc.

J A1  kc (c A1  c A2 ) . . . (3.43)

where J A1 is the flux of A from the surface A1 relative to the whole bulk phase, kc is
( DAB  M )
an experimental mass transfer coefficient in kg mol/s.m2. (kg mol/m3) or
( z2  z1 )
simplified as m/s, and cA2 is the concentration at point 2 in kg mol A/m3 or more usually
the average bulk concentration c A2 . This defining of a convective mass-transfer
coefficient kc is quite similar to the convective heat-transfer coefficient h.
3.10.2 Mass-Transfer Coefficient for Equimolar Counter Diffusion
Figure 3.3 presents the phenomenon of equimolar counterdiffusion of gases. Gases A and
B are kept in the two chambers connected by a tube at a total pressure p. Assume
molecular diffusion to be under a steady state condition. In order to maintain uniform
concentration of gases, both the chambers are stirred continuously. Let, the partial
pressures be pA1 > pA2 and pB1 > pB2.
Molecules of A diffuse to the right and B to the left. Since the total pressure p is constant
throughout, the net moles of A diffusing to the right must be equal to the net moles of B
to the left. If this is not so, the total pressure would not remain constant. This means that

J *Az   J Bz
*
. . . (3.44)

PA1 PA2

1 2
PB1 PB 2
*
P J A P
*
J B

P
PA1
PB2
PA, PB, or P
PA2
PB1

Figure 3.3 : Equimolar Counter Diffusion of Gases A and B

75
Introduction to Heat The subscript z is dropped when the direction is obvious. Writing Fick’s law for B for
and Mass Transfer constant c

dcB
J B*   DBA . . . (3.45)
dz
Now since p = pA + pB constant, then

c  c A  cB . . . (3.46)

Differentiating both sides,

dc A   dcB . . . (3.47)

Equating Eqs. (3.18) to (3.45),

dc A dc
J *A   DAB   J B*   ( ) DBA B . . . (3.48)
dz dz
Substituting Eq. (3.47) into Eq. (3.48) and canceling like terms,

DAB  DBA . . . (3.49)

This shows that for a binary gas mixtures of A and B the diffusivity coefficient DAB for A
diffusing in B is the same as DBA diffusing into A.

3.10.3 Mass-Transfer Coefficient for A Diffusing through Stagnant,

Non-diffusing B
The case of diffusion of A through stagnant or non-diffusing B at steady state often
occurs. In this case, one of the boundaries at the end of the diffusion path is impermeable
to component B and hence the species B cannot pass through it. Figure 3.4 presents the
evaporation of a pure liquid such as benzene (say, species A) at the bottom of a narrow
tube, where a large amount of inert or non-diffusing air (species B) is passed over the top.
The benzene vapour (A) diffuses through the air (B) in the tube. The boundary at the
liquid surface at point is impermeable to air, since air is insoluble in benzene liquid.
Hence, air (B) cannot diffuse into or away from the surface. At point 2, the partial
pressure pA2 = 0, since a large volume of air is passing by.

Air (B)
2 PA2

z0

z2 – z1
z
ZF
NA

1 PA1
Liquid
Benzene (A)

Figure 3.4 : Diffusion of Benzene (A) through Non-diffusing Air (B)

Another example is absorption of NH3 (A) vapour which is in air (B) by water as shown
in Figure 3.5. The water surface is impermeable to the air as air is very slightly soluble in
water. Thus, since B cannot diffuse, NB = 0.

76
 1 NH3 (A)
air (B)
NA
Mass Transfer
z2 – z1

Liquid Water

Figure 3.5 : Ammonia in Air being Absorbed into Water

For the species A diffusing through stagnant, non-diffusing species B where NB = 0,

Eq. (3.44) gives for steady state
dx A c A
N A   c DAB  (N A  0) . . . (3.50)
dz c
p c p
Keeping the total pressure p constant, substituting c  , p A  x A p and A  A
RT c p
into Eq. (3.50)
DAB dp A p
NA    A NA . . . (3.51)
RT dz p

Rearranging Eq. (3.51) and integrating we get,

 p  D dp A
N A  1  A    AB . . . (3.52)
 p  RT dz
z2 p A2
DAB dp A
NA  dz  
RT  p
. . . (3.53)
z1 p A1 1 A
p

DAB p p  pA 2
NA  ln . . . (3.54)
RT ( z 2  z1 ) p  p A1

Eq. (3.54) is the final expression to be used to calculate the flux A.

Now, p  p A1  p B1  p A2  p B 2 . . . (3.55)

Hence, pB1  p  p A1 . . . (3.56)

And, pB 2  p  p A2 . . . (3.57)

The log mean part of the inert species B can be written as

pB 2  pB1 p A1  p A2
pBM   . . . (3.58)
p   p  p A2 
ln  B 2  ln  
 pB1   p  p A1 
Substituting Eq. (3.58) into Eq. (3.54),
DAB p
NA  ( p A1  p A 2 ) . . . (3.59)
RT ( z 2  z1 ) p BM

Example 3.1
A mixture of He and N2 gas is contained in a pipe at 298 K and 1 atm total
pressure which is constant throughout. At one end of the pipe at point 1 the partial
pressure pA1 of He is 0.60 atm and at the other end 0.2 m (20 cm) pA2 = 0.20 atm.
77
Introduction to Heat Calculate the flux of He at steady state if DAB of the He-N2 mixture is
and Mass Transfer 0.687  10– 4 m2/s (0.687 cm2/s). Use SI and cgs units.
Solution
Since total pressure p is constant, then c is constant, where c is as follows for a gas
from the perfect gas law.
PV  n RT

n P
 c
V RT
where n is kg mol A plus B, V is volume in m3, T is temperature in K,
R is 8314.3 m3. Pa/kg mol.K or R is 82.057 cm3 atm/g mol.K.
For steady state the flux J*Az is constant. Also, DAB for a gas is constant. We can
write,
z2 c A2
J Az  dz   DAB  dc A
z1 c A1

DAB (c A1  c A 2 )
J Az 
z2  z1

Also, from the perfect gas law, pA V = nA RT, and

p A1 n A
c A1  
RT V
DAB ( p A1  p A 2 )
Now, J Az 
RT ( z 2  z1 )

This is the final equation to use, which is in the form easily used for gases. Partial
pressures are pA1 = 0.6 atm = 0.6  1.01325  105 = 6.08  104 Pa and
pA2 = 0.2 atm = 0.2  1.01325  105 = 2.027  104 Pa. then, using SI units,

(0.687  10  4 ) (6.08  10 4  2.027  10 4 )

J Az 
8314 (298) (0.20  0)

= 5.63  10– 6 kg mol A/s. m2

If pressures in atm are used with SI units,

(0.687  10  4 ) (0.60  0.20)

J Az 
(82.06  10  3 ) (298) (0.20  0)

= 5.63  10– 6 kg mol A/s. m2

For cgs units
0.687 (0.60  0.20)
J Az 
82.06 (298) (20  0)

= 5.63  10– 7 g mol A/s. m2

Example 3.2
Ammonia gas (A) is diffusing through a uniform tube 0.10 m long containing N2
gas (B) at 1.0132  105 Pa and 298 K. The diagram is similar to Figure 3.3. At
point 1, pA1 = 1.013  104 Pa and at point 2, pA2 = 0.507  104 Pa. The diffusivity
DAB = 0.230  10– 4 m2/s. Calculate :
78
 (i) The flux J*A at steady state, and Mass Transfer

(ii) Repeat for J*B.

Solution
P = 1.01132  105 Pa, z2 – z1 m, and T = 298 K.
Substituting into the following equation

DAB ( p A1  p A 2 ) (0.23  10  4 ) (1.013  10 4  0.507  10 4 )

J A  
RT ( z 2  z1 ) 8314 (298) (0.10  0)

= 4.70  10– 7 kg mol A/s.m2

For component B

pB1  P  p A1  1.0132  10 5  1.013  10 4  9.119  10 4 Pa

and pB 2  P  p A2  1.0132  10 5  0.507  10 4  9.119  10 4 Pa,

DAB ( pB1  pB 2 ) (0.23  10  4 ) (9.119  10 4  9.625  10 4 )

J B  
RT ( z 2  z1 ) 8314 (298) (0.10  0)

=  4.70  10– 7 kg mol B/s.m2

The negative value for J*B means the flux goes from point 2 to 1.
Example 3.3
Water in the bottom of a narrow metal tube is held at a constant temperature of 293
K. The total pressure of air (assumed dry) is 1.01325  105 Pa (1.0 atm) and the
temperature is 293 K (20oC). Water evaporates and diffuses through the air in the
tube and the diffusion path z2 – z1 is 0.1524 m (0.5 ft) long. The diagram is similar
to Figure (6.22(a)). Calculate the rate of evaporation at steady in
lb mol/h.ft2 and kg mol/s.m2. The diffusivity of water vapour at 293 K and 1 atm
pressure is 0.250  10– 4 m2/s. Assume that the system is isothermal. Use SI and
English units.
Solution
The diffusivity is converted to ft2/h by using the suitable conversion factor

DAB  0.250  10  4 (3.875  10 4 )  0.969 ft 2/h

17.54
Vapour pressure of water at 20oC is 17.54 mm or p A1   0.0231 atm
760
0.0231 (1.01325  10 5 )  2.341  10 3 Pa, pA2 = 0 (pure air). Since the temperature
is 20oC (68oF), T = 460 + 528oR = 293 K.
R = 0.730 ft3 atm/lb mol.oR. To calculate the value of pBM,
pB1  P  p A1  1.00  0.0231  0.9769 atm
pB 2  P  p A2  1.00  0  1.00 atm
pB 2  pB1 1.00  0.9769
pBM    0.988 atm  1.001  10 5
p   1.00 
ln  B 2  ln  
 pB1   0.9769 

( pB1  pB 2 )
Since pB1 is closed to pB2, the linear mean could be used and would
2
be very closed to pBM.
Substituting into Equation with z2 – z1 = 0.5 ft, 79
Introduction to Heat DAB P 0.969 (1.0) (0.0231  0)
and Mass Transfer NA  ( p A1  p A 2 ) 
RT ( z 2  z1 ) p BM 0.730 (528) (0.5) (0.988)
= 1.175  10– 4 lb mol/h.ft2
(0.250  10  4 ) (1.01325  10 5 ) (2.341  10 3  0)
NA 
8314 (293) (0.1524) (1.001  10 5)
= 1.595  10– 7 kg mol/s.m2
Exercise 3.1
(a) A gas of CH4 and He is contained in a tube at 101.32 kPa pressure and
298 K. At one point the partial pressure of methane is pA1 = 60.79 kPa and at
a point 0.02 m distance away, pA2 = 20.26 kPa. If the total pressure is
constant throughout the tube, calculate the flux of CH4 (methane) at steady-
state for equimolar counter-diffusion.
(b) The gas CO2 is diffusing at steady state through a tube 0.20 m long having a
diameter of 0.01 m and containing N2 at 298 K. The total pressure is
constant at 101.32 kPa. The partial pressure of CO2 at one end is 456 mm
Hg and 76 mm Hg at the other end. The diffusivity DAB is 1.67  10– 5 m2/s
at 298 K. Calculate the flux of CO2 in cgs and SI units for equimolar
counter-diffusion.
Exercise 3.2
(a) Ammonia gas (A) and nitrogen gas (B) are diffusing in counterdiffusion
through a straight glass tube 2.0 ft (0.610 m) long with an inside diameter of
0.080 ft (24.4 mm) at 298 K and 101.32 kPa. Both ends of the tube are
connected to large mixed chambers at 101.32 kPa. The partial pressure of
NH3 in one chamber is constant at 20.0 kPa and 6.666 kPa in the other
chamber. The diffusivity at 298 K and 101.32 kPa is 2.30  10– 5 m2/s.
(i) Calculate the diffusion of NH3 in lb mol/h and kg mol/s.
(ii) Calculate the diffusion of N2.
(iii) Calculate the partial pressures at a point 1.0 ft (0.305 m) in the tube
and plot pA, pB and P versus distance z.
(b) Ammonia gas is diffusing through N2 under steady-state conditions with N2
non-diffusing since it is insoluble in one boundary. The total pressure is
1.013  105 Pa and the temperature is 298 K. The partial pressure of NH3 at
one point is 1.333  104 Pa and at the other point 20 mm away it is
6.666  103 Pa. The DAB for the mixture at 1.013  105 Pa and 298 K is
2.30  10– 5 m2/s.
(i) Calculate the flux of NH3 in kg mol/s.m2.
(ii) Do the same as (i) but assume that N2 also diffuses, i.e. both
boundaries are permeable to both gases and the flux is equimolar
counter-diffusion. In which case is the flux greater?
(c) Methane gas is diffusing in a straight tube 0.1 m long containing helium at
298 K and a total pressure of 1.01325  105 Pa. The partial pressure of CH4
at one end is 1.400  104 Pa and 1.333  103 Pa at the other end. Helium is
insoluble in one boundary, and hence is non-diffusing or stagnant. The
diffusivity is given in Table 6.2. Calculate the flux of methane in
kg mol/s.m2 at steady state.

80
 Mass Transfer
3.11 SUMMARY
Let us summarise what we have learnt in this unit.
Present unit describes the mass transport phenomena. Physical origin and rate equation
for mass transfer are explained in details. Mass transfer may occur due to the molecular
diffusion in a stationary medium due to a concentration gradient. Apart from this, mass
transfer may occur due to the bulk motion of fluid. This is known as convective mass
transfer. Based on the types of mass transfer mechanism, various mass transfer
coefficients may be defined and evaluated. A few special cases of mass transfer between
two species are also described. Some solved examples are given in the unit to clarify the
conceptions. A few unsolved problems are given at the end of the unit, which will help
you to have more insight of the theory.

3.12 KEY WORDS

Fick’s Law : Defines a second important transport property,
namely, the binary diffusion coefficient or mass
diffusivity, DAB.
Diffusional Mass Transfer : Mass transfer in a stationary medium due to the
concentration gradient.
Convective Mass Transfer : Mass transfer due to bulk fluid motion.
Solute : A solid added to a solvent

3.13 ANSWERS TO SAQ’s

Refer the preceding text in this unit for all the Answers to SAQs.

FURTHER READINGS
F. P. Incropera and D. P. Dewitt (2004), Fundamentals of Heat and Mass Transfer, John
Wiley and Sons, 5th Edition.
W. M. Rohsenow and H. C. Choi (1961), Heat Mass and Momentum Transfer,
Prentice-Hall Englewood Cliffs, N. J.
E. R. G. Eckert and R. M. Drake Jr. (1959), Heat and Mass Transfer, McGraw-Hill,
New York.
P. K. Nag (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.
P. Ghoshdastidar (2004), Heat Transfer, Oxford University Press.
J. P. Holman (2002), Heat Transfer, Tata McGraw-Hill, New Delhi, 9th Edition.
E. Fried (1969), Thermal Conduction Contribution to Heat Transfer at Contacts,
Thermal Conductivity, R. P. Tye [Ed], Volume 2, Academic Press, London.
R. B. Bird (1956), Advance Chem. Eng., 1, 170.
R. B. Bird, W. E. Stewart, and E. N. Lightfoot (1060), Transport Phenomena, Wiley,
New York.
J. O. Hirschfelder, C. F. Curtiss, and R. B. Bird (1954), Molecular Theory of Gases and
Liquids, Wiley, New York.
R. E. Treybel (1980), Mass Transfer Operations, 3rd Edition, McGraw-Hill Books
Company, New York.
81
Introduction to Heat C. J. Geankoplis (2004), Transport Processes and Unit Operations, 3rd Edition,
and Mass Transfer Prentice-Hall of India Pvt. Ltd, New Delhi.
M. N. Ozisik (1985), Heat Transfer – A Basic Approach, McGraw-Hill Book Company.

HEAT AND MASS TRANSFER

This course consists of 6 blocks.
Block 1, Introduction to Heat and Mass Transfer, deals with the basic concepts of heat
and mass transfer and gives a bird’s eye view of the mechanisms. It builds up the
necessary background for the understanding of the physical significances of the
conduction, convection and radiation heat transfer. The phenomena of mass transfer,
mechanism of mass transfer and various types of mass transfer are discussed.
Block 2, Conduction, deals with the governing equations of conduction heat transfer.
Application of Fourier’s law to steady and unsteady state conduction has been analysed.
The governing equation of heat conduction is presented in rectangular, cylindrical as well
as spherical coordinates. Formulation and solution are done for extended surfaces.
Performance parameters such as fin efficiency and effectiveness are discussed in details.
Block 3, Convection, deals with heat transfer by convection and discusses the physical
and mathematical basis for the understanding of convective transport and to reveal
various heat transfer correlations. The analysis of convection is complicated because the
fluid motion is affected by pressure drop, drag force and heat transfer. The literature of
convective heat transfer is overwhelming and ever growing.
Block 4, Radiation, deals with the third mode of heat transfer, i.e. radiation has been
considered. Thermal radiation is that electromagnetic radiation emitted by a body as a
result of its temperature. The principles of radiation, radiation intensity and emissive
power are discussed. Concept of blackbody and laws of radiation are discussed.
Block 5, Mass Transfer, deals with mass transfer processes occur in a variety of
applications in mechanical, chemical and aerospace engineering. Similarly, mass transfer
is also considered in many of the processes in physics, chemistry and biology. Typical
examples include transpiration cooling of jet engines and rocket motors, the ablative
cooling of space vehicles during reentry into the atmosphere, mass transfer from laminar
and turbulent streams onto the surface of a conduit, evaporation and condensation in heat
exchangers, membrane separations, etc.
Molecular diffusion and Ficks law of diffusion are discussed.
Block 6, Heat and Mass Transfer Applications, deals with applications of heat and mass
transfer in practical engineering fields. Different types of heat exchangers utilized in
industries are discussed. The very purpose of the entire course will remain incomplete if
we do not discuss about heat and mass transfer associated with chemical processes.
Chemical processes such as evaporation, liquid-liquid extraction, adsorption and
membrane separation are in use in industry where lot of mass transfer is involved. This
block mainly devoted for applications of heat and mass transfer in chemical processes.

INTRODUCTION TO HEAT AND MASS

TRANSFER
Unit 1 introduces the basic concepts of heat and mass transfer and gives a bird’s eye view
of the mechanisms. A discussion of units and dimension is also presented. Unit 2 builds
up the necessary background for the understanding of the physical significances of the
82
conduction, convection and radiation heat transfer. Emphasis is placed on the Mass Transfer
mathematical formulation of practical heat conduction problems. This matter is
illustrated with numerous representative examples. Interaction of conduction, convection
and radiation is also discussed. Unit 3 deals with the phenomena of mass transfer.
Mechanism of mass transfer and various types of mass transfer are discussed. The unit is
supported with solved problems for better insight of the mass transfer mechanism. All the
three units include some unsolved problems and SAQs which will help you
understanding the basic heat and mass transfer phenomena.

83
Indira Gandhi MRW-002
National Open University
School of Engineering & Technology HEAT TRANSFER

BLOCK 2

Conduction
 Indira Gandhi
National Open University
MRW – 002
School of Engineering & Technology Heat Transfer

Block

2
CONDUCTION
UNIT 4
Governing Equations of Heat Conduction 5

UNIT 5
Numerical Methods to Solve Heat Conduction Problems 51

UNIT 6
Heat Transfer from Extended Surface 99
CONDUCTION

Present Block is devoted to conduction heat transfer. The entire block consists of
three units.

Unit 4, introduces the governing equations of conduction heat transfer.

Application of Fourier’s law to steady and unsteady state conduction has been
analysed. The governing equation of heat conduction is presented in rectangular,
cylindrical as well as spherical coordinates. Heat transfer and determination of
temperature distribution for plane wall, cylinder and sphere under steady state
condition are discussed. Transit heat transfer concept is introduced with heat
generation number, Biot number and Fourier number. A detail analysis of
transient heat transfer by lumped capacitance method is described.

Unit 5, describes various methods for solving heat conduction equations.

Analytical and graphical methods are described. Limitation of these methods are
highlighted and it is emphasized that numerical methods is used in complicated
geometries, boundary conditions and variable thermal properties. Finite difference
method is described both for the steady and transient conduction problems.

Unit 6, describes application of extended surfaces for heat transfer devices.

Different types of fins and their applications are cited in this unit. Formulation
and solution are done for extended surfaces. Performance parameters such as fin
efficiency and effectiveness are discussed in detail.

The units are supported with solved problems. All the three units include some
unsolved problems and SAQ’s which will help you understanding the heat
conduction mechanism.
 Governing Equations of
UNIT 4 GOVERNING EQUATIONS OF HEAT Heat Conduction

CONDUCTION
Structure
4.1 Introduction
Objectives
4.2 General Equation of Heat Conduction
4.2.1 Rectangular Coordinate System
4.2.2 Cylindrical Coordinates
4.2.3 Spherical Coordinates
4.3 Steady State Heat Conduction in Simple Geometrical Systems
4.3.1 Plane Wall
4.3.2 Cylindrical Wall
4.3.3 Spherical Shell
4.4 Transient Conduction
4.4.1 Lumped Capacitance Method
4.4.2 General Lumped Capacitance Analysis
4.5 The Semi-infinite Solid
4.6 Multi-dimensional Effects
4.7 Summary
4.8 Key Words
4.9 Answers to SAQs

1.1 INTRODUCTION
In conduction mode heat is transferred through a complex sub-microscopic mechanism
that involves flow of free electrons and lattice vibration. Conduction is predominant in
case of solid or liquid metals. In case of liquids and gases, once heat begins to flow, even
if no external force is applied, density gradients are set up and convective currents are set
in motion. Heat is then transported on a macroscopic scale as well as on a microscopic
scale with convection currents generally being the more effective. In the following sub-
units conduction heat transfer has been discussed in details for steady as well as transient
states.
Objectives
After studying this unit, you should be able to
 formulate heat transfer by conduction in different geometries under steady
state condition,
 evaluate temperature at different locations in rectangular, cylindrical and
spherical coordinate system,
 differentiate the steady state and transient nature of heat conduction,
 evaluate heat transfer by lumped capacitance method and its limitations, and
 solve some problems on steady and transient heat transfer.

4.2 GENERAL EQUATION OF HEAT CONDUCTION

General equation of heat conduction is explained in three coordinate systems.
5
Conduction 4.2.1 Rectangular Coordinate System
Differential Form
Let us consider an infinitesimal volume element of sides x, y and z as shown in
Figure 4.1. The considerations here will include the non-steady condition of
temperature variation with time t.
Z
dQy + dz

dQx + dy

dQx + dx
dQx

z

y

dQy
x
x

dQz

Figure 4.1 : Volume Element for Determining Heat Conduction Equation

According to the Fourier heat conduction law, the heat flowing into the left most
face of the element in the x-direction
dT
dQx   k y z . . . (4.1)
x
The value of the heat flow out of the right face of the element can be obtained by
expanding dQx in a Taylor series and retaining only the first two terms as an
approximation :

dQx  dx  dQx  (dQx ) x  . . . . . . (4.2)
x
The net heat flow by conduction in the x-direction is therefore
   T 
dQx  dQx  dx   (dQx ) x    k y z  x
x x  x 

 2T
k x y z . . . (4.3)
x 2
Similarly, the net heat flows in the y-and z-direction are
 2T
dQ y  dQ y  dy  k x y z . . . (4.4)
y 2

 2T
dQz  dQz  dz  k x y z . . . (4.5)
z 2
Here, the solid has been assumed to be isotropic and homogeneous with properties
uniform in all directions. Let us consider that there is some heat source within the
solid, and heat is produced internally as a result of the flow of electrical current or
nuclear or chemical reactions. Let qG is the rate at which heat is generated

6
 Governing Equations of
W 
internally per unit volume  3  . Then the total rate of heat generation in the Heat Conduction
m 
elemental volume is qG x y z. The net heat flow owing to conduction and the
heat generated within the element together will increase the internal energy of the
volume element. The rate of accumulation of internal energy IE within the control
volume is
T
IE  c x . y . z . . . (4.6)
t
where c is the specific heat and  is the density of the solid.
An energy balance can be achieved on the volume element as :
Rate of energy storage within the solid
= Rate of heat influx – Rate of heat outflux + Rate of heat generation
T
or  c x y z  (dQx  dQy  dQz )  (dQx  dx  dQy  dy  dQz  dz )
t
 qG x y z

  2T  2T  2T 
 k x y z  2  2  2   q G x y z . . . (4.7)
 x y z 


T   2T  2T  2T 
or c k 2  2  2   qG . . . (4.8)
t  x y z
 
 2T  2T  2T qG 1 T
    . . . (4.9)
x 2 y 2 z k  t

k
where  is the thermal diffusivity of the solid given by .
c
If the temperature of a material is not a function of time, the system is in the steady
state and does not store any energy. The steady state form of a three-dimensional
conduction equation in rectangular coordinates is
 2T  2T  2T qG
2
 2
 2
 0 . . . (4.10)
x y z k
If the stem is in steady state and no heat is generated internally, the conduction
equation simplifies to
 2T  2T  2T
  0 . . . (4.11)
x 2 y 2 z 2
Eq. (4.11) is known as the Laplace equation. It occurs in a number of areas in
addition to heat transfer, for example, in diffusion of mass or in electromagnetic
fields. The operation of taking the second derivatives of the potential in a field has
therefore been given a short symbol, 2, called the Laplacian operator. For the
rectangular coordinate system Eq. (4.11) becomes
 2T  2T  2T
2
 2
 2
 2 T  0 . . . (4.12)
x y z
Since the operator 2 is independent of coordinate system, the above form will be
useful when we study conduction in cylindrical and spherical coordinates. The heat
conduction can thus be written as
qG 1 T
2 T   . . . (4.13)
k  t 7
Conduction Vector Method
Equation can also be derived vectorially. Let us consider a control surface S
enclosing a volume V as shown in Figure 4.2.
q

ds

d

qG

Control surface S
q

Figure 4.2 : Heat Conduction through a Volume Element

The net rate of heat outflow across the surface S is given by  q . nds where n is
s
the normal direction. Converting the surface integral to volume integral

 q . n ds   div q dv . . . (4.14)
s v
Where q is the heat flux per unit area. The net rate of heat inflow to the control

volume (CV) is   div q dv .

v
If there is a volumetric heat source inside the CV the rate of heat generation is

 qG dv . . . (4.15)
v
The rate of energy accumulation within the CV is
 
t  e  dv 
t  c T  dV . . . (4.16)
v v
e being the specific energy (J/kg). By energy balance,

t   c T dv   qG dv   div q dV . . . (4.17)
v v v
where q is the heat flux per unit area. Writing the energy equation for the
elemental volume dV within the CV
T
c dV  qG dV  div q dV . . . (4.18)
t
Since dV is now independent, it can be removed from the above equation.
Therefore,
T
c  qG  div q . . . (4.19)
t
Now, div q   . q . . . (4.20)
and q k T . . . (4.21)

div q   (  k  T )   k  2 T . . . (4.22)

for constant k.
8
 Substituting in Eq. (4.19), Governing Equations of
Heat Conduction
T
c  qG  k  2 T . . . (4.23)
t
qG 1 T
Therefore, 2 T   . . . (4.24)
k  t
Which is the same as Eq. (4.13).
4.2.2 Cylindrical Coordinates
For a general transient three-dimensional heat conduction problem in the cylindrical
coordinates with T = T (r, q, z, t), let us consider an elementary volume as shown in
Figure 4.3.
dV  d r r d  dz . . . (4.25)
r dQr +dr
dQθ +dθ
dr
dQz +dz
ds

n dQz
q
dQθ

dQr dθ z
r

dθ
θ
r

Figure 4.3 : Heat Conduction in a Cylindrical Volume Element (r, , z)

By Fourier’s law,
T
dQr   k (r d  z ) . . . (4.26)
r

dQr  dr  dQr  (dQr ) dr . . . (4.27)
r

  T 
dQr  dQr  dr    k r d  dz  dr
r  r 

 2T T
 kr d  dz dr 2
 k d  dz dr . . . (4.28)
r r

T
Similarly, dQ   k dz dr . . . (4.29)
r 

  T 
dQ  d Q  d     k dz dr  d . . . (4.30)
  r  
9
Conduction 
dQ  d   d Q  (d Q ) r d  . . . (4.31)
r 

T  2T T
 dr r d  dzc  kr d  dr 2
 k d  dz dr
t r r

 2T  2T
 k dz dr d   k dr rd  dz
2  z2

1  2T
 k dz dr d  . . . (4.32)
r 2
T
dQz   k dr rd  . . . (4.33)
z

dQz  dz  dQz  (dQz ) dz . . . (4.34)
z
  T 
dQz  dQz  dz     k dr r d   dz
z  z 

 2T
 k dr r d  dz . . . (4.35)
z 2
Rate of heat generation from an internal heat source
 qG dr rd  dz . . . (4.36)

Rate of energy accumulation within the CV

T
  ( dr r d ) c . . . (4.37)
t
By energy balance, from Eqs. (4.28)-(4.37),
T  2T T  2T
 dr rd  d z c  k r d  dr 2  k d  dz dr  k dz dr d  2
t r r 

 2T
 k dr r d  dz  q G dr r d  dz . . . (4.38)
z 2

T  2T 1 T 1  2T  2T
c k  k  k  k . . . (4.39)
t r 2 r r r 2 2 z 2

 2T 1 T 1  2T  2T q 1 T
or qG   2  2  G  . . . (4.40)
r 2 r r r  2
z k  t

1  T 1  2T  2T q 1 T
or r  2 2
 2  G  . . . (4.41)
r r r r  z k  t
This is the general heat conduction equation in cylindrical co-ordinates. If we compare
this equation with Eq. (4.13), the Laplacian is

1   T  1  2T  2T
2 T  r   2 . . . (4.42)
r r  r  r 2 2 z
If heat flows only in radial direction, T = T (r, t), Eq. (4.41) reduces to
1   T  qG 1 T
r   . . . (4.43)
r r  r  k  t
10
If the temperature distribution does not vary with time, then at steady state, Governing Equations of
Heat Conduction
1   T  qG
r  0 . . . (4.44)
r r  r  k

In this case the equation for the temperature contains only a single variable r and is
therefore an ordinary differential equation.
When there is no volumetric energy generation and temperature is a function of the
radius only, the steady-state conduction for cylindrical coordinates is

d  dT 
r 0 . . . (4.45)
dr  dr 

4.2.3 Spherical Coordinates

For spherical coordinates, as shown in Figure 4.4 the temperature is a function of the
three space coordinates r,  and  and time t, i.e. T = T (r, , , t). The general form of
the conduction equation in spherical coordinates can be found as
z

dr

dθ

d


Figure 4.4 : Spherical Coordinate System for the General Conduction Equation

1   2 T  1   T  1  2T q 1 T
2 r 
r   2  sin    2 2 2
 G . . . (4.46)
r  r  r sin      r sin   k  t

Where the Lapacian includes the first three terms of the above equation in the spherical
coordinates.

4.3 STEADY STATE HEAT CONDUCTION IN SIMPLE

GEOMETRICAL SYSTEMS
We will now derive solutions to the conduction equation as obtained in the previous
section for simple geometrical systems with and without heat generation.
4.3.1 Plane Wall
In Unit 2 we saw that the temperature distribution for one-dimensional steady conduction
through a wall is linear. We will verify the result by simplify the more general equation
(Eq. 4.13)
qG 1 T
2 T   . . . (4.47)
k  t

11
Conduction T T T
For steady state,  0 . Since T is only a function of x,  0 and  0 . There is
t y z
no internal heat generation, qG = 0. Therefore, the above equation reduces
d 2T
0 . . . (4.48)
dx 2
Integrating the ordinary differential equation twice yields the linear temperature
distribution
T ( x )  C1 x  C 2 . . . (4.49)

For a wall as shown in Figure 4.5, at x = 0, T = T1 and at x = b, T = T2

T1

T2

Figure 4.5 : Heat Conduction Through a Plane Wall

T1  T2
T  x  T2 . . . (4.50)
b
Which agrees with the linear temperature distribution deduced by increasing Fourier’s
law
dT
Qk   kA . . . (4.51)
dx
Let us now consider a heat source generating throughout the system. If the thermal
conductivity is constant and the heat generation is uniform, Eq. (4.13) reduces to

d 2T qG
2
 0 . . . (4.52)
dx k

dT q
On integration,   G  C1 . . . (4.53)
dx k
A second integration gives
qG 2
T ( x)   x  C1 x  C 2 . . . (4.54)
2k
where C1 and C2 are constants.
At x = 0, T = T1 and at x = b, T = T2 substituting in Eq. (4.52),
T1  C2 . . . (4.55)

qG 2
T2   b  C1 b  T1 . . . (4.56)
2k
12
 T2  T1 qG Governing Equations of
C1   b . . . (4.57) Heat Conduction
b 2k
Therefore, the temperature distribution is
qG 2 T2  T1 q
T ( x)   x  x  G x  T1 . . . (4.58)
2k b 2k
It may be seen that Eq. (4.49) is now modified by two terms containing the heat
generation and that the temperature distribution is no longer linear. If the two surface
temperatures are equal (T1 = T2), then the temperature distribution becomes
2
qG 2  x  x  
T ( x)   b       T1 . . . (4.59)
2k  b  b  
Which is a parabolic and symmetric about the central plane with a minimum Tmax
b
at x  .
2
dT q  1 2x 
  G b2   2   0 . . . (4.60)
dx 2k b b 
2x 1
or 2
 . . . (4.61)
b b
b
or x . . . (4.62)
2
q Gb2
and TMAX  T1  . . . (4.63)
8k
In dimensionless form, on dividing Eq. (4.59) by Eq. (4.63)
x
70  . . . (4.64)
b
T ( x )  T1
 4 (   2 ) . . . (4.65)
Tmax  T1
Let us consider the case where heat is transferred from the two sides of the wall to the
surrounding fluid at T (Figures 4.6-4.7). For simplicity, let us assume that both the wall
surfaces are at Tw (Figure 4.7). At steady state and for one-dimensional heat flow,

qG = heat generated per

unit volume
b/2
∞

T1
T2
x

b
Figure 4.6 : Heat Flow through a Wall with Heat Generation

13
Conduction dT 2 qG
2
 0 . . . (4.66)
dx k
Let the excess temperature be
  T  T . . . (4.68)

d 2 d 2T
So that  . . . (4.69)
dx 2 dx 2
∞

h
k
k T
T∞
T T
θo θ
T∞ T
o 
b/2 b/2
h

∞
Figure 4.7 : Heat Transfer from Two Sides of a Wall having a Heat Source

d 2 qG
Therefore, 2
 . . . (4.70)
dx k
d q
  G x  C1 . . . (4.71)
dx k
qG 2
and  x  C1 x  C2 . . . (4.72)
2k
Which is parabolic. The central plane is the plane of symmetry where the solid
d
temperature is the maximum and  0 and is taken as the reference plane with x = 0.
dx
d
At x = 0,  0 , C1 = 0 . . . (4.73)
dx
b  d  qG b
At x  ,  b  
2  dx  k 2
2

Again, at wall
   q b q b
qk   k    k G  G  h (Tw  T  )  h w . . . (4.74)
 x  b k 2 2
2

qG b
w  . . . (4.75)
2h
From Eq. (4.72)
qG 2
 x  C2 . . . (4.76)
2k
b
When x  ,   w
2
14
 Governing Equations of
qG b 2
qw    C2 . . . (4.77) Heat Conduction
2k 4
qG b 2 qG b qG 2
C2   w    b . . . (4.78)
2k 4 2h 8k
Therefore, the final temperature distribution is
qG 2 qG b qG
 x    T . . . (4.79)
2k 2h 8k
qG 2 q b
T  (b  4 x 2 )  G  T . . . (4.80)
8k 2h
This is the temperature distribution. At the mid-plane, x = 0 and T = Tmax.
qG 2 qG b
T  b   T . . . (4.81)
8k 2h
SAQ 1
(a) Show that the temperature profile for heat conduction through a plane wall
of constant thermal conductivity is a straight line.
(b) Show that the temperature profile for heat conduction through a plane wall
with a heat source and constant thermal conductivity is parabolic.
4.3.2 Cylindrical Wall
Cylinder without Heat Generation
As shown in the Figure 4.8, heat is assumed to flow only radially in a hollow pipe
consisting fluid inside the heat is transferred through a hollow pipe. We want to
determine the temperature distribution and the heat transfer rate in a long hollow
cylinder of length L if the inside and outside surface temperatures are Ti and To
respectively and there is no internal heat generation.
qk
To
T1
T
To
T = T(r)
ro
K = uniform
r
ro qG = 0 r1
dr
r1
T1

(a) (b)
Figure 4.8 : Radial Heat Conduction through a Hollow Cylinder

Since the temperatures at two surfaces are constant, the temperature distribution in
wall is not a function of time, and the conduction equation is given by Eq. (4.44)

d  dT 
r 0 . . . (4.82)
dr  dr 

On integration
dT
r  C1 . . . (4.83)
dr

15
Conduction dT C1
or  . . . (4.84)
dr r
A second integration yields
T  C1 ln r  C2 . . . (4.85)

At r  ri , T  Ti

Ti  C1 ln ri  C2 . . . (4.86)

At r  r0 , T  T0

T0  C1 ln r0  C2  C1 ln r0  Ti  C i ln ri . . . (4.87)

T0  Ti
C1  . . . (4.88)
r 
ln  0 
 ri 
T0  Ti
and C2  Ti  ln ri . . . (4.89)
 r0 
ln  
 ri 
Substituting C1 and C2 in Eq. (4.87),
T0  Ti T  Ti
T  ln r  Ti  0 ln ri . . . (4.90)
 r0   r0 
ln   ln  
 ri   ri 

r
ln  
T (r )  Ti
  ri  . . . (4.91)
T0  Ti r 
ln  0 
 ri 
The rate of heat transfer by conduction
dT C T  Ti
Qk   kA   k 2 r L 1   k 2 L 0 . . . (4.92)
dr r r 
ln  0 
 ri 
2 k L (Ti  T0 )
Qk  . . . (4.93)
r 
ln  0 
 ri 
Then the thermal resistance offered by the wall is

r 
ln  0 
T  T0 r
Rth  i   i . . . (4.94)
Qk 2 k L

As shown in Figure 4.9, the rate of heat conduction through a composite

cylindrical wall with convection at the inside and outside surfaces is given by
T Th  Tc
Q  . . . (4.95)
 Rth  r2   r3 
ln   ln  
1 r r 1
  1  2
hi 2 r1 L 2 k1 L 2 k 2 L h0 2 k3 L
16
 where Th and Tc are the hot and cold fluid temperatures, hi and ho are the inside Governing Equations of
Heat Conduction
and outside heat transfer coefficients and k1 and k2 are the thermal conductivities of
the two walls in series.
Tc, ∞
T3
T2

T1 B
AL
Th, ∞
r1 = n hc,o
r2
r3 = r0
q L
hc,1

Th, ∞
T1 T2
T3
Tc, ∞

Th ∞ T1 T2 T3

1 ln(r2/r1) ln(r3/r2) 1
hc1 2r1L 2kAL 2kBL hc,02r0L

Figure 4.9 : Temperature Distribution in a Composite Cylinder

with Convection at the Interior and Exterior Surfaces
Now Q  U 0 A0 (Th  Tc ) . . . (4.96)
where Uo is the overall heat transfer coefficient based on outside area Ao  2 ro L
given by
1 1
  Rth . . . (4.97)
U 0 A0 r  r 
ln  2  ln  3 
1 r r 1
  1  2
hi Ai 2 k1 L 2 k 2 L h0 A0
where ri = r1 and ro = r3. To find the wall temperature T1, T2 and T3, we can use
Q
Th  T1  Q R1  . . . (4.98)
hi Ai
Q 2 k1 L
T1  T2  . . . (4.99)
r 
ln  2 
 r1 
Q
T2  T3  . . . (4.100)
ho Ao
Cylinder with Heat Generation
Let us consider the Figure 4.10, a long solid cylinder of radius R with internal heat
generation, such as an electric coil in which heat is generated as a result of electric
current in the wire or a cylindrical nuclear fuel element in which heat is generated
by nuclear fission.
The one-dimensional heat conduction equation in cylindrical coordinates is
1 d  dT  qG
r  0 . . . (4.101)
r dr  dr  k

d  dT  qG r
r  . . . (4.102)
dr  dr  k
On integration, 17
Conduction
dT q r2
r  G  C1 . . . (4.103)
dr 2k
dT q r C
 G  1 . . . (4.104)
dr 2k r
qG

hc
hc

q
L

T T
r
T∞ T∞
O
R
q

Figure 4.10 : Temperature Distribution in a Cylindrical Rod with Heat Generation

By second integration,

qG r 2
T   C1 ln r  C 2 . . . (4.105)
4k
dT
At r  0,  0 (with origin at the centre line of the cylinder). But from
dr
dT
Eq. (4.104),   , which is impossible.
dr
 dT  q R C
r  R,    G  1 . . . (4.106)
 dr  r  R 2k R

Heat generated in the cylindrical rod  qG  R 2 L .

This heat is conducted to the surface and then converted away.
 dT 
qG  R 2 L   k 2 L R   . . . (4.107)
 dr  r  R

 dT  q R
   G . . . (4.108)
 dr  r  R 2k

From Eqs. (4.104) and (4.108), C1 = 0

qG 2
T  r  C2 . . . (4.109)
4k
At r  R , T  Tw
qG 2
C2  Tw  R . . . (4.110)
4k
Substituting in Eq. (4.105)
qG 2 q
T  r  Tw  G R 2 . . . (4.111)
4k 4k
18
 Governing Equations of
q q R2  r2 
or T (r )  G ( R 2  r 2 )  T w  G 1  2   TW . . . (4.112) Heat Conduction
4k 4k  R 

This is the maximum temperature variation along the wall radius. The maximum
temperature occurs at r = 0
qG R 2
Tmax   Tw . . . (4.113)
4k
In dimensionless form Eq. (4.112) becomes
2
T (r )  Tw r
1   . . . (4.114)
Tmax  Tw R 
For a hollow cylinder with uniformly distributed heat source and specified surface
temperature T = Ti at r = ri and T = T0 at r = r0, Eq. (4.105) gives
qG 2
Ti   ri  C1 ln ri  C 2 . . . (4.115)
4k
qG 2
To   ro  C1 ln ro  C 2 . . . (4.116)
4k
Evaluating C1 and C2 from the above two equations and substituting in
Eq. (4.105).
We can obtain the temperature distribution as
r
ln  
q
T (r )  G (ro2  r 2 )   ro   qG (r 2  r 2 )  T  T   T . . . (4.117)
 ro   4k
o i o i o
4k 
ln  
 ri 
If a solid cylinder is immersed in a fluid at T and the convection heat transfer
coefficient is hc and T = Tw at r = R, the heat conduction from the cylindrical is
equal to the rate of convection at the surface, or
 dT 
k   hc (Tw  T ) . . . (4.118)
 dr  r  R

From Eq. (4.108)

 dT  h (T  T ) q R
   c w  G . . . (4.119)
 dr  r  R k 2k

qG R
Tw  T  . . . (4.120)
2hc
From Eq. (4.112)

qG R 2  r2  qG R
T (r )  1  2   T  . . . (4.121)
4k  R  2 hc

In dimensionless form,

T (r )  T q R  h R   r  2 
 G 2  c 1     . . . (4.122)
T 4hc T  k   R  

and maximum temperature is

19
Conduction Tmax q R  h R
1 G  2 c  . . . (4.123)
T 4hc T  k 

There are two dimensionless parameters in the above equation which are important
q R h R
in conduction, viz. G the heat generation number, and c the Biot number,
hc T k
which appears in problems with simultaneous conduction and convection.
The Biot number is the ratio of conduction resistance to convection resistance or
ro
Rk h r
Bi   k  c o . . . (4.124)
Rc 1 k
hc

The limits of Biot numbers are

ro 1
Bi  0 when Rk   0 or, Rc  
k hc

1 r
Bi   when Rc   0 or, Rk  o  
hc k

The Biot number approaches zero when the conductivity of solids very large
(k  ) or the convection coefficient of heat transfer is very low (hc  0), i.e.
when the solid is practically isothermal and the temperature change is mostly
caused in the fluid by convection at the interface. On the contrary, the Biot number
approaches infinity when the thermal resistance predominates (k  0) or the
convection resistance is very low (hc  0).
SAQ 2
(a) Show that the maximum temperature in a cylindrical rod with heat
 kW  T q R  h R
generation qG  3  is given by max  1  G  2  c .
m  T 4hc T  k 

(b) What are the limiting values of Biot number?

(c) What is heat generation number?
4.3.3 Spherical Wall
As shown in Figure 4.11 a hollow sphere with uniform temperature at the inner and outer
surfaces, the temperature distribution without heat generation in the steady state can be
obtained by simplifying Eq. (4.46). Under these conditions the temperature is only a
function of the radius r, and the conduction equation in the spherical coordinates is
1 d  2 dT 
r 0 . . . (4.125)
r 2 dr  dr 
On integration, and putting at r = ri, T = Ti and r = ro, T = To
ro  ri 
T (r )  Ti  (To  Ti ) 1  r  . . . (4.126)
ro  ri  

20
 Governing Equations of
Heat Conduction
r0 T0

T = T(r)
T1 K = uniform
qG = 0
r1

qk

Figure 4.11 : Heat Conduction in a Hollow Sphere without Heat Generation

The rate of heat transfer through the spherical wall is

dT T  To
Qk   k 4 r 2  i . . . (4.127)
dr ( ro  ri )
4 k ro ri
The thermal resistance for spherical wall is then
ro  ri
Rth  . . . (4.128)
4 k ro ri

4.4 TRANSIENT CONDUCTION

Many heat transfer problems are time dependent. Such unsteady, or transient, problems
typically arise when the boundary conditions of a system are changed. For example, if the
surface temperature of a system is altered, the temperature at each point in the system
will also begin to change. The changes will continue to occur until a steady-state
temperature distribution is reached. Consider a hot metal billet that is removed from a
furnace and exposed to a cool air stream. Energy is transferred by convection and
radiation from its surface to the surroundings. Energy transfer by conduction also occurs
from the interior of the metal to the surface, and the temperature at each point in the billet
decreases until a steady-state condition is reached. Such time dependent effects occur in
many industrial heating and cooling processes. In the following sub-units, transient heat
transfer analysis has been discussed.
4.4.1 Lumped Capacitance Method
A simple, yet common, transient conduction problem is one for which a solid experiences
a sudden change in its thermal environment. Consider a hot metal forging that is initially
at a uniform temperature Ti and is quenched by immersing it in a liquid of lower
temperature T < Ti as shown in Figure 4.12. If the quenching is said to begin at time t =
0, the temperature of the solid will decrease for time t > 0, until it eventually reaches T.
This reduction is due to convection heat transfer at the solid-liquid interface. The essence
of the lumped capacitance method is the assumption that the temperature of the solid is
spatially uniform at any instant during the transient process. This assumption implies that
temperature gradients within the solid are negligible.
From Fouriers law, heat conduction in the absence of a temperature gradient implies the
existence of infinite thermal conductivity. Such a condition is clearly impossible.
However, although the condition is never satisfied exactly, it is closely approximated if
the resistance to conduction within the solid is small compared with the resistance to heat
transfer between the solid and its surroundings. For now we assume that this is, in fact,
the case.
In neglecting temperature gradients within the solid, we can no longer consider the
problem from within the framework of the heat equation. Instead, the transient 21
Conduction temperature response is determined by formulating an overall energy balance on the solid.
This balance must relate the rate of heat loss at the surface to the rate of change of the
internal energy. Thus, with respect to the Figure 4.12,
 Eout  Est . . . (4.134)

dT
or  h As (T  T )   Vc . . . (4.135)
dt

Ti t<0
T = Ti

Liquid

Eout = qconvenction

T(t) Esi

T∞  Y1 t  0
T = T(t)

Figure 4.12 : Cooling of a Hot Metal Forging

Introducing the temperature difference

  T  T . . . (4.136)

 d    dT 
And recognizing that      , it follows that
 dt   dt 

 Vc d 
 . . . (4.137)
h As dt

Separating variables and integrating from the initial condition, for which t = 0 and
T (0) = Ti, we then obtain
 Vc  d t

h As  i 
  0 dt . . . (4.138)

where i  Ti  T . . . (4.139)

Evaluating the integrals it follows that

 Vc 
ln  i  t . . . (4.140)
h As 

 T  T  hA  
or   exp   s  t  . . . (4.141)
i Ti  T    Vc  
Eq. (4.140) may be used to determine the time required for the solid to reach some
temperature T, or, conversely, Eq. (4.141) may be used to compute the temperature
reached by the solid at same time t.
The foregoing results indicate that the difference between the solid and fluid
temperatures must decay exponentially to zero as t approaches infinity. This behaviour is
 Vc 
shown in Figure 4.13. From Eq. (4.141) it is also evident that the quantity   may
 h As 
be interpreted as a thermal time constant s. This time constant may be expressed as

22
 Governing Equations of
 1  Heat Conduction
t    ( Vc)  Rt Ct . . . (4.142)
 h As 
1
Vc
t = = Rt Ct
hAs

 T  T

1 T1  T

0.368

0 t. 1 t. 2 t. 3 t. 4

t
Figure 4.13 : Transient Temperature Response of Lumped Capacitance Solids
for Different Thermal Time Constants s

where Rt is the resistance to convection heat transfer and Ct is the lumped thermal
capacitance of the solid. Any increase in Rt or Ct will cause a solid to respond more
slowly to changes in its thermal environment and will increase the time required to reach
thermal equilibrium ( = 0). This behavior is analogous to the voltage decay that occurs
when a capacitor is discharged through a resistor in an electrical RC circuit.
To determine the total energy transfer Q occurring up to some time t, we simply write
t t
Q  0 q dt  h As  0  dt . . . (4.143)

Substituting for  from Eq. (4.140) and integrating, we obtain

  t 
Q  (Vc ) i 1  exp    . . . (4.144)
  t 
The quantity Q is, of course, related to the change in the internal energy of the solid, and
from Eq. (4.134)
 Q   E st . . . (4.145)
For quenching Q is positive and the solid experiences a decrease in energy. Eqs. (4.140),
(4.141) and (4.144) also apply to situations where the solid is heated ( < 0), in which
case Q is negative and the internal energy of the solid increases.
Validity of the Lumped Capacitance Method
From the foregoing results it is easy to see why there is a strong preference for
using the lumped capacitance method. It is certainly the simplest and most
convenient method that can be used to solve transient conduction problems. Hence,
it is important to determine under what conditions it may be used with reasonable
accuracy.
As shown in Figure 4.14, to develop a suitable criterion consider steady-state
conduction through the plane wall of area A.
Although we are assuming steady-state condition, this criterion is readily extended
to transient processes. One surface is maintained at a temperature Ts, 1 and the other
surface is exposed to a fluid of temperature T < Ts, 1. The temperature of this
surface will be some intermediate value, Ts, 2, for which T < Ts, 2 < Ts, 1. Hence,
under steady-state conditions the surface energy balance is 23
Conduction kA
(Ts ,1  Ts , 2 )  hA (Ts , 2  T ) . . . (4.146)
L
where k is the thermal conductivity of the solid. Rearranging, we then obtain
 L 
Ts ,1  Ts , 2   R hL
kA
    cond   Bi . . . (4.147)
Ts , 2  T  1  Rconv k
 
 hA 

qcond qconv

Bi << 1
Ts. 1 Ts. 2

Bi = 1

Ts. 2

Bi >> 1 Ts. 2

T∞ , h

L
X

Figure 4.14 : Effect of Biot Number on Steady-State Temperature Distribution

in a Plane Wall with Surface Convection

 hL 
The quantity   is a dimensionless parameter appearing in Eq. (4.147). It is
 k 
termed the Biot number, and it plays a fundamental role in conduction problems
that involve surface convection effects. According to Eq. (4.147) and as illustrated
in Figure 4.14, the Biot number provides a measure of the temperature drop in the
solid relative to the temperature difference between the surface and the fluid. Note
especially the conditions corresponding to Bi < < 1. The results suggest that, for
these conditions, it is reasonable to assume a uniform temperature distribution
across a solid at any time during a transient process. This result may also be
associated with interpretation of the Biot number as a ratio of thermal resistances,
Eq. (4.147). If Bi < < 1, the resistance to conduction within the solid is much less
than the resistance to convection across the fluid boundary layer. Hence, the
assumption of a uniform temperature distribution is reasonable.
We have introduced the Biot number because of its significance to transient
conduction problems. Consider the plane wall of Figure 4.15, which is initially at a
uniform temperature Ti and experiences convection cooling when it is immersed in
a fluid of T < Ti. The problem may be treated as one-dimensional in x, and we are
interested in the temperature variation with position and time, T (x, t). This
variation is a strong function of the Biot number, and three conditions are shown in
Figure 4.15.
For Bi < < 1 the temperature gradient in the solid is small and T (x, t)  T (t).
Virtually all the temperature difference is between the solid and the fluid, and the
solid temperature remains nearly uniform as it decreases to T. For moderate to
large values of the Biot number, however, the temperature gradients within the
solid are significant. Hence, T = T (x, t). Note that for Bi > > 1, the temperature
difference across the solid is much larger than that between the surface and the
fluid.
24
 Lumped capacitance method is a preferred method for solving transient conduction Governing Equations of
Heat Conduction
problems. Hence, when confronted with such a problem, the very first thing that
one should do is calculate the Biot number. If the following condition is satisfied
h Lc
Bi   0.1 . . . (4.148)
k

T∞, h
T(x, O) = T1 T(x, O) = T1

T∞ T∞
T∞, h T∞ T∞
-L L -L L -L L
-L L
x
Bi << 1 Bi = 1 Bi >> 1
T = T(t) T = T(x, t) T = T(x, t)

Figure 4.15 : Transient Temperature Distributions for different Biot Numbers

in a Plane Wall Symmetrically Cooled by Convection

The error associated with using the lumped capacitance method is small. For
convenience, it is customary to define the characteristic length of Eq. (4.148) as
V
the ratio of the solids volume to surface area, Lc  . Such a definition
As
facilitates calculation of Lc for solids of complicated shape and reduces to the
r
half-thickness L for a plane wall of thickness 2 L as shown in Figure 4.15, to 0
2
r
for a long cylinder, and to 0 for a sphere. However, if one wishes to implement
3
the criterion in a conservative fashion, Lc should be associated with the length
scale corresponding to the maximum spatial temperature difference. Accordingly,
for a symmetrically heated (or cooled) plane wall of thickness 2 L, Lc would
remain equal to the half-thickness L. However, for a long cylinder or sphere, Lc
r r
would equal the actual radius r0, rather than 0 or 0 .
2 3

V
Finally, we note that, with Lc  , the exponent of Eq. (4.141) may be expressed
As
as
h As t ht h Lc k t h Lc t
  2
 . . . (4.149)
Vc c Lc k c Lc k L2c
h As t
or  Bi . Fo . . . (4.150)
Vc
t
where Fo  . . . (4.151)
L2c
is termed the Fourier number. It is a dimensionless time, which, with the Biot
number, characterizes transient conduction problems. Substituting Eq. (4.150) into
Eq. (4.141), we obtain
25
Conduction  T  T
  exp ( Bi . Fo) . . . (4.152)
i Ti  T
SAQ 3
(a) What do you mean by Biot number and Fourier number?
(b) What are the conditions for validity of lumped capacitance method?

4.4.2 General Lumped Capacitance Analysis

Although transient conduction in a solid is commonly initiated by convection heat
transfer to or from an adjoining fluid, other processes may induce transient thermal
conditions within the solid. For example, a solid may be separated from large
surroundings by a gas or vacuum. If the temperatures of the solid and surroundings differ,
radiation exchange could cause the internal thermal energy, and hence the temperature, of
the solid to change. Temperature changes could also be induced by applying a heat flux
at a portion, or all, of the surface and/or by initiating thermal energy generation within
the solid. Surface heating could, for example, be applied by attaching a film or sheet
electrical heater to the surface, while thermal energy could be generated by passing an
electrical current through the solid.
Figure 4.16 depicts a situation for which thermal conditions within a solid may be
influenced simultaneously by convection, radiation, an applied surface heat flux, and
internal energy generation.
Surroundings
T sur

P, c, V, T(0) = Ti

q”rad

Eg, Est T, h

q”s

q”conv

As, h As(c,r)
Figure 4.16 : Control Surface for Generalized Lumped Capacitance Analysis

It is presumed that, initially (t = 0), the temperature of the solid (Ti) differs from that of
the fluid, T, and the surroundings, Tsur, and that both surface and volumetric heating
(qs and q) are initiated. The imposed heat flux qs and the convection-radiation heat
transfer occur at mutually exclusive portions of the surface, As (h) and As (c, r), respective
and convection-radiation transfer is presumed to be from the surface. Moreover, although
convection and radiation have been prescribed for the same surface, the surfaces may, in
fact, differ (As, c  As, r). Applying conservation of energy at any instant t
dT
qs As , h  E g  (qconv
  qrad
 ) As (c , r )  Vc . . . (4.153)
dt
dT
or, qs As , h  E g  h (T  T )    (T 4  Tsur
4 
) As (c , r )  Vc . . . (4.154)
  dt
Eq. (4.154) is a nonlinear, first-order, nonhomogenous, ordinary differential equation that
cannot be integrated to obtain an exact solution. However, exact solutions may be
obtained for simplified versions of the equation. For example, if there is no imposed heat
flux or generation and convection is either nonexistent (a vacuum) or negligible relative
26 to radiation. Eq. (4.154) reduces to
 dT Governing Equations of
Vc    As , r  (T 4  Tsur
4
) . . . (4.155) Heat Conduction
dt
Separating variables and integrating from the initial condition to any time t, it follows
that
 As , r  t T dT
 Vc 0 dt  T
i
4
Tsur  T4
. . . (4.156)

Evaluating both integrals and rearranging, the time required to reach the temperature T
becomes
Vc  Tsur  T T  Ti
t 3 ln  ln sur
4 As , r  Tsur  Tsur  T Tsur  Ti

  T   1  Ti  

 2  tan 1    tan    . . . (4.157)
  Tsur   Tsur  
This expression cannot be used to evaluate T explicitly in terms of t, Ti, and Tsur nor does
it readily reduce to the limiting result for Tsur = 0 (radiation to deep space). However,
returning to Eq. (4.156), its solution for Tsur = 0 yields
Vc  1 1 
t  3  3  . . . (4.158)
3 As , r   T Ti 
An exact solution to Eq. (4.154) may also be obtained if radiation may be neglected and h
d  dT
is independent of time. Introducing a reduced temperature   T – T, where  ,
dt dt
Eq. (4.154) reduces to a linear, first-order, non-homogenous differential equation of the
form
d
 a  b  0 . . . (4.159)
dt
 h As , c   qs As , c  E g  
where a    and b     . Although Eq. (4.159) may be solved by
 Vc   Vc  
summing its homogeneous and particular solutions, an alternative approach is to
eliminate the non-homogeneity by introducing the transformation
b
    . . . (4.160)
a
d  d 
Recognising that  , Eq. (4.160) may be substituted into Eq. (4.159) to yield
dt dt
d 
 a   0 . . . (4.161)
dt
Separating variables and integrating from 0 to t (i to ) , it follows that

 exp ( at ) . . . (4.162)
i
or substituting for  and ,
b
T  T   
 a   exp ( at ) . . . (4.163)
b
Ti  T   
a

27
Conduction b
T  T a
Hence  exp ( at )  1  exp ( at )  . . . (4.164)
Ti  T Ti  T
As it must, Eq. (4.164) reduces to Eq. (4.141) when b = 0 and yields T = Ti at t = 0. As
b
t  , Eq. (4.164) reduces to (T  T )    , which could also be obtained by
a
performing an energy balance on the control surface of Figure 4.16 for steady-state
conditions.

4.5 THE SEMI-INFINITE SOLID

A semi-infinite solid is characterized by a single identifiable surface as shown in
Figure 4.17. In principle, such a solid extends to infinity in all but one direction. If a
sudden change of conditions is imposed at this surface, transient, one-dimensional
conduction will occur within the solid. The semi-infinite solid provides a useful
idealization for many practical problems. It may be used to determine transient heat
transfer near the surface of the earth or to approximate the transient response of a finite
solid, such as a thick slab. For this second situation the approximation would be
reasonable for the early portion of the transient, during which temperatures in the slab
interior (well removed from the surface) are uninfluenced by the change in surface
conditions.
Case (1) Case (2) Case (3)
T(x,o) = Ti T(x,o) = Ti T(x,o) = Ti
T(o,t) = Ts -k T/xx=o = q”o -k/xx=o = h(T∞ 0 – T(O,i)

T∞, h
Ts q”
o

x x x
x x x

T (x, t)
Ts T∞

t t t
t t

Ti Ti Ti
x x x
Figure 4.17 : Transient Temperature Distributions in a Semi-Infinite Solid for
Three Surface Conditions (a) Constant Surface Temperature, (b) Constant Surface Heat Flux,
and (c) Surface Convection

The interior boundary condition is of the form

T ( x  , t )  Ti . . . (4.165)

Closed-form solutions have been obtained for three important surface conditions,
instantaneously applied at t = 0. These conditions are shown in Figure 4.17. They include
application of a constant surface temperature application of a constant surface heat flux
q0 and exposure of the surface to a fluid characterized by T  Ti and the convection
coefficient h.
The solution for Case 1 may be obtained by recognizing the existence of a similarity
variable , through which the heat equation may be transformed from a partial
differential equation, involving two independent variables (x and t), to an ordinary
differential equation expressed in terms of the single similarity variable. To confirm that

28
 x Governing Equations of
such a requirement is satisfied by   1
, we first transform the pertinent Heat Conduction

(4 t ) 2
differential operators, such that
T dT  1 dT
  1 2
. . . (4.166)
x d  x  4 t  d 

 2T d  T   1 d 2T
  . . . (4.167)
x 2 d   x  x 4 t d 2

T dT  x dT
  1
. . . (4.168)
t d  t d
2t (4 t ) 2

With assumption of no internal heat generation and constant thermal conductivity the
heat equation becomes

d 2T dT
2
  2 . . . (4.169)
d d

With x = 0 corresponding to  = 0, the surface condition may be expressed as

T (  0)  Ts . . . (4.170)

And with x  , as well as t = 0, corresponding to   , both the initial condition and

the interior boundary condition correspond to the single requirement that
T (n  )  Ti . . . (4.171)

Since the transformed heat equation and the initial/boundary conditions are independent
x
of x and t,   1
is indeed, a similarity variable. Its existence implies that,
(4 t ) 2
irrespective of the values of x and t, the temperature may be represented as a unique
function of .
The specific form of the temperature dependence, T (), may be obtained by separating
variables in Eq. (4.169), such that

 dt 
d 
 d     2 d  . . . (4.172)
 dT 
 
 d 
Integrating, it follows that

 dT  2
ln       C1 . . . (4.173)
 d 
dT
or  C1 exp (  2 ) . . . (4.174)
d
Integrating a second time, we obtain

T  C1 0 exp ( u 2 ) du  C2 . . . (4.175)

Where u is a dummy variable. Applying the boundary condition at  = 0, Eq. (4.170), it

follows that C2 = T5 and 29
Conduction 
T  C1 0 exp ( u 2 ) du  T s . . . (4.176)

From the second boundary condition, Eq. (4.171), we obtain


Ti  C1 0 exp ( u 2 ) du  Ts . . . (4.177)

or, evaluating the definite integral,

2 (Ti  Ts )
C1  1
. . . (4.178)
2
Hence, the temperature distribution may be expressed as

 
T  Ts  2  

T  Ts
 1
 2


0 exp ( u 2 ) du  erf  . . . (4.179)
 
Where the Gaussian error function, erf , is a standard mathematical function that is
tabulated in Appendix A. The surface heat flux may be obtained by applying Fourier’s
law at x = 0, in which case
T d (erf  ) 
qs   k   k (Ti  Ts ) . . . (4.180)
x x0 d x 0

  1
 2  2

qs  k (Ts  Ti )  1  exp (   ) (4 t ) 2 . . . (4.181)
 2 
  0

k (Ts  Ti )
qs  1
. . . (4.182)
(  t) 2

Analytical solutions may also be obtained for the Case 2 and Case 3 surface conditions,
and results for all three cases are summarized as follows.
Case 1 : Constant Surface Temperature : T (0, t) = Ts
T ( x, t )  Ts x
 erf . . . (4.183)
Ti  Ts 2 t

k (Ts  Ti )
qs (t )  . . . (4.184)
 t

Case 2 : Constant Surface Heat Flux : qs ( t ) = q0

1
 t  2
2q0  
   exp   x   q0 x erf c  x 
2
T ( x, t )  Ti      . . . (4.185)
k  2 t 
 4 t  k  

Case 3 : Surface Convection

T
k  h [T  T (0, t )]
x x0

30
 Governing Equations of
T ( x, t )  Ti  x    h x h 2 t 
 erf c    exp  Heat Conduction
 
T  Ti  2 t  
    k k 2 

  x h t 
 erf c    . . . (4.186)
 k 
  2 t 

The complementary error function, erf c w, is defined as erf c w  1 – erf c w.

Temperature histories for the three cases are shown in Figure 4.17, and distinguishing
features should be noted. With a step change in the surface temperature, Case 1,
temperatures within the medium monotonically approach Ts with increasing t, while the
magnitude of the surface temperature gradient, and hence the surface heat flux, decreases
1

as t 2 . In contrast, for a fixed surface heat flux (Case 2), Eq. (4.185) reveals that
1
T (0, t )  Ts (t ) increases monotonically as . For surface convection (Case 3), the
t2
surface temperature and temperatures within the medium approach the fluid temperature
T with increasing time. As Ts approaches T there is, of course, a reduction in the
surface heat flux, qs (t )  h [T  Ts (t )] . Specific temperature histories computed from
Eq. (4.186) are plotted in Figure 4.18. The result corresponding to h =  is equivalent to
that associated with a sudden change in surface temperature, Case 1. That is, for h = ,
the surface instantaneously achieves the imposed fluid temperature (Ts = T), and with
the second term on the right-hand side of Eq. (4.185) reducing to zero, the result is
equivalent to Eq. (4.183).

1.0

0.5
T∞ T(x,t)
h
∞
3
2 x
1
T ∞ Ti
T – Ti

0.5
0.4
0.1
0.3
0.2
0.005
0.1

0.05
H t = 0.05 5
k
0.001
0 0.5 1.0 1.5
x
2t 1

Figure 4.18 : Temperature Histories in a Semi-Infinite Solid with Surface Convection

An interesting permutation of Case 1 results when two semi-infinite solids, initially at
uniform temperatures TA, i and TB, i, are placed in contact at their free surfaces
(Figure 4.19).
If the contact resistance is negligible, the requirement of thermal equilibrium dictates that,
at the instant of contact (t = 0), both surfaces must assume the same temperature Ts, for
which TB, i < Ts < TA, i. Science Ts does not change with increasing time, it follows that
the transient thermal response and the surface heat flux of each of the solids are
determined by Eqs. (4.179) and (4.182), respectively.
The equilibrium surface temperature of Figure 4.19 may be determined from a surface
energy balance, which requires that
31
Conduction qs, A  qs, B . . . (4.187)

Substituting from Eq. (4.182) for qs, A and qs, B and recognising that the x coordinate of
Figure 4.19 requires a sign change for qs, A , it follows that

 k A (Ts  TA , i ) k B (Ts  TB , i )
1
 1
. . . (4.188)
  A t  2   B t  2

or, solving for Ts,

( k  c)1/A 2 TA, i  ( k  c)1/ 2

B TB , i
Ts  . . . (4.189)
( k  c)1A2  ( k  c)1B2

B
TA, i kB, pB, cB

q”s,B
t

Ts
t
q”s,A
A

kA, pA, cA TB, i

Figure 4.19 : Interfacial Contact between Two Semi-Infinite Solids at Different Initial Temperatures

1
Hence the quantity m  is a weighing factor that determines whether Ts will
(k  c ) 2
more closely approach TA, i  (m A  mB ) or TB , i  (mB  m A ) .

SAQ 4
(a) What do you mean by a semi-infinite solid? What is its speciality?
(b) What are the three surface conditions in solution of a transient heat
conduction problem for a semi-infinite solid?

4.6 MULTIDIMENSIONAL EFFECTS

Consider immersing the short cylinder of Figure 4.20, which is initially at a uniform
temperature Ti, in a fluid of temperature T  Ti. Because the length and diameter are
comparable, the subsequent transfer of energy by conduction will be significant for both
the r and x coordinate directions. The temperature within the cylinder will therefore
depend on r, x, and t.
Assuming constant properties and no generation, the appropriate form of the heat
equation is
1   T   2T 1 T
r   . . . (4.190)
r r  r  x 2  t

32
 where x has been used in place of z to designate the axial coordinate. A closed-form Governing Equations of
Heat Conduction
solution to this equation may be obtained by the separation of variables method.
Although we will not consider the details of this solution, it is important to note that the
end result may be expressed in the following form :

T (r , x, t )  T T ( x, t )  T T (r , t )  T
 . . . . (4.191)
Ti  T Ti  T Plane Ti  T Infinite
wall cylinder

T∞, h ro

r ro
+L T(r, x, t)
(r, x)
x
x
T∞, h L x L
T∞, h r
= x
L L
Midplane

-L
θ(r, x.t) θ(r, t) θ(x, t)
T∞, h = ×
θi θi θi
r
* * * * *
θ =c(r ,t )×P(x ,t )
Figure 4.20 : Two-Dimensional Transient Conduction in a Short Cylinder,
(a) Geometry, (b) Form of the Product Solution
That is, the two-dimensional solution may be expressed as a product of one-dimensional
solutions that correspond to those for a plane wall of thickness 2 L and an infinite
cylinder of radius r0. For F0 > 0.2, these solutions are provided by Figures D.1 and D.2
for the plane wall and Figures D.3 and D.4 for the infinite cylinder.
Results for others multidimensional geometries are summarized in Figure 4.21. In each
case the multidimensional solution is prescribed in terms of a product involving one or
more of the following one-dimensional solutions :
T ( x, t )  T
S ( x, t )  . . . (4.192)
Ti  T Semi-infinite
solid

T ( x, t )  T
P ( x, t )  . . . (4.193)
Ti  T Plane
wall

T (r , t )  T
C (r , t )  . . . (4.194)
Ti  T Infinite
cylinder

The x coordinate for the semi-infinite solid is measured from the surface, whereas for the
plane wall it is measured from the mid plane. In using Figure 4.21 the coordinate origins
should carefully be noted. The transient, three-dimensional temperature distribution in a
rectangular parallelepiped, Figure 4.21(h), is then, for example, the product of three
one-dimensional solutions for plane walls of thicknesses 2L1 and 2L3. That is
T ( x1 , x2 , x3 , t )  T
 P ( x1 , t ) . P ( x2 , t ) . P ( x3 , t ) . . . (4.195)
Ti  T
The distances x1, x2 and x3 are all measured with respect to a rectangular coordinate
system whose origin is at the center of the parallelepiped.

33
Conduction S(x,t) P(x,t) C(r,t)

x r
x

2L1 ro
(a) Semi-infinite Solid (b) Plane Wall (c) Infinite Cylinder

S(x1,t)P(x2,t) P(x1,t)P(x2,t) C(x1,t)S(x,t)

r
x2
x2
x1 x
x1

2L1

2L2 ro
2L2

(d) Semi-infinite Plate (e) Infinite Rectangular Bar (f) Semi-infinite Cylinder

S(x3,t)P(x1,t)P(x2,t) S(x1,t)P(x2,t)P(x3,t)
C(r,1)P(x,t)

x3

2L1 2Lt x
x2 r
x1
x3

x1
x2 2L1 2L2
2L2
2L2 r
(g) Semi-infinite Rectangular Bar (h) Rectangular Parallelepiped (i) Short Cylinder

Figure 4.21 : Solutions for Multi-Dimensional Systems Expressed

as products of One Dimensional Results

The amount of energy Q transferred to or from a solid during a multi-dimensional

transient conduction process may also be determined by combining one-dimensional
results.
Example 4.1
The air inside a chamber at T, i = 50oC is heated convectively with
hi = 20 W/m2 K by a 200-mm thick wall having thermal conductivity of 4 W/m K
and a uniform heat generation of 1000 W/m3. To prevent any heat, generated
within the wall, from being lost to the outside of the chamber at T, 0 = 25oC with
h0 = 5 W/m2 K, a very thin electrical strip heater is placed on the outer wall to
provide a uniform heat flux q0 .

34
 (a) Sketch the temperature distribution in the wall on T – x coordinates for the Governing Equations of
Heat Conduction
condition where no heat generated within the wall is lost to the outside
chamber. Identity T, i and T, 0 on the plot.
(b) What are the temperatures at the wall boundaries T0 and TL (at x = 0 and
x = L) for the conditions of Part (a).
(c) Determine the value of q0 that must be supplied by the strip heater so that
all heat generated within the wall is transferred to the inside of the chamber
for the temperature T0 as computed in part (b).
Solution
d 2T qG
2
 0
dx k
dT qG
 x  C1 . . . (1)
dx k
qG 2
(a) T  x  C1 x  C 2
2k
(b) At x = 0, T = T0,
dT
0 . . . (2)
dx
From Eq. (1), C1 = 0
qG 2
T  x  C2
2k
At x = L, T = TL
qG 2
TL   L  C2
2k
qG L2
or C2  TL 
2k
The temperature distribution in the wall is thus,
qG 2
T  (L  x 2 )  TL . . . (3)
2k
At x = 0, from Eq. (3),
qG 2
T  T0  Tmax  L  TL
2k
1000  (0.2) 2
  TL . . . (4)
24
Again, at the inside wall surface, from Eq. (1),
 dT  q L
hi (TL  T , i )   k    G k
 dx x  L k

qG L 1000  0.2
TL  T , i    10o C
hi 20

TL  50  10  60 o C C Ans. (a)

From Eq. (4),

35
Conduction
T0  Tmax  5  60  65 o C Ans. (b)

(c) q0  h0 (T0  T, 0 )  5 (65  25)

= 200 W/m2 Ans. (c)

The temperature profile is shown in Figure 4.22 given below :
L

Outside Inside
Chamber chamber
T∞,0,h0
q

Air

T∞, 0 T0 T∞, 0 = 50C

x
h0

H0, = 5W/m2K

h1, = 20W/m2K
L = 0.2,
K = 4W/mk
qo = 1 kW/m3

Figure 4.22

Example 4.2
In a cylindrical fuel rod of a nuclear reactor heat is generated internally,
accordingly according to the equation

  r 2 
qG  q0 1    
  r0  
 

where qG is the local rate of heat generation per unit volume at radius r, ro is the
outside radius, and q0 is the rate of heat generation per unit volume at the centre
line. Calculate the temperature drop from the centre line to the surface for a 2.5 cm
outer diameter rod having k = 25 W/m K, if the rate of heat removal from the
surface is 1650 kW/m2.
Solution
In cylindrical coordinates, the radial variation of temperature in at steady state
when there is heat generation is given by

36
 r0

Governing Equations of
Heat Conduction

qG

0
r
Tro

h 0 h

Figure 2.23

1 d  dT  qG
r 
r dr  dr  k

d  dT  qG r  r2  r
or r    q0  1  
dr  dr  k  r02  k

dT q  r2 r4 
or r  0   2   C1
dr k  2 4r0 

dT q r  r2  C
or k   0 1  2   1 . . . (1)
dr 2  2r0  r

 dT 
At r  0,  k  cannot be defined,
 dr 
 C1 = 0

dT q r  r2 
k   0 1  2 
dr 2  2r0 

Heat transfer from the rod

 dT  q r  1  q0 r0
qF   k    0 0 1   
 dr  r  r0 2  2 4

Given : qF = 1650 kW/m2, r0 = 1.25 cm

q0  1.25  10  2
1650 
4
1650  100  16
q0   528  103 kW/m 3
5
This is the volumetric heat generation at the centre line of the rod. Now,

dT q r  r2 
  0 1  2 
dr 2k  2r0 
37
Conduction  r2
q0 r4 
T    2   C2
2k 
 2 8r0 
At r = 0, T = Tc the centre line temperature.
C2 = Tc

q0  r2 r4 
T    2   Tc
2k 
 2 8r0 
At, r = r0, the temperature drop,

q0 r02  1 1  3 q0 r02
Tc  Tr0    
2k  2 8  16 k

3 528  10 6 (W/m 3 )  (1.25) 2  10 4 m 2

 
16 25 W/mK

= 618.7oC.
Example 4.3
A bar of square cross-section connects two metallic structures. One structure is
maintained at a temperature 200oC and the other is maintained at 50oC. The bar,
20 mm  20 mm, is 100 mm long and is made of mild steel (k = 0.06 kW/m K).
The surroundings are at 20oC and the heat transfer coefficient between the bar and
the surroundings is 0.01 kW/m2K.
Derive an equation for the temperature distribution along the bar and hence
calculation the total heat flow rate from the bar to the surroundings.
Solution
From Figure 2.24 below :
A = 0.2 cm 0.2 cm
T = 20 oC

Q1 Q T2 = 50 oC
T1 = 200o C
2

L = 0.1 m
K = 0.06 kW/m K
H = 0.01 k/Wm2 K

Figure 2.24
dT
Q1   kA
dx
d Q1
Q2  Q1  dx
dx
d  dT  d 2T
Q1  Q2     kA  dx  kA dx
dx  dx  dx 2
 h P dx (T  T )

Letting   T  T ,

d 2 hP
2
 0
dx kA

38
 Governing Equations of
or ( D 2  m2 )   0 Heat Conduction
1
 hP  2
where m 
 kA 
The general solution is

  C1 e mx  C2 e  mx . . . (1)

At x  0,   1  C1  C 2  180 o C . . . (2)

At x  1,    2  C1 e m1  C 2 e  m1  30 o C . . . (3)

1 1 1
 hP  2  0.01  4  0.02  2  4  2
m     
 kA   0.06  0.02  0.02   0.12 

ml  0.577, e m1  1.78 ,

e  m1  0.561  5.77 m 1

30  C1  1.78  C2  0.561

3.173 C1  C2  53.476

From Eq. (2)

C1  C2  180.00

On subtraction,
2.173 C1   126.524

C1   58.22

From Eq. (2),

C2  238.22

The temperature distribution is

   58.22 e5.77 x  238.22 e 5.77 x

 d 
Q1   kA     kA ( 58.22  5.77e 5.77 x  238.22  5.77e 5.77 x ) x 0
 dx  x  0

 0.06  4  10  4 (335.93  1374.53)

= 0.0410 kW = 41.0 W

 d 
Q2   kA     kA ( 58.22  5.77  1.78  238.22  5.77  0.561)
 dx  x 1

 0.06  4  10  4 (598  771.5)

= 0.0328 kW = 32.8 W
Heat flow rate from the bar to the surroundings
 Q1  Q2  41.0  32.8  8.2 W
39
Conduction Example 4.4
A load of peas at a temperature of 25oC is to be cooled down in a room at a
constant air temperature of 1oC.
(a) How long the peas will require to cool down to 2oC when the surface heat
transfer coefficient of the peas is 5.81 W/m2 K?
(b) What is the temperature of the peas after a lapse of 10 min from the start of
cooling?
(c) What air temperature must be used if the peas were to be cooled down to
5oC in 30 min? The peas are supposed to have an average diameter of 8 mm.
Their density is 750 kg/m3 and specific heat 3.35 kJ/kg K.
Solution
We have for a lumped heat-capacity system
hAt
 T  T
  e cV  e  Bi . Fo
i Ti  T

Since the diameter of the peas is only 8 mm, we can neglect any temperature
variation within the pea.
3
d
  4  
V  2   d  750  0.008  1
 2
A d 3 2 3 2
3  4  
2
2 1 ht 5.81 t
(a) ln  
25  1 c 3.35  103

5.81 t
ln 24  3.178 
3.35  103

t  1832.4 s  30.54 min

5.81  600
T 1 
(b) e 3350  0.353
25  1

T  9.48o C
5.81  30  60
5  T 
(c) e 3350  0.044
25  T

1.1  0.044 T  5  T

0.956 T  3.9

T  4.08o C

Example 4.5
A steel tube of length 20 cm with internal and external diameters of 10 and 12 cm
is quenched from 500oC to 30oC in a large reservoir of water at 1ooC. Below 100oC
the heat transfer coefficient is 1.5 kW/m2 K. Above 1000oC it is less owing to a
film of vapour being produced at the surface, and an effective mean value between
500oC and 100oC is 0.5 kW/m2 K. Neglecting internal thermal resistance of the
steel tube, determine the quenching time.
40
Solution Governing Equations of
Heat Conduction
di = 10 cm, d0 = 12 cm, l = 20 cm, T = 10oC
A   (d 0  d i ) l   (10  12)  20  1382 cm 2
 2 
V  (d 0  d i2 ) l  (144  100)  20  691 cm3
4 4
Cooling from 500oC to 100oC
T  T hAt
 ln 
Ti  T cV

100  10 490 500  1382  10  4  t

 ln  ln 
500  10 90 7800  470  691  10  6
 t = 62.12 s
Cooling from 100oC to 30oC
30  10 1500  1382  10  4  t
 ln 
100  10 7800  470  691  10  6
 t =18.38 s
Total time for quenching = 62.12 + 18.38 = 80.5 s.
Example 4.6
Steel ball bearings (k = 50 W/m K,  = 1.3  10– 5 m2/s) having a diameter of
40 mm are heated to a temperature of 650oC and then quenched in a tank of oil at
55oC. If the heat transfer coefficient between the ball bearings and oil is
300 W/m2 K, determine :
(a) the duration of time the bearings must remain in oil to reach a temperature
of 200oC,
(b) the total amount of heat removed from each bearing during this time, and
(c) the instantaneous heat transfer rate from the bearings when they are first
immersed in oil and when they reach 200oC.
Solution
To determine if the bearings have negligible resistance, we first check the
magnitude of the Biot number.
r
h
hL 300  0.02
Bi   3  0.04
k k 3  50
Since Bi < 0.1, internal resistance may be neglected.
t t 1.3  10  5 t
Fo     0.2925 t
L2 r
2
 0.02 
2
   
 3  3 
(a) The time required for the ball bearings to reach 200oC is

 e  Bi . Fo
i
200  55 145
 e  0.04  0.2925 t   e  0.0117 t
650  55 595
 t = 120.67 s
which corresponds to a Fourier number of 35.3.
(b) Total amount of heat removed from each bearing 120.67 s 41
Conduction t

 hA (Ti  T ) e
 Bi . Fo
Q  hA (T  T )  dt
0

t hAt
 t
 hA i  e reV dt  hA i (1  e  Bi . Fo )
Bi . Fo
0

120.5
 300  4 (0.02) 2 (650  55) (1  e  0.04  35.3 ) 
0.04  35.31

= 57.9  104 W s or J = 57.9 kJ.

(c) Instantaneous heat transfer rate at t = 0 (or Fo = 0) is
Q  hA (Ti  T )  300  4  (0.02) 2  (650  55)
= 897 W
and at t = 120 s (Fo = 35.3),
Q  hA (Ti  T ) e  Bi Fo

 300  4  (0.02) 2 (650  50) e  (0.04) (35.3)

= 218 W.
Example 4.7
A semi-infinite aluminium cylinder (k = 237 W/m K,  9.71  10– 5 m2/s) of
diameter 20 cm is initially at a uniform temperature of 200oC. The cylinder is now
placed in water at 15oC where heat transfer takes place by convection with
h = 120 W/m2 K. Determine the temperature at the centre of the cylinder 15 cm
from the end surface 5 min after the start of the cooling.
Solution
We will solve the problem using the one-term solution for the cylinder and the
analytic solution for the semi-infinite medium.
hr0 120  0.1
Bi    0.05
k 237
t 9.71  10  5  5  60
Fo    2.913  0.2
r2 (0.1)2
Thus, one-term solution is applicable for cylinder, and Bi = 0.05, A1 = 1.0124 and
1 = 0.3126
2 2
c  A1 e  1 F 0  1.0124 e  (0.3126)  (2.913)

= 0.7616
The solution for the semi-infinite solid can be determine from
 
 x   hx h 2 t 
1   semi  inf ( x, t)  erf c    exp   
1  k k 2 
  
 2 ( t ) 
2

  1 
  x h (t ) 2  
erf c  1
 
  k 
 2 ( t )
2
  

42
 Governing Equations of
Heat Conduction

Tl = 200 oC
h

T∞ = 15 oC

D = 0.2 m
L = 15 cm T∞

Figure 2.25

x 0.15 m
where  1
 1
 0.88
2 ( t ) 2 2 (9.17  10  5  5  60) 2
1 1
5
h ( t ) 2 120 [9.71  10  5  60] 2
  0.086
k 237
hx 120  0.15
  0.0759
x 237
h 2 t
2
 (0.086) 2  0.0074
k
   semi inf  1  erf c (0.88)  exp (0.0759  0.0074) erf c (0.88  0.086 )

 1  0.2133  exp (0.0833)  0.170  0.974

Tc  T
  semi  inf  cy1  0.974  0.762  0.742
Ti  T

Tc  15  (200  15)  0.742  152.3 oC .

Exercise 4.1
A solid sphere of radius b = 5 cm and thermal conductivity k = 20 W/moC is
heated uniformly throughout the volume at a rate of 2000 W/m3, and heat is
dissipated by convection from its outer surface into the ambient air at T = 25oC
with a heat transfer coefficient h = 20 W/m2.oC. Determine the steady state
temperature at the centre and the outer surface of the sphere.

Exercise 4.2
A hollow cylinder with inner radius 30 mm and outer radius 50 mm is heated at the
inner surface at a rate of 105 W/m2 and dissipates heat by convection from the
outer surface into a fluid at temperature 100oC with a heat transfer coefficient of
400 W/m2 K. There is no energy generation, and the thermal conductivity of the
solid is assumed to be constant at 15 W/m K. Calculate the temperatures of the
inside and outside surfaces of the cylinder.

43
Conduction Exercise 4.3
An aluminium plate (k = 160 W/m K,  = 2790 kg/m3, cp = 0.88 kJ/kg K) of
thickness 30 mm and at a uniform temperature of 225o is suddenly immersed at
time t = 0 in a well-stirred fluid at a constant temperature of 25oC. The heat
transfer coefficient between the plate and the fluid is 320 W/m2 K. Determine the
time required for the centre of the plate to reach 50oC.
Exercise 4.4
A cubical piece of aluminium with the same properties, as given in problem 4.5, is
10 mm on a side and is heated from 50oC to 300oC by a direct flame. How long
should the aluminium remain in the flame if the flame temperature is 800oC and
the convective heat transfer coefficient between the flame and aluminium is
190 W/m2 K?
Exercise 4.5
Steel cylinder of diameter 0.25 m and length 0.8 m initially at 25oC is placed in a
furnace, where T = 1000oC. Determine the temperature at the centre and on the
surface of the cylinder after a lapse of 1 h. Assume k, , cp and h.
Exercise 4.6
During quenching, a cylindrical rod made of 1080 steel, 1 cm in diameter and
20 cm in length is first heated to 750oC and then immersed in a water bath at
100oC. The heat transfer coefficient can be taken as 250 W/msqC. The density,
sp heat and thermal conductivity of the steel are  = 7801 kg/m3, c = 473 J/kgoC
and k = 43 W/moC, respectively. Calculate the time required for the rod to reach
300oC.

4.9 SUMMARY
In the present unit, both the steady state and transient heat transfer due to conduction are
discussed. General method to solve a steady state conduction heat transfer problem is
illustrated. All the formulations in Cartesian, cylindrical and spherical coordinates are
presented. Solution of problems for simple cases, such as plane wall, long cylinder and
spheres are given. Subsequently, discussion is presented for transient heat transfer
problems. Lumped capacitance method is presented in details. Transient problems in
semi-infinite medium is given at the end. Some solved examples and SAQs are given for
better understanding of the unit.

4.10 KEY WORDS

Steady State Conduction : Parameters are independent of time
Transient Conduction : Conduction parameters are time dependent
Semi-Infinite Solid : Characterized by a single identifiable surface
Rectangular Coordinate : Coordinates in x, y and z direction.
Cylindrical Coordinate : Coordinates in radial r, circumferential  and axial
direction z of a cylinder.
Spherical Coordinate : Coordinates in radial r, circumferential  and
azumthal direction  of a sphere.
BIOT Number : Ratio of conduction resistance to convection
resistance.

4.11 ANSWERS TO SAQs

44 Please refer the relevant preceding text in this unit for answers to SAQs.
 Governing Equations of
APPENDIX-A Heat Conduction

w erf w w erf w w erf w

0.00 0.00000 0.36 0.38933 1.04 0.85865
0.02 0.02256 0.38 0.40901 1.08 0.87333
0.04 0.04511 0.40 0.42839 1.12 0.88679
0.06 0.06762 0.44 0.46622 1.16 0.89910
0.08 0.09008 0.48 0.50275 1.20 0.91031
0.10 0.11246 0.52 0.53790 1.30 0.93401
0.12 0.13476 0.56 0.57162 1.40 0.95228
0.14 0.15695 0.60 0.60386 1.50 0.96611
0.16 0.17901 0.64 0.63459 1.60 0.97635
0.18 0.20094 0.68 0.66378 1.70 0.98379
0.20 0.22270 0.72 0.69143 1.80 0.98909
0.22 0.24430 0.76 0.71754 1.90 0.99279
0.24 0.26570 0.80 0.74210 2.00 0.99532
0.26 0.28690 0.84 0.76514 2.20 0.99814
0.28 0.30788 0.88 0.78669 2.40 0.99931
0.30 0.32863 0.92 0.80677 2.60 0.99976
0.32 0.34913 0.96 0.82542 2.80 0.99992
0.34 0.36936 0.100 0.84270 3.00 0.99998
The Gaussian error function is defined as
w
2 2

 e w dv
erf w 
 0

The complementary error function is defined as

erf w  1  erf w

100
50
30
1.0 20
10
0.7 9
7
6
0.5
0.4 3
2.5
0.3 2.0

0.2
0
0.8 1.0
1.4 1.0
0.1 0.3 0.5
0.1 0.8
0 1 2 3 4 0.5 100
0.4 90
0.3 80
70
0.2 16 45 50 60
14 40
0.1 1.0 12
18
20
25 35
30
0.07 0.8 10 Bi -1 = k/hl
0.05 9
8
0.04 0.7
0.03 7

0.02 0.6 6
4
0.5
0.01 0.4
3

0.007 0.3
2.5

0.005 0.2
2.0
0.004 0.1
1.4
1.8

0.003
0.002 0
0.05
1.2 1.6

0.001
0 1 2 3 4 6 8 10 12 14 16 18 20 24 26 28 30 40 50 60 70 80 90 110 130 150 300 400 500 600
D1
45
Conduction 1.0
0.2

0.9

0.8 0.4

0.7

x/L
0.6 0.6

0.5

0.4

0.8
0.3

0.2 0.9

0.1

0 1.0
0.01 0.02 0.05 0.1 0.2 0.5 1.0 2 3 5 10 20 50

(k/hL) = Bi -1 D2

1.0 100
50
30
0.7 18
12
0.5
8
0.4 6
5
4
0.3 3.5

0.2 2.0 2.5

3.0
0 0.6 1.2 1.6
0.8 1.0
0.1 0.8
0 1 2 3 4 0.5 Bi -1 = k/hl

0.4
0.3 100

0.2 16 80
90
14 70
0.1 1.0 12
18
20 25
60
50
0.0 0.8 10 45
40
7
0.0 9
30
35
8
0.0
5 0.7
0.0
4 7

3
0.0 0.6 6
2 0.5
4

0.01 0.4
3

0.007 0.3
2.5

0.005 0.2
2.0
0.004 0.1
1.4
1.8
0.003
0.002 0
0.05
1.2 1.6

0.001
0 1 2 3 4 6 8 10 12 14 16 18 20 24 26 28 30 40 50 60 70 80 90 110 130 150 300 400 500 600

D3

46
 Governing Equations of
Heat Conduction
1.0
0.2
0.9

0.8 0.4

0.7

x/L
0.6 0.6

0.5

0.4

0.8
0.3

0.2 0.9

0.1

0 1.0
0.01 0.02 0.05 0.1 0.2 0.5 1.0 2 3 5 10 20 50
D4
(k/hro )= Bi-1

REFERENCES
F. P. Incropera and D. P. Dewitt (2004), Fundamentals of Heat and Mass Transfer,
5th Edition , John Willay and Sons.
W. M. Rohsenow and H. C. Choi (1961), Heat Mass and Momentum Transfer,
Prentice-Hall Englewood Cliffs, New Jersey.
E. R. G. Eckert and R. M. Drake Jr. (1959), Heat and Mass Transfer, McGraw-Hill,
New York.
P. K. Nag (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.
J. P. Holman (2002), Heat Transfer, 9th Edition, Tata McGraw-Hill, New Delhi.

47
Conduction

48
 Numerical Methods to
UNIT 5 NUMERICAL METHODS TO SOLVE Solve Heat Conduction
Problems
HEAT CONDUCTION PROBLEMS
Structure
5.1 Introduction
Objectives

5.2 Analytical Solution

5.3 Graphical Method
5.3.1 Determination of Heat Transfer Rate by Graphical Method
5.3.2 The Conduction Shape Factor

5.4 Finite Difference Scheme

5.4.1 Central, Forward and Backward Difference Schemes with Uniform Grid
5.4.2 Forward Differences
5.4.3 Backward Differences
5.4.4 Condition for using Forward, Backward and Central Difference Expressions
5.4.5 Difference Expressions of Higher Accuracy

5.5 Numerical Errors

5.5.1 Round-off Error
5.5.2 Truncation Error
5.5.3 Discretisation Error

5.6 Accuracy of a Solution : Optimum Step Size

5.7 Numerical Methods for Conduction Heat Transfer
5.7.1 Numerical Methods for a One Dimensional Steady Time Problems
5.7.2 Numerical Methods for Two Dimensional Steady State Problems

5.8 Transient One Dimensional Problems

5.9 Summary
5.10 Key Words
5.11 Answers to SAQs

1.1 INTRODUCTION
Analytical method is applicable to simple geometries and boundary conditions only. In
case of complicated geometries, boundary conditions, and temperature dependent thermal
properties, analytical method can not be used. Further, most of the practical problems
encountered are associated with complicated geometries, boundary conditions as well as
variable thermal properties. In such situation, numerical methods are extensively used to
find the heat transfer rate and temperature distribution. One of the most extensively used
numerical methods for conduction heat transfer problems is the finite difference method.
In the present unit analytical solution for simple geometry is first presented. Graphical
method for certain category of problems are also described. Finite difference method is
then elaborately described.
Objectives
After studying this unit, you should be able to
 know the importance of analytical, graphical and numerical methods in
solving various heat conduction problems,

51
Conduction  appreciate importance of numerical method for solving conduction heat
transfer problems,
 formulate finite difference equations for a conduction heat transfer problem,
and
 use various methods to solve the finite difference equation.

5.2 ANALYTICAL SOLUTION

Analytical method is applicable to simple geometries and boundary conditions only. The
components of the heat flow per unit area (heat flux) q in the x and y directions are
obtained from Fourier’s law
dT
qx   k . . . (5.1)
dx
dT
qy   k . . . (5.2)
dx
The heat flux depends on the temperature gradient and is, therefore, a vector, while
temperature is a scalar.
The heat flux q at a given point (x, y) is the resultant of the components qx and qy at
that point and its directed perpendicular to the isotherm (Figure 5.1).
qy qn”

qx”
y
n
Isotherm
x
Figure 5.1 : Heat Flow in Two Dimensions

If the temperature distribution in a system is known, the rate of heat flow can easily be
calculated. Therefore, heat flow analyses usually concentrate on determining the
temperature field.
Let us consider a simple case of a thin rectangular plate (Figure 5.2), free of heat sources
and insulated at the top and bottom surfaces. For steady state two-dimensional heat
conduction in the absence of any heat source, and uniform thermal conductivity, Laplace
equation applies
 2T  2T
 0 . . . (5.3)
x 2 y 2

T
Since  0 , the temperature is a function of x and y only. If k is uniform, the
z
temperature distribution must satisfy Eq. (5.3), a linear and homogeneous partial
differential equation that can be integrated by assuming a product solution for T (x, y) of
the form
T  X (x) y ( y ) . . . (5.4)

where X = X (x), is a function of x only, and Y = Y (y), a function of y alone.

Differentiating Eq. (5.4) twice, first with respect to x and then with respect to y,

52
 Numerical Methods to
 2T 2 X
Y . . . (5.5) Solve Heat Conduction
x 2 x 2 Problems

 2T  2Y
 . . . (5.6)
y 2 y 2
y

T = Tm sin (x/L)

T=0 T=0

0 X
L

T=0

Figure 5.2 : Rectangular Adiabatic Plate with Sinusoidal Temperature Distribution on One Edge

Substituting in Eq. (5.3)

2 X  2Y
Y X 0 . . . (5.7)
x 2 y 2

1  2 X 1  2Y
or,  2
 2
 2 (say) . . . (5.8)
X x Y y

The variables are now separated. The LHS is a function of x only, while the RHS is a
function y alone. Since neither side can change as x and y vary, both must be equal to a
constant, say 2. We have, therefore, two ordinary differential equations

d2X
2
 2 X  0 . . . (5.9)
dx

d 2Y
2
 2 Y  0 . . . (5.10)
dy

The general solution of Eq. (5.9) is

X  C1 cos  x  C2 sin  x . . . (5.11)

and the general solution of Eq. (5.10) is

Y  C3 e   y  C 4 e  y . . . (5.12)

Therefore, from Eq. (5.4),

T  X . Y  (C1 cos  x  C 2 sin  x ) (C3 e  y  C 4 e  y) . . . (5.13)

where C1, C2, C3 and C4 are constants calculated from the boundary conditions. As shown
in Figure 5.2, boundary conditions to be satisfied are
T = 0 at x = 0 . . . (5.14)
T = 0 at x = L . . . (5.15)
T = 0 at y = 0 . . . (5.16) 53
Conduction x
T  Tm sin   at y = b . . . (5.17)
 L 
Substituting these conditions in Eq. (5.13), from the 3rd condition (at y = 0)
(C1 cos  x  C2 sin  x) (C3  C4 )  0 . . . (5.18)

From the first condition (at x = 0)

C1 (C3 e   y  C4 e  y )  0 . . . (5.19)
From the second condition

(C1 cos  L  C2 sin  L) (C3 e   y  C4 e  y )  0 . . . (5.20)

Eq. (5.18) gives C3 =  C4 and Eq. (5.19) gives C1 = 0. Using these results in Eq. (5.20)
2C2 C3 sin  L sinh  y  0 . . . (5.21)
n
To satisfy this condition, sin  L = 0 or,   , where n = 1, 2, 3, . . . There exists
L
therefore, a different solution for each integer n, and each solution has a separate
integration constant Cn. Summing these solutions, we get from Eq. (5.13),
 nx n y
T   C n sin sinh . . . (5.22)
n 1 L L
The last boundary condition needs that at y = b
x x b
Tm sin  C1 sin sinh . . . (5.23)
L L L
Tm
Hence, C1  . . . (5.24)
 b 
sinh  
 L 
The solution therefore becomes (From Eq. (5.22))
y
sinh  
T ( x, y )  Tm  L  sin  x . . . (5.25)
 b  L
sinh  
 L 
Corresponding temperature field is shown in Figure 5.3.
Isotherms

Heat flow lines

Figure 5.3 : Isotherms and Heat Flow Lines for the Plate in Figure 5.2

The solid lines are isotherms and the dashed lines are heat flow lines, which are
orthogonal.

54
SAQ 1 Numerical Methods to
Solve Heat Conduction
(a) Write down the Laplace equation for steady state two dimensional heat Problems
conduction in the absence of any heat source, and uniform thermal
conductivity.
(b) What is the significance of isotherms?
(c) Why are the isotherms and adiabatics orthogonal?
5.2.1 Extension to 3-D Problems
The separation of variable method can be extended to three-dimensional problems, by
assuming that T = X Y Z. Substituting this expression for T in equation
 2T  2T  2T
  0 . . . (5.26)
x 2 y 2 z 2
Separating the variables, integrating the total differential equations and using the given
boundary conditions, the solution for the temperature distribution can be obtained.

5.3 GRAPHICAL METHOD

Graphical method can rapidly provide an approximate estimate of the temperature
distribution and heat flow in geometrically complex 2-D systems. However, this method
is limited to isothermal and adiabatic boundary conditions only.
The objective of the graphical solution is to systematically construct a network of
isotherms and adiabatics. The flux lines are analogous to the streamlines in a potential
flow, i.e. tangent to the direction of heat flow at any point. Therefore, no heat can flow
across the constant flux lines. The isotherms are analogous to constant potential lines and
heat flows perpendicular to them. Thus, isotherms and adiabatics intersect at right angles.
Consider a square, two-dimensional channel whose inner and outer surfaces are
maintained at temperature T1 and T2, respectively. A cross section of the channel is
shown in Figure 5.4.

b y
a x
x
Adiabat
T1
ss
T2 T2 qi d
y
c
Tj
(c)
Symmetry lines
T1
(a)
qi
qi Tj
2 Isotherms

(b)
Figure 5.4 : Two-Dimensional Conduction in a Square Channel of Length L
(a) Symmetry Planes, (b) Flux Plot and (c) Typical Curvilinear Square

Following procedure is adapted to construct the flux plot :

Step 1
Identify all relevant lines of symmetry. Such lines are determined by thermal as
well as geometrical conditions. As for example, for the square channel in
Figure 5.4(a), such lines include the designated vertical and horizontal and
55
Conduction diagonal lines. For this system it is therefore possible to consider only one-eighth
of the configuration as shown in Figure 5.4(b).
Step 2
Line of symmetry are adiabatic in the sense that there can be no heat transfer in a
direction perpendicular to the lines. They are there for heat flow lines and should
be treated as such. Since there is no heat flow in a direction perpendicular to a heat
flow line, it can be termed as adiabat.
Step 3
After all the known lines of constant temperature are identified, an attempt should
be made to sketch lines of constant temperature within the system. Note that
isotherms should always be perpendicular to adiabats.
Step 4
Heat flow lines should be drawn with an eye towards creating a network of
curvilinear squares. This is done by having heat flow lines and isotherms intersect
at right angles and by requiring that all sides of each square be of approximately of
same length. It is often impossible to satisfy this second requirement exactly, and it
is more realistic to strive for equivalence between the sums of the opposite sides of
each square (Figure 5.4(c)). Assuming x-coordinate to the direction of heat flow
and the y-coordinate to the direction normal to this flow, the requirement may be
expressed as
ab  cd ac  bd
x   y  . . . (5.27)
2 2
Several iterations must often be made to draw a network of curvilinear squares.
This trial and error process involves adjusting the isotherms and adiabats until
satisfactory curvilinear squares are obtained for most of the network. Once the flux
plot has been obtained, it may be used to infer the temperature distribution in the
medium.
SAQ 2
(a) Explain the steps to be considered for construction of flux plot in graphical
method.
(b) Explain the graphical method of solving a two-dimensional heat conduction
problem.
5.3.1 Determination of Heat Transfer Rate by Graphical Method
Let, qi = Energy flow rate through a lane (region between two adiabats).
If the flux plot is properly constructed, the value of qi will be approximately the same for
all the lanes and the total heat transfer may be expressed as :
M
q   qi  M qi . . . (5.28)
i 1

where M is the number of lanes associated with the plot. From the curvilinear plot of
Figure 5.4(c) and application of Fourier’s law of heat conduction, qi may be expressed as
T j T j
qi  k Ai  k (y . l ) . . . (5.29)
x x
where Tj is the temperature difference between successive isotherms, Ai is the
conduction heat transfer area for the lane, and l is the length of the channel normal to the
page. As the temperature increment is approximately same for all the adjoining isotherms,
the overall temperature difference between boundaries, T1  2 , may be expressed as

56
 N Numerical Methods to
T1  2   T j  N T j . . . (5.30) Solve Heat Conduction
j 1 Problems

where N is the total number of temperature increments. Combining Eqs. (5.28)-(5.30) and
recognizing that x  y for curvilinear squares, we obtain
Ml
q k T1  2 . . . (5.31)
N
The manner in which a flux plot may be used to obtain the heat transfer rate for a
M 
two-dimensional system is evident from Eq. (5.31). The ratio of   may be obtained
N
from the plot.
Recall that the specification of N is based on Step 3 of the foregoing procedure, and the
value, which is an integer, may be a consequence of following Step 4. Note that M is not
necessarily an integer, since a fraction may be needed to arrive at a satisfactory network
of curvilinear squares. For the network of Figure 5.4(b), N = 6 and M = 5. Of course, as
the network, or mesh of curvilinear squares is made finer, N and M increase and the
M 
estimate of   becomes more accurate.
N
5.3.2 The Conduction Shape Factor
Eq. (5.31), may be used to define shape factor S, of a two-dimensional system. Heat
transfer rate with shape factor may be expressed as
q  S k T1  2 . . . (5.32)

where for a flux plot

Ml
S . . . (5.33)
N
From Eq. (5.32), it follows that a two-dimensional conduction resistance may be
expressed as
1
Rl , 2  Dcond  . . . (5.34)
Sk
Shape factors have been obtained for numerous two-dimensional systems and results are
summarised in Figure 5.5.
System Schematic Restrictions Shape Factor
Case 1 T2
Isothermal z
sphere 2 D
D
buried in a z D
T1 2 1
semi-infinite D 4z
medium

Case 2 T2
Horizontal L>D
isothermal Z 2 L
L>D
cylinder of 1  2 z 
length L L
3D cosh  
buried in a T1
D z D
2
semi-infinite
medium

57
Conduction Step 3
T2
Vertical
cylinder in a 2 L
semi-infinite L L>D  4L 
medium T1 ln  
 D
D

Case 4
T1 D1 D2
Conduction 2 L
between two L > D1, D2 3 2 2
 1  4 w  D1  D2 
T2
cylinders of cosh  
W L>w  
length L in 2 D1 D1
 
infinite
medium
Case 5
Horizontal
 T2 
circular
cylinder of
length L, Z 2 L
z>D
midway  8L 
between L>w ln  
T1 D
parallel Z  D 
 
planes of
equal length T2
and infinite
width
Case 6
Circular T2
cylinder of
length L D
2 L
w>D
oriented in  1.08 w 
square solid
W L>w ln  
T1
 D 
of equal
length

Case 7
Eccentric d T1
circular
T2
cylinder of D
length L in a 2 L
D>d 2 2 2
cylinder of 1  D  a  4x 
equal length L>D cosh  
 2D d 
 
Z

Case 8
Conduction T2
through the
edge of L
D
adjoining D > D5 0.54 D
walls

T1
L

58
 Numerical Methods to
Case 9
Solve Heat Conduction
L Problems
Conduction
through L
corner of
three walls L < length
with a and width of 0.15 L
temperature L wall
difference
T1 – 2
across the
walls
Case 10 D
T1
Disk of
diameter D
and T1 on a
semi-infinite None 2D
medium of k
thermal
conductivity T2
k and T2

Case 11
L 2 L
Square W
channel of  1.4 W 
w 0.785 ln  
length L T2 w
T1 2 L
W
w  1.4 W 
w 0.930 ln    0.050
w
w

Figure 5.5 : Conduction Shape Factor for Selected Two-Dimensional Systems

SAQ 3
(a) What do you mean by conduction shape factor? How is it estimated?
(b) How does the conduction shape factor help in solving conduction heat
transfer problems graphically?

5.4 FINITE DIFFERENCE SCHEME

Basic approach in solving a problem with the finite difference method is to discretise the
derivatives appearing in the governing differential equation at a finite number of
uniformly or non-uniformly spaced grid points that fill the computational domain. The
governing differential equation is then transformed into a system of difference equations.
This means that if there are 100 grid points, where variables are unknown, there will be
100 equations to solve per variable. The necessity of using a computer arises because of
the huge number of arithmetic operations that are required to be carried out in a
reasonable time for solving a large number of equations. The simplification inherent in
the use of algebraic equations rather than differential equations is what makes numerical
methods so powerful and widely applicable.
5.4.1 Central, Forward and Backward Difference Schemes
with Uniform Grid
Consider a smooth function y = f (x) as in Figure 5.6.
The Taylor series for the said function at xi + h expanded about xi is

59
Conduction
yi h 2 y  h3
y ( xi  h)  yi  yi h   i  ... . . . (5.35)
2! 3!
where h = x and yi, yi + 1, yi – 1 are the ordinates corresponding to xi, xi + h,, xi – h,
respectively. The function at (xi – h) is similarly given by
yi h 2 y  h3
y ( xi  h)  yi  yi h   i ... . . . (5.36)
2! 3!
Tangent at xi

Approximation of
the actual slope
Y = f(x)

yi
yi - 1 yi + 1

h = x h = x

T2

xi - h x xi + h

Figure 5.6 : Uniformly Spaced Grid Points on a Continuous and Differentiable Function y = f (x)

Subtracting Eq. (5.36) from Eq. (5.35) we get,

2 yi h3
y ( xi  h)  y ( xi  h)  2 yi h   ... . . . (5.37)
3!

yi h3
or 2 yi h  y ( xi  h)  y ( xi  h)   ... . . . (5.38)
3

y ( xi  h)  y ( xi  h) yih 2
Therefore, yi    (higher order terms) . . . (5.39)
2h 6
yi  1  yi  1
or yi   (error of order h 2 ) . . . (5.40)
2h
yi  1  yi  1
yi   0 (h 2 ) . . . (5.41)
2h
The notation o (h2) means that in arriving at Eq. (5.41), terms of the order of h2 and
higher have been neglected. o (h2) is called the truncation error. The truncation error is
the difference between the exact mathematical expression and its numerical
approximation.

 dy 
Eq. (5.41) is called the central-difference approximation of y   i.e.  at xi with an
 dx 
error of order h . In Figure 5.6, the approximation is depicted by the slope of the dashed
2

line. The actual derivative is shown by the solid line drawn tangent to the curve at xi. The
difference can be viewed as due to the truncation error resulting from using a truncated
Taylor series.
Now, adding Eqs. (5.35) and (5.36), we get

yi h 4
y ( xi  h)  y ( xi  h)  2 yi  yi h 2   ... . . . (5.42)
12
60
 Numerical Methods to
yi h 4
or yi h 2  y ( xi  h )  y ( xi  h )  2 yi   (higher-order terms) . . . (5.43) Solve Heat Conduction
12 Problems

yi  2  2 yi  yi  1
yi  2
 0 (h 2 ) . . . (5.44)
h
Eq. (5.44) is the central-difference approximation of the second derivative of the function
 d2y 
with respect to x  i.e. 2  evaluated at xi with an error of order h2. Alternatively,
 dx 

Eq. (5.44) may be expressed as
yi  1  yi yi  yi 1

yi  h h . . . (5.45)
h
y 1  y 1
i i
2 2
or yi  . . . (5.46)
h
h
where y  1 and y  1 represent the slopes of the tangents to the curve at xi  and
i
2
i
2
2
h
xi  , respectively.
2
The central-difference expressions reveal that the first and second derivatives of the
function involve values of the function on both sides of the x-value at which the
derivative of the function is to be evaluated.
5.4.2 Forward Difference
From Taylor series expansions, it is also easy to obtain expressions for the derivatives
which are entirely in terms of values of the function at xi and points to the right of xi.
These are called forward-difference expressions.
Starting from the Taylor series expansion as given in Eq. (5.35), we get

yi h 2 yi h3
yi h  y ( xi  h )  yi    ... . . . (5.47)
2! 3!

y ( xi  h)  yi y  h 2 yih 3
or yi h   i   ... . . . (5.48)
h 2! 3!
Dropping terms of the order of h and higher,
y ( xi  h)  yi
yi   0 (h ) . . . (5.49)
h
yi (2h) 2 yi (2h)3
Similarly, y ( xi  2h)  yi  yi (2h)    ...
2! 3!
 yi  yi (2h )  2 yi h 2  0 (h 3 ) . . . (5.50)

yi h 2
Also, y ( xi  h)  yi  yi h   0 (h 3) . . . (5.51)
2!
Multiplying Eq. (5.51) by 2 and substracting from Eq. (5.50) gives
y ( xi  2h)  2 y ( xi  h)  yi
yi   0 (h ) . . . (5.52)
h2

61
Conduction yi  2  2 yi  1  yi
yi   0 (h ) . . . (5.53)
h2
5.4.3 Backward Difference
Following the approach given in Section 5.4.2, one can easily obtain derivative
expressions which are entirely in terms of the values of the function at xi. These are
known as backward-difference expressions, which are given below for yi and yi .
yi  yi  1
yi   0 (h ) . . . (5.54)
h
yi  2 yi  1  yi  2
yi   0 (h ) . . . (5.55)
h2
5.4.4 Conditions for using Forward, Backward, and Central-Difference
Expressions
 Forward-difference expressions are used when data to the left of a point, at
which a derivative is desired, are not available.
 Backward-difference expressions are used when data to the right of the
desired point are not available.
 Central-difference expressions are used when data on both sides of the
desired point are available and are more accurate than either forward- or
backward-difference expressions.
SAQ 4
(a) Explain the forward difference, central difference and backward difference
schemes.
(b) Under what conditions will you use these schemes?
5.4.5 Difference Expressions of Higher Accuracy
By retaining a greater number of terms in the Taylor series, it is possible to obtain
forward-, backward-, and central difference expressions for a higher accuracy. The
following expressions show central-difference expressions for yi and yi with an error
of 0 (h4) and forward- and backward-difference expressions for the same with an error of
0 (h2). It is apparent that for a greater accuracy more number of neighbouring points are
involved. For example, Eq. (5.54) is a two-point forward-difference scheme for yi , while
Eq. (5.55) is a three-point forward-difference scheme. Central difference with an error of
0 (h4) :
 yi  2  8 yi  1  8 yi  1  yi  2
yi  . . . (5.56)
12h
 yi  2  16 yi  1  30 yi  16 yi  1  yi  2
yi  . . . (5.57)
12h 2
Forward difference with an error of 0 (h2) :
 yi  2  4 yi  1  3 yi
yi  . . . (5.58)
2h
 yi  3  4 yi  2  5 yi  1  2 yi
yi  . . . (5.59)
h2
Backward difference with an error of 0 (h2) :
3 yi  4 yi  1  yi  2
yi  . . . (5.60)
2h
62
 2 yi  5 yi  1  4 yi  2  yi  3 Numerical Methods to
yi  . . . (5.61) Solve Heat Conduction
h2 Problems

Now, the question is : When does one use a higher-order difference scheme? There is no
set answer to this. It depends on the accuracy requirement of a problem, and the analyst
will have to use his own judgment.

5.5 NUMERICAL ERRORS

Three most important errors that commonly occur in numerical solutions are the
(a) round-off error,
(b) truncation error, and
(c) discretization error.
5.5.1 Round-Off Error
The round-off error is introduced because of the inability of the computer to handle a
large number of significant digits. Typically, in single-precision, the number of
significant digits retained ranges from 7 to 16, although it may vary from one computer
system to another. The round-off error arises due to the fact that a finite number of
significant digits or decimal places are retained and all real numbers are rounded off by
the computer. The last retained digit is rounded off if the first discarded digit is equal to
or greater than 5. Otherwise, it is unchanged. For example, if five significant digits are to
be kept in place, 5.37527 is rounded off to 5.3753, and 5.37524 to 5.3752.
5.5.2 Truncation Error
The truncation error is due to the replacement of an exact mathematical expression by a
numerical approximation. This error has been discussed earlier in this chapter with
respect to finite-difference approximations. Basically, it is the difference between an
exact expression and the corresponding truncated form (for example, truncated Taylor
series) used in the numerical solution.
5.5.3 Discretization Error
The discretization error is the error in the overall solution that results from the truncation
error assuming the round-off error to be negligible. Therefore,
(Discretization error) = (exact solution) – (numerical solution with no round-off solution)
SAQ 5
What are different types of numerical errors? Expalin with suitable examples.

5.6 ACCURACY OF A SOLUTION : OPTIMUM STEP SIZE

The accuracy of a numerical solution is determined by its total error, which is the sum of
round-off error and truncation error. Hence, (total error) = (round-off error) + (truncation
error). However, it is obvious that the round-off error increases as the total number of
arithmetic operations increases. Again, the total number of arithmetic operations
increases if the step size decreases (that is, when the number of grid points increases).
Therefore, the round-off error is inversely proportional to the step size. On the other hand,
the truncation error decreases as the step size decreases (or as the number of grid points
increases). Because of the aforementioned opposing effects, an optimum step size is
expected, which will produce minimum total error in the overall solution.
5.6.1 Method of Choosing Optimum Step Size – Grid Independence Test
A numerical analyst has to be extremely careful as regards the accuracy of a solution. To
get the most accurate solution (that is, the solution with the least total error), one has to 63
Conduction perform a grid independence test. The test is carried out by experimenting with various
grid sizes and watching how the solution changes with respect to the changes in the grid
size. Finally, a stage will come when changing the grid spacing will not affect the
solution. In other words, the solution will become independent of grid spacing. The
largest value of grid spacing for which the solution is essentially independent of the step
size is chosen so that both the computational time and effort and the round-off error are
minimised.

5.7 NUMERICAL METHODS FOR CONDUCTION

HEAT TRANSFER
Many difficult problems arise in conduction, for example, variable thermal conductivity,
distributed energy sources, radiation boundary condition, for which analytical solution
are not available. Approximate solutions for these are obtained by numerical methods.
The basic approach is to arrive at the relevant governing differential equation based on
the physics of the particular problem. They are then converted into the required finite-
difference forms. To begin with, the numerical solution procedure for the problem of a
simple one-dimensional steady-state heat conduction in a cooling fin described. It is to be
noted that a simple, closed-form straightforward analytical solution for this problem is
available. The idea is to show the use of the numerical method and to compare the
numerical solution with its analytical counterpart.
5.7.1 Numerical Methods for a One-Dimensional Steady-State Problem
Consider one-dimensional, steady-state heat conduction in an isolated rectangular fin as
shown in Figure 5.7. The base temperature is maintained at T = T0 and the tip of the fin is
insulated. The fin is exposed to a convective environment (neglecting radiation heat
transfer from the fin) which is at T (T  T0 ) . The average heat transfer coefficient of
the fin to the surroundings is h. The length of the fin is L and the coordinate axis begins
at the base of the fin. The one-dimensionality arises from the fact that the thickness of the
fin is much small as compared to its length, and the width can be considered either too
long or the sides of the fin to be insulated.

T = T0 H, T

dT
=0
dx

x
L

Figure 5.7 : Physical Domain of a Rectangular Fin

Governing Differential Equation

The energy equation for the fin in the steady state (assuming constant k) is

d 2T hP
2
 (T  T )  0 . . . (5.62)
dx kA
where P and A are the perimeter and the cross-sectional area of the fin, respectively.
Boundary Conditions
Since Eq. (5.62) is a linear, second-order ordinary differential equation, two
boundary conditions are needed to completely describe this problem. Boundary
conditions are as follows :
BC-1 : at x = 0, T = T0 . . . (5.63)
64
 dT Numerical Methods to
BC-2 : at x = L, 0 . . . (5.64) Solve Heat Conduction
dx Problems

Non-Dimensionalisation
Non-dimensionalising Eqs. (5.62)-(5.64) and using the dimensionless variables
T  T x
 ,X  . . . (5.65)
T0  T L

d 2
We obtain  ( mL) 2   0 . . . (5.66)
dx 2
hP
Where m2  . . . (5.67)
kA
and  (0)  1 . . . (5.68)

 (1)  0 . . . (5.69)

d
where   . . . (5.70)
dX
Discretisation
Eq. (5.66) is discretised at any interior grid point i (see Figure 5.8) using central
d 2
difference for as follows :
dX 2
 d 2 
 2
  ( mL) 2    0 . . . (5.71)
  i
 dX  i
Interior grid point

1 2 M -1 M

i-1 i i+1

Figure 5.8 : Computational Domain of the Fin with Equally Spaced Grid Points

i  1  2i  i  1
or 2
 ( mL) 2 i  0 . . . (5.72)
( X )

or i  1  D i  i  1  0, i  2, . . . , M . . . (5.73)

where D  2  (mL ) 2 (X ) 2 .

Handling of the Boundary Condition
At x = L, i.e. at i = M, Eq. (5.73) reduces to
M 1  D M  M 1 0 . . . (5.74)

A careful look at Eq. (5.74) reveals that M + 1 represents a fictitious temperature 

at point M +1, which lies outside the computational domain. There is a remedy to
tackle this issue.

65
Conduction Remedy : Image Point Technique
It is assumed that the  versus X curve extends beyond X = 1 so that at X = 1, the
d
condition  0 is satisfied. In other wards, the  versus X curve can be
dX
imagined to look as in Figure 5.9. The dotted line represents the mirror-image
extension of the solid line, indicating that a minimum exists at X = 1. Figure 5.9(a)
shows a mirror-image extension of the fin.

M - 1 M + 1
M

x x

(a)

=1 =1

XM - 1 XM XM + 1

(b)

Figure 5.9 : (a) Mirror-image Extension of the  versus X Curve near the Fin Tip and
(b) Mirror-image Extension of the Fin

Therefore, the boundary condition at X = 1 can be approximately satisfied by

taking
M 1  M 1 . . . (5.75)

d
Eq. (5.75) also follows from the central-difference approximation of at i = M.
dX
Substituting Eq. (5.75) into Eq. (5.74), we get
2 M 1  D M  0 . . . (5.76)

Therefore, we can write that

i  1 for i  1 (known) . . . (5.77)

i  1  D i  i  1  0 for i = 2, . . . , m – 1 . . . (5.78)

2 M 1  D  M  0 for i = M . . . (5.79)

Hence, we have a set of M – 1 linier simultaneous algebraic equations and M – 1

unknowns, which can be easily solved by standard numerical methods. For the
case in which M = 5, we have N = M – 1 = 4 equations to solve. The four equations
can be written in the matrix form as

 D 1   1   1
 1 D 1    0
   2    . . . (5.80)
 1 D 1  3   0
    
 2 D    4   0
66
 It should be noted that 1 corresponds to the temperature at grid point 2 in Numerical Methods to
Solve Heat Conduction
Figure 5.8, and so on for 2, 3, 4. An alternative to the image-point scheme is to Problems
d
use a second-order backward difference for at i = M.
dX
Methods of Solution
In Eq. (5.80), the coefficient matrix has three diagonals - the main diagonal,
sub-diagonal and super-diagonal, and hence the name tridiagonal matrix (TDM).
The set of equations in Eq. (5.80) is called tridiagonal system of equations. See
Figure 5.10 for a pictorial representation of TDM.
Main diagonal
Super diagonal
Sub-diagonal

Tri-diagonal coefficient matrix Unknown Right-hand

A Column vector Column vector
X C
Figure 5.10 : Pictorial Representation of a Tridiagonal Coefficient Matrix,
an Unknown Vector and a Known Right Hand Vector

The set of equations in Equation 5.80 can be solved by any of the three methods :
 Gaussian elimination,
 Thomas algorithm (or tridiogonal matrix algorithm or simply TDMA),
and
 Gauss-Seidel iterative method.
Gaussian Elimination (GE)
This method reduces a given set of N equations to an equivalent triangular set, so
that one of the equations has only one unknown. This unknown is determined and
the remaining unknowns are obtained by the process of back substitution. The
basic approach is shown in a step-by-step form as given. The set of equations to be
solved are written in a matrix form in Eq. (5.81)
 a11 a12 a13   x1   c1 
a a23   x   c 
 21 a22  2  2 . . . (5.81)
 a31 a32 a33   x3   c3 

a11 is called the pivot, below which the terms are to be made zero.
Step I
 a11 a12 a13   x1   c1 
 (1) (1)   x    c (1) 
 a22 a23   2  2  . . . (5.82)
 (1) (1)   x3   c (1) 
 a32 a33   3 
(1)
The superscript represents the step number. a22 is now the pivot for the
next operation.
67
Conduction Step II
 a11 a12 a13   x1   c1 
 (1) (1)   x    c (1) 
 a22 a23   2  2  . . . (5.83)
 (2)   x3   c (2) 
 a33   3 
Solution Accuracy
The round-off error may significantly affect the accuracy if a large number of
equations is involved. In addition, the round-off error is cumulative because the
errors are carried on from one step to the other during the elimination process.
Consequently, GE is generally used if the number of equations is typically less
than 20 when the coefficient matrix is dense. For a sparse coefficient matrix,
however a large number of equations can be solved. The TDM system is a good
example of a sparse coefficient matrix. If Gaussian elimination is applied to this
system, only one of the a s is eliminated from the column containing the pivot
element in each step, since the remaining elements below the diagonal are zero.
Therefore, only one elimination process is employed at each step. The number of
operations needed for solving a tridiagonal system is of order N, that is, O (N) as
compared to O (N3) for a system with a dense coefficient matrix. Therefore, a
much smaller number of operations and consequently much lower round-off errors
arise in the solution of the systems. Obviously, the computer time is much less for
solution by TDMA. Thus, large tridiagonal systems are generally solved by this
method.
Thomas Algorithm or TDMA
The set of equations in Eq. (5.78) can be readily solved by the Gaussian
elimination method with a maximum of three variables per equation. The solution
can be expressed very concisely. Eq. (5.78) is actually a special form of the system
(using N = M – 1).
b1 T1  c1 T2  d1

a2 T1  b2 T2  c2 T3  d 2

a3 T2  b3 T3  c3 T4  d3


ai Ti  1  bi Ti  ci Ti  1  di


aN  1 TN  2  bN  1 TN  1  cN  1 TN  dN 1

a N TN 1  bN TN  d N . . . (5.84)

First, let us demonstrate the validity of a recursion solution of the form (Carnahan
et al. 1969)
ci
Ti   i  Ti  1 . . . (5.85)
i

in which the constants i and i are to be determined. The substitution of Eq. (5.85)
into Eq. (5.84) gives
 ci  1 
ai   i  1  Ti   bi Ti  ci Ti  1  di . . . (5.86)
  
 i  1 
Rearranging Eq. (5.86), we obtain

68
 di  ai  i  1 ci Ti  1 Numerical Methods to
Ti   . . . (5.87) Solve Heat Conduction
ai ci  1 ai ci  1 Problems
bi  bi 
i  1 i  1

Eq. (5.87) verifies the form of Eq. (5.84), subject to the following recursion
relations :
ai ci  1
i  bi  . . . (5.88)
i  1

di  ai  i  1
i  . . . (5.89)
i

Also, from the first equation of Eq. (5.84),

d1 c1
T1   T2 . . . (5.90)
b1 b1

From which we get

d1
1  b1 , 1  . . . (5.91)
1

Finally, the substitution of the recursion solution into the last equation of
Eq. (5.84) yields
 cN 1 
d N  aN   N 1  TN 
d N  a N TN  N  1 
TN 
1
   . . . (5.92)
bN bN

from which

d N  aN  N  1
TN   N . . . (5.93)
aN cN  1
bN 
N  1

In a nutshell, the complete algorithm for the solution of the tridiagonal system is

TN   N . . . (5.94)

ci Ti  1
Ti   i  , i  N  1, N  2, . . . , 1 . . . (5.95)
i

where, s and s are determined from the recursion formulae

d1
i  bi , 1  . . . (5.96)
1

ai ci  1
i  bi  , i  2, 3, . . . , N . . . (5.97)
i  1

di  ai  i  1
i  , i  2, 3, . . . , N . . . (5.98)
i

The above algorithm is also known as the Thomas algorithm.

69
Conduction Finally, it is to be noted that Eq. (5.78) might also be solved by the Gauss-Seidel
iteration scheme discussed next.
The Gauss-Seidel Iterative Method (GS)
For a large number of equations (typically of the order of several hundred) iterative
methods such as Jacobi, Gauss-Seidel, which initiate the computations with a
guessed solution and iterate to the desired solution of the systems of equations
within a specified convergence criterion, using improved guesses in the second and
third iterations till the final one, are often more efficient. In the GS method only
the values of the latest iteration are stored, and each iterative computation of the
unknown employs the most recent values of the other unknowns. In this method,
unlike in direct methods, such as Gaussian elimination, the round-off error does
not accumulate. The round-off error after each iteration simply produces a less
accurate input for the next iteration. Therefore, the resulting round-off error in the
numerical solution is only what arises in the computation for the final iteration.
However, the solution is not exact but is obtained to an arbitrary, specified,
convergence criterion.
Convergence Criteria for the GS Method
Typical convergence criteria used are

 xi( p  1)  xi( p )   for i = 1, 2, . . . , N

xi( p  1)  xi( p )
   for i = 1, 2, . . . , N
xi( p )

where is a very small number, e.g. 0.01, 0.001, 0.00001. Criterion 2 is applicable if
an estimate of the magnitude of the unknowns xi is not available and none of the
unknowns is expected to be zero.
Conditions for Convergence in the GS Method : Scarborough Criterion
Convergence is guaranteed for linear systems if
N
aii   | aij | for all i,
j  1, j  i

N
and if aii   | aij | for at least one i,
j  1, j  i

that is, when the system is diagonally dominant. This is also known as the
Scarborough criterion. This is a sufficient condition, which means that
convergence may still be possible even if the above condition is not satisfied.
Fortunately, it turns out that in fluid flow and heat transfer problems;
finite-difference formulation indeed leads to a diagonally dominant coefficient
matrix, which is the reason why for large systems the Gauss-Seidel method is so
widely used.
Application of the GS Iterative Method
In order to demonstrate the iteration process, the following system of three linear
equations is solved by the GS iterative method using a pocket calculator :
10 x1  x2  2 x3  44

2 x1  10 x2  x3  51

x1  2 x2  10 x3  61 . . . (5.99)
Clearly, the coefficient matrix in Eq. (5.99) is diagonally dominant because
70
 10  1  2 Numerical Methods to
Solve Heat Conduction
Problems
10  2  1

10  1  2

Therefore, the Scarborough criterion is satisfied and hence, one is certain to get a
converged solution using the GS iterative method. As a first guess, let us take

 x1 , x2 , x3 0  0, 0, 0 
We take  = 0.02
1  1
Then, x1(1)  44  x2(0)  2x3(0)   [44  (0)  2 (0)]  4.40
10   10

1  1
x2(1)  51  2x1(1)  x3(0)   [51  2 (4.40)  0]  4.22
10   10
1  1
x3(1)  61  x1(1)   2x2(1)  [61  4.40  2 (4.22)]  4.81
10   10
Now, check for convergence after the first iteration :
0  4.40  4.40  

0  4.22  4.22  

0  4.81  4.81  

We find that there is no convergence. One more iteration gives

 x1 , x2 , x3 2  3.01, 4.01, 4.99 

Again check the convergence after the second iteration :

3.01  4.04  1.39  

4.01  4.22  0.21  

4.81  4.99  0.18  

Still there is no convergence. One more iteration gives

 x1 , x2 , x3 3  3.00, 4.00, 5.00 

We again check for convergence after the third iteration :

3.01  3.00  0.01  

4.01  4.00  0.01  

4.99  5.00  0.01  

We see that convergence is reached. Hence, further computation stops. Therefore,

it required three iterations to obtain a converged solution. Incidentally, the
converged solution is also the exact solution of this set of equations. The reason is
that the number of unknowns is very small in this case.

71
Conduction Relaxation : Over-Relaxation and Under-Relaxation
One of the problems with the GS method is that it is relatively slow to converge to
the solution. The rate of convergence can often be improved by the relaxation
method which is explained next.
Let us consider the example that has been solved by the GS method.

1
10

44  x2(0)  2 x3(0)  x1(0) 
x1(1)  x1(0)  . . . (5.100)
change produced by the current iteration

Now the change produced by the current iteration can be increased if we multiply
it by a factor  (  1) . However,  also has an upper limit. For   2 , the change
is so great that instead of convergence, divergence occurs, that is, the solution
never converges. Therefore, Eq. (5.100) can now be written as

1
x1(1)  x1(0)   
10
 

44  x2(0)  2x3(0)  x1(0) 



or x1(1)  (44  x2(0)  2 x3(0) )  x1(0) 1    . . . (5.101)
10
From Eq. (5.101), it is readily seen that for

  0, x1(1)  x1(0) (no progress)

  1, x1(1)  x1(1) (basic GS iteration)

GS

0 <  < 1, under-relaxation  interpolation between x1(0) and x1(1)

GS

1 <  < 2  over-relaxation  extrapolation beyond x1(1)

GS

In a compact form, the relaxation method may be written as

 i 1 N 
  ci   aij x (j p  1)   aij x j( p ) 
    (1   ) x ( p ) . . . (5.102)
j 1 j j 1
xi( p  1) i
aii
for i = 1, 2, . . . , N
Successive under-relaxation (SUR) or under-relaxation is generally used for non-
linear equations and for systems that result in a divergent GS iteration. Successive
over-relaxation is widely used for accelerating convergence in linear systems.
Optimum Relaxation Factor opt
The question is : What value of the over-relaxation factor should be used? There is
no set rule to determine this. One has to do numerical experimentation to find out
the relaxation factor which gives the highest rate of convergence. This is called the
optimum relaxation factor opt which lies between 1 and 2 and varies from one
problem to another. For the simple case of Laplace’s equation ( 2 T  0) in a
square with Dirichlet boundary conditions (that is, known temperature on the
boundaries), Young (1954) and Frankel (1950) show that opt equals the smaller
root of

t 2  2  16  16  0 . . . (5.103)

72
 Numerical Methods to
 
with t  2cos   , where n is the total number of increments into which the side Solve Heat Conduction
n Problems
of the square is divided. In other words, n is the number of grid spacing. The
number of iterations required for a given convergence criterion falls very rapidly
when the parameter is in the immediate vicinity of opt, and it is generally better to
overestimate opt than to underestimate it (Carnahan, et al. 1969).
Solution of Eq. (5.80) by all Three methods
We shall now solve Eq. (5.80) having a tridiagonal coefficient matrix by the
Gaussian elimination, TDMA, and Gauss-Seidel iterative method and choose the
proper method.
Recall Eq. (5.80) is given below :
 D 1   1   1
 1 D 1    0
   2   
 1 D 1  3   0
    
 2 D    4   0

where D  2  (mL ) 2 (X ) 2 . Let us consider a fin with mL = 2. Also,

1
X   0.25 . Therefore, D  2  (2) 2 (0.25) 2  2.25 . Hence, Eq. (5.80) now
4
becomes
 2.25 1   1  1 
 1 2.25 1     0
   2    . . . (5.104)
 1 2.25 1   3   0
     
 2 2.25  4   0
Solution by Gaussian Elimination
Step I
The pivot is 2.25. So, first eliminate all terms below the pivot.
 2.25 1   1   1 
 1.81 1     0.444
R1    2   
 R2
2.25  1 2.25 1   3   0 
     
 2 2.25  4   0 
Step II
Now, the pivot is 1.81. So eliminate all terms below the pivot.
 2.25 1   1   1 
 1.81 1     0.444
R2   2 
 R3   
1.81  1.7 1   3   0.246
     
 2 2.25  4   0 
Step III
Now, the pivot is 1.7. So eliminate all terms below the pivot.
 2.25 1   1   1 
 1.81 1     0.444
R3  1.176  R 4    2   
 1.7 1   3   0.246
     
 1.07  4   0.290
73
Conduction Step IV
The last step as now the triangular coefficient matrix is produced. R1, R2,
R3, R4 are used to denote the first, second, third, and fourth rows,
respectively. The unknowns are obtained by back substitution.
Back Substitution
0.290
4   0.271
1.07
1.7 3  4  0.246  3  0.304

1.812  3  0.444  2  0.413

2.25 1  2  1  1  0.628
Therefore, the unknown temperatures are
1  0.628, 2  0.413, 3  0.304, 4  0.271
The total number of arithmetic operations (multiplications and divisions) to obtain
the solution is 13 (9 for elimination and 4 for back substitution).
Solution by TDMA
Recall the tridiagonal matrix algorithm given by Eqs. (5.94) and (5.97). With
respect to Eq. (5.104),
d1  1, d 2  0, d3  0, d 4  0

b1  2.25, b2  2.25, b3  2.25, b4  2.25

a2   1, a3   1, a4   2

c1   1, c2   1, c3   1
From the TDMA,
4   4 ,

1  b1  2.25

d1 1
1    0.444
1 2.25

a c   (1) (  1) 
2  b2   2 1   2.25     1.805
 1   2.25 

a c   (  1) (  1) 
3  b3   3 2   2.25     1.695
 2   1.805 

a c   (  2) (  1) 
4  b3   4 3   2.25   1.69   1.07
 3   

( d 2  a2 1 ) [0  (  1) (0.444)]
2    0.246
2 1.805

( d 2  a3  2 ) [0  (  1) (0.246)]
3    0.145
3 1.695

( d 4  a4  3 ) [0  (  2) (0.145)]
4    0.271
4 1.07

4   4  0.271
74
 c3 4 [(  1) (0.271)] Numerical Methods to
3   3   0.145   0.304 Solve Heat Conduction
3 1.695 Problems

c2 3 [(  1) (0.304)]
2   2   0.246   0.414
2 1.805

c1 2 [(  1) (0.414)]
1  1   0.444   0.628
1 2.25

Therefore, 1  0.628, 2  0.414, 3  0.304, 4  0.271

The total number of arithmetic operations required to obtain the solution is 10.
Solution by the Gauss-Seidel Iteration

Let [1 , 2 , 3 , 4 ]0  [1, 1, 1, 1] and  = 0.001.

Iteration 1 : [1 , 2 , 3 , 4 ]1  0.888, 0.839, 0.817, 0.726 

Iteration 2 : [1 , 2 , 3 , 4 ] 2  0.817, 0.726, 0.645, 0.573 

Iteration 3 : [1 , 2 , 3 , 4 ]3  0.767, 0.627, 0.6, 0.533 

Iteration 4 : [1 , 2 , 3 , 4 ] 4  0.723, 0.588, 0.498, 0.442 

Iteration 5 : [1 , 2 , 3 , 4 ]5  0.705, 0.535, 0.434, 0.385 

Iteration 6 : [1 , 2 , 3 , 4 ]6  0.682, 0.496, 0.391, 0.347 

Iteration 7 : [1 , 2 , 3 , 4 ]7  0.664, 0.468, 0.362, 0.321 

Iteration 8 : [1 , 2 , 3 , 4 ]8  0.652, 0.45, 0.342, 0.304 

Iteration 9 : [1 , 2 , 3 , 4 ]9  0.664, 0.438, 0.329, 0.292 

Iteration 10 : [1 , 2 , 3 , 4 ]10  0.639, 0.43, 0.32, 0.284 

Iteration 11 : [1 , 2 , 3 , 4 ]11  0.635, 0.424, 0.314, 0.279 

Iteration 12 : [1 , 2 , 3 , 4 ]12  0.632, 0.42, 0.31, 0.275 

Iteration 13 : [1 , 2 , 3 , 4 ]13  0.631, 0.418, 0.308, 0.273 

Iteration 14 : [1 , 2 , 3 , 4 ]14  0.63, 0.416, 0.306, 0.272 

Iteration 15 : [1 , 2 , 3 , 4 ]15  0.629, 0.415, 0.305, 0.271 

Convergence is reached on the 15th iteration. Hence, further computation stops.

Total number of iterations required = 15
Total number of arithmetic operations in each iteration = 5
Therefore, total number of arithmetic operations = 75
The Number of Arithmetic Operations for each Method : A Comparison
As can be seen from Table 5.1, it is obvious that TDMA is the fastest method and
the Gauss-Seidel iteration is the worst method for solving a tridiagonal system of
equations. Therefore, the choice falls on the TDMA for the solution of Eq. (5.80).
75
Conduction Table 5.1 : A Comparative Study of the Number of Arithmetic Operations
Needed to Solve Eq. (5.80) by Three Methods

Name of the Method Operations Number of Arithmetic

Gaussian elimination 13
Tridiagonal matrix algorithm (TDMA) 10
Gauss-Seidel iteration
(initial guess = [1, 1, 1, 1]  = 0.001) 75

Hence, for solving a tridiagonal system of linear equations, Thomas algorithm

(TDMA) is preferred.
Checking for Accuracy
The accuracy of a numerical solution is usually checked in one of the three ways :
 Comparison with the analytical solution: For most practical problems
analytical solutions do not exist. But, this is a good way of checking
the accuracy of a new numerical method.
 Comparison with the limiting case analytical solution: This is possible
when the analytical solution for some limiting value of a parameter
governing the solution is available.
 Comparison with experimental results: This is most desirable for
complex problems, such as turbulence, combustion, non-Newtonian
fluid flow, and heat transfer, which require many assumptions for the
purpose of modeling.
Comparison of the Present Numerical Result with the Corresponding
Analytical Solution
For the present problem of heat conduction in a fin, an analytical solution is
available. Therefore, a comparison of the numerical results with the exact solution
(for mL = 2) will enable us to obtain an estimate of the numerical error. Table 5.2
gives a comparison of the numerical and analytical solutions.
Table 5.2 : A Comparison of the Numerical and Analytical Solutions
of the Fin Problem
Temperature  Absolute
 cosh mL (1  X )  Percent Error
Location X    with Respect to
 cosh mL  the Exact
Numerical Solution

0 a
1 1
0
0.25 0.628 0.625
0.48
0.50 0.414 0.410
0.97
0.75 0.304 0.2995
1.50
1 0.271 0.266
1.88

Dirichlet boundary condition and hence not computed Table 5.2 clearly reveals
that X = 0.25 is not good enough and the grid spacing needs to be finer. In other
words, a higher number of grid points are necessary to obtain a more accurate
solution. However, one has to be also careful in increasing the number of grid
76 points as this will result in a higher round-off error. Therefore, a grid independence
 test, which gives an optimum X, is called for. A point to note is that even with a Numerical Methods to
Solve Heat Conduction
relatively coarse grid the accuracy is quite good. This means that with a slight Problems
decrease in the grid spacings, the numerical solution will be even closer to its
analytical counterpart. Another interesting feature of Table 5.2 is the gradually
increasing error for increasing X. This is possible because of the fact that at the left
boundary (X = 0) the Dirichlet condition is imposed and, therefore, for both
numerical and exact solutions the same temperature is used for the calculation of
temperature at X = 0.25. Hence, the temperature of the grid point closest to the left
boundary (i.e. at X = 0.25) computed by the numerical method is most accurate
and the error accumulates as the distance of a grid point with respect to the left
boundary increases.
Convective Boundary Condition
If the tip of the fin was convective instead of insulated, the discretisation equation
at i = M would have to be modified. The dimensionless boundary condition at the
fin tip in the changed scenario would be written in the mathematical form as
d  he L
 0 . . . (5.105)
dX k
where he is the convection heat transfer coefficient from the tip of the fin to the
surroundings.
Using the image-point technique as discussed earlier and the central-difference
d
scheme for discretisation of at i = M, Eq. (5.105) is expressed as
dX
M 1  M 1 he L
 M  0 . . . (5.106)
2 X k
2he L  X
or M 1  M 1  M . . . (5.107)
k
Substituting the expression for  M 1 from Eq. (5.107) into Eq. (5.74), we obtain

 2h L X 
2 M 1  M  D  e 0 . . . (5.108)
 k 

Therefore, only the last equation is changed. The method of solution remains the
same as before. To check the accuracy of Eq. (5.107), substituting he = 0
(corresponding to the insulation condition), we obtain
2 M 1  D M  0 . . . (5.109)

Which is the same as that obtained for the insulated tip.

5.7.2 Numerical Methods for Two-Dimensional Steady-State Problems
Consider the case of steady heat conduction in a long square slab (2L  2L) in which heat
qW
is generated at a uniform rate of . The problem can be assumed to be a
m3
two-dimensional one as the dimension of the slab is much longer in the direction normal
to the cross-sectional plane. Therefore, the end efforts can be neglected. All four sides are
maintained at T = T, temperature of the surrounding fluid, assuming a large heat transfer
coefficient.

77
 T = T
Conduction

2L

T = T q”’ T = T

T = T
2L

Figure 5.11 : Physical Domain of the Slab with Square Cross Section (2L  2L)

Consideration of Symmetry
A close look at the physics of the problem reveals that the problem is
geometrically and thermally symmetric. Therefore, from the temperature
distribution in any quarter of the physical domain, by mirror-imaging one can get
the solution for the entire region. Figure 5.12 shows the computational domain (top
right-hand quarter). The use of symmetry enables the numerical analyst to obtain
the solution much faster as the number of grid points is greatly reduced.
y

T = T
L

T
=0 q”’ T = T
x

x
0 L

T
=0
y
Figure 5.12 : Computational Domain (Top Right Hand Quarter) Considering Symmetry
Governing Differential Equation
The governing non-dimensional energy equation (assuming constant k) is
 2  2
  1 0 . . . (5.110)
X 2 Y 2
T  T x y
where  , X  ,Y 
2
 q L  L L
 
 k 
Boundary Conditions
The non-dimensional boundary conditions are as follows :

At X = 0, 0 . . . (5.111(a))
X
At X = 1,   0 . . . (5.111b))
78
  Numerical Methods to
At Y = 0, 0 . . . (5.111(c)) Solve Heat Conduction
Y Problems
At Y = 0,   0 . . . (5.111(d))
Discretisation
The computational domain including the notations for the interior grid points is
shown in Figure 5.13.
=0

yN

(I, j + 1)

X Y

 =0
=0 (I, 1 – j) (I, 1) (I + 1, j)
x

(I, j – 1)

y1 = 0
X1 = 0 XM

=0
y
Figure 5.13 : Interior Grid Points in the Computational Domain

 2  2
Eq. (5.110) is discretised using the central difference for and at the
dX 2 Y 2
interior grid point (i, j) as follows :
i  1, j  2i , j  i  1, j i  1, j  2i , j  i  1, j
2
 2
 1 0 . . . (5.112)
( X ) ( Y )
Taking  X   Y , Eq. (5.112) reduces to

 i  1, j  i , j 1  4i , j  i , j 1  i  1, j  ( X ) 2 . . . (5.113)
Boundary Condition along X = 0
Using the image-point technique,
i  1, j  i  1, j . . . (5.114)
and setting i = 0, Eq. (5.112) becomes
 2, j  i, j 1  4i, j  i, j 1  ( X ) 2 . . . (5.115)

Boundary Condition along Y = 0

Using the image-point technique
i , j 1  i , j 1 . . . (5.116)
and setting j = 1, Eq. (5.112) becomes
 i  1, 1  2i , 2  4i , 1  i  1, 1  ( X ) 2 . . . (5.117)
Handling of Corner Points
Corner points need special attention because they belong to both horizontal and
vertical surfaces. Therefore, the boundary conditions at both the surfaces apply
there. However, if one or both of the surfaces have the Dirichlet condition
(specified temperature), then there is no problem because the corner point can be
assumed to have a specified temperature. But, if both surfaces have Neumann
(insulation) and/or Robbins (convective) conditions, then the corner point needs to
be handled separately since both conditions exist there. With respect to the present
problem, out of the four corner points, only the bottom-left corner points is 79
Conduction exposed to Neumann conditions in X and Y directions. Other three have either one
or both surfaces exposed to the Dirichlet condition. Figure 5.14 shows the image
points for the bottom-left-hand corner represented by the grid point (1, 1).

(1.2)

(2.1)
(0.1) (1.1)

(1.0)

Figure 5.14 : Image Points for the Bottom Left Hand Corner Point (1, 1)
Using the image-point technique,
0, 1  2, 1 . . . (5.118)

1, 0  1, 2 . . . (5.119)

  2  0, 1  21, 1  2, 1

Now,  2  . . . (5.120)
 X 1, 1 ( X )2

  2  1, 0  21, 1  1, 2

 2   . . . (5.121)
 Y 1, 1 ( Y )2

Substituting Eq. (5.118) into Eq. (5.120), and Eq. (5.119) into Eq. (5.121). We get
  2  22, 1  21, 1
 2  . . . (5.122)
 X 1, 1 ( X )2

  2  21, 2  21, 1
 2   . . . (5.123)
 Y 1, 1 ( Y )2

Setting i = 1, j = 1 and substituting Eqs. (5.122) and (5.123) into Eq. (5.112), we
obtain for  X   Y ,

22, 1  41, 1  21, 2  (  X ) 2  0 . . . (5.124)

Methods of Solution
1
Let us consider an example in which  X  . The grid points that are unlebelled
4
(Figure 5.15) are all at temperature  = 0 as imposed by the boundary condition.
Thus, there are 16 unknown temperature to find (Figure 5.16).

80
 y
Numerical Methods to
 = 0 (known)
Solve Heat Conduction
Problems
1

(1, 4) (2, 4) (3, 4) (4, 4)

(1, 3) (2, 3) (3, 3) (4, 3)

 = 0 (known)

(1, 2) (2, 2) (3, 2) (4, 2)

(1, 1) (2, 1) (3, 1) (4, 1)

0 x
1
0
Figure 5.15 : Labelling of Grid Points using Double Subscript Notation

 = 0 (known)

4 8 12 16
Line of symmetry

3 7 11 15
 = 0 (known)

2 6 10 14

1 5 9 13
0
1
0
Figure 5.16 : Labelling of the Same Grid Points using Single Subscript Notation
Since there are 16 unknowns, there will be 16 equations to solve. Using the matrix
representation for these equations, Eq. (5.124(a)) is obtained from
Eqs. (5.113)-(5.117) and Eq. (5.124)
4 2 0 0 2   1  1 / 16
 1 4 1 0 0 2   2  1 / 16
0 1 4 1 0 0 2
    1 / 16
   3 
0 0 1 4 0 0 0 2   4  1 / 16
 1 0 0 0 4 2 0 0 1   5  1 / 16
 1 0 0 1 4 1 0 0 1     1 / 16
   6 
 1 0 0 1 4 1 0 0 1   7  1 / 16
 1 0 0 0 4 0 0 0 1   8  1 / 16
 1 0 0 0 4 2 0 0 1   9  1 / 16
 1 0 0 1 4 1 0 0 1
   1 / 16
   10   
 1 0 0 1 4 1 0 0 1   11  1 / 16
 1 0 0 1 4 0 0 0 1  12  1 / 16
 1 0 0 0 4 2 0 0   13  1 / 16
   
 1 0 0 1 4 1 0  14  1 / 16
 1 0 0 1 4 1  15  1 / 16
 1 0 0 1 4  16  1 / 16

Equation (5.124(a))
A close look at Eq. (5.124(a)) reveals the following :
 The division of X and Y into relatively coarse subdivisions leads to
many equations (4 × 4 = 16). In a practical situation, the number of
equations may be hundred or more.
81
Conduction  The coefficient matrix is banded, which means that the non-zero
components only appear in a band on either side of the main diagonal.
There are 9 diagonals (Figure 5.17). So, the bandwidth is large as
compared to TDM. The advantage of having a banded matrix (this is
true also for TDM) is that special sub-routines can be written to solve
the problem in less computer time than if the matrix was filled with
non-zero components.

Bandwidth

Figure 5.17 : Pictorial Representation of the Banded Coefficient Matrix Showing Nine Diagonals

 The zero matrix components outside the band need not be stored in
the computer. This is of great significance in large problems where
the computer memory size becomes a limiting factor.
Choice of the Proper Method
Eq. (5.124(a)) can be solved in two ways :
(a) By Gaussian elimination, and
(b) By Gauss-Seidel iteration.
Let us weight the pros and cons of both methods before we make our final choice.
It is interesting to note that the banded coefficient matrix in Eq. (5.124(a)) has
124 components within the band rather than the 256 spaces that would have been
required to store the entire matrix. One could even reduce the bandwidth by
recognizing the physical and geometrical symmetry across one of the diagonals of
the square as shown in Figure 5.17. This means that i , j   j , i or 2  5 ,
3  9 , 4  13 , and so on. This would reduce the number of equations from
16 to 10.
Furthermore, it may be noted that many of the components within the band itself
are zero. In this case, 60 of 124 band components are zero. These components
must still be stored, however, if Gaussian elimination is to be used, because during
the elimination process they will, in general, change to non-zero values. If
computer storage is critical, one might prefer to use a method that does not require
storing these zero component in the band. The Gauss-Seidel iteration method is
one way of doing this. In addition to this, the round-off error is minimum in the
Gauss-Seidel method. Therefore, in view of the aforesaid two overwhelming
merits, in spite of the clear-cut advantage of Gaussian elimination because of its
non-iterative nature, the GS method is chosen to solve Eq. (5.124(a)).
Check for Accuracy
For the present problem, the accuracy of the numerical results can be checked by
comparing it with the corresponding analytical solution available. Subsequently, a
grid independence test must be done to obtain the desired results. The analytical
(or exact) solution in the dimensionless form is given below:

82
  (  1) n cosh  Y Numerical Methods to
1
 ( X , Y )  [1  X 2 ]  2  n
. . . (5.125) Solve Heat Conduction
2 n0  3n cosh  n Problems

(2n  1) 
where  n  , where n = 0, 1, 2, . . .
2

5.8 TRANSIENT ONE-DIMENSIONAL PROBLEMS

Consider a hot infinite plate (Figure5.18) of finite thickness 2L. The plate is suddenly
exposed to a cool fluid at T. The initial temperature of the plate is Ti (Ti  T ) . The heat
transfer coefficient is large. We wish to find the temperature of the plate as a function of
space and time using a numerical method.

T = T T(x,o) = T, T = T
K,p,c
Constant

2L
Figure 5.18 : Physical Domain of the One-dimensional Transient Conduction in an Infinite Plane Slab
The problem can be modeled as a one-dimensional, unsteady-state problem
T T
because   0 as the plate is infinitely long in y- and z-directions.
y z
Consideration of Symmetry
Since the problem is a thermally and geometrically symmetric one, only one half
of the plate can be taken as the computational domain with the insulation boundary
condition at x = L (Figure 5.19).

T = T

T(x,o) = T,
T
0
x

L
x

Figure 5.19 : Computational Domain of the One Dimensional Transient Conduction Problem
Governing Differential Equation
For constant thermophysical properties k, , c the non-dimensional energy
equation for the plate is

  2
 . . . (5.126)
 X 2 83
Conduction T  T x at
where  , X  ,t  2
Ti  T L L
Initial and Boundary Conditions
The Initial and boundary conditions are as follows :
IC
At  = 0,  = 1 for all x . . . (5.127(a))
For  > 0, BC 1
At X = 0,  = 0 . . . (5.127(b))
BC 2

At X = 1, 0 . . . (5.127(c))
X
Discretisation
For any interior grid point, the finite difference formulation give,

i i  1  2i  i  1
 . . . (5.128)
 ( x)2
For i = 1, . . . , M
The equation for X = 1 is obtained by using the image-point technique, i.e. by
substituting  M  1   M  1 in Eq. (5.128) for i = M. Therefore,

 M 2 M  1  2 M
 . . . (5.129)
 ( X ) 2

For the sake of demonstration, let us take four equal subdivisions in the X-direction
1
(Figure 5.20). Therefore,  X  .
4
y

(0) (1) (2) (3) (4)

x
0 1 1 3 1

4 2 4
Figure 5.20 : Equally Spaced Grid points in the x-direction of the Computational Domain

At i = 0,  = 0 (known)
At i = 1, 2, 3, . . .
From Eq. (5.128), we obtain
d 1 1 1
 2
( 0  21   2 )  ( 21   2 ) . . . (5.130)
d ( X ) ( X ) 2

d 2 1
 (1  2   3 ) . . . (5.131)
d ( X ) 2
84
 d 3 1 Numerical Methods to
 ( 2  2 3   4 ) . . . (5.132) Solve Heat Conduction
d ( X ) 2 Problems

At i = 4, from Eq. (5.129), we obtain

d 4 1
 (2 3  2 4 ) . . . (5.133)
d ( X ) 2

Thus, we have four simultaneous ordinary differential equations [Eqs. 5.130-5.131]

to solve. This system of ordinary differential equations may be classified as initial
value problems. This is because these equations are to be solved for the unknowns
as a function of time, beginning with an initial value for each of the unknowns. In
this case the initial values are obtained from the initial temperature distribution in
the plate, which is given by
1 (0)  2 (0)  3 (0) 4 (0)  1 . . . (5.134)

Methods of Solution
There are three methods by which this initial- value problem can be solved. These
are the
(a) Euler (or explicit),
(b) Crank-Nicholson, and
(c) Pure implicit methods.
The Euler Method (also known as the Explicit Method)
Since the given problem is an initial-value one, we will know the solution p and
will seek p + 1 at some later point in time  p  1   p   . In the Euler method of
solution, the solution at a future time p + 1 is obtained by computing the derivative
at the present time p and then by moving ahead in time in the following way :
p
p 1 d
p
    . . . (5.135)
d
The Euler scheme is pictorially represented in Figure 5.21 for the grid points 1, 2,
3, 4. Eq. (5.135) can be written as
p
d 1
1p  1  1p   . . . (5.136(a))
d
p
d 2
2p  1  2p   . . . (5.136(b))
d
p
d 3
3p  1  3p   . . . (5.136(c))
d
p
d 4
4p  1  4p   . . . (5.136(d))
d

85
Conduction

Obtained from earlier computations

beginning with the condition at  = 0

Linear extrapolation
p of slope at p
p+1



 p p+1
Figure 5.21 : Pictorial Representation of the Euler Method
Substituting Eq. (5.136(a)) into Eq. (5.130), Eq. (5.136(b)) into Eq. (5.131),
Eq. (5.136(c)) into Eq. (5.132), Eq. (5.136(d)) into Eq. (5.133), we obtain
1p  1  1p 1
 ( 21p   2p ) . . . (5.137)
 ( X ) 2

2p  1  2p 1
 2
(1p  21p   3p ) . . . (5.138)
X ( X )

3p  1  3p 1
 2
( 2p  2 3p   4p ) . . . (5.139)
 ( X )

4p  1  4p 1
 2
(2 3p  2 4p ) . . . (5.140)
 ( X )
Eqs. (5.137)-(5.140) are then rearranged to give
 2  p 
1p  1   1  2
1  p . . . (5.141)
 
( X )  ( X ) 2 2


  2  p 
2p  1  1p   1   2  p . . . (5.142)
( X ) 2  ( X ) 2 (  X ) 2 3
 

  2  p 
3p  1  2
 p
2   1   3  p
2 4
. . . (5.143)
( X )  ( X ) 2  ( X )

2  2  p
4p  1  3p   1   4 . . . (5.144)
( X ) 2  ( X ) 2
 
Eqs. (5.141)-(5.144) can be written in the following matrix form :
p 1 p
 1  1  2 r r   1 
   4 1  2r r   
 2     2 . . . (5.145)
 3   r 1  2r r   3 
    
 4   2r 1  2r   4 


where r  .
( x ) 2

86
 The TDM on the right-hand side of Eq. (5.145) is known (and constant) once the Numerical Methods to
Solve Heat Conduction
size of the time step  is chosen. The known value of  at p (that is p) are Problems
multiplied by this TDM to obtain the new values of  at p + 1. This matrix
multiplication is quite easy to carry out on the computer since only the non-zero
term will contribute to the calculation. Thus, p + 1 values are obtained explicitly in
terms of p values and hence the name, explicit method. Note that the solution of
simultaneous algebraic equation is not necessary in this scheme, which makes it a
very attractive method. Once p + 1 are obtained, they are stored in p, and the
computation is repeated for the next time step. This procedure continues until the
result at the desired time is obtained or till the steady state is reached. However, a
major drawback of this method is that for r > 0.5, that is when 1 – 2r is negative,
the solution become unstable. Therefore, a stability limit of r  0.5, is imposed,
which result in considerable restriction on the time step  for a particular value
of X.
The Crank-Nicholson Method
In the Euler method the value of the derivative at the beginning of the time interval
was used to progress in time. A more accurate method would be to use the
arithmetic mean value of the derivatives at the beginning and at the end of the time
1
interval, i.e. use the time derivative at p  , a time which is midway between
2
p and p + 1. Therefore,

 d p p  1
p 1 p 1 d
       . . . (5.146)
2  d  d 

Substituting Eqs. (5.130)-(5.133) in Eq. (5.146), we obtain

1p  1  1p 1  ( 21p   2p )  ( 21p  1   2p  1) 

 . . . (5.147)
 2 (  x) 
2 

2p  1  2p 1  (1p  2 2p   3p )  (1p  1  2 2p  1   3p  1) 

 . . . (5.148)
 2 (  x) 
2 

3p  1  3p 1
  ( p  2 3p   4p )  ( 2p  1  2 3p  1   4p  1)  . . . (5.149)
 2  2 
2 (  x)

4p  1  4p 1  (2 3p  2 4p )  (2 3p  1  2 4p  1) 

 . . . (5.150)
 2 (  x) 2  

Eqs. (5.147)-(5.150) are rearranged, which results in a set of four simultaneous

algebraic equations in 1p  1 , 2p  1 , 3p  1 , 4p  1 , represented in the matrix form
as
p
 1  r r / 2   1 
 r / 2 1  r r / 2   
   2
  r / 2 1  r  r / 2  3 
   
 r 1 r   4 
p
 1 r r / 2    1 
   

 r / 2 1 r r / 2    2 . . . (5.151)
  r / 2 1 r r / 2   3 
   
 r 1  r   4 
87
Conduction Similar to the case in the Euler method, the right-hand side can be computed
directly because all the components are known. This results in a column matrix as
before. The difference arises in the fact that this does not give an explicit results
for the unknowns on the left-hand side; rather an implicit TDM system of algebraic
equations results. This system of algebraic equations must then be solved in each
time step. The Crank-Nicholson method, although a stable method, gives
erroneous results in the early time if the time step is too large. However, the error
damps out as time progresses towards the steady state.
The Pure Implicit Method
In contrast with the Euler or the Crank-Nicholson method, in the pure implicit
scheme, the time derivative at the new time is used to move ahead in time. Thus,
p 1
d
p  1  p   . . . (5.152)
d
From Eqs. (5.130)-(5.133) and Eq. (5.152), we obtain
1p  1  1p 1
 [( 21p  1   2p  1)] . . . (5.153)
 ( X ) 2

2p  1  2p 1
 2
[(1p  1  2 2p  1   3p  1 )] . . . (5.154)
 ( X )

3p  1  3p 1
 [( 2p  1  2 3p  1   4p  1 )] . . . (5.155)
 ( X ) 2

4p  1  4p 1
 2
[(2 3p  1  2 4p  1 )] . . . (5.156)
 ( X )
p 1 p
1  2 r r   1   1 
  r 1  2r r     
   2   2 . . . (5.157)
  r 1  2r r   3   3 
     
 2 r 1  2 r   4   4 
Eq. (5.157) is an implicit set of equations to solve for the new temperatures at each
time step. The pure implicit scheme is an unconditionally stable scheme, that is,
there is no restriction on the time step, which is in sharp contrast with the Euler
and the Crank-Nicholson method. However, the Euler and the pure implicit
methods have the same order of accuracy, while the Crank-Nicholson method is
more accurate than either of the two for the same time step. The accuracy and
stability of each of the three methods are detailed in the sections to follow.
Accuracy of the Euler, Crank-Nicholson, and Pure Implicit Methods
d
In the Euler method, at any grid point i, is evaluated at p, that is,
d
p
d ip  1  ip
 . . . (5.158)
d i 
Eq. (5.158) is forward difference in time. Therefore, the order of accuracy in time
d
is 0 ( ). In the pure implicit method, is evaluated at p + 1, that is,
d
p 1
d ip  1  ip
 . . . (5.159)
d i 

88 Eq. (5.159) actually arises from Eq. (5.160) shown below

 p 1 Numerical Methods to
d  p  1  i( p  1)  1 Solve Heat Conduction
 i . . . (5.160)
d  Problems
i

Although the RHS of Eq. (5.159) looks the same as that of Eq. (5.158), the former
is actually backward difference in time, which is obvious from Eq. (5.160).
Therefore, the order of accuracy in time is 0 ( ). In the Crank-Nicholson method,
d 1
is evaluated at p  , that is,
d 2
p 1
d 2 ip  1  ip
 . . . (5.161)
d i   
2 
 2 
p 1
d 2 ip  1  ip
 . . . (5.162)
d i 
Again, although the RHS of Equation 5.161 looks the same as that of Eqs. (5.158)
and (5.159), the former is actually central difference in time, which is obvious
from Eq. (5.160). Therefore, the order of accuracy in time is 0 (  ) 2 . This
explains why the Crank-Nicholson scheme is one order more accurate in time as
compared to the Euler or pure implicit scheme.
In all the three methods, the space derivatives are discretised using the
central-difference scheme. Therefore, the order of accuracy in space in the Euler,
pure implicit, and Crank-Nicholson methods is 0 (  X ) 2 .
The Euler method, the Crank-Nicholson method, and the pure implicit method are
also called FTCS (forward-time, central space), CTCS (central-time, central space),
and BTCS (backward-time, central space), respectively.
To summarize, the order of accuracy of each method can be written as follows :
Euler or explicit : 0 [(  X ) 2 , ( )] FTCS

Crank-Nicholson : 0 [(  X ) 2 , ( ) 2 ] CTCS

Pure Implicit : 0 [(  X ) 2 , ( )] BTCS

Stability : Numerically Induced Oscillations
From the preceding discussion, it is apparent that all the three schemes will give
better results if time steps are made smaller. In practice, however, one would
usually like to take as large a time step as one can to reduce the computational
effort and time. In addition to decreasing the accuracy of the solution, large time
steps can introduce some unwanted, numerically induced oscillations into the
solution, making it physically unrealistic. Such solutions are not acceptable and the
method that produces such a solution is called unstable method. This brings us to
the formal definition of a stable numerical scheme, which is the one for which
errors from any source (round-off, truncation, mistakes, etc.) are not permitted to
grow in the sequence of numerical procedures as the calculation proceeds from one
step to the next.
The Case of One Grid Point
Consider the case of only one grid point, that is, the grid point on the insulation
boundary of a plate (Figure 5.22). Therefore, X  1, r   . Hence, we have
only one equation to solve, that is,
d 1
  21 . . . (5.163)
d

89
Conduction

Known
temperature

(O) (1)

Figure 5.22 : The Case of Only One Grid Point

Subjects to the initial conditions
1 (0)  1 . . . (5.164)
The analytical solution of Eq. (5.163) is
1  e  2 . . . (5.165)
The following three equations are obtained corresponding to the three numerical
schemes :
Euler
1p  1  (1  2 r ) 1p . . . (5.166)
Crank-Nicholson
(1  r ) 1p  1  (1  r ) 1p . . . (5.167)
Pure Implicit
(1  2 r ) 1p  1  1p . . . (5.168)
Each of these may be put in the following general form :
p  1  p . . . (5.169)
where  is defined by
Euler
  1  2r . . . (5.170)

Crank-Nicholson
1 r
 . . . (5.171)
1 r
Pure Implicit
1
 . . . (5.172)
1  2r
The value of  determines the character of the solution. This is self-explanatory
from Figure 5.23, which shows  as a function of r (r =   for this special case)
for each of the three numerical methods we have considered.
A close inspection of Figure 5.23 reveals that as r  0, that is, if the time step is
made smaller and smaller, all three schemes become identical. As the time step is
increased, in each case, the solutions begin to deviate from one another. The Euler
method can have steady decay, stable oscillations, or unstable oscillations. The
Crank-Nicholson method can have either steady decay or stable oscillations. The
pure implicit method has only a steadily decaying type of solution. From the graph,
it is also seen that the stability limit for the Euler method is 0.5 while that for the
Crank-Nicholson method is 1.0. The Euler method becomes totally unstable at r =
1.0. While the Euler method is called conditionally stable, the Crank-Nicholson
90
 method is called unconditionally stable because the oscillations ultimately damp Numerical Methods to
Solve Heat Conduction
out with time. The pure implicit method is truly unconditionally stable method.
Problems
Steady unbounded
Growth
+1

Pure implicit
Steady
decay

0
0.5 1 2 r

Crank–Nicholson Stable
oscillations
Euler
-1
Unstable
oscillation

Figure 5.23 : Stability Curves for the Case of One GrPd Point
Figure 5.24 compares the three numerical solutions to the corresponding exact
solution (drawn qualitatively) 1  e  2 for r = 1.2 which exceeds the stability
limit for both the Euler method and the Crank-Nicholson method.

r = 1.2 r = 1.2 r = 1.2

2 2 2
Euler Pure implicit
1 1 Crank-Nicholson 1
Exact Exact Exact
1 1 1

0 0 0
1 2 1 2 1 2
  
-1 -1 -1

-2 -2 -2

Figure 5.24 : Comparison of Numerical Solutions based on the Euler, Crank-Nicholson, and Pure
Implicit Methods with Corresponding Exact Solution for the case of One Grid Point

The figures reveals that oscillations in the Euler solution grow without bound, and
oscillations are seen in the Crank-Nicholson solution but gradually damp out for
large time. The pure implicit solution does not show any oscillations.
The Case of More than One Grid Point
The example of one grid point may be extended to the more general case in which
there are more than one grid points, that is, more than one equation. The matrix
representation for any of three numerical schemes can be written as

A p  1  B  p . . . (5.173)
where A and B are matrices that depend on the particular method. Eq. (5.102) can
be written as

 p  1  A 1 B  p . . . (5.174)
Note that in the right-hand side of Eq. (5.174) A – 1 B is a square matrix. We also
know that associated with every square matrix (let us call this matrix S) are a
special set of vectors, called eigenvectors, and a related set of scalars, called
eigenvalues. Formally, the vector x is an eigenvector of S if and only if x is a
non-zero vector and  is a scalar (which may be zero), such that
Sx =  x . . . (5.175) 91
Conduction The scalar  is an eigenvalue of S if and only if there exists a non-zero vector x
such that Eq. (5.175) holds. The eigenvalues  of the matrix A– 1 B play a similar
role to the  in the case of one grid point. If there are N simultaneous equations
being handled, there will be N eigenvalues of A– 1 B. These values will determine
the character of the solution. Now,

A 1 B  p   p . . . (5.176)

or ( A 1 B   I )  p  0 . . . (5.177)

where I is an N  N unite matrix. To get a non-trivial solution of Eq. (5.177),

det ( A 1 B   I )  0 . . . (5.178)

We may now multiply both side of Eq. (5.178) by det (A) to get

det ( A) det ( A  1 B   I )  0 . . . (5.179)

or det ( AA 1 B  A  I )  0 . . . (5.180)

or ( det ( B   A)  0 . . . (5.181)

There will be three general classes of solutions which will arise in this problem.
Case I
If all eigenvalues are between 0 and 1, there will be no oscillations. The
solution will gradually approach a steady-state value.
Case II
If one of the eigenvalues falls between 0 and – 1, numerically induced
oscillation will appear.
Case III
If one of the eigenvalues is less than – 1, the oscillations will be unstable.
An example of the Two-Grid-Point Case for the Euler Method
As an example, let us consider the two grid point case for Euler method. The
matrices A and B are then given by

1 0 
A 
0 1 

1  2 r r 
B
 2r 1  2r 

1  2 r r 
Then, B  A 
 2r 1  2r   

The determinant of the above matrix is given by

det ( B   A)  (1  2 r  ) 2  2 r 2  0 . . . (5.182)

Solving Eq. (5.182), we get

1  1  r (2  2)

 2  1  r (2  2)

92
 The value of 1 will determine the character of the solution since it is this value Numerical Methods to
Solve Heat Conduction
that is most likely to be negative (because of the larger coefficient or r). Problems
The  versus r plots (Figure 5.25) show the same general trend of Figure 5.23, but
the curves have shifted to the left so that the critical values of r are smaller than
those for the one-grid-point case. The upper limit for stable oscillations of the
2
Euler method is now  0.586 [since  crit   1  1  r (2  2) ] as
(2  2)
compared to 1.0 in the one grid point case.

+1

Pure implicit

1 0 r
1 1 2

Crank–Nicholson

Euler
-1 0.586

Figure 5.25 : Stability Curves for the case of Two Grid Points

SAQ 6
(a) Explain discretization and stability with respect to the numerical method of
solving a two-dimensional heat conduction problem.
(b) Discuss Gauss-Siedel iteration technique to determine temperature at a
nodal point in a two-dimensional solid.
(c) What is relaxation method? What do you mean by residuals?
(d) Briefly describe the Thomas algorithm.
Exercise 1
Consider the following one dimensional steady state conduction problem

d 2T ( x ) 1
2
 g0 in 0  x  L
dx k

dT ( x )
0 at x  0
dx
dT ( x )
k  hT (x )  0 at x  L
dx
Write the finite difference formulation of this heat transfer problem by dividing the
region 0 < x < L into four equal parts.

Exercise 2
In a parallel plate fuel element for a gas cooled reactor, the heat generation in the
fuel element has approximately a cosine distribution. The simplest steady state
model for the temperature distribution in the fuel element may be taken as

93
Conduction d 2T ( x ) 1 x
2
 g o cos 0 in 0  x  L
dx k 2L
dT
0 at x  0
dx
T  0 at x  L
where L is the half thickness of the fuel element. By dividing the region into four
equal parts, calculate the temperature distribution with finite difference for
k  12 W/m . oC , L  5  10  3 m , and g  6  108 W/m 2 . Compare the
numerical results with the exact solution.
Exercise 3
Consider a straight fin of rectangular cross section having thermal conductivity
k = 40 W/m . oC, length L = 30 cm, thickness t = 0.5 cm, and a large width
perpendicular to the plane of Figure 5.26. The base is at T0 = 130oC, and the fin tip
is regarded insulated. The fin dissipates heat by convection with a heat transfer
coefficient h = 400 W/m2 . oC into a fluid at T = 30oC.

T = 130o c Convection
Insulated

C = 0.5 cm

L = 3 cm
Convection

Figure 5.26
(a) By using a one-dimensional mesh of size x = 0.3 cm, calculate by
finite difference the temperature distribution along the fin.
(b) Estimate the heat transfer rate through the fin per 10 – cm width
perpendicular to the plane of the paper.
(c) Compare the numerical results with those obtained from the analytical
solution of the one-dimensional equation.
Exercise 4
By writing an energy balance on a differential volume element, derive the finite
difference form of the heat conduction equation
 2T  2T g
2
 2
 0
x y k
for the nodal point A in each of the accompanying Figure 5.27 for boundary
conditions indicated.
Convection into a
Convection into a
medium at T
T2 medium at T
T T2
A
A

T1
T1

Convection into a
Insulated medium at T

(a) (b)

94
 Numerical Methods to
Solve Heat Conduction
Problems
Insulated
Convection into a
medium at T
T2
T2
A
A
T
T1 Insulated T
T1 Insulated

(c) (d)
Figure 5.27
Exercise 5
Write the finite difference formulation for two dimensional, steady state heat
conduction with no heat generation for a square region of side L by using mesh
L
size x  y  for the boundary conditions shown in Figure 5.28. Also, express
3
the resulting finite difference equations in matrix form for the nine unknown node
temperatures Tm, m = 1 – 9.

T
 0.5T  0
y
L
T1 T2 T3

T T4 T5 T6
0
y
T = 20OC

T7 T8 T9

x
0 L
T = 200OC

Figure 5.28

Exercise 6
A large and very thick brick wall ( = 5  10– 7 m2/s) which is initially at a uniform
temperature Ti = 125oC is suddenly exposed to cooling by maintaining its surface
at x = 0 at T0 = 25oC. To calculate the temperature transients at depths small in
comparison to the thickness, the wall can be regarded as a semi infinite medium
confined to the region x  0. By using a suitable method and mesh size x = 0.3
cm, calculate the temperature at x = 1.2 cm from the surface t = 1 and 5 min after
the exposure.
Exercise 7
A marble slab (k = 2 W/m . oC,  = 1  10– 6 m2/s) that is L = 4 cm thick is initially
at a uniform temperature Ti = 200oC. Suddenly one of its surfaces is lowered to
0oC and is maintained at that temperature, while the other surface is kept insulated.
Develop an explicit finite difference scheme for the determination of the
temperature distribution in the slab as a function of position and time as well the
heat flux at the boundary surface.

95
Conduction
5.9 SUMMARY
In the present unit, different solution methods of conduction heat transfer problems are
discussed. Analytical method is restricted to simple geometry and boundary conditions.
Graphical method can be applied to some complex problems but its workability is
restricted. One of the best methods and most frequently used numerical scheme is the
finite difference method for solution of conduction heat transfer problems. In this method
the entire domain is divided into a grid and differential equations are rewritten as
difference equation. Depending upon the nature of the problem, various iterative methods
can be used. Accuracy and grid independence test are important for numerical problems
which are discussed in details. Solution for 1-D and 2-D conduction problems are done
by analytical as well as finite difference methods. You can also understand how finite
difference method can be applied to solve transient conduction problems. Computer
knowledge and computation skill are essential in solving the problems.

5.10 KEY WORDS

Error : Difference between the exact (true) value and
actual value.
Computational Domain : Space considered for computation.
Grid Independent : Final parameter under consideration is
independent of grid size.

5.11 ANSWERS TO SAQs

Please refer the relevant preceding text in this unit for answers to SAQs.

REFERENCES
J. P. Holman, (2002), Heat Transfer, Tata McGraw-Hill, New Delhi, 9th Edition.
M. N. Ozisik, (1985), Heat Transfer A Basic Approach, McGraw-Hill International
Edition.
P. S. Ghoshdastidar, (2004), Heat Transfer, Oxford University Press, New Delhi.

96
 Heat Transfer from
UNIT 6 HEAT TRANSFER FROM EXTENDED Extended Surfaces

SURFACES
Structure
6.1 Introduction
Objectives
6.2 Extended Surfaces
6.3 A General Conduction Analysis
6.4 Fins of Uniform Cross-sectional Area
6.4.1 Case A
6.4.2 Case B
6.4.3 Case C
6.4.4 Case D
6.5 Fin Performance
6.6 Fins of Non-uniform Cross-sectional Area
6.7 Overall Surface Efficiency
6.8 Summary
6.9 Key Words
6.10 Answers to SAQs

6.1 INTRODUCTION
The term extended surface is commonly used to depict an important special case
involving heat transfer by conduction within a solid and heat transfer by convection
(and/or radiation) from the boundaries of the solid. Until now, we have considered heat
transfer from the boundaries of a solid to be in the same direction as heat transfer by
conduction in the solid. In contrast, for an extended surface, the direction of heat transfer
from the boundaries is perpendicular to the principal direction of heat transfer in the solid.
Objectives
After studying this unit, you should be able to
 understand the wide application of extended surfaces in different heat
transfer devices,
 distinguish different types of extended surfaces,
 formulate the heat transfer equations with fins,
 appreciate need of an extended surface,
 evaluate heat transfer and temperature distribution for a system with fins,
and
 solve some problems on extended surfaces.

6.2 EXTENDED SURFACES

Consider a strut that connects two walls at different temperatures and across which there
is fluid flow (Figure 6.1). With T1 > T2, temperature gradients in the x-direction sustain
heat transfer by conduction in the strut. However, with T1 > T2 > T there is concurrent

99
Conduction heat transfer by convection to the fluid, causing qx, and hence the magnitude of the
dT
temperature gradient, , to decrease with increasing x.
dx
T2

qx, 2
L

T

h
qconv
a
Fluid

T1

O
T1 T(x) T2
qx, 1

T1 > T2 >T
Figure 6.1 : Combined Conduction and Convection in a Strut

Although there are many different situations that involve such combined
conduction-convection effects, the most frequent application is one in which an extended
surface is used specifically to enhance heat transfer between a solid and an adjoining
fluid. Such an extended surface is termed a fin.
Consider the plane wall of Figure 6.2(a). If Ts is fixed, there are two ways in which the
heat transfer rate may be increased. The convection coefficient h could be increased by
increasing the fluid velocity, and/or the fluid temperature T could be reduced. However,
there are many situations for which increasing h to the maximum

T

H T
Q = hA
(T -T)
h
T

T A
(a) (b)
Figure 6.2 : Use of Fins to Enhance Heat Transfer from a Plane Wall,
(a) Bare Surface, (b) Finned Surface

possible value is either insufficient to obtain the desired heat transfer rate or the
associated costs are prohibitive. Such costs are related to the blower or pump power
requirements needed to increase h through increased fluid motion. Moreover, the second
option of reducing T is often impractical. Examining Figure 6.2(b), however, we see that
there exists a third option. That is, the heat transfer rate may be increased by increasing
the surface area across which the convection occurs. This may be done by employing fins
that extend from the wall into the surrounding fluid. The thermal conductivity of the fin
material has a strong effect on the temperature distribution along the fin and therefore
influences the degree to which the heat transfer rate is enhanced. Ideally, the fin material
should have a large thermal conductivity to minimize temperature variations from its
base to its tip. In the limit of infinite thermal conductivity, the entire fin would be at the
temperature of the base surface, thereby providing the maximum possible heat transfer
enhancement.

100
There are many engineering applications of fin such as cooling of engine heads, radiators, Heat Transfer from
Extended Surfaces
condenser coils of refrigerators, lawn mowers, electric generators, electric power
transformers, etc. Various types of fins are used for different applications. Some
configurations of fins are given in Figure 6.3.

x x

(a) (b)

x
(c) (d)
Figure 6.3 : Fin Configurations (a) Straight Fin of Uniform Cross Section,
(b) Straight Fin of Non-Uniform Cross Section, (c) Annular Fin and (d) Pin Fin.

A straight fin is any extended surface that is attached to a plane wall. It may be of
uniform cross-sectional area, or its cross-sectional area may vary with the distance x from
the wall (Figures 6.3(a) (b)). An annular fin is one that is circumferentially attached to a
cylinder, and its cross section varies with radius from the wall of the cylinder (Figure
6.3(c)). The foregoing fin types have rectangular cross sections, whose area may be
expressed as a product of the fin thickness t and the width w for straight fins or the
circumference 2r for annular fins. In contrast, a pin fin, or spine, is an extended surface
of circular cross section. Pin fins may also be of uniform or non-uniform cross section
(Figure 6.3(d)). In any application, selection of a particular fin configuration may depend
on space, weight, manufacturing, and cost considerations, as well as on the extent to
which the fins reduce the surface convection coefficient and increase the pressure drop
associated with flow over the fins.
SAQ 1
(a) What is a fin?
(b) Give some examples of fins.
(c) Why it is necessary to attach fins in a heat exchanger?

101
Conduction
6.3 A GENERAL CONDUCTION ANALYSIS
Consider the extended surface of Figure 6.4. The analysis is simplified if certain
assumptions are made. We choose to assume one-dimensional conditions in the
longitudinal (x) direction, even though conduction within the fin is actually
two-dimensional.
dq
Convection
dAs
qx
Ac (x)
qx-dx

dx

X
x

z
y

x
Figure 6.4 : Energy Balance for an Extended Surface

The rate at which energy is convected to the fluid from any point on the fin surface must
be balanced by the rate at which energy reaches that point due to conduction in the
transverse (y, z) direction. However, in practice the fin is thin and temperature changes in
the longitudinal direction are much larger than those in the transverse direction. Hence,
we may assume one-dimensional conduction in the x-direction. We will consider steady-
state conditions and also assume that the thermal conductivity is constant, that radiation
from the surface is negligible, that heat generation effects are absent, and that the
convection heat transfer coefficient h is uniform over the surface.
Applying the conservation of energy to the section in Figure 6.4.

q x  q x  dx  dq conv . . . (6.1)

From Fourier’s law we know that

dt
q x   k Ac . . . (6.2)
dx
where Ac is the cross-sectional area, which may vary with x. Since the conduction heat
rate at x + dx may be expressed as

dq x
q x  dx  q x  dx . . . (6.3)
dx
It follows that

dT d  dT 
q x  dx   kAc k  Ac  dx . . . (6.4)
dx dx  dx 

The convection heat transfer rate may be expressed as

dqconv  h d As (T  T ) . . . (6.5)

where dAs is the surface area of the differential element. Substituting the foregoing rate
equations into the energy balance, Eq. (6.1), we obtain

d  dT  h dAs
 Ac  (T  T )  0 . . . (6.6)
dx  dx  k dx

102
 Heat Transfer from
d 2T  1 dAc  dT  1 h dAs  Extended Surfaces
or     (T  T )  0 . . . (6.7)
dx 2  Ac dx  dx  Ac k dx 
Eq. (6.7) provides a general form of the energy equation for an extended surface. Its
solution for appropriate boundary conditions provides the temperature distribution, which
may be used with Eq. (6.2) to calculate the conduction rate at any x.

6.4 FINS OF UNIFORM CROSS-SECTIONAL AREA

To solve Eq. (6.7) it is necessary to be more specific about the geometry. We begin with
the simplest case of straight rectangular and pin fins of uniform cross section (Figure 6.5).
Each fin is attached to a base surface of temperature T (0) = Tb and extends into a fluid of
temperature T.

x Q h

dx
L
QO Qx Qx + dx

t h
TO l qconv
h Th
Q
qf D
T x
L AC
T1

(a) Straight Fin (b) Thin Rod

Figure 6.5 : Straight Fin of Uniform Cross-section (a) Rectangular, and (b) Pin Fin

For the prescribed fins, Ac is a constant and As = Px where As is the surface area measured
dA dAs
from the base to x and P is the fin perimeter. Accordingly, with c  0 and  P,
dx dx
Eq. (6.7) reduces to
d 2T hP
2
 (T  T )  0 . . . (6.8)
dx kAc

To simplify the form of the equation, we transform the dependent variable by defining an
excess temperature  as
 ( x)  T ( x)  T . . . (6.9)

d  dT
where, since T is a constant,  . Substituting Eq. (6.9) into Eq. (6.8), we then
dx dx
obtain
d 2
2
 m2   0 . . . (6.10)
dx
hP
where m2  . . . (6.11)
kAc

Eq. (6.10) is a linear, homogeneous, second-order differential equation with constant

coefficients. Its general solution is of the form

 ( x)  C1 e mx  C2 e  mx . . . (6.12)
By substitution it may readily be verified that Eq. (6.12) is indeed a solution to
Eq. (6.10).
103
Conduction To evaluate the constants C1 and C2 of Eq. (6.12), it is necessary to specify appropriate
boundary conditions. One such condition may be specified in terms of the temperature at
the base of the fin (x = 0)
 (0)  Tb  T  b . . . (6.13)

The second condition, specified at the fin tip (x = L), may correspond to one of four
different physical situations. These are described below as Case A, Case B, Case C and
Case D.
6.4.1 Case A
The first condition considers convection heat transfer from the fin tip. Applying an
energy balance to a control surface about this tip (Figure 6.6), we obtain
dT
h Ac [T ( L )  T ]   k Ac . . . (6.14)
dx xL

d
or h  ( L)   k . . . (6.15)
dx xL

Fluid, T
qconv
Tb

qb = qf dT hAc [T(L) - T]

kA c x L
dx

Figure 6.6 : Conduction and Convection in a Fin of Uniform Cross Section

That is, the rate at which energy is transferred to the fluid by convection from the tip
must equal the rate at which energy reaches the tip by conduction through the fin.
Substituting Eq. (6.12) into Eqs. (6.13) and (6.15), we obtain, respectively
b  C1  C2 . . . (6.16)

and h (C1 e  mL  C2 e  mL )  km (C2 e  mL  C1 e mL ) . . . (6.17)

Solving for C1 and C2, we get,

 h 
cosh m ( L  x)    sinh m ( L  x)
  mk 
 . . . (6.18)
b  h 
cosh mL    sinh mL
 mk 

The temperature distribution ( vs x) is shown schematically in Figure 6.7.

b
θ (x)

θ
O L

Figure 6.7 : Temperature Distribution along the Length of a Fin with Uniform Cross-section

104
It may be noted that the magnitude of the temperature gradient decreases with increasing Heat Transfer from
Extended Surfaces
x. This trend is a consequence of the reduction in the conduction heat transfer qx (x) with
increasing x due to continuous convection losses from the fin surface.
We are particularly interested in the amount of heat transferred from the entire fin. From
Figure 6.7 it is evident that the fin heat transfer rate qf may be evaluated in two
alternative ways, both of which involve use of the temperature distribution. The simpler
procedure, and the one that we will use, involves applying Fourier’s law at the fin base.
dT d
That is, q f  qb   k Ac   k Ac . . . (6.19)
dx x0 dx x0

Hence, knowing the temperature distribution,  ( x), q f may be evaluated, giving

 h 
sinh mL    cosh mL
 mk 
q f  h P k Ac b . . . (6.20)
 h 
cosh mL    sinh mL
 mk 
However, conservation of energy dictates that the rate at which heat is transferred by
convection from the fin must equal the rate at which it is conducted through the base of
the fin. Accordingly, the alternative formulation for qf is
qf  A f
h [T ( x )  T ] d As . . . (6.21)

qf  A f
h  (x ) d As . . . (6.22)

where Af is the total, including the tip, fin surface area. Substitution of Eq. (6.18) into
Eq. (6.22) would yield Eq. (6.20).
6.4.2 Case B
The second tip condition corresponds to the assumption that the convective heat loss
from the fin tip is negligible, in which case the tip may be treated as adiabatic and
d
0 . . . (6.23)
dx xL

Substituting from Eq. (6.12) and dividing by m, we obtain

C1 e mL  C2 e  mL  0 . . . (6.24)
Using this expression with Eq. (6.16) to solve for C1 and C2 and substituting the results
into Eq. (6.12), we obtain
 cosh m ( L  x)
 . . . (6.25)
b cosh mL
Using this temperature distribution with Eq. (6.19), the fin transfer rate is then
q f  h P k Ac b tanh mL . . . (6.26)
6.4.3 Case C
In the same manner, we can obtain the fin temperature distribution and heat transfer rate
for Case C, where the temperature is prescribed at the fin tip. That is the boundary
condition is  ( L)   L , and the resulting expressions are of the form
 L 
  sinh mx  sinh m ( L  x)
 
 b . . . (6.27)
b sinh mL
L
cosh mL 
b
q f  h P k Ac b . . . (6.28)
sinh mL
105
Conduction 6.4.4 Case D
Fin of infinite length
The very long fin, case as L  , L  0 and it is easily verified that

 e  mx . . . (6.29)
b

q f  h P k Ac b . . . (6.30)

SAQ 2
Derive an expression for heat transfer and temperature distribution for a
rectangular fin with its tip at adiabatic condition.

6.5 FIN PERFORMANCE

Fins are used to increase the heat transfer from a surface by increasing the effective
surface area. However, the fin itself represents a conduction resistance to heat transfer
from the original surface. For this reason, there is no assurance that the heat transfer rate
will be increased through the use of fins. An assessment of this matter may be made by
evaluating the fin effectiveness f.
It is defined as the ratio of the fin heat transfer rate to the heat transfer rate that would
exist without the fin. Therefore,
qf
f  . . . (6.31)
h Ac , b b

where Ac, b is the fin cross-sectional area at the base. In any rational design the value of f
should be as large as possible, and in general, the use of fins may rarely be justified
unless f  2.
Subject to any one of the four tip conditions that have been considered, the effectiveness
for a fin of uniform cross section may be obtained by dividing the appropriate expression
for qf by h Ac , b b . Although the installation of fins will alter the surface convection
coefficient, this effect is commonly neglected. Hence, assuming the convection
coefficient of the finned surface to be equivalent to that of the unfinned base, it follows
that, for the infinite fin approximation (Case D), the result is
1
 kP  2
F    . . . (6.32)
 hAc 
Several important trends may be inferred from this result.
Obviously, fin effectiveness is enhanced by the choice of a material of high thermal
conductivity. Aluminum alloys and copper come to mind. However, although copper is
superior from the standpoint of thermal conductivity, aluminum alloys are the more
common choice because of additional benefits related to lower cost and weight. Fin
effectiveness is also enhanced by increasing the ratio of the perimeter to the cross-
sectional area. For this reason the use of thin, but closely related fins is preferred, with
the provision that the fin gap not be reduced to a value for which flow between the fins is
severely impeded, thereby reducing the convection coefficient.
Eq. (6.32) also suggests that the use of fins can be better justified under conditions for
which the convection coefficient h is small. Hence, it is evident that the need for fins is
stronger when the fluid is a gas rather then a liquid and when the surface heat transfer is
by free convection. If fins are to be used on a surface separating a gas and a liquid, they
are generally placed on the gas side, which is the side of lower convection coefficient. A
106
common example is the tubing in an automobile radiator. Fins are applied to the outer Heat Transfer from
Extended Surfaces
tube surface, over which there is flow of ambient air (small h), and not to the inner
surface, through which there is flow of water (large h). Note that, if f > 2 is used as a
criterion to justify the Implementation of fins, Eq. (6.32) yields the requirement that
 kP 
   4.
 h Ac 
Eq. (6.32) provides an upper limit to f which is reached as L approaches infinity.
However, it is certainly not necessary to use very long fins to achieve near maximum
heat transfer enhancement.
Fin performance may also be quantified in terms of a thermal resistance. Treating the
difference between the base and fluid temperatures as the driving potential, a fin
resistance may be defined as
b
Rt , f  . . . (6.33)
qf

The result is extremely useful, particularly when representing a finned surface by a

thermal circuit. Note that, according to the fin tip condition, an appropriate expression for
qf may be obtained from Eqs. (6.20), (6.26), (6.28) and (6.30).
Dividing Eq. (6.33) into the expression for the thermal resistance due to convection at the
exposed base,
1
Rt , b  . . . (6.34)
h Ac , b

and substituting from Eq. (6.31), it follows that

Rt , b
f  . . . (6.35)
Rt , f

Hence the fin effectiveness may be interpreted as a ratio of thermal resistances, and to
increase f it is necessary to reduce the conduction/convection resistance of the fin. If the
fin is to enhance heat transfer, its resistance must not exceed that of the exposed base.
Another measure of fin thermal performance is provided by the fin efficiency f. The
maximum driving potential for convection is the temperature difference between the base
(x = 0) and the fluid, b  Tb  T . Hence, the maximum rate at which a fin could
dissipate energy is the rate that would exist if the entire fin surface were at the base
temperature. However, since any fin is characterised by a finite conduction resistance, a
temperature gradient must exist along the fin and the above condition is an idealisation.
A logical definition of fin efficiency is, therefore,
qf qf
f   . . . (6.36)
qmax h A f b

where Af is the surface area of the fin. For a straight fin of uniform cross-section and an
adiabatic tip
M tanh mL tanh mL
f   . . . (6.37)
h P L b mL

From Eq. (6.37), it is observed that f approaches its maximum and minimum values of
1 and 0, respectively, as L approaches 0 and .
As discussed earlier, Eq. (6.20) gives a cumbersome expression for heat transfer from a
straight rectangular fin with an active fin tip. An approximate prediction for the heat
transfer for the same fin may be obtained by using the adiabatic tip boundary condition.
107
Conduction t
In such a case, the length of the fin in Eq. (6.37) is corrected to Lc  L    for a
 2
D
rectangular fin. The corrected value of length can be taken as Lc  L    in case of
4
pin fin.
This correction is based on assuming equivalence between heat transfer from the actual
fin with tip convection and heat transfer from a longer, hypothetical fin with an adiabatic
tip.
Hence, with tip convection, the fin heat rate may be approximated as
q f  M tanh mLc . . . (6.38)
The corresponding efficiency is
tanh mLc
f  . . . (6.39)
mLc

 ht   hD 
Errors associated with the approximation are negligible if   or    0.0625 .
k   2k 
If the width of a rectangular fin is much larger then its thickness, w   t , the perimeter
may be approximated as P = 2w, and
1 1
 hP  2  2h  2
mLc    Lc    Lc . . . (6.40)
 kAc   kt 

Multiplying numerator and denominator by L1/c 2 and introducing a corrected fin profile
area, Ap  Lc t , it follows that
1
 2h 2 3 2
mLc    Lc . . . (6.41)
 kAp 
 
Hence, as shown in Figures (6.8) and (6.9), the efficiency of a rectangular fin with tip
1
 h 2
convection may be represented as a function of L3c 2   .
 k Ap 
 
100

y-x2 Lc = L
80 Ap = L1/3
y x
1/2
L
60

f (%)
Lc = L+1/2
40 Ap = Lc t 1/2
L y-x

y Lc = L
20 1/2 Ap = L1/2
L

O 0.5 1.0 1.5 2.0 2.5

Lc3/2 (h/kAp)1/2
Figure 6.8 : Efficiency of Straight Fins ( Rectangular, Triangular and Parabolic Profiles)

108
 Heat Transfer from
110 Extended Surfaces

80

60
f (%) 1 = r2c/r1

40 2
r2c = r2+h/2 3
t Lc = L+h2
Ap = Lct 5
20 L
r1
r2
0 0.5 1.0 1.5 2.0 2.5

Lc3/2 (h/kAp)1/2
Figure 6.9 : Efficiency of Annular Fins of Rectangular Profile

SAQ 3
(a) Define fin efficiency?
(b) What is fin effectiveness?
(c) What are the parameters to consider for evaluating the fin effectiveness?
Discuss their effect on performance of a fin.
(d) When is the use of fin is not justified?

6.6 FINS OF NON-UNIFORM CROSS-SECTIONAL AREA

Analysis of the thermal behavior is more complicated for the case of fins with non-
uniform cross section. In such cases the second term of Eq. (6.7) must be retained, and
the solutions are no longer in the form of simple exponential or hyperbolic functions.
For example, consider the annular fin shown Figure 6.10. It is assumed that the fin
thickness is uniform (t is independent of r), the cross-sectional area, Ac  2 r t , varies
with r.
r2

r1

Th
s

T - h
Figure 6.10 : Annular Fin

Replacing x by r in Eq. (6.7) and expressing the surface area as As  2 (r 2  r12 ) , the
general form of the fin equation reduces to
d 2T 1 dT 2h
2
  (T  T )  0 . . . (6.42)
dr r dr kt
2h
or, with m 2  and   T  T ,
kt
109
Conduction d 2 1 d
2
  m2   0 . . . (6.43)
dr r dr

The foregoing expression is a modified Bessel equation of order zero, and its general
solution is of the form

 ( r )  C1 I 0 ( mr )  C2 K 0 ( mr ) . . . (6.44)

where I0 and K0 are modified, zero-order Bessel functions of the first and second kinds,
respectively. If the temperature at the base of the fin is prescribed,  (r1 )  b , and an
d
adiabatic tip is presumed,  0 , C1 and C2 may be evaluated to yield a temperature
drr2
distribution of the form

 I (mr ) K1 (mr2 )  K 0 (mr ) I1 (mr2 )

 0 . . . (6.45)
b I 0 (mr1 ) K1 (mr2 )  K 0 (mr1 ) I1 (mr2 )

d [ I 0 (mr )]  d [ I 0 (mr )]
where I1 (mr )  and K1 (mr )  are modified, first-order
d (mr ) d (mr )
Bessel functions of the first and second kinds, respectively. The Bessel functions are
tabulated at the end of the unit (Appendix-I).
With the fin heat transfer rate expressed as

dT d
q f   k Ac , b   k (2 r1 t ) . . . (6.46)
dr r  r1 dr r  r1

It follows that

K1 (mr1 ) I1 (mr2 )  I1 (mr1 ) K1 (mr2 )

q f  2 k r1 t b m . . . (6.47)
K 0 (mr1 ) I1 (mr2 )  I 0 (mr1 ) K1 (mr2 )

From which the fin efficiency becomes

qf 2r1 K1 ( mr1 ) I1 ( mr2 )  I1 ( mr1 ) K1 ( mr2 )

f   . . . (6.48)
h 2 (r22  r12 ) b m (r22  r12 ) K 0 (mr1 ) I1 (mr2 )  I 0 (mr1 ) K1 (mr2 )

This result may be applied for an active (convective) tip, if the tip radius r2 is replaced by
t
a corrected radius of the form r2c  r2    . Results are represented graphically in
2
Figure 6.9.
Knowledge of the thermal efficiency of a fin may be used to evaluate the fin resistance,
where, from Eqs. (6.33) and (6.36), it follows that

1
Rt , f  . . . (6.49)
h Af  f

Although results for the fins of uniform thickness or diameter were obtained by assuming
an adiabatic tip, the effects of convection may be treated by using a corrected length (Eq.
(6.39)) or radius (Eq. (6.48)). The triangular and parabolic fins are of non-uniform
thickness, which reduces to zero at the fin tip. The volume of a straight fin is simply the
product of its width and profile area, V  w A p .

Fin design is often motivated by a desire to minimise the fin material and/or related
manufacturing costs required to achieve a prescribed cooling effectiveness. Hence, a
straight triangular fin is attractive because, for equivalent heat transfer, it requires much
110
less volume (fin material) than a rectangular profile. In this regard, heat dissipation per Heat Transfer from
Extended Surfaces
q q
unit volume,   is largest for a parabolic profile. However, since   for the
V
 f V  f
parabolic profile is only slightly larger than that for a triangular profile, its use can rarely
be justified in view of its larger manufacturing costs. The annular fin of rectangular
profile is commonly used to enhance heat transfer to or from circular tubes.

6.7 OVERALL SURFACE EFFICIENCY

In contrast to the fin efficiency f, which characterises the performance of a single fin,
the overall surface efficiency 0 characterises an array of fins and the base surface to
which they are attached. Representative arrays are shown in Figure 6.10, where S
designates the fin pitch. In each case the overall efficiency is defined as
qt qt
0   . . . (6.50)
qmax h At b

where qt is the total heat rate from the surface area At associated with both the fins and
the exposed portion of the base (often termed the prime surface). If there are N fins in the
array, each of surface area Af, and the area of the prime surface is designated as Ab, the
total surface area is
At  NA f  Ab . . . (6.51)

The maximum possible heat rate would result if the entire fin surface, as well as the
exposed base, were maintained at Tb.
The total rate of heat transfer by convection from the fins and the prime (unfinned)
surface may be expressed as
qt  N  f h A f b  h Ab b . . . (6.52)

where the convection coefficient h is assumed to be equivalent for the finned and prime
surfaces and f is the efficiency of a single fin. Hence,

 NA f 
qt  h  N  f A f  ( At  NA f ) b  h At 1  (1   f )  b . . . (6.53)
 At 
Substituting Eq. (6.53) into Eq. (6.50), it follows that
NA f
0  1  (1   f ) . . . (6.54)
At

From knowledge of 0, Eq. (6.50) may be used to calculate the total heat rate for a fin
array.
Recalling the definition of the fin thermal resistance, Eq. (6.33) and Eq. (6.50) may be
used to infer an expression for the thermal resistance of a fin array. That is,
b q
Rt , 0   . . . (6.55)
qt 0 h At

where Rt, 0 is an effective resistance that accounts for parallel heat flow paths by
conduction/convection in the fins and by convection from the prime surface. Figure 6.11
illustrates the thermal circuits corresponding to the parallel paths and their representation
in terms of an effective resistance.

111
Conduction (NfhAf)-2

q11 Nqf

Tb T
qb qb
Tb
Tb

[h(A2-NAf)]-2

q1
Tb T
T h (fhAf)-1

Figure 6.11 : Fin Array and Thermal Circuit with Fins Integral to the Base

If fins are machined as an integral part of the wall from which they extend, there is no
contact resistance at their base. However, more commonly, fins are manufactured
separately and are attached to the wall by a metallurgical or adhesive joint. Alternatively,
the attachment may involve a press fit, for which the fins are forced into slots machined
on the wall material. In such cases (Figure 6.12), there is a thermal contact resistance,
Rt, c, which may adversely influence overall thermal performance.

m (NfhAf)-1
RfNAc b

q1
Nqf
Tb
qb Tb T
qb

q1
Tb T
(fhAf)-1
T h

Figure 6.12 : Fin Array and Thermal Circuit with Fins Attached to the Base

An effective circuit resistance may again be obtained, where, with the contact resistance,
b 1
Rt , o (c )   . . . (6.56)
qt o (c ) hAt

It is readily shown that the corresponding overall surface efficiency is

NA f  f 
o (c )  1  1   . . . (6.57)
At  C1 

 Rt, c 
where C1  1   f h A f   . . . (6.58)
 Ac , b 
 
In manufacturing, care must be taken to render Rt , c   Rt , f .

Example 6.1
12 (Twelve) number of fins each having thermal conductivity k = 75 W/mK and
0.75 mm thickness protrude 25 mm from a cylindrical surface of 50 mm diameter
and 1 m length placed in an atmosphere of 40oC. If the cylindrical surface is
maintained at 150oC and the heat transfer coefficient is h = 23 W/m2 . K, calculate
(a) the rate of heat transfer,
112 (b) the percentage increase in heat transfer due to fins,
 (c) the temperature at the centre of the fins, and Heat Transfer from
Extended Surfaces
(d) fin efficiency and fin effectiveness.
Solution
Refer to the following figure :
The perimeter of one fin

P  2 ( L  b )  2 (1  0.75  10  3)  2 m

A  bL  0.75  10  3  1  0.00075 m 2

1 1
 hP  2  23  2 2 1
m     28.6 m
 kA   0.00075  75 
ml  28.6  0.025  0.715
tanh ml  tanh 0.715  0.61
T = 40oc

TW = 150oC

l = 25mm
L = 1m

R = 23 w/m2k

b = 0.75

Heat transfer from one fin

Qo  m k A o tanh ml  28.6  75  0.00075  (150  40)  0.61  108 W
Heat transfer from 12 fins
Q  12  Qo  12  108  1296 W
Heat transfer from uniform portion of the surface
Qs  h A  0
 23  [(  0.05  l)  (12  0.00075  l)]  (150  40)
 23  0.148  110 = 374.44 W
Total heat transfer from the cylindrical surface
QT  Q  Qs  1296  374.44  1670.44 W Ans. (a)
If the cylinder were without fins, the heat transfer would have been
Q  23  (  0.05  l )  110  397.4 W
Increase in heat transfer due to fins
1670.44  397.44
  100  320.3% Ans. (b)
397.44
 cosh m (1  x)

o cosh ml
Tc  40 cosh (28.6  0.0125) 1.062
or,    0.845
150  40 cosh 0.715 1.257
Centre temperature of the fins, Tc = 133oC Ans. (c) 113
Conduction If tip loss is neglected
tanh ml 0.61
fin    0.853
ml 0.715
or, fin  85.3% Ans. (d)
Fin effectiveness
1670.44
  4.2 Ans. (e)
397.44
As fin effectiveness is more than 2, use of fins is justified.
Example 6.2
One end of a long rod is inserted into a furnace while the other projects into
ambient air. Under steady state the temperature of the rod is measured at two
points 75 mm apart and found to be 125oC and 88.5oC, respectively, while the
ambient temperature is 20oC. If the rod is 25 mm min diameter and h = 23.36
W/m2K, find the thermal conductivity of the rod material.
Solution
The temperature distribution is given by
  C1 e  mx  C2 e mx
1
 hP  2
where m 
 kA 
At x  0,    o
At x  ,   0
 C2 = 0
  C1 e  mx
Let, l be the distance between the two points where temperatures are measured.
Then, 1  o e  mx1

and 2  0 e  m ( x1  l )

1 e  mx1
  mx  ml
 e ml
2 e 1 .e
125  20 105
or,   e ml
88.5  20 68.5
ml = 0.427
1
1  2 1
0.427  hP  2  h  d   h 2
or, m  5.696       2 
0.075  kA   k d2   kd 
 4 
h
m2  4   32.44
kd
4  23.36
k  115.2 W/m.K
0.025  32.44

114
Example 6.3 Heat Transfer from
Extended Surfaces
A bar of square cross section connects two metallic structures. One structure is
maintained at a temperature 200oC and the other is maintained at 50oC. The bar,
20 mm  20 mm, is 100 mm long and is made of mild steel (k = 0.06 kW/m . K).
The surroundings are at 20oC and the heat transfer coefficient between the bar and
the surroundings is 0.01 kW/m2 . K.
Derive an expression for the temperature distribution along the bar and hence
calculate the total heat flow rate from the bar to the surroundings.
Solution
dT
Q1   kA
dx
dQ1
Q2  Q1  dx
dx
d  dT  d 2T
Q1  Q2     kA  dx  kA dx  h P dx (T  T  )
dx  dx  dx 2
Let,   T  T

d 2 hP
2
 0
dx kA
1
 hP  2
or, ( D 2  m 2 )   0 where m   
 kA 
T = 20oc

x
T2 = 50oc
T1 = 200oc
Q1 Q2

x
h = 0.01 kw/m2 k
l = 0.1 m

K = 0.06 kw/m.k.
20mm w/m2k

20mm

x-x

The general solution is

  C1 e mx  C2 e  mx

At x  0,   1  C1  C 2  180 o C

At x  1,    2  C1 e ml  C 2 e  ml  30 oC

1 1 1
 hP  2  0.01  4  0.02  2  4  2
m     
 kA   0.06  0.02  0.02   0.12 

ml = 0.577

e ml  1.78
115
Conduction
e  ml  0.561

30  C1  1.78  C2  0.561

3.173 C1  C2  53.476

From Eq. (2)

C1  C2  180.00

On subtraction,
2.173 C1   126.524

C1 =  58.22
From Eq. (2)
C2 = 238.22
The temperature distribution is

   58.22 e5.77 x  238.22 e 5.77 x

 d 
Q1   kA     kA ( 58.22e 5.77 x  238.22e 5.77 x ) x  0  5.77
 dx x  0

 0.06  4  10  4  (58.22 e5.77 x  238.22 e 5.77 x ) x  0  5.77  41.0 W

 d 
Q2   kA  
 dx  x  l

  0.06  4  10  4  ( 58.22  5.77  1.78  238.22  5.77  0.561)  32. 8 W

Heat flow rate from the bar to the surroundings

Q  Q1  Q2  41.0  32.8  8.2 W

Example 6.4
Determine the heat transfer rate from the rectangular fin of length 20 cm, width
40 cm and thickness 2 cm. The tip of the fin is not insulated and the fin has a
thermal conductivity of 150 W/m . K. The base temperature is 100oC and the fluid
is at 20oC. The heat transfer coefficient between the fin and the fluid is
30 W/m2 . K.
Solution
The extended length
A 40  2
Lc  l   20   20.95 cm  0.21 m
P 84

A  40  2  80 cm 2

As  Lc P  0.21  0.84  0.1764 m 2

1 1
 hP  2  30  0.84  2 1
m     4.583 m
 kA   150  0.008 
116
 Q0  m k A 0 tanh ml Heat Transfer from
Extended Surfaces
 4.583  150  0.008  80  tanh (4.583  0.21)  328.0 W

tanh ml tanh (4.583  0.21)

Also, fin    0.775
ml (4.583  0.21)

Q0  fin  As  h  (T0  T )

 0.775  30  0.1764  (100  20)  328 W

Example 6.5
Derive the condition for the maximum heat flow for a given weight of a
rectangular fin with minimum weight.
Solution
In some applications such as aircraft, heat transfer from engine should be
maximum with minimum weight of the engine. For a given weight, the maximum
heat transfer is desirable.
Weight of one fin  b  l  L  
where  is the density of the material.
Let, A1  b  l = area of fin cross-section normal to L.
The length of the fin L is fixed at a given dimension. Here, we have to change the
other two dimensions (i.e. b and l) so as to obtain maximum heat flow for a given
area A1 (Figure 6.3(a)).
1
 2h  2
We have m 
 kb 
AbL
and A1  b  l
If the tip loss is neglected,
1
Q  (h p k A) 2  0 tan ml  m k A  0 tanh ml
1  1 
 2h  2   2 h  2 A1 
   k b L  0 tanh  
 kb   kb  b 
 
1  
1 1
 2h  2  A1 
 (2hk ) 2 b2 L  0 tanh    3 
 k   
 b2 
For a given area A1, Q will be maximum when
1  1  1
dQ   2h  2 A1  1 1
 0  (2 hk ) L  0 tanh  
2 b2
db  k  3  2
 b 2 
1
1 5
1 1  2h  2  3 
 (2hk ) 2 L 0 b 2    A1   b 2
 1   k   2
 2 h  A1 
cosh 2  
2
 k  3
 b 2 
117
Conduction 1
 2h  2
Putting u 
 kb 
A1
and  ml
b
1
tanh u 3 A1  2 h  2 1
We get, 1
   0
2 b  k  cosh2 u
2
2b 2
1
A  2h  2 1
or tan u  3 1   0
b  kb  cosh 2 u

or cosh u sinh u  3u  0
u
or sinh  3u
2
e 2u  e  2u
or  6u
2
 4u 2 8u 3 16u 4 32u 5 
or  1  2u      . . .

 2 6 24 120 
 4u 2 8u 3 16u 4 32u 5 
  1  2u      . . .   12u
 2 6 24 120 
 

16u 3 64u 5
or 4u    12u
6 120

or u 4  5u 2  15  0
1
 5  (25  60) 2
or u2   2.1
2
or u  ml  1.419
1
 2h  2
Now,   l  1.419
 kb 
1
1  2k  2
or  1.419  
b  hb 
2
This is the condition for the maximum heat flow for a given weight of fin, giving
the optimum ratio of fin height to half the fin thickness.
Exercise 6.1
Show that the Fin Efficiency for a Rectangular Fin is given by
1
 2hlc2  2
tanh  
fin   kb 
1
 2hlc2  2
 
 kb 
118
 lb Heat Transfer from
where lc  = corrected length Extended Surfaces
2
A
or lc  l 
P
Exercise 6.2
Show that the Total Heat Transfer from a Finned Wall is given by
Q  h 0  A  (l  fin ) Afin 

where A = Total area of finned and unfinned surfaces,

Afin = Area of the finned surface, and
fin = Fin efficiency.
and 0  T0  T

Exercise 6.3
A very long rod 5 mm in diameter has one end maintained at 100oC. The surface of
the rod is exposed to ambient air at 25oC with a convection heat transfer
coefficient of 100 W/m2 . K.
(a) Determine the temperature distribution along rods constructed from
pure copper, 2024 aluminium alloy, and type AISI 316 stainless steel.
What are the corresponding heat losses from the rods?
(b) Estimate how long the rods must be for the assumption of infinite
length to yield an accurate estimate of the heat loss.
[Hint : Evaluate the properties at mean temperature, i.e.
T  T
T  b  62.5o C , kcu  398 W/m.K ,
2
kal  180 W/m.K and for stainless steel, k ss  14 W/m.K ]
Exercise 6.4
A long circular aluminium rod is attached at one end to a heated wall and transfers
heat by convection to a cold fluid.
(a) If the diameter of the rod is tripled, by how much would the rate of
heat removal change?
(b) If a copper rod of the same diameter is used in place of the aluminium
rod, by how much would the rate of heat removal change?
Exercise 6.5
Consider two long, slender rods of the same diameter but different materials. One
end of each rod is attached to a base surface maintained at 100oC, while the surface
of the rods are exposed to ambient air at 20oC. By traversing the length of each rod
with a thermocouple, it was observed that the temperatures of the rods were equal
at the positions xA = 0.15 m and xB = 0.075 m, where x is measured from the base
surface. If thermal conductivity of the rod A is known to be
kA = 70 W/m . K, determine the thermal conductivity for the rod B.

6.8 SUMMARY
Extended surfaces are used in many heat transfer appliances to increase the heat transfer
rate from the device. There are different types of extended surfaces, also know as fins,
such as rectangular, triangular, annular, pin fin etc. I the present unit, heat transfer
equation through a fin is formulated and solutions are presented for different possible 119
Conduction conditions. Performance indices of fin are also discussed. Condition under which use of
fin is acceptable is also described. The solution for heat transfer rate and temperature
profile with and without fins are illustrated with examples.

6.9 KEY WORDS

Fin : Enhance heat transfer between a solid and an
adjoining fluid.
Fin Effectiveness : Ratio of the fin heat transfer rate to the heat
transfer rate that would exist without the fin.
Fin Efficiency : It is a measure of the thermal performance of a fin.
Adiabatic : No heat transfer.

6.10 ANSWERS TO SAQs

Please refer the relevant preceding text in this unit for answers to SAQs.

Appendix-I

Bessel function of First Kind

x J0 (x) J1 (x)

0.0 1.0000 0.0000

0.1 0.9975 0.0499
0.2 0.9900 0.0995
0.3 0.9776 0.1483
0.4 0.9604 0.1960

0.5 0.9385 0.2423

0.6 0.9120 0.2867
0.7 0.8812 0.3290
0.8 0.8463 0.3688
0.9 0.8075 0.4059

1.0 0.7652 0.4400

1.1 0.7196 0.4709
1.2 0.6711 0.4983
1.3 0.6201 0.5220
1.4 0.5669 0.5419

1.5 0.5118 0.5579

1.6 0.4554 0.5699
1.7 0.3980 0.5778
1.8 0.3400 0.5815
1.9 0.2818 0.5812

2.0 0.2239 0.5767

2.1 0.1666 0.5683
2.2 0.1104 0.5560
2.3 0.0525 0.5399
2.4 0.0025 0.5202

120
 Heat Transfer from
Extended Surfaces
Modified Bessel Function of First and Second Kinds
x e– x I0 (x) e– x I1 (x) ex K0 (x) ex K1 (x)
0.0 1.0000 0.0000  
0.2 0.8269 0.0823 2.1407 5.8334
0.4 0.6974 0.1368 1.6627 3.2587
0.6 0.5993 0.1722 1.4167 2.3739
0.8 0.5241 0.1945 1.2582 1.9179
1.0 0.4657 0.2079 1.1445 1.6361
1.2 0.4198 0.2152 1.0575 1.4429
1.4 0.3831 0.2185 0.9881 1.3010
1.6 0.3533 0.2190 0.9309 1.1919
1.8 0.3289 0.2177 0.8828 1.1048
2.0 0.3085 0.2153 0.8416 1.0335
2.2 0.2913 0.2121 0.8056 0.9738
2.4 0.2766 0.2085 0.7740 0.9229
2.6 0.2639 0.2046 0.7459 0.8790
2.8 0.2528 0.2007 0.7206 0.8405
3.0 0.2430 0.1968 0.6978 0.8066
3.2 0.2343 0.1930 0.6770 0.7763
3.4 0.2264 0.1892 0.6579 0.7491
3.6 0.2193 0.1856 0.6404 0.7245
3.8 0.2129 0.1821 0.6243 0.7021
4.0 0.2070 0.1787 0.6093 0.6816
4.2 0.2016 0.1755 0.5953 0.6627
4.4 0.1966 0.1724 0.5823 0.6453
4.6 0.1919 0.1695 0.5701 0.6292
4.8 0.1876 0.1667 0.5586 0.6142
5.0 0.1835 0.1640 0.5478 0.6003
5.2 0.1797 0.1614 0.5376 0.5872
5.4 0.1762 0.1589 0.5279 0.5749
5.6 0.1728 0.1565 0.5188 0.5633
5.8 0.1696 0.1542 0.5101 0.5525
6.0 0.1666 0.1520 0.5019 0.5422
6.4 0.1611 0.1479 0.4865 0.5232
6.8 0.1561 0.1441 0.4724 0.5060
7.2 0.1515 0.1405 0.4595 0.4905
7.6 0.1473 0.1372 0.4476 0.4762
8.0 0.1434 0.1341 0.4366 0.4631
8.4 0.1398 0.1312 0.4264 0.4511
8.8 0.1365 0.1285 0.4168 0.4399
9.2 0.1334 0.1260 0.4079 0.4295
9.6 0.1305 0.1235 0.3995 0.4198
10.0 0.1278 0.1213 0.3916 0.4108

1  2 
I x  1 ( x)  I x  1 ( x)    I n ( x)
 x  121
Conduction
REFERENCES
J. P. Holman, (2002), Heat Transfer, 9th Edition, Tata McGraw-Hill, New Delhi.
M. N. Ozisik, (1985), Heat Transfer A Basic Approach, McGraw-Hill
International Edition.
F. P. Incropera and D. P. Dewitt, (2004), Fundamentals of Heat and Mass Transfer,
5th Edition , John Wley and Sons.
P. K. Nag, (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.

122
Heat Transfer from
Extended Surfaces

123
Conduction
CONDUCTION
Present block is devoted to conduction heat transfer. The entire block consists of three
units.
Unit 4 introduces the governing equations of conduction heat transfer. Application of
Fourier’s law to steady and unsteady state conduction has been analysed. The governing
equation of heat conduction is presented in rectangular, cylindrical as well as spherical
coordinates. Heat transfer and determination of temperature distribution for plane wall,
cylinder and sphere under steady state condition are discussed. Transient heat transfer
concept is introduced with heat generation number, Biot number and Fourier number. A
detail analysis of transient heat transfer by lumped capacitance method is described.
Unit 5 describes various methods for solving heat conduction equations. Analytical and
graphical methods are described. Limitation of these methods are highlighted and it is
emphasized that numerical methods is used in complicated geometries, boundary
conditions and variable thermal properties. Finite difference method is described both for
the steady and transient conduction problems.
Unit 6 describes application of extended surfaces for heat transfer devices. Different
types of fins and their applications are cited in this unit. Formulation and solution are
done for extended surfaces. Performance parameters such as fin efficiency and
effectiveness are discussed in detail.
The units are supported with solved problems. All the three units include some unsolved
problems and SAQ’s which will help you understanding the heat conduction mechanism.

124
Indira Gandhi MRW-002
National Open University
School of Engineering & Technology HEAT TRANSFER

BLOCK 3

Convection
 Indira Gandhi
National Open University
MRW – 002
School of Engineering & Technology Heat Transfer

Block

3
CONVECTION
UNIT 7
Convection Heat Transfer 5

UNIT 8
Boundary Layer Formation 47
CONVECTION

Objective of the present block is to discus about the Heat Transfer by Convection.
Both the Units i.e. 7 and 8 are devoted to discuss the physical and mathematical
basis for the understanding of convective transport and to reveal various heat
transfer correlations.

The engineering applications, the pressure drop or drag force associated with the
flow inside ducts or over bodies is also of interest. Therefore, appropriate
correlations are presented to predict the pressure drop or drag force in flow.

The analysis of convection is complicated because the fluid motion is affected by

pressure drop, drag force and heat transfer. The literature of convective heat
transfer is overwhelming and ever growing. In recent years, with the availability
of high speed, large capacity digital computers, great advances have been made in
the analysis or very complicated heat transfer problems. However, a large number
of simple engineering problems can be handled with the use of standard heat
transfer correlations. To achieve this objective, basic concepts associated with the
flow over a body, flow inside a duct and turbulence are discussed. Effect of fluid
viscosity is dealt in particular for flow and heat transfer problems with the help of
boundary layer theory.
 Convection Heat
UNIT 7 CONVECTION HEAT TRANSFER Transfer

Structure
7.1 Introduction
Objectives

7.2 Fundamentals of Convection

7.3 Concept of Boundary Layer
7.3.1 Flow Over a Flat Plate
7.3.2 Drag Coefficient and Drag Force
7.3.3 Thermal Boundary Layer

7.4 Heat Transfer Coefficient

7.5 Conservation Equations of Mass, Momentum and Energy for Laminar Flow
Over a Flat Plate
7.5.1 Vector Method
7.5.2 Conservation of Momentum Equations
7.5.3 Conservation of Energy Equation

7.6 Principle of Similarity Applied to Heat Transfer

7.7 Derivation of Dimensionless Parameters from the Differential Equations
7.7.1 Forced Convection
7.7.2 Free Convection
7.7.3 Unsteady Heat Conduction

7.8 Evaluation of Convection Heat Transfer Coefficients

7.9 Dimensional Analysis
7.9.1 Rayleigh’s Method
7.9.2 Buckingham -Theorem

7.10 Summary
7.11 Key Words
7.12 Answers to SAQ’s

7.1 INTRODUCTION
Convective heat transfer or, simply, convection is the study of heat transport processes
effected by the flow of fluids. Convective heat transfer has grown to the status of a
contemporary science because of man’s desire to understand and predict the extent to
which a fluid will act as “carrier” or “conveyer belt” for energy and matter. It is clearly a
field at the interface between heat transfer and fluid mechanics. For this study of
convective heat transfer problems must rest on a solid understanding of basic heat
transfer and fluid mechanics principles.
Objectives
After studying this unit, you should be able to
 understand the mechanism of heat transfer by convection,
 classify various convective heat transfer processes,
 understand velocity and temperature boundary layers,
 derive the governing equations such as continuity, momentum and energy
equations,
5
Convection  appreciate importance of similarity principle and non dimensional numbers,
and
 apply dimensional analysis to find correlation of different variables.

7.2 FUNDAMENTALS OF CONVECTION

When energy transfer takes place between a solid surface and a fluid system in motion,
the process is known as convection. Convection is classified as
 Natural convection
 Forced convection
If the fluid motion is caused due to density difference, it is called free convection. If it
impressed by a pump or a compressor, it is called the forced convection. It is not possible
to separate the problem of heat transfer from that of the motion of the fluid, and so a
study of the hydrodynamic behaviour of the fluid is very much necessary in order to gain
an understanding of the heat transfer phenomena within a moving fluid.
A fluid may be defined as a material that supports no shearing when at rest or in a state of
uniform motion. Fluids exhibiting a linear relation between the rate of strain and the
applied shear stress are called Newtonian fluids. Common substances such as the gases,
water and oil are of this type. Certain suspensions that do not conform to a linear stress
rate and strain law are called non-Newtonian fluids. We will, however, be concerned with
the Newtonian fluids.
A fluid will be treated as a continuum, i.e. we will ignore the fact that the fluid is made
up of discrete particles (atoms, molecules, ions or electrons) and consider that the
smallest volume of fluid with which we are concerned is sufficiently large so that
macroscopic properties such as density, pressure and temperature have their usual
meanings and the motion of the fluid in contact with a solid surface is identical with that
of the surface (no slip motion).
In general, the behaviour of the flow depends on the properties of the fluids and on the
boundary conditions imposed. To analyse this behaviour requires the application of the
principle of conservation of mass (continuity equation), Newton’s laws of motion
(momentum equations) and the laws of thermodynamics (energy equation) along with the
phenomenological laws like Fourier’s law, Fick’s law and Newton’s law of viscosity.
Fluids include both liquids and gases. While liquids are incompressible, gases are
compressible, having their densities varying with pressure greatly, and also with
temperature.

7.3 CONCEPT OF BOUNDARY LAYER

Let us consider two plates a distance S apart as shown in Figure 7.1. The lower plate is at
rest, while the upper plate moves with a constant velocity U 0 . The space between the
plates is filled with a fluid.

Experience shows that in order to maintain the velocity U 0 of the upper plate a force is
necessary, and this force is directly proportional to velocity U 0 and inversely
proportional to the distance S. The force per unit area of the plate is equal to the shear
stress . Therefore,

U0
  . . . (7.1)
S
U0
or   . . . (7.2)
S
6
where  is the constant of proportionality. Convection Heat
Transfer
Moving
Uo

y
S

Stationary
Figure 7.1 : Flow between Two Parallel Plates (Conette Flow)

The fluid layers immediately adjacent to the two plates possess velocities equal to those
of the plates, namely U 0 and zero, respectively, while in the rest of the fluid the velocity
varies in a linear manner. Here  is a property of the fluid, called dynamic or absolute
viscosity. Phenomenological law represented by Eq. (7.2) is known Newton’s law of
viscosity. The fluids obeying this law are known as Newtonian fluids, as mentioned
earlier. Nonviscous or inviscid fluids are known as perfect or ideal fluids.
A more general form of Newton’s law is
du
  . . . (7.3)
dy

du
where is the velocity gradient at the wall (y = 0) as shown in Figure 7.2. When a
dy
fluid flows over a solid surface, there is a stagnant film immediately adjacent to the wall
where the fluid velocity is zero and through which heat is conducted.
Fluid flow
T
U y y  0
y
U(y) Surface of plate
T(y)

qc T Tw

Heat flow
Figure 7.2 : Velocity and Temperature Distribution in Laminar Forced Convection Flow
Over a Heated Flat Plate at Temperature TW

In an elastic solid, shear stress is proportional to the shear strain. In a viscous fluid, shear
N ms NS
stress is proportional to the rate of shear strain. The dimension of  is 2
. or .
m m m2
kgm s kg dynes
It is also . 2 or . In cgs units, it is , which was called “poise” after the
2
s m ms cm 2
French physicist L.J.M. Poiseuille. However, it used to be expressed in centipoises (cp).

The viscosity of a liquid is much larger than that of a gas, i.e. liq   gas . As temperature
increases, the viscosity of a liquid decreases, because of the decrease of cohesive forces
between the molecules as the liquid becomes lighter. But as temperature increases, the
viscosity of a gas increases, because the molecules travel faster as a result of which there
is increase in the transfer of molecular momentum.
7
Convection There is another frequently used property, the kinematic viscosity, v, defined as


v . . . (7.4)


kgm3
The dimension of  is or m2/s. It is also called momentum diffusivity.
mskg

dyne  sec cm 3 gmcm sec cm 3 cm 2

In cgs units, it used to be in or . . or which used
cm 2 gm sec 2 cm2 gm s
to be referred as “stoke” after the British physicist G.G. Stoke. However, it used to be
expressed in centistokes (cs).
Let us assume that some fluid is flowing over a solid surface as shown in Figure 7.3(a). If
we imagine a curve in the fluid, the tangent at every point of which indicates the
direction of the velocity of the fluid particle, then the curve is known as a streamline.
When one streamline slides smoothly over another streamline, and if this is maintained in
the entire flow is known as laminar. When there is transverse flow of fluid particles and
streamlines are interwoven with one another, the flow is known as turbulent as shown in
Figure 7.3(b).
Streamlines

(a)
Smoke

Turbulent Flow

Laminar Flow

(b)

Figure 7.3 : (a) Laminar flow, (b) Laminar and Turbulent Flow

It was the British scientist Osborne Reynolds who first differentiated these two types of
flow in a series of experiments conducted in 1883. From various experiments on the
flows of fluids through pipes, Reynolds discovered that the absolute viscosity , the mass
density of the fluid , and the diameter of the pipe D are the three other factors besides
the velocity of the fluid, controlling the transition from laminar to turbulent flow. The
results were confirmed by various experiments of French scientist M. Couette in 1890.
Reynolds and Couette arranged the four quantities in a dimensionless form, which is
called the Reynolds number ( Re ) in honour of Osborne Reynolds. It is defined as

 um D um D
Re   . . . (7.5)
 v
where um is the mean velocity of the fluid,  = absolute viscosity,  = mass density,
D = pipe diameter,  = kinematic viscosity.
8
The value of Reynolds number at which the flow pattern changes from laminar to Convection Heat
Transfer
turbulent motion is called critical Reynolds number R ec . For smooth circular pipes, R ec
is usually taken as 2300. If Re > 2300, the flow is turbulent. If Re < 2300, the flow is
laminar.
SAQ 1
(a) Define Reynolds number.
(b) What is the difference between free and forced convection?
(c) What do you mean by viscosity? What is its unit in SI system?
7.3.1 Flow Over a Flat Plate
When a fluid flows over a flat plate and its velocity is measured at various points normal
to the surface in the immediate vicinity of the wall, a velocity profile is obtained, as
shown in Figure 7.4. The velocity begins with the value zero at the wall and increases
within a thin layer of thickness  to the value of free-stream velocity u.
Laminar region Transition Turbulent

U
y 

U U Surface
of plate
Viscous
sublayer

x
x=o U
du
dy

Figure 7.4 : Velocity Profiles in Laminar, Transition and Turbulent Boundary

Layers in Flow over a Flat Plate

This distance from the wall , is called the boundary layer thickness where there is
velocity gradient and above which velocity is uniform and there is no viscous effect.
Viscosity effect is thus confined only in the boundary layer, and the main flow outside
the boundary layer, called the potential flow, is considered frictionless, where for each
streamline Bernoulli equation applies.
In the flow of a fluid over a flat plate held parallel to the direction of flow (Figure 7.4),
the vertical scale is purposely enlarged in order to show the detail of the flow pattern.
When the fluid passes the leading edge of the plate, the velocity gradient and the viscous
boundary shear are high. The fluid is moving in the laminar regime, and the boundary
layer is thin. This is called laminar boundary layer. As the fluid travels further down
stream along the plate, the retardation of fluid flow progresses due to viscous shear, and
the boundary layer grows in thickness. As a result the velocity gradient gradually
decreases. Meanwhile the boundary shear is reduced as the thickness increases. When the
boundary layer becomes thick enough, the particles begin to move out of the smooth
layers or laminae, the laminar motion becomes unstable, and finally the flow becomes
turbulent. However, under the turbulent boundary layer, there is still a thin layer of fluid
immediately next to the solid boundary, and this is still flowing in the laminar pattern.
This is called the laminar sublayer. The layer of transition from the laminar sublayer to
the turbulent layer is called the buffer layer. Since a laminar boundary layer cannot
suddenly change into a turbulent one, a transition zone exists between them.
The behaviour of flow in the boundary layer with the distance x from the leading edge is
governed by the magnitude of Reynolds number given by
u x
Re x  . . . (7.6)
v 9
Convection where u is the free-stream velocity, x is the distance from leading edge and  is the
kinematic viscosity.
The boundary layer growth for flow over a flat plate is shown in Figure.7.5.
The orderly motion of fluid continues along the plate until a critical distance is reached or
Reynolds number attains a critical value R exc , when fluid eddies begin to form
characterizing the end of the laminar boundary layer and the beginning of transition from
the laminar to turbulent boundary layer. This value of critical Reynolds number for flow
over a flat plate is
u xC 5
Re x   5 × 10 . . . (7.7)
v
Turbulent layer

Laminar boundary Transition Turbulent boundary

Layer region Layer
u
u U(x, y)
u
Boundary layer
Thickness  (x)

0 xc Buffer
 (x)
layer
uxc Boundary Viscous
Rcc  v
sublayer
Layer thickness
Figure 7.5 : Boundary Layer Growth for Flow over a Flat Plate

This critical value, however, strongly depends on the surface roughness and the
turbulence level of the free stream. For example, with large disturbances in the free
stream, the transition may begin at a Reynolds number as low as 105 , and for flows
which are free from disturbances, it may not start until a Reynolds number of 106 or
more.
The boundary layer concept for flow over a curved body is illustrated in Figure 7.6. Here,
the x-coordinate is measured along the curved surface of the body. By starting from the
stagnation point and at each x -location, the y-coordinate is measured normal to the
surface of the body. The free-stream velocity u  x  is not constant, but varies with
distance x along the curved surface. The boundary layer thickness   x  increases with
distance x along the surface. However, because of the curvature of the surface, after some
u
distance x, the velocity profile u (x, y) exhibits a point of inflection, i.e. becomes
y
zero at the wall surface. Beyond the point of inflection, the flow reversal takes place, and
the boundary layer is said to be detached from the surface. Beyond the point of flow
reversal, the flow patterns are complicated and the boundary layer analysis does not hold
good.
Unseparated
flow u
Boundary layer
u
y
u (x)
u u
Point of x
inflection
x Separated
y u
0
flow
y y  0

(a)

10
 Convection Heat
Transfer
Edge of
Boundary layer

Reverse flow

Separation

(b)
Figure 7.6 : Boundary Layer Growth for Flow Along (a) Curved Body and (b) Circular Cylinder

As shown in Figure.7.6, if a line 1-1 is drawn in the boundary layer and parallel to the
boundary surface so that the area 234 = area 256, the distance between the line 1-1 and
the boundary line is called the displacement thickness   . Here   represents the  
distance by which an equivalent uniform stream would have to be displaced from the
surface to give the same volume flow of the fluid.
Vol. flow

u u
3
(x)

5
1 1
6 2 4
* *

u
Figure 7.7 : Displacement Thickness *

The volume flow in the boundary layer


Q   u d y  u       . . . (7.8)
0

 
1  u 
 
u   u  u d y   1
0 0  u
d y

. . . (7.9)

7.3.2 Drag Coefficient and Drag Force

Let the velocity profile u (x, y) in the boundary layer is known. The viscous shear stress
 x acting on the wall at any location x is defined by

 u 
x     . . . (7.10)
  y  y 0
and it can be determined from the known velocity profile. In practice, shear stress, or the
local drag force per unit area  x , is related to local drag coefficient C f x by the relation

 u2
 x  Cf . . . (7.11)
x
2
Thus knowing the drag coefficient, one can calculate the drag force exerted by the fluid
flowing over the flat plate. From Eq. (7.10) and (7.11), we have

11
Convection
2v  u 
C fx    . . . (7.12)
u2  y  y 0

Thus the total drag coefficient C f x can be determined from Eq. (7.12) if the velocity
profile u (x, y) in the boundary layer is known. The mean value of the drag coefficient
over the plate length L is
L
1
C f   C f x dx . . . (7.13)
L0

Then the drag force F acting on the plate of length L and width b is equal to

 u2
F  bLC f . . . (7.14)
2
7.3.3 Thermal Boundary Layer
Analogous to the concept of velocity boundary layer, one can visualise the development
of a thermal boundary layer with temperature varying from Tw to T in the boundary
layer thickness  t as shown in Figure 7.8. Let us consider that a fluid at a uniform
temperature T
u
T
u
q’’ T T
 y
T(y) T(y)
u x
T Tw
 Velocity boundary layer
T = Thermal boundary layer

Figure 7.8 : Growth of Velocity and Thermal Boundary Layers in Flow over
a Flat Surface of Arbitrary Shape

flows along a flat plate maintained at a constant temperature Tw . If we define the

dimensionless temperature  (x, y) as
Tw  T
 ( x, y )  . . . (7.15)
Tw  T
where T (x, y) is the local temperature in the fluid, then at y = 0,   0 and at
y  ,   1 . Therefore, at each location x along the plate one finds a location
y   t  x  in the fluid where  equals 0.99. The locus of such points where  = 0.99 is
called the thermal boundary layer  t  x  .

The relative thickness of the thermal boundary layer  t  x  and the velocity boundary
layer   x  depend on the Prandtl number of the fluid. For fluids having Pr = 1,
 t  x  =   x  . For fluids having Pr <<1, such as liquid metals,  t >>  , whereas for
fluids having Pr >>1,  t <<  .

7.4 HEAT TRANSFER COEFFICIENT

If the temperature profile T (x, y) in the thermal boundary layer is known, then the heat
flux
12
 Convection Heat
 T 
q  x   k   . . . (7.16) Transfer
 y  y  0
where k is the thermal conductivity of the fluid. In practice, a local heat transfer
coefficient h (x) is defined as
q  x   hx Tw  T  . . . (7.17)

which, as we saw before, is the Newton’s law of cooling. From Eqs. (7.16) and (7.17),

 T y  y 0   
hx  k k  . . . (7.18)
T  Tw  y  y 0

The local heat transfer coefficient hx can be determined by knowing the dimensionless
temperature distribution   x, y  in the thermal boundary layer. It decreases along the
length (Figure 7.9).
hx

hx

x
dx

0 x L

1 L
Figure 7.9 : Mean Heat Transfer Coefficient hm   h x dx
L0

The mean heat transfer coefficient as shown in Figure 7.9, over the length L of the plate
is
L
1
hm  hc   hx dx . . . (7.19)
L0

The total heat transfer rate Q from the plate of length L and width b is

Q  bLhm Tw  T  . . . (7.20)

The transfer of heat from the solid wall to the fluid takes place by a combination of
conduction and mass transport. In laminar flow, heat is transferred by molecular
conduction from streamline to streamline. In turbulent flow, the conduction mechanism is
aided by innumerable eddies which carry lumps of fluid across the streamlines. These
fluid particles act as carriers of energy and transfer energy by mixing with other fluid
particles

T  Tw
Q  k f A  hA Tw  T   . . . (7.21)
f

Since k f of fluids is generally small (expect liquid metals), the rate of heat transfer very
much depends on the rate of mixing of fluid particles. Higher the Reynolds number,
higher will be the rate of mixing, lower the value of  t and higher the values of h and Q.
13
Convection SAQ 2
(a) Explain with the help of a diagram the concept of hydrodynamic and
thermal boundary layer for a laminar flow over a thin flat plate.
(b) What is critical Reynolds number?
(c) What is the physical significance of Prandtl number?
(d) Sketch the hydrodynamic and thermal boundary layer for a laminar flow for
a flow over a flat plate for (1) Pr  1 , (2) Pr  1 and (3) Pr  1 .

7.5 CONSERVATION EQUATIONS OF MASS, MOMENTUM

AND ENERGY FOR LAMINAR FLOW OVER A FLAT
PLATE
To derive the conservation of mass or continuity equation, let us consider a control
volume within the boundary layer as shown in Figure 7.10 and assume that steady-state
conditions prevail. The rates of mass flow into and out of the control volume in the
x-direction are
M x'    u  dydz . . . (7.22)

  
and M x'    u    u  dx  dydz . . . (7.23)
 x 
 v 
 v  dy 
 y 

dx
u
dy  u 
u dy  + dx 
 x 
dx
Surface of flat plate
pplate (x, y)


Figure 7.10 : Control Volume (dx  dy  1)for Conservation of Mass in an Incompressible
Boundary Layer in Flow Over a Flat Plate

Net mass flow in x-direction,


M x'  M x"    u  dxdydz . . . (7.24)
x
Similarly, net mass flow in y- and z-directions

M y'  M "y    v  dxdydz . . . (7.25)
y

and M z'  M "y     w  dxdydz . . . (7.26)
y
Net rate of mass accumulation within the control volume

 dxdydz  . . . (7.27)
t

   
    u     v     w  dxdydz
 x y z 

14
  Convection Heat
Therefore,  div V = 0 . . . (7.28) Transfer
t
For steady state,

0 . . . (7.29)
t
 div V  0 . . . (7.30)
For an incompressible fluid,

Div V  0 . . . (7.31)
u v w
i.e.   0 . . . (7.32)
x y z
For a two-dimensional flow,
u v
 0 . . . (7.33)
x y
7.5.1 Vector Method
Let us consider a fixed control surfaces S enclosing a volume V as shown in Figure 7.11.

S V

dV

dS
m

Figure 7.11 : Control Volume

The rate of accumulation of mass inside S must be equal to the rate of inflow across the
control surface minus the rate of outflow.
Rate of accumulation of mass in the control volume



t 
v
 dV

Net rate of outflow across the control surface

 S
 V dS

The surface integral is transformed to volume integral as

 div  V  dV
V

By mass balance,

t   dV =    
div V dV . . . (7.34)
V V

Writing the conservation of mass equation for the differential volume dV with the
integral sign removed 15
Convection 
t
 
  dv    div V dV . . . (7.35)

Since dV is now independent, it is stuck off both sides of the above equation, and we
obtain

 div V  0 . . . (7.36)
t
By expansion of Eq. (7.36)
   
  u     v    w   0 . . . (7.37)
t x y z

  u v w    
or      u v w 0 . . . (7.38)
t  x y z  x y z

The expressions u , etc. describe the changes in density suffered by the differential
x
element as a result of displacements of the type dx = u dt, etc. to a new position, at which
the density has a different local value. The expressions are said to denote the components

of the convection rate of change of density, while the term denotes the local time
t
rate of changes of density. The sum of local and convective components gives the total or
D
“substantial” rate of change of density, for which the notation is usually employed.
Dt
Thus,
D     
 u v w . . . (7.39)
Dt t x y z
The continuity equation may be written as
D  u v w 
    0 . . . (7.40)
Dt  x y z 
D
or    div V . . . (7.41)
Dt
Similarly, if the velocity u  u  x , y , z , t  , then the acceleration of the fluid particle is

Du u x u y u z u
   
dt x t y t z t t
u u u u
=u v w + . . . (7.42)
x y z t

  Localdifferential
Convectivedifferential

SAQ 3
(a) Derive the continuity equation for a 2-D incompressible flow
(b) What is the significance of the continuity equation?
7.5.2 Conservation of Momentum Equations
The conservation of momentum equation is obtained from application of Newton’s
second law of motion to the element. Let us consider in the flow of a fluid within the
laminar boundary layer an elementary parallelepiped of sides dx, dy and dz as shown in
Figure 7.12.

16
 p Convection Heat
p+ dy
y y Transfer
u

 + d
u + du  + d

dy
u p
p p+ dx
gx x


dy

y

gy
dz

x
dx dx
x
z
p
Figure 7.12 L: Forces Acting on a Fluid Element

Three forces act on the element: inertia force, pressure force and viscous or friction force.
Inertia Force
Fg  g x  dx dy dz . . . (7.43)

Pressure Force
 p 
Fp  p dy dz   p  dx  dy dz
 x 
p
 dx dy dz . . . (7.44)
x
Viscous Force
The velocity of fluid particles at the bottom surface of the element is less than that
of the particles within the element. Therefore, the shear stress will develop which
would tend to oppose the flow and the shear force is  dx dz. On the top surface
of the element the particles above (y + dy) move at a velocity exceeding that of the
particles within the element and hence would tend to accelerate and the shear force
would be
  
  dy  dxdz
 y 

Therefore, the net shear force is dx dy dz
y
   u 
or F  dx dy dz     dx dy dz
y y  y 

 2u
 dx dy dz . . . (7.45)
y 2
This applies only to one-dimensional flow. In three dimensions,
  2u  2u  2u 
F    2  2  2  dx dy dz
 x y z 

  2 u dx dy dz . . . (7.46)
Adding Eqs. (7.43), (7.44) and (7.45), the x-axis component of the resultant of all
the forces acting upon the considered elemental volume. 17
Convection p
Fx  Fg  Fp  F   g x  dv  dv   2u dv . . . (7.47)
v
where dv = dx dy dz
 p 
or Fx   g x     2u  dv . . . (7.48)
 x 
By Newton’s second law of motion
Du  p 
Fx    dv    gx     2u  dv . . . (7.49)
Dt  x 
Du p
or    gx    2 u . . . (7.50)
Dt x
The components of the resultant force along the y- and z-directions can similarly
be obtained
Dv p
  gy    2 v . . . (7.51)
Dt y
Dw p
  gz     2 w . . . (7.52)
Dt z
This system of three Eqs. (7.50)-(7.52) is known as Navier-Stokes differential
equation for incompressible viscous liquids. For compressible fluids it can be
shown as
Du p 1 
  gx    2u   div V
Dt x 3 x
p  1  
 gx       2u  divV  . . . (7.53)
x  3 x 
Dv p  1  
  gy      2v  divV  . . . (7.54)
Dt y  3 y 
Dw p  1  
  gz       2w  divV  . . . (7.55)
Dt z  3 z 
DV 1
or   G  grad p   (  2 V  grad div V ) . . . (7.56)
Dt 3
For incompressible fluids

Div V  0 . . . (7.57)
The Navier-Stokes equations, together with the continuity equation, form the basis
of the mechanics of viscous fluids. They together represent four equations for the
four unknowns u, v, w and p. In the case of compressible fluids we encounter an
additional unknown  , but we also have at our disposal the equation of state.

SAQ 4
Derive the momentum equation of a 2D incompressible flow. Explain the different
terms involved with the final equations.
7.5.3 Conservation of Energy Equation
We will now device the conservation of energy applied to the fluid element. The equation
governing the conduction of heat in a stationary medium, in the absence of heat sources,
18 is
 T Convection Heat
  2 T . . . (7.58) Transfer
t
T
where is the local differential or the rate of change of temperature at a point which
t
is stationary.
If the medium is in motion, as in the case of a fluid, the total or substantial rate of change
of temperature is required, which is given below.
DT T T T T
 u v z
Dt t x y z

  2T  2T  2T 
  2 T    2  2  2  . . . (7.59)
 x y z 
In presence of large pressure gradients or for fluids moving at high velocities, two
additional terms must be included to account for the compression work and for the
dissipation of energy due to friction. The complete energy equation for a compressible
fluid may thus be written as
DT DP
 cp   k  2T   V  . . . (7.60)
Dt Dt
where  V  denotes the dissipation function, first used by Lord Rayleigh, and given by

 u  2  v  2  w  2   u v  2
 V   2             
 x   y   z    y x 
2
 v w   u w  2
 
2
       divV . . . (7.61)
 z y   z x  3
The effect of viscous dissipation can be significant if the fluid is very viscous, as in
journal bearings, or if the fluid shear rate is very high. From Eq. (7.59), the following
two-dimensional expression for the energy equation without dissipation is obtained

T T   2T  2T 
u v  2  2  . . . (7.62)
x y  x y 

T T
Since the boundary layer is quite thin, under normal conditions  . Also the
y x
pressure term in the momentum equation, Eq. (7.50), is zero for flow over a flat plate,
 u 
since    0 . Eq. (7.50) thus reduces to
 x 

u u   2u  2 v 
u v  v 2  2  . . . (7.63)
x y  y y 
Then the similarity between the momentum and energy equations becomes apparent, as
given below

u u   2u 
u v  v 2  . . . (7.64)
x y  y 

19
Convection
T T   2T 
and u v  2  . . . (7.65)
x y  y 
Where v is the kinematic or momentum diffusivity and  is the thermal diffusivity, the
dimensions of both being m2/s. The ratio of these two transport properties is called the
Prandtl number Pr. Therefore,

v  c p cp
Pr    . . . (7.66)
 k k
If v   , then Pr = 1, and the momentum and energy equations are identical. For this
condition, nondimensional solutions of u (y) and T (y) are identical if the boundary
conditions are similar. Thus it is apparent that the Prandtl number controls the relation
between the velocity and temperature distributions.
The conditions which are important to study in the analysis of a heat transfer process are
Geometric Conditions
Round, rectangular, smooth or rough.
Physical Conditions
Properties of the fluid like oil, water, air, etc. c p ,  , k and  .

Boundary Conditions
Velocity and temperature distribution.
Time Condition
Steady, unsteady, periodic, etc.
Any flow or heat transfer problem can be solved by solving mass, momentum and energy
equations with appropriate boundary conditions, but in actual cases we often find
analytical solutions very complex and difficult. Broad simplifying assumptions are
frequently needed to arrive at a solution, and the experimental results vary widely from
the theoretical data.
SAQ 5
Derive the energy equation and explain the significance of different terms
appearing in the final equation.

7.6 PRINCIPLE OF SIMILARITY APPLIED TO HEAT

TRANSFER
The concept of similarity is derived from geometry. Two bodies are considered similar
when there corresponding linear dimensions are in a constant ratio to one another (Figure
7.13), such that in the two triangles
11 12 13
   C , similarity constant
11' 1'2 13'
The concept of physical similarity demands in addition to the above that all the other
physical quantities involved in a given pair of systems, e.g. force, time intervals,
velocities and temperatures, are respectively proportional to one another. For two similar
systems
2  T
 f , 2  f , 2  fT . . . (7.67)
1 1 T1
20 and so on,
 Convection Heat
Transfer
l2
l3
l2 l3

l1 l1

Figure 7.13 : Geometric Similarity of Two Triangles

where f  , f  and fT are similarity parameters for density, viscosity and temperature
respectively.
The physical laws discovered on the basis of the study of a model would apply not only
to the original system, but also to an infinite number of other systems, provided they are
physically similar to the model.
The methods of similarity allow us to generalize the experimental results with the aid of
model rules. These model rules help to establish dimensionless parameters which must
have the same values for both the model and the original system.
By Newton’s laws of motion,
mu
P  mf  . . . (7.68)
t
m1u1
so that P1  . . . (7.69)
t1

m2 u2
and P2  . . . (7.70)
t2

P2 m2 u2 t1
 . . . (7.71)
P1 m1 u1 t2

fu
or f p  fm . . . (7.72)
ft
f p ft Pt Pt
so that  1 1
 2 2  Ne . . . (7.73)
f m fu m1u1 m2u 2
where Ne  Newton number.
The dimensioned quantities are grouped together to yield meaningful dimensionless
parameters.

7.7 DERIVATION OF DIMENSIONLESS PARAMETERS

FROM THE DIFFERENTIAL EQUATIONS
The laws of similarity, as applied to heat transfer, were obtained for the first time by
Nusselt, who derived the dimensionless parameters appropriate to forced and free
convection from the differential equations together with the respective boundary
conditions.
7.7.1 Forced Convection
Let us consider two pipes of different diameters as shown is Figure 7.14, each carrying a
steady flow of a different fluid.
21
Convection  p1
Water
1

 p2

Oil
2

Figure 7.14 : Application of Similarity Principle

Both fluids are incompressible. Each flow is generated by pressure difference applied at
the ends of the pipe. Each flow is classified as forced convection. The fluids exchange
heat with the walls of the pipes, the actual direction of the heat flow being immaterial, i.e.
one fluid may be cooled while the other may be heated by the pipe. The physical
properties of the fluid are assumed to be constant, independent of temperature. The flow
of heat is considered to be steady.
The entire flow and heat transfer processes are described by the continuity, momentum
and energy equations :
Continuity Equation

 div V  0 . . . (7.74)
t
Momentum Equation
(x-component)
Du p 1 
  g    2u   divV . . . (7.75)
Dt x 3 x
Energy Equation
DT
  2 T . . . (7.76)
Dt
By denoting the quantities appropriate to each pipe by subscripts 1 and 2,
Continuity Equation

Incompressible and steady flow div V  0

For the first pipe,

u1 v1 w1

  0 . . . (7.77)
x1 y1 z1
x-component of the momentum equation
u
Forced convection,  g  0; Steady state, 0
t
u1 u u
u1  v1 1  w1 1
x1 y1 z1

1 p1   2u  2u  2u 
  v1  21  21  21  . . . (7.78)
1 x1  x1 y1 z1 

22
Energy Equation Convection Heat
Transfer
T1 T T
u1  v1 1  w1 1
x1 y1 z1

  2T1  2T1  2T1 

 1  2  2  2  . . . (7.79)
 x1 y1 z1 

We can describe the heat flow at the wall in terms of heat transfer coefficient, and
a mean temperature excess of the fluid  m above the measured wall temperature as
shown in Figure 7.15.

T

T T

T

Figure 7.15 : Heat Transfer at the Wall

T  Tw
q  h Tw  T   k  hm . . . (7.80)
t

T  Tw
q  h Tw  T   k  hm . . . (7.81)
t

  
q1  h11  k1  1  . . . (7.82)
 n1  w
Identical equations and boundary conditions apply to the second pipe except for
the suffix 1 being replaced by 2. We now postulate that the two systems are
physically similar. Thus for length scale, the proportionality factor f L is defined
as
L2
fL  . . . (7.83)
L1
Similarly, the other proportionality factors are listed below :
For the velocities u, v, w
u2
fu  . . . (7.84)
u1
For the pressure p
p2
fp  . . . (7.85)
p1
For the density 23
Convection 2
fp  . . . (7.86)
1
For the kinematic viscosity
v2
fv  . . . (7.87)
v1
For the temperature T and 
T2
fT  . . . (7.88)
T1
For the thermal diffusivities
2
f  . . . (7.89)
1
For the heat transfer coefficients h
h2
fh  . . . (7.90)
h1
For the thermal conductivities k
k2
fk  . . . (7.91)
k1

If we now introduce these proportionality factors (or similarity constants) into the
equations appropriate to the second pipe and take common factors outside the
brackets, we obtain for the second pipe (putting u2  f u .u1 , T2  f T .T1 , v2  f v .v1
and so on) :
Continuity Equation

fu  u1 v1 w1 

   0 . . . (7.92)
fL 
 1x y1 z1 

x-Momentum Equation

f L2  u1 u u 
 u1  v1 1  w1 1  . . . (7.93)
fL  x1 y1 z1 

fp 1 p1 f v f u   2 u1  2 u1  2 u1 
  2 v1  2   
2 
fL f  1 x1 fL x
 1  y 1
2
z 1 

Energy Equation

fu fT  T1 T T 
 u1  v1 1  w1 1 
fL  x1 y1 z1 

f fT   2T1  2T1  2T1 

 1  2  2  2  . . . (7.94)
f L2  x1 y1 z1 
Equation describing the heat flow :

f k fT  1 
f h fT h1 m1   k  . . . (7.95)
f L  n1  w
24
These Eqs. (7.92)-(7.95) describe the processes in the second pipe, but they Convection Heat
Transfer
become identical with those of the first pipe Eqs. (7.77)-(7.79) if the following
conditions are satisfied :
fu2 fp f f
  v 2u . . . (7.96)
fL fL f  fL

f u fT f f
  2T . . . (7.97)
fL fL
f k fT
f h fT  . . . (7.98)
fL
It may be noticed that the continuity equation yields no condition for the
fu
proportional constants, since the value of is arbitrary. If we now substitute for
fL
the proportionality factors from Eqs. (7.84)-(7.91), we obtain the conditions which
must be satisfied for physical similarity to exist between our two systems and to
permit one system to be considered as a model of the other.
From Eq. (7.96) :
u22 L1 p2 L1 1 v2 u2 L12
  . . . (7.99)
u12 L2 p1 L2 2 v1 u1 L22
u2 L2 u1 L1 uL
  . . . (7.100)
v2 v1 v

1u12  2 u22  u 2
  . . . (7.101)
p1 p2 p
From Eq. (7.97) :

u2 T2 L1 2 T2 L12
 . . . (7.102)
u1 T1 L2 1 T1 L22

u2 L2 u1 L1 uL
  . . . (7.103)
2 1 
From Eq. (7.98) :

h2 T2 k2 T2 L1
 . . . (7.104)
h1 T1 k1 T1 L2

h2 L2 h1 L1 hL
  . . . (7.105)
k2 k1 k
The resulting dimensionless parameters are designated with the names of
outstanding scientists in the field
uL
 Re or N Re , Reynolds number
v
uL
 Pe or N Pe , Peclet number


25
Convection hL
 Nu or N Nu , Nusselt number
k

( u 2 )
The dimensionless product does not represent a true similarity parameter,
p
since the pressure in a channel with prescribed dimensions and velocities, adjusts
itself to the correct value in accordance with Eq. (7.78). Thus the flow is
determined by the Reynolds number alone, if geometrically similar boundaries are
assumed.
In addition to the three similarity parameters derived above, we may choose to use
any combinations of them, as long as a total number of three independent
parameters is preserved. Thus, for example, the ratio Pe/Re = v /  = Pr or N pr ,
Prandtl number is particular useful, because it contains only the properties of the
fluid.
Our original problem is thus solved, since if the three dimensionless parameters Re,
Pr and Nu are equal in both systems, then the two systems are physically similar
and constant properties exist between all quantities concerned. This may be
immediately extended to include all geometrically similar systems (i.e. all circular
pipes) for which the dimensionless parameters have the same values. If we can
obtain the solution for any one system by any means, e.g. empirically, it must be
possible to write this solution in the form
F (Re, Pr, Nu) = 0 . . . (7.106)
and it is valid for all the systems characterized by the same values of the
dimensionless parameters. To obtain an expression for a given variable, for
instance h, from Eq. (7.106), we solve the equation explicitly for the parameter
which contains the variable, i.e. Nu.
Nu = F (Re, Pr) . . . (7.107)
k
or h F (Re, Pr) . . . (7.108)
L
Hence, the heat transfer coefficient may be predicted for all similar systems from a
single model experiment. Eqs. (7.107) and (7.108) also express the fact that the
flow of the fluid is not affected by the heat transfer. The momentum equation
Eq. (7.78) yields as a single dimensionless parameter the Reynolds number.
7.7.2 Free Convection
Eq. (7.106) is valid for forced convection only. In the case of free convection, the
buoyancy force experienced by a fluid system at a higher temperature, i.e. at a lower
density, than the surrounding fluid must be introduced into the momentum equation as a
body force.
Let us consider, as an example, a vertical wall at a temperature higher than that of the
surrounding fluid as shown in Figure 7.16. The fluid layer heated up by the wall suffers a
decrease in density and as a result experiences an up thrust relative to the surrounding
fluid.

26
 Convection Heat
T - T Transfer

T

T

 

Figure 7.16 : Temperature and Velocity Distribution in Natural Convection

1
The change in the specific volume vs  with temperature may be expressed in terms

of coefficient of thermal expansion

1  vs 
 . . . (7.109)
vs  T  p

where vs is the specific volume.

We will assume that only density varies with temperature, while other properties
 
 , c p , k are still assumed constant. For an ideal gas,

pvs  RT . . . (7.110)

RT
or vs  . . . (7.111)
p

 vs  R vs
 T   T  T . . . (7.112)
 p

1
 . . . (7.113)
T
( 0   ) g
Buoyancy force per unit mass .


Now, v  v0 1    . . . (7.114)

where   T  T0  excess temperature

1 1
 1    . . . (7.115)
 0
or  0     . . . (7.116)

 0     . . . (7.117)

 g
Buoyancy force per unit mass is   g .

27
Convection Buoyancy force per unit volume, G   g 

This is directed upward. By introducing this body force in the momentum equation,
Eq. (7.78),
u1 u u 1 p1
u1  v1 1  w1 1    g 1  1 1
x1 y1 z1  1 x1

  2u  2u  2u 
 v1  21  21  21  . . . (7.118)
 x1 y1 z1 

The subscript 1 refers to the model of a system, while subscript 2 refers to the system
itself. To describe the system itself, we require, in addition to the proportionality factors
2 g
of Eqs. (7.84)-(7.91), two more such factors defined by f   and f g  2 . In
1 g1
accordance with the procedure adopted for forced convection earlier, we obtain the
following condition

f u2 fp f f
  f g fT f   v 2u . . . (7.119)
fL fL f  fL

Since the pressure differences in a free convection flow are generally negligible, the
second term of Eq. (7.119) does not yield any dimensionless parameter. Unlike forced
convection there exists no prescribed velocity. The fluid velocity is zero both at the plate
and at a large distance from it, outside the boundary layer. The ratio f u is thus largely
meaningless and may be eliminated. For Eq. (7.119),

f u2 f f
 f g fT f   v 2u . . . (7.120)
fL fL

Equating the first and third terms, f u  f v / f L , and substituting

f v2
f g fT f   3 . . . (7.121)
fL

g 2  2  2 v22 L13
 . . . (7.122)
g1 1 1 v12 L32

g 2  2 2 L32 g1 11 L13 g  L3

   Gr . . . (7.123)
v22 v12 v2

g  L3
where Gr = Grashof number = .
v2
Eqs. (7.97) and (7.98) representing energy and heat transfer equations are equally valid
for free convection.

f u fT f f
  2T . . . (7.124)
fL fL

f k fT
and f h fT  . . . (7.125)
fL

28
 fv Convection Heat
Substituting f u  in the first equation, Transfer
fL

f v fT f f
2
  2T . . . (7.126)
fL fL

f v  f . . . (7.127)

v2  2
or  . . . (7.128)
v1 1

v2 v1 v
   Pr . . . (7.129)
2 1 

f k fT
Again, f h fT  . . . (7.130)
fL

hL
which yields  Nu, the Nusselt number. Thus, the heat transfer in free convection
k
can be described by the equation
Nu = F (Gr, Pr) . . . (7.131)

7.7.3 Unsteady Heat Conduction

Fourier’s differential equations for unsteady heat conduction can be treated in the same
manner in which we have handled the equations of forced and free convection. For
one-dimensional heat flow, we have

T  2T
 2 . . . (7.132)
t x
For physical similarity between a system and its model, we have to introduce an
t2
additional proportionality factor for time f t  . Following the same procedure, we
t1
obtain

fT f f
  2T . . . (7.133)
ft fL

t
 constant . . . (7.134)
L2
It is the Fourier number F0 , which is often referred to as dimensionless time.

SAQ 6
(a) Define Nusselt number, Grashof number and Prandtl Number.
(b) What is the principle of similarity?

29
Convection
7.8 EVALUATION OF CONVECTION HEAT TRANSFER
COEFFICIENTS
In convection heat transfer the key unknown is the heat transfer coefficient. Five general
methods are available for its evaluation :
 Dimensional analysis combined with experiments.
 Exact mathematical solutions of the boundary layer equations.
 Approximate analyses of the boundary layer equations by integral methods.
 Analogy between heat and momentum transfer.
 Numerical analysis.
All five of these techniques have contributed to our understanding of convection heat
transfer. Yet no single method can solve all the problems because each one has
limitations that restrict its scope of application.
Dimensional Analysis
It is mathematically simple and has found a wide range of applications. The chief
limitation of this method is that the results obtained are incomplete and quite
useless without experimental data. Dimensional analysis makes little contribution
to our understanding of the transfer process, but facilitates the interpretation and
extends the range of experimental data by correlating them in terms of
dimensionless groups.
To correlate the experimental data there are two methods for determining
dimensionless groups :
 Enlisting the variables pertinent to a phenomenon and rationally
grouping them. This technique is simple to use, but if a pertinent
variable is omitted, erroneous results ensue.
 Dimensionless groups and similarity conditions are deduced from the
differential equations describing the phenomenon. This method, as
presented in the previous section, is preferable if the phenomenon can
be described mathematically.
Exact Mathematical Solutions
Exact mathematical analyses requires simultaneous solution of the equations
describing the fluid motion and the transfer of energy in the moving fluid. The
physics of the problem must be well understood to describe it mathematically.
Complete mathematical equations can be written only for laminar flow under
simple boundary conditions. Exact solutions are, however, important because the
assumptions made can be specified precisely and their validity can be verified
experimentally. Availability of high speed computers has increased the range of
problems amenable to mathematical solution.
Approximate Analysis of Boundary Layer
It avoids the detailed mathematical description of the flow in the boundary layer.
Instead, a plausible but simple equation is used to describe the velocity and
temperature profiles in the boundary layer. The problem is then analyzed on a
microscopic basis by applying the equation of motion and the energy equation to
the fluid in the boundary layer. This method is relatively simple and yields
solutions within engineering accuracy to problems that cannot be treated by an
exact mathematical analysis.

30
Analogy Between Heat and Momentum Transfer Convection Heat
Transfer
It is a useful tool for analyzing turbulent transfer processes. Our knowledge of
turbulent-exchange mechanisms is not good enough to write mathematical
equations describing the temperature distribution directly, but the transfer
mechanism can be explained in terms of a simplified model. One such model
explains a mixing motion in a direction perpendicular to the mean flow accounting
for the transfer of momentum as well as energy, similar to that used to picture the
motion of gas molecules in the kinetic theory. Experimental results are
substantially in agreement with analytical predictions based on the model.
Numerical Methods
They can solve approximately the equations of motion and energy. The
approximation results from the need to express the field variables (velocity,
temperature and pressure) at discrete points in time and space rather than
continuously. However, the solution can be made sufficiently accurate if care is
taken in discretising the exact solutions. One of the most important features of
numerical methods is that once the solution procedure has been programmed,
solutions for different boundary conditions, property variables and so on can be
easily computed. Generally, numerical methods can handle complex boundary
conditions easily.

7.9 DIMENSIONAL ANALYSIS

In engineering, we represent physical concepts by symbols or dimensions, as for instance
length by L and velocity by V. Through experience, we have learned that we can select a
certain number of dimensions as fundamental, and express all other dimensions in terms
of products of powers of these fundamental dimensions. The dimensions of commonly
used quantities in heat transfer analysis are listed in Table 7.1 with four fundamental
dimensions : mass M, length L, time t and temperature T. For example, the dimension of
heat transfer coefficient is W/ m 2 K. In MLtT system, it is (Nm/s) (1/ m 2 K) or, (kg m/ s 2 )
(1/s m k) or kg/ s 3 K which is M/ t 3 T (or M t 3T 1 ). Similarly, all other quantities can be
expressed in fundamental dimensions, as given in Table 7.1.

The methods of dimensional analysis are founded upon the principle of dimensional
homogeneity, which states that all equations describing the behaviour of physical
systems, must be dimensionally consistent, i.e. each term with reference to a given set of
fundamental dimensions must have the same dimensions. When the equations governing
a process are known and solvable, dimensional analysis suggests logical grouping of
quantities for presenting the results. When the mathematical equations governing certain
processes are unknown or too complex, dimensional analysis lays the foundation of an
efficient experimental program for obtaining the results, by reducing the number of
variables requiring investigation and by indicating a possible form of the semi-empirical
correlations that may be formulated. It should be borne in mind that dimensional analysis
by itself cannot provide quantitative answers, and thus cannot be a substitute for the exact
or the approximate mathematical solutions. It is nevertheless an important tool to learn to
use, especially in instances when mathematical analysis is impractical or when some
rapid, qualitative answers are needed.
Table 7.1 : Important Heat Transfer Physical Quantities and their Dimensions

Quantity Symbol Dimensions in MLtT

Length L,x,y L

Time t t

Mass M M
31
Convection 2
Force F ML t

Temperature T T

2 2
Heat transfer Q ML t

Velocity u , v, u  Lt
1

2
Acceleration a,g LT

2 2
Work W ML t

2 1
Pressure p Mt L

Density  ML
3

Internal energy e L2 t 2

Enthalpy h L2 t 2

Specific heat c L2 t 2 T 1
Absolute  1 1
viscosity ML t

Kinematic
viscosity
v L2 t 1

Thermal 3 1
conductivity
k ML t T

Thermal 
diffusivity L2 t 1

Thermal 3 1 2
resistance
R Tt M L

Coefficient of
 T 1
expansion

Surface tension  Mt
2

Shear stress  ML t
1 2

Heat transfer 3 1
coefficient
h Mt T

1
Mass flow rate m Mt
The application of similarity principle to the continuity, momentum and energy equations
for identifying the dimensionless parameters that govern the concerned process, which
was discussed earlier, is also based on the principle of dimensional homogeneity.
However, when the governing equations of a problem are unknown, an alternate
approach in the application of dimensional analysis is necessary. At the very start, it is
required to know, or more typically to guess, the independent variables that determine the
behaviour of a particular dependent variables of interest. These can usually be found by
logic or intuition developed from previous experiences with problems of a similar nature,
but there is no way to ensure that all essential quantities have been included or not.
Rayleigh first used this method and the rules of algebra to combine the many variables of
32
a problem into dimensionless groups. We are providing in the next section two examples Convection Heat
Transfer
of application of Rayleigh’s method.
7.9.1 Rayleigh’s Methods
Method-I
Let us consider the frictional resistance of fluid flow per unit area of the inside
surface of the pipe. A reasonable assumption can be made that the resistance which
causes pressure drop of the fluid  p  is a function of tube diameter (D), fluid
density    , fluid velocity (u) and fluid viscosity    , or
p  f [u , D ,  ,  ] . . . (7.135)
Let p  Cu a D b  c  d . . . (7.136)
where C is a dimensionless constant. The dimensional equation of the above
expression in fundamental dimensions M, L and t are
MLt 2
  L  ML  ML 
a c d
b
2
 Lt 1 3 1 1
t . . . (7.137)
L
ML1t 2  La b3 c d t  a d M c d . . . (7.138)
For the homogeneity of
M : 1 = c + d,
L :  1 = a + b – 3 c – d and,
t :  2 =  a – d.
On solving these equations we have b =  d, c = 1 – d and a = 2 – d.
p  Cu 2  d D  d  1 d  d
d
    u2
 C u  2
  C . . . (7.139)
  uD  Re dD
 uD
where Re d   Reynolds number. The values of constants C and D have to

be determined by experiments.
Method-II
Let us consider forced convection heat transfer between a fluid flowing through a
pipe and its wall. We enlist the variables pertinent to the phenomenon by logic or
intuition and group them into dimensionless parameters. Let us assume that the
heat transfer coefficient h is a function of pipe diameter (D), fluid velocity (u), and
the fluid properties of density    , viscosity    , thermal conductivity (k) and

 
specific heat c p , or

h  f  D , u ,  ,  , k , c p  . . . (7.140)

h  BD a u b  c  d k e c pf . . . (7.141)
where B is a constant. Expressing the quantities in terms of fundamental
dimensions M, L, t and T

  ML  ML 
b c d
Mt 3T 1  BLa Lt 1 3 1 1
t

 MLt  L t 
e f
3
T 1 2 2
T 1 . . . (7.142)

or, Mt 3T 1  BLa b3 c d  e 2 f t  b d 3 e 2 f M c d  eT  e f . . . (7.143)

For dimensional homogeneity of
M:1=c+d+e 33
Convection L : 0 = a + b – 3c – d + e + 2f
t : -3 = – b – d – 3e – 2f
T:–1=–e–f
On solving the above equations, a  c  1, b  c , d   c  f and e  1  f .
Therefore, h  BD c 1u c  c   c  f k 1 f c pf . . . (7.144)
c f
hD  uD    c p  
 B    . . . (7.145)
k     k 
or, Nud  B Re cd Pr f . . . (7.146)

uD  cp 
where Re d   Reynolds number, Pr = = Prandtl number and
 k
hD
Nud   Nusselt number. Thus, in forced convection heat transfer, Nusselt
k
number is a function of Reynolds number and Prandtl number. The constants B, c
and f have to be evaluated from the experimental data.
SAQ 7
(a) Explain Rayleigh’s method.
(b) How the Rayleigh’s method is applied in deriving the functional relationship
of pressure drop of a fluid per unit area of the inside surface of a pipe?
(c) Show by Rayleigh’s method that in forced convection heat transfer the
Nusselt number is a function of Reynolds number and Prandtl number.
7.9.2 Buckingham -Theorem
A simple and more systematic way of determining the dimensionless groups was
suggested by Buckingham and has come to be known as “pi-theorem”. If a small number
of physical quantities is involved, the Rayleigh method is simpler. But if the number of
physical quantities increases beyond a given limit, the procedure is tedious, and the
pi-theorem may be used advantageously. More physical quantities simply mean a few
more  -terms. Each  -term can be solved exactly the same way as in the case of fewer
physical quantities. According to the Buckingham  -theorem, any physical equation
may be described by
  Q1 , Q2 , Q3 ,..., Qm   0 . . . (7.147)

which is a function of m common quantities Q1 , Q2 ,..., Qm . If n fundamental dimensions

M, L, t, and T, etc. are chosen, then the equation may be transformed into a new equation
containing  m  n  dimensionless terms represented by  as follows :

  1 ,  2 ,  3 ...,  m n  . . . (7.148)

where each  -term consists of (n + 1) quantities of Qs . To determine the  -terms, n of

the Q quantities have to be so chosen that they contain all the fundamental dimensions M,
L, t and T, and these are taken as repeating variables which form a dimensionless number
with each of the remaining variables.
We are giving below a few examples illustrating the application of Buckingham
 -theorem.
(a) We want to find out the relationship of pressure drop of a fluid per unit
length of a pipe through which it is flowing. We know or guess that p
depends upon the tube diameter (D), and velocity (u), density    and
viscosity    of the fluid. The physical equation may be written as
34
   u , D,  ,  , p   0 . . . (7.149) Convection Heat
Transfer
There are five variables, or m = 5. Let us suppose that the three fundamental
dimensions are M, L and t, or n = 3. Therefore, the number of dimensionless
 -terms = m – n = 5 – 3 = 2. Therefore, Eq. (7.149) can be written as
 1 ,  2   0 . . . (7.150)

where each  -term consists of n + 1 = 3 + 1 = 4 common quantities. Let the

repeating variables (n) be u, D and  (which contain all the fundamental
dimensions M, L and t), and these will form a dimensionless number with
each of the remaining variables, i.e. p and  .

Let 1   a ub Dc  p . . . (7.151)

and  2   eu f D g  . . . (7.152)

 1   ML3   Lt 1   L   Mt 2 L2 
a b c
Therefore, . . . (7.153)

M 0 L0 t 0  M a 1 L3 a b c 2 t b 2 . . . (7.154)
In order to make  1 dimensionless, the exponents of M, L and t must all be
equal to zero. For dimensionless homogeneity
a 1 0
 3a  b  c  2  0
b20
On solving we obtain a   1, b   2, c  1 .

pD
 1   1 u 2 D  p  . . . (7.155)
u2

 2   ML3   Lt 1   L  ML
e f
g 1 1
Similarly, t . . . (7.156)

or M 0 L0 t 0  M e 1 L3 e  f  g 1t  f 1
e 1 0
 3e  f  g  1  0
 f 1 0
e  1, f  1, g  1

or  2   1 u 1 D 1   . . . (7.157)
 Du
Therefore, Eq. (7.150) can be written as
 pD  
 2 ,  0 . . . (7.158)
  u  Du 
It can be written as
pD 1
   Re d  . . . (7.159)
u2 2
 Du
where Re d 

35
Convection
u2
or p    Re d  . . . (7.160)
2D
If the length of the pipe is L, the total pressure drop is
fL  u 2
p1  p2  pL  . . . (7.161)
D 2
where f is the Darcy-Weisbach friction factor for a smooth pipe, which is a
function of Reynolds number, i.e. f    Re d  .

(b) Forced Convection Heat Transfer

The heat transfer coefficient can be expressed as a functional relation
h  f  D,  , u ,  , c p , k  . . . (7.162)

or,   D,  , u ,  , c p , k , h   0 . . . (7.163)

Here, the number of variables, m = 7. Let the number of fundamental

dimensions be n = 4, which are M, L, t and T. The number of
 -terms = m – n = 3.
Therefore,   1 ,  2 ,  3   0 . . . (7.164)

Each  -term is composed of (n + 1) i.e. five quantities of which four are

repeating variables, forming a dimensionless number with each of the
remaining variables
   , D, u , c p ,  , k , h   0 . . . (7.165)

Let  , D, u and c p be taken as repeating variables which contain all the

four fundamental dimensions M, L, t and T and which will make a
dimensionless number with each of  , k and h.

Let  1   a Db u c c dp . . . (7.166)

M 0 L0 t 0 T 0  ML1t 1 ( ML3 ) a L  Lt 1
b
  L t
c 2 2
T 1 
d

 M 1 a L13a b  c  2 d t 1c  2 d T  d . . . (7.167)

Therefore, 0 = 1 + a
0  1  3a  b  c  2d  0
0  1  c  2d
0 d

Solving these, we obtain, a =  1, b =  1, c =  1 and d = 0.


 1   1 D 1u 1c 0p  . . . (7.168)
 Du
Similarly, 2  k  a1 D b1 u c1 c dp1

   L  Lt  L t 
a1 b1 c1 d1
 MLt 3T 1 ML3 1 2 2
T 1 . . . (7.169)

M 0 L0T 0 t 0  M 1 a1 L13a1 b1  c1  2 d1t 3c1 2 d1T 1 d1 . . . (7.170)

1  a1  0
36
 1  3a1  b1  c1  2d 1  0 Convection Heat
Transfer
 3  c1  2d 1  0
 1  d1  0
Solving these equations, we obtain
a1  1 b1  1 c1  1 d 1  1
k
 2  k  1 D 1u 1c p1  . . . (7.171)
 Duc p

 3  h a D b u c c dp
2 2 2 2

  ML   L  Lt  L t 
a2 b2 c2 d2
 Mt 3T 1 3 1 2 2
T 1 . . . (7.172)

M 0 L0 t 0T 0  M 1 a2 L3a2 b2  c2  2 d2 t 3c2 2 d2T 1 d2 . . . (7.173)

Therefore, 1  a2  0

 3a 2  b2  c 2  2d 2  0
 3  c 2  2d 2  0
1  d2  0
On solving, a2  1 b2  0 c2  1 d 2  1

h
 3  h 1 D 0 u 1c p1  . . . (7.174)
cpu

  1 ,  2 ,  3   0 . . . (7.175)

It can be written as
 3  B 1m 2n . . . (7.176)

m n
h     k 
 B    . . . (7.177)
cpu   Du    duc p 
m n
hD k     k 
.  B    . . . (7.178)
k  c p uD   Du    Duc p 
m n 1
hD     k  
 B   
 Du   c 
k  p Du 
 m  n 1 1 n
  Du   cp 
 B    . . . (7.179)
    k 
or, Nud  B Reda Pr b . . . (7.180)

which is the same as Eq. (7.146).

37
Convection (c) Free Convection Heat Transfer
The heat transfer coefficient depends upon the buoyancy force per unit mass
 g   , density    , vertical height (L), viscosity    , thermal
 
conductivity (k) and specific heat c p . Thus, it can be written

   , L,  , k , c p , g  , h   0 . . . (7.181)

The number of variables, m = 7. The number of fundamental dimensions,

n = 4, as before. The number of  -terms = 7  4  3. Each  -term will
consist of (4 + 1) or 5 variables. Let  , L,  and k be the repeating
variables, which contain all the fundamental dimensions. Now,

  1 ,  2 ,  3   0 . . . (7.182)

Then,  1   a Lb  c k d g


 ML3  L  ML
a b 1 1 c
t  MLt 3
T 1  Lt 
d 2

 M a c  d L3a b c  d 1t c  3d  2T  d

 M 0 L0 t 0T 0 . . . (7.183)
acd 0
 3a  b  c  d  1  0
 c  3d  2  0
d 0
On solving we get a  2, b  3, c   2 and d  0

g  L3  2
 1   2 L3  2 k 0 g   . . . (7.184)
2
 2   a Lb m c k d c P
1 1 1 1


 ML3  L  ML
a1 b1 1 1 c1
t  MLt 3
T 1  L t
d1 2 2
T 1 
 M a1  c1  d1 L3a1  b1  c1  d1  2 t  c1 3d1 2 T  d 1
 M 0 L0 t 0T 0 . . . (7.185)
a1  c1  d 1  0
 3a1  b1  c1  d 1  2  0
 c1  3d 1  2  0
 d1  1  0
Solving a1  0, b1  0, c1  0 and d1  1.

 cP
 2   0 L0  1 k 1c P  . . . (7.186)
k
 3   a Lb  c k d h
2 2 2 2

38
 
 ML3  L  ML
a2 b2 1 1 c 2
t  MLt 3
T 1  Mt
d2 3
T 1  Convection Heat
Transfer

 M a2  c2  d 2 1 L3a2  b2  c2  d 2 t  c2 3d 2 3T  d 2 1

 M 0 L0 t 0T 0 . . . (7.187)
a2  c2  d 2  1  0
 3a 2  b2  c 2  d 2  0
 c 2  3d 2  3  0
 d2 1  0
Solving, a2  0, b2  1, c2  0 and d 2  1.
Table 7.2 : Dimensionless Group for Heat Transfer and Fluid Flow
Group Definition Physical Interpretation

hc L Dimensionless heat transfer coefficient;

Nusselt Number  Nu L  ratio of convection heat transfer to
kf conduction in a fluid layer of thickness L

Peclet Number  PeL  Re L Pr Product of Reynolds and Prandtl numbers

cP  v
Prandtl Number (Pr)  Ratio of molecular momentum diffusivity
k 
Rayleigh Number (Ra) GrL Pr Product of Grashof and Prandtl numbers

U xL
Reynolds Number  Re L  Ratio of inertia to viscous forces
v

hc Nu L

Stanton Number U  C P Re L Pr Dimensionless heat transfer coefficient

hL
 3   0 L1 0 k 1 h  . . . (7.188)
k
Therefore, Eq. (7.182) becomes
 g  L3  2  c p hL 
 , ,  0 . . . (7.189)
  2
k k 
a b
hL  g  L3    c p 
or,  B    . . . (7.190)
k  V
2
  k 
g  L3
where Gr = Grashof number =
V2
cp
Pr = Prandtl number =
k
Nu = Nusselt number = (hL)/k
a
Nu = BG r P r a . . . (7.191)
Dimensionless analyses have been performed on many heat transfer systems,
and Table 7.2 summarizes the most important dimensionless groups used in
design.
39
Convection SAQ 8
(a) State Buckingham  theorem.
(b) What are the merits and demerits of Buckingham  theorem.
(c) What are the repeating variables? How are these chosen?
Example 7.1
An application involving exact solution of momentum and energy equations is met
within a situation called Coquette flow with heat transfer. Coquette flow provides a
simple model for flow between two parallel plates. The lower plate is stationary
and the upper plate at a distance L is moving with a velocity U. the lower and
upper plates are maintained at uniform temperatures T0 and TL respectively. This
model can be applied to the case of a shaft rotating in its stationary bearing with a
heavy lubricating oil in the clearance. If the clearance is small, the situation can be
considered as that of flow between two parallel plates. Find also the surface heat
fluxes to the plates.
Solution
With reference to Figure 7.17 given below, the continuity equation for steady-state
in compressible flow
Moving plate
u
TL

 = 0, W = L

u=0
T0

Stationary plate
Figure 7.17

u
 0 or u  f  x 
x
The velocity does not vary with x.
The x-momentum equation.

2u 2u 2u

or   0
x 2 y 2 x 2

 2u
0
y 2
u
 C1 y
y
or, u  C1 y  C2, which is linear,
At y  0, u  0,
 C2  0
y  L, u  U U  C1 L or C1  U / L
The velocity distribution in Couette flow is then given by
40
 U Convection Heat
u  y  y Transfer
L
u y
or, 
U L
The energy equation can be written as
DT
k  2T     c p
Dt
2
 2T  u   T T T T 
k 2      c p  u v w 
y  y   t x y z 
2
 2T U 
k 2    0
y L
2
d 2T  U 
2
  
dy kL
2
dT  U 
    y  C3
dy kL
2
 U 
T     y 2  C3y  C 4
2k  L 
At y  0, T  T0 ,
 C4  T0
At y  L, T  TL ,
2
 U  2
 TL     L  C 3L  T0
2k  L 
TL  T0  U 2
C3  
L 2k L
2
 U  T  T0  U2
 T  y     y2  L y y  T0
2k  L  L 2k L
 2  y y2  y
or T  y   T0  U   2   T L  T0 
2k L L  L
In dimensionless form,

T  y   T0
2
y  2 y/L y /L
  U
2
 
TL  T0 L 2k TL  T0 
y U 2  y 
 1  1  
L  2k TL  T0   L 

y cp
Let   , Pr  ,
L k
U2
E = Eckert number =
c p TL  T0  41
Convection T  y   T0  1 
and     ;     1  Pr . E 1   
TL  T0  2 
This is the temperature distribution in Couette flow. Because of viscous dissipation,
the maximum temperature occurs in the fluid and heat transfer occurs both to the
hot and cold plates. For Pr . E = 0, there is no flow (U = 0) and hence there is no
viscous dissipation, and the temperature distribution is linear. The surface heat
fluxes can be obtained by Fourier’s law:
dT  1 1  T T 
q  y   k  k  U 2   2 2y   L 0 
dy  2k L L  L 

 U 2 TL  T0  k U 2 k
q y 0  k    TL  T0  Bottom plate
2k L L 2L L
 U2 k U 2 k
qyL  k  TL  T0    TL  T0  Top plate
2k L L 2L L
These are the heat fluxes.
Example 7.2
A heavy lubricating oil (   0.8 Ns/m 2 , k  0.15 W/mK) flows in the
clearance between a shaft and its bearing. It the bearing and shaft are kept at 10oC
and 30oC respectively and the clearance between them is 2 mm, determine the
maximum temperature in the oil and the heat flux to the plates for a velocity
U = 6 m/s.
Solution
Because of small clearance between the journal and its bearing, the flow between
them may be assumed as Couette flow. The surface heat fluxes are
U 2 K
q 0    TL  T0 
2L L
0.8  6 2 0.15
 3
  30  10 
2  2 10 2 10 3
 7200  1500  8700 W/m 2   8.7 kW/m 2
U 2 k
q L   TL  T0   7200  1500
2L L
 5700 W/m 2  5.7 W/m 2
dT
The maximum temperature in the oil will occur where 0
dy
2
 U 
   y  C3  0
kL
2
 U  TL  T0  U 2
  y  
kL L 2k L

TL  T0 kL2  U 2 kL2
y 
L U 2 2k L U 2

42
 Convection Heat
 k 1
2  L
 T  T0   L Transfer
 U 2 

 0.15 1
y  30  10    L  0.604L
 0.8  36 2

 2  y y 2  TL  T0
Tmax  T0  U  2 
 y
2k L L  L

0.8
 36 0.604  0.604    30  10 0.604
2
 10 
2  0.15  

 10  22.944  12.8  45.74 0 C.

Exercise 9.1
(a) With the help of Buckingham   theorem show that
(i) for forced convection heat transfer

Nud  B Reda Pr b

(ii) for free convection heat transfer

Nu  BGr a Pr b
(b) The efficiency  of a fan depends on density  , dynamic viscosity  of
the fluid, angular velocity  , diameter D of the rotor and the discharge Q .
Express  in terms of dimensionless parameters.

(c) The resulting force R of a supersonic plane during flight can be considered
as dependent upon the length of the aircraft l , velocity V , air viscosity  ,
air density  and bulk modulus of air K . Express the functional
relationship between these variables and the resistance force.
(d) Using Buckingham   theorem, show that the velocity through a circular
orifice is given by

D  
V  2 gH   , 
 H VH 
where H is the head causing flow, D is the diameter of the orifice,  is
the coefficient of viscosity,  is the mass density, and g is the acceleration
due to gravity.
Exercise 9.2
(a) A ship 300 m long moves in sea-water, whose specific weight is
1030 kgf/m3. A 1:100 model of this ship is to be tested in a wind tunnel.
The velocity of air in the wind tunnel around the model is 30 m/s and the
resistance of the model is 60 kgf. Determine the velocity of the ship in sea
water and also the resistance of the ship in sea water. Assume the following
data :
Specific weight of air = 1.24 kgf/m3
Kinematic viscosity of sea water = 0.012 strokes
Kinematic viscosity of air = 0.018 strokes
43
Convection (b) Air at 25oC and at atmospheric pressure flows over a flat plate at 3 m/s. If
the plate is 1 m wide and the wall is maintained at 75oC, calculate the
following at locations x  1 m and x  xc from the leading edge :

(i) hydrodynamic and thermal boundary layer thickness,

(ii) local and average friction coefficients,
(iii) local and average heat transfer coefficients,
(iv) the total rate of heat transfer, and
(v) the total drag force due to friction.
Properties of air at 50oC are   1.093kg/m 3 , c p  1.005 kJ/kgK,
v  17.95  10  6 m 2/s and k  0.0282 W/mK .
(c) Air at 27oC and 1 atm pressure flows over a flat plate with a velocity of
2 m/s. Estimate
(i) the boundary layer thickness at distances of 20 cm and 40 cm from
the leading edge of the plate,
(ii) the mass flow that enters the boundary layer between x = 20 cm and
x = 40 cm. Take  of air at 27oC as 1.85  10 5 kg/ms. Assume unit
depth in z-directions. If the plate is heated over its entire length to a
temperature of 60oC, calculate the heat transfer in,
(iii) the first 20 cm of the plate,
(iv) the first 40 cm of the plate, and
(v) Compute the drag force exerted on the first 40 cm of the plate.
Properties of air at 316.5 K are v  17.36  10  6 m 2 /s ,
k  0.02749 W/mK, Pr = 0.7 and c p  1.006 kJ/kg K.

7.10 SUMMARY
In the present unit background of convective heat transfer has been discussed.
Derivations of continuity, momentum and energy equations are given. Concept of
hydrodynamic as well as temperature boundary layers are given at length. Detail
discussion is provided on similarity principles. The unit provides detail discussion on
dimensional analysis with workout problems.

7.11 KEY WORDS

Convection : A mode of heat transfer due to molecular diffusion
and bulk motion of fluid.
Newtonian Fluids : Fluids exhibiting a linear relation between the rate
of strain and the applied shear stress.
Viscosity : Property of a fluid, indicator of viscous loss or
frictional loss at the interface as the fluid flow over
a solid surface.
Inviscid Fluid : Fluid having no viscosity.
Reynolds Number : Dimensionless number indicating ratio of inertia to
viscous force.

44
Nusselt Number : Dimensionless heat transfer coefficient; ratio of Convection Heat
Transfer
convection heat transfer to conduction heat
transfer in a fluid layer.

7.12 ANSWERS TO SAQs

Please refer the relevant preceding text in this unit for answers to SAQs.

REFERENCES
J. P. Holman (2002), Heat Transfer, 9th Edition Tata McGraw-Hill, New Delhi.
M. N. Ozisik (1985), Heat Transfer A Basic Approach, McGraw-Hill International
Edition.
F. P. Incropera and D. P. Dewitt (2004), Fundamentals of Heat and Mass Transfer,
5th Edition John Wiley and Sons.
P. K. Nag (2002), Heat Transfer, Tata McGraw Hill Publishing Company Limited,
New Delhi.
A. Bejan (1985), Convective Heat Transfer, 2nd Edition, John Wiley and Sons, INC.

45
 Boundary Layer
UNIT 8 BOUNDARY LAYER FORMATION Formation

Structure
8.1 Introduction
Objectives
8.2 Analytical Solution for Laminar Boundary Layer Flow Over a Flat Plate
8.3 Convection Heat Transfer
8.4 Approximate Integral Boundary Layer Analysis
8.5 Turbulent Flow over a Flat Plate : Analogy between Momentum and Heat
Transfer
8.6 Reynolds Analogy for Turbulent Flow over a Flat Plate
8.7 Constant Heat Flux Boundary Condition
8.8 Boundary Layer Thickness in Turbulent Flow
8.9 Forced Convection Inside Tubes and Ducts
8.10 Analysis of Laminar Forced Convection in a Long Tube
8.11 Heat Transfer Coefficient for Laminar Flow in a Tube
8.12 Velocity Distribution in Turbulent Flow through a Pipe
8.13 Analogy Between Heat and Momentum Transfer in Turbulent Flow
8.14 Empirical Correlations
8.15 Summary
8.16 Key Words
8.17 Answers to SAQ’s

8.1 INTRODUCTION
In this unit convective heat transfer is discussed in details based on boundary layer theory.
The concept of boundary layer is a tool specially to treat the effect of fluid viscosity in a
demarcated domain delinking the potential and viscous regimes in a flow field. The
discussion is presented based on laminar boundary layer and extended the theory to
turbulent boundary layer. Both the hydrodynamic and thermal boundary layer concepts
are given in appropriate places. Convective heat transfer for flow over different
geometries are discussed here.
Objectives
After studying this unit, you should be able to
 treat different type of problems with the boundary layer concept,
 distinguish laminar and turbulent heat transfer, and
 formulate and evaluate correlations pertaining to heat transfer and
hydrodynamics for external and internal flow.

8.2 ANALYTICAL SOLUTION FOR LAMINAR BOUNDARY

LAYER FLOW OVER A FLAT PLATE

To determine the forced convection heat transfer coefficient h c and the friction
coefficient C f for incompressible steady laminar flow over a flat surface, we must
satisfy the continuity, momentum and energy equations simultaneously. For
two-dimensional flow :
47
Convection Continuity
u v
 0 . . . (8.1)
x y
Momentum
u v  2u
u v v 2 . . . (8.2)
x y y
Energy
T T  2 T
u v  . . . (8.3)
x y y 2
8.2.1 Boundary-layer Thickness and Skin Friction Coefficient
Eq. (8.2) must be solved simultaneously with the continuity equation, Eq. (8.1), in order
to determine the velocity distribution, boundary layer thickness and skin friction
coefficient. Let us first make an order-of-magnitude analysis of the differential equations
to obtain the functional form of the solution. Within the boundary layer we may say that
the velocity u is of the order of free-stream velocity u  . Similarly, the y-dimension is of
the order of the boundary layer thickness  . Thus,
u ~ u . . . (8.4)
y~  . . . (8.5)
and we might write the continuity equation in an approximate form as
u v
 0 . . . (8.6)
x y
u v
 0 . . . (8.7)
x 
u 
v ~  . . . (8.8)
x
Then, by using this order of magnitude for v , the analysis of momentum equation would
yield
u u  2u
u v  2 . . . (8.9)
x y y
u u  u u
u    2 . . . (8.10)
x x  
x
2 ~ . . . (8.11)
u
1/ 2
 x 
or, ~  . . . (8.12)
 u 
Dividing by x to express the result in dimensionless form gives
1/ 2
    1
~   . . . (8.13)
x  u x   Re x 
1/ 2

This is the functional relationship of  with the local Reynolds number. Since the
velocity profiles have similar shapes at various distances from the leading edge of the flat
y
plate, the important variables is , and we assume that the velocity may be expressed as

a function of this variable.
48
 Boundary Layer
u  y
 g  . . . (8.14) Formation
u  
Introducing the order-of-magnitude estimate for  from Eq. (8.12),
u
 g   . . . (8.15)
u
1/ 2
y u 
where  1/ 2
 y   . . . (8.16)
 Vx   Vx 
 
 u 
Here,  is called the similarity variable and g   is the function for which we seek a
solution. In accordance with the continuity equation, a stream function  may be
defined so that

u . . . (8.17)
y

v . . . (8.18)
x
Inserting Eq. (8.17) in Eq. (8.15) gives
1/ 2
 Vx 
   u  g  dy   u    g  d . . . (8.19)
 u 
1/ 2
 Vx 
or   u    f   . . . (8.20)
 u 
where f     g  d .

From Eqs. (8.18) and (8.20), we obtain

1/ 2
1  Vu   df 
v      f  . . . (8.21)
2 u   d 
u u  2u
Expressing , and 2 in terms of  and inserting the resulting expressions in the
d y y
momentum equation, Eq. (8.2), we obtain the ordinary, nonlinear, third-order differential
equation
d2 f d3 f
f  2 0 . . . (8.22)
d 2 d 3
It may be solved numerically for the function f   subject to the three boundary
conditions :
Physical Coordinates Similarity Coordinates
df
1. u = 0 at y = 0 0 at  0
d
2. v0 at y0 f = 0 at  0
u df
3. 0 at y  1.0 at  
y d 49
Convection The solution to Eq. (8.22) was first obtained by Blasius in 1908[1]. The important results
are shown in Figures (8.1) and (8.2).

1.0

 = 1.0 cm
+  = 2.0 cm
Slop = 0.332  = 2.5 cm

u  = 4.0 cm
u  = 5.0 cm
 = 7.5 cm
 = 10.0 cm
0.5
 = 12.5 cm
U 8 m/s
 = 15.0 cm
Thin plate  = 17.5 cm
(No.1)
Blasius

0
0 1 2 3 4 5 6 7

y pUx
x μ

Figure 8.1 : Velocity Profile in a Laminar Boundary Layer according

to Blasius, with the Experimental Data of Human [2]

100.0

10.0
Transition Range
Cfx x103

1.0

0.1

10 102 103 104 105 106

Rex

Figure 8.2 : Local Friction Coefficient Varying with Reynolds Number along the Distance from
Leading Edge for Laminar Flow over a Flat Plate

Figure 8.1 shows the Blasius velocity profiles in the laminar boundary layer on a flat
plate in dimensionless form together with the experimental data of Hansen [10]. The
ordinate is the ratio of local and free-stream velocities, and the abscissa is a
1/ 2
 y  x
dimensionless distance parameter    u  . The velocity u reaches 99% of the
x  

50
 1/ 2 Boundary Layer
 y  x Formation
free-stream velocity u  at    u   5.0 . If we define the hydrodynamics
x  
boundary layer thickness  as that distance from the surface at which the local velocity
u reaches 99% of the free-stream velocity u  , then

 5
 . . . (8.23)
x Re x 1 / 2

where Re x 
 u x  or
 u x  . Eq. (8.23) satisfies the qualitative description of the
 V
boundary layer growth, as was shown by order-of-magnitude analysis in Eq. (8.13). It
explains that at x = 0,   0 , and as x increases  increases. Again, where u  increases,
 decreases. The shear stress at the wall can be obtained from the velocity gradient at y
= 0 in Figure 8.1. We find that

 u / u  
 0.332 . . . (8.24)
  y / x Re x 
1/ 2
y 0

 u  u
   0.332  Re x 1 / 2 . . . (8.25)
 y  y  0 x

The wall shear stress becomes

 u  u
 w      0.332   Re x 1 / 2 . . . (8.26)
 y  y  0 x

It may be noted that the wall shear stress at the leading edge (x = 0) is very large, and it
decreases with increasing distance (x) from the leading edge. Dividing both sides of
u 2
Eq. (8.26) by the dynamic pressure of the free-stream,  , we obtain
2
w u 2 0.664
C fx   0 . 332  Re 1/ 2
 . . . (8.27)
u  Re x 1 / 2
x
u  / 2
2
x 2

where C f x  v '1 is the dimensionless local drag or friction coefficient. Figure 8.2 shows

the variation of C f x  v '1 with Re x . The average friction coefficient is obtained by

integrating Eq. (8.27) over the whole length of the plate, or
L
1 1.328
C f   C f x dx  2(C f x ) x L  . . . (8.28)
L0 (Re L )1 2

u L
where Re L  .
V
SAQ 1
(a) Define drag coefficient and drag force.
(b) What is boundary layer thickness?
(c) What do you mean by laminar and turbulent boundary layer?

51
Convection
8.2 CONVECTION HEAT TRANSFER
The velocities u and v in the energy conservation Eq. (8.3) have the same values at any
point ( x, y ) as in the momentum Eq. (8.2). For the case of the flat plate, Pohlhausen [1]
used the velocities calculated previously by Blasius to obtain the solution for the heat
transfer problem. If the momentum and energy equations are compared, we find them to
be similar if V   and the surface temperature Tw is constant. A solution for u (x, y) is
also a solution for T (x, y) which can easily be checked if the symbol T is replaced by the
symbol u while the boundary conditions for both are identical. If we define a
dimensionless temperature

T    Tw
    . . . (8.29)
T  Tw

u
then   0 and  0 at y = 0
u

u
  1 and  1 at y  
u

where Tw is the wall or surface temperature and Tw is the free-stream temperature. The
 V
condition V   corresponds to a Prandtl number   of unity. For Pr = 1, the
  
velocity distribution is identical to the temperature distribution. The transfer of
momentum is analogous to the transfer of heat when Pr = 1. For gases Prandtl number
ranges from 0.6 to 1.0. The analogy is, therefore, satisfactory for gases. Liquids, however,
have Prandtl numbers considerably different from unity, and the preceding analysis is not
applicable directly to liquids.
Pohlhausen’s results can be modified empirically to include fluids having different values
of Prandtl number. As shown in Figure. 8.3 the temperature profiles computed
theoretically have been plotted for various Prandtl numbers.

0.10

0.8

0.6
T - Tw
T  Tw
0.4

0.2

0
0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6 4.0

y U x
=
x ν
Figure 8.3 : Temperature Distribution in a Fluid Flowing over a Heated Plate
for Various Prandtl Numbers

We define a thermal boundary layer thickness  t as the distance from the wall at which
the temperature difference between the wall and the fluid T  Tw  reaches 99% of the
free-stream value T  Tw  . From the temperature profiles, we observe that for Pr < 1,
 t >  , and for fluids having Pr > 1,  t <  . According to Pohlhausen, the curves are
52
 Boundary Layer
replotted and shown in Figure 8.4 using the correction factor Pr 1 / 2 in the abscissa, which Formation
 y 1/ 2
is now    Re x  Pr1/ 3 .
 x

 (T - Tw) / (T - Tw)

y 1/3  = 0.332
 Re x Pr 
x 
y=0
1.0

T - Tw
Temperature distribution
T  Tw

0.5

1.0 2.0 3.0 4.0

y 1/ 3
Re x Pr
x
Figure 8.4 : Temperature Distribution for Laminar Flow over a
Heated Plate at Uniform Temperature

The dimensionless temperature gradient at the surface (y = 0) is

 T  Tw  / T  Tw 

 0.332 . . . (8.30)
1/ 2
  y / x   Re x  Pr1/ 3 
 
Therefore, at any specified value of x,

 T  Re 1x/ 2 Pr 1 / 3
   0 . 332 T  Tw  . . . (8.31)
 t  y  0 x

The local rate of convection heat transfer per unit area becomes

 T 
qc:   k  
 y  y  0

Re 1x/ 2 Pr 1 / 2
  k 0.332 T  Tw  . . . (8.32)
x
The total rate of heat transfer from a plate of width b and length L is
L
Q  q dx  0.664k Re
:
c
1/ 2
L Pr 1 / 3 bTw  T  . . . (8.33)
x 0

The local heat transfer coefficient obtained from Eq. (8.32),

qc: k
hcx   0.332 Re1L/ 2 Pr 1 / 3 . . . (8.34)
Tw  T x

The local Nusselt number is

hcx x
Nu x   0.332 Re 1L/ 2 Pr 1 / 3 . . . (8.35)
k
53
Convection The average Nusselt number over the entire plate of length L is
L
Nu L   Nu x dx  2 Nu x  x  L  0.664 Re 1L/ 2 Pr 1 / 3 . . . (8.36)
0

and the average heat transfer coefficient

k
h c  2hcx  x  L  0.664 Re 1L/ 2 Pr 1 / 3 . . . (8.37)
L
The physical properties in Eqs. (8.31)-(8.37) vary with temperature. Experimental data
are found to agree satisfactory with the results predicted analytically using the above
equations if the properties are evaluated at a mean temperature T * given by
T  T 
T *  w  , often called the film temperature.
2
SAQ 2
(a) What do you mean by film temperature?
(b) How do you calculate the average heat transfer coefficient?

8.4 APPROXIMATE INTEGRAL BOUNDARY LAYER

ANALYSIS
We had earlier developed the exact mathematical solution of the differential equations
describing the laminar flow of a fluid over a flat surface in deriving the boundary layer
thickness and the heat transfer coefficient. To circumvent the problems involved in
solving the partial differential equations of the boundary layer, Theodore von Karman
suggested the approximate integral method in which he considered a control volume that
extends from the wall to beyond the boundary layer. Let us consider a control volume
(CV) bounded by the two planes AB and CD normal to the surface, a distance dx apart,
and a parallel plane in the free stream at a distance l from the surface as shown in
Figure 8.5. Let us consider unit width of plate in z-direction.
y

D
B

y=
u u
Velocity
boundary layer
u (y)

x
A C

dx

Figure 8.5 : Control Volume for Integral Momentum Conservation Analysis

Momentum flow across face AB into the CV in is

1
  u 2 dy
0

Similarly, momentum flow across face CD is

d  
1 1
  u dy    u 2 dy dx
2

0
dx  0 
Fluid entering across BD at the rate
54
 Boundary Layer
d  
1
  udy dx Formation
=
dx  0 

This quantity is the difference between the rate of flow leaving across CD and that
entering across AB. Since the fluid entering across BD has a velocity component in the
x-direction equal to u  , the flow of x-momentum across the upper face into the CV is

d  
1
u   udy dx
dx  0 

Net x-momentum transfer
I = outflow – inflow

d  
1 1 1
or, I    u 2 dy    u 2
dy  dx    u 2 dy . . . (8.38)
0
dx  0  0

For y   , u  u  and the integrand I will be zero. We have to consider the integrand
only within the limits from y = 0 to y   . There will be no shear across face BD outside
du
the boundary layer where  0 is zero. A shear force  w acts at the
dy
fluid-solid interface, and there will be pressure forces acting on faces AB and CD.
Net forces acting on the CV are
 dp  dp
p   p  dx     wdx   dx   wdx . . . (8.39)
 dx  dx
By Newton’s second law of motion

dp d  
 dx   w dx      u u   u dy  dx . . . (8.40)
dx dx  0 
dp
For flow over a flat plate the pressure gradient in the x-direction, can be neglected.
dx

d
 u  u  u dy   w
dx 0
Therefore, . . . (8.41)

The above equation is often called von Karman’s momentum integral equation.
Assuming a four-term polynomial for the velocity distribution [3]
u  y   a  by  cy 2  dy 3 . . . (8.42)

Where the constants are evaluated from the boundary conditions :

at y = 0, u = 0 and so a = 0
 2u
u = v = 0 and 0
y 2
u
y   , u  u  and 0
y
From these conditions we find
3 u u
a  0, b  , c  0, d   3
2  2
Substituting in Eq. (8.42)
55
Convection
3 u u y 3
u y . . . (8.43)
2  2 3
3
u 3 y 1 y 
    . . . (8.44)
u 2  2   
Substituting Eq. (8.44) for the velocity distribution in the integral momentum Eq. (8.41),

d
dx 0
 
 uu   u 2 dy   w . . . (8.45)

d  2  3 y 1 y3  2  3 y 1 y3  
 2

dx 0 
L.H.S.    u   3 
 u    3 
dy
 2 2   2  2   

d  2  3 y 1 y3 

  9 y2 3 y4 1 y6
dx 0 
  u      u 2    dy
 2  23   4  2
2  4
46 

d  2  3  2 1 4 9 3 3 5 1  7 
  u    
     dy
dx   2 2 2 4  4 3 2 5 4 7 
3 2 4 6

d  2 39 
  u  
dx  280 
 du  3 u
Again, w       . . . (8.46)
 dy  y 0 2 

d  2 39  3 u
Therefore,   u  . . . (8.47)
dx  280  2 
39 d  3 u
 u2   . . . (8.48)
280 dx 2 
1/ 2 1/ 2
 280 vx   vx 
    4.64   . . . (8.49)
 13 u   u 

When x  0,   0 and    x 
1/ 2

 4.64
Also,  . . . (8.50)
x  Re x 1/ 2

u x
where local Reynolds number Re x  .
v
Eq. (8.50) gives a value of  , only 8% below that of the exact analysis, given by
Eq. (8.23) where
 5
 . . . (8.51)
x  Re x 1/ 2

The results, of approximate analysis are thus satisfactory in practice.

Substituting for  from Eq. (8.50) in Eq. (8.46),

56
 Boundary Layer
3 u  Re x 
1/ 2
Formation
w    . . . (8.52)
2 4.64 x
1 2
Dividing both sides by  u
2

3u   Re x   2
1/ 2
w
 Cf   . . . (8.53)
1/ 2  u
2
9.28 x  u2
x

v 0.647
 Re x 
1/ 2
C f x  0.647  . . . (8.54)
 Re x 
1/ 2
u x

The exact analysis gave us (Eq. (4.91))

0.664
C fx  . . . (8.55)
 Re x 
1/ 2

The integral energy equation can be derived in a similar fashion. A CV extending beyond
the limits of both the temperature and the velocity boundary layers is to be used in the
derivation as shown in Figure 8.6.
y
B D
ys

T
u

Velocity
Temperature boundary layer
u(y) Boundary layer thickness, 
Thickness, t
T(y)
x
A C
Tw

dx

Figure 8.6 : Control Volume for Integral Energy Conservation Analysis

Energy entering the CV across face AB

1
  c p uTdy
0

Energy leaving the CV across face CD

d  
1 1
  c p uTdy    c p uTdy dx
0
dx  0 

Energy carried into the CV across the upper face

d  
1

dx  0
 c p T   udy dx


Heat conducted across the wall at the interface

57
Convection
 T 
  kdx 
 y  y  0
Making an energy balance
1 1
 T 
0 c p uTdy  kdx y   0 c p uTdy
 
y 0

d   d  
1 1
   c p Tdy dx  c p T   udy dx  0 . . . (8.56)

dx  0  
dx  0 
 
At 1   t , T  T , and integration need be taken up to y   t .

d t k  T 
 u T  T dy    . . . (8.57)
dx 0 c p  y  y  0

d t  T 
or  u T  T dy     . . . (8.58)
dx 0  y  y  0
This is known as the integral energy equation of the laminar boundary layer at low speed.
Let us assume the temperature distribution in the thermal boundary layer as
T  y   e  fy  gy 2  hy 3 . . . (8.59)
The boundary conditions are :
 2T
at y  0, T  Tw ,  0 (since u  v  0 )
y 2
dT
at y   t T  T and 0
dy
From these conditions, we get
3 T  T w
e  Tw , f  ,g 0
2 t
T  T w
h
2 t3
3 T  T w T T
T  Tw  y   3 w y3 . . . (8.60)
2 t 2 t
3
T  Tw 3 y 1 y 
or,     . . . (8.61)
T  Tw 2  t 2   t 
This is the temperature distribution in the thermal boundary layer. Substituting in energy
equation, Eq. (8.58),
t t

 T  T udy   T  T   T  Tw udy

0 0

 3 y 1 y
t

3
 3y 1  y 3 
 T  Tw u   1         dy
0 
2  t 2   t    2 2    


58
  3  3 Boundary Layer
  9  2  3  4  1  3  3  4  1
 T  Tw u     y    y  
 4 3
y   3
 y   y   3 3
  4 
 y dy

Formation
 2  4 t  2  4 t 
3
   t     t  
 3  2t 3  2t 3  2t 1  4t 3  4t 1  4t 
 T  Tw u         . . . (8.62)
 2 2 4  4  8  3 20  3 28  3 
t
If we let   the above expression can be written as

 3 3 4
T  Tw u     2   
 20 280 
For fluid having Pr  1,   1, the second term in the parentheses can be neglected
compared to the first. Substituting in the approximate integral equation, Eq. (8.58),
d  
T  Tw u   2  2    T 
dx 20  y  y  0
3 1
  T  Tw 
2 t
3  T  Tw 
 . . . (8.63)
2 
1 
u   3  . . . (8.64)
10 x
From Eq. (8.48),
 140  1 140  10 1
  ;  . . . (8.65)
x 13  u  13 u  u  3
13 1
3  t  0.976  Pr 1 / 3 . . . (8.66)
14 Pr

If we compare with Eq. (8.31),  Pr1/ 3 we find that except for the numerical constant
t
(0.976 instead of 1.0), the above result is in agreement with the exact calculation of
Pohlhausen.
The rate of heat transfer per unit area is
 T  3 k
q c"   k     T  Tw  . . . (8.67)
 y  y  0 2 t

Substituting Eqs. (8.50) and (8.66) for  and  t ,

3 T  Tw 
q c"   k
2 0.976 Pr 1 / 3
3 k T  Tw  Pr Re x
1/ 3 1/ 2
 . . . (8.68)
2 0.976  4.64 x
The local heat transfer coefficient
q c" k
hc   0.331 Re 1x/ 2 Pr 1 / 3 . . . (8.69)
T w  T x
The local Nusselt number
hc x
Nu x   0.331 Re 1x/ 2 Pr 1 / 3 . . . (8.70)
k 59
Convection This result is in excellent agreement with Eq. (8.35), the result of an exact analysis,
which gives
Nu x  0.332 Re 1x/ 2 Pr 1 / 3 .
SAQ 3
(a) Derive the integral momentum and energy equations.
(b) Calculate the drag force, boundary layer thickness and heat transfer
coefficient for flow over a flat plate assuming cubic velocity and
temperature profile.
(c) What will be the variation in boundary layer thickness if the velocity profile
chosen is u  a0  a1 y ?

8.5 TURBULENT FLOW OVER A FLAT PLATE: ANALOGY

BETWEEN MOMENTUM AND HEAT TRANSFER
The flow in the boundary layer is more often turbulent, rather than laminar. In laminar
flow where the streamlines slide over one another, momentum and heat transfer take
place by molecular diffusion. In turbulent flow, the transport mechanism is aided by
innumerable eddies. Irregular velocity fluctuations are superimposed upon the motion of
the main stream, and these fluctuations are primarily responsible for the transfer of heat
as well as momentum. The rates of momentum and heat transfer in turbulent flow and the
associated friction and heat transfer coefficient are many times larger than in laminar
flow because of better mixing in which groups of particles collide with one another at
random, establish cross-flow on a macroscopic scale and effectively mix the fluid.
Instantaneous streamlines in turbulent flows are highly torn and uneven, and it is difficult
to trace the path of individual fluid elements. But if the flow at a point is averaged over a
period of time, and compared to the period of a single fluctuations, the time-mean
properties and the velocity of the fluid are constant, if the average flow remains steady. It
is, therefore, possible to describe each fluid property and the velocity in turbulent flow in
terms of a mean value, which does not vary with time and a fluctuating component,
which is a function of time.
Let us consider a two-dimensional flow as shown in Figure 8.7 in which the mean
velocity is parallel to the x-direction. The instantaneous velocity components u and v can
then be expressed in the form
u  u  u ' , v  v' , p  p  p' , T  T  T ' . . . (8.71)
and so on,
du
y l
dy
u +u’

+’
l

u
u (y)
+u’

u- u’
du
l
dy
x

Figure 8.7 : Mixing Length for Momentum Transfer in Turbulent Flow

where the bar over a symbol denotes the time-mean value and the prime denotes the
instantaneous deviation from the time-mean value. According to the model used to
60 describe the flow,
 t* Boundary Layer
1 Formation
u  *  udt . . . (8.72)
t 0

where t * is a time interval that is large compared to the period of the fluctuations. The
time variations of u and u ' is qualitatively shown in Figure 8.8. From Eq. (8.72) and
Figure 4.26 it is apparent that the time average of u ' is zero (i.e., u ' 0 ). Similarly,
 
v  0 and v'  0 .

u’

t
Figure 8.8 : Time Variation of Instantaneous Velocity in Turbulent Flow

The fluctuating velocity components continuously transport mass and so momentum

across a plane normal to the y-direction. Instantaneous rate of transfer of x-momentum in
y-direction per unit area


  v '  u  u ' 
where the minus sign has a special significance which will be discussed later.
Time-average of the instantaneous rate of x-momentum transfer per unit area
t*

 v'u  u 'dt
1
t   * . . . (8.73)
t 0

This is also called “apparent turbulent shear stress” or “Reynolds stress”,  t . Breaking the
term in Eq. (8.73) into parts
t* t*

 v  u dt  v  u   u v
1 1
0 v udt  t *
' '
t   * ' ' '
. . . (8.74)
t 0

Since u is a constant and the time-average of v  is zero, the first term of the above
'

equation will be zero. Therefore,

 t   u 'v ' . . . (8.75)

where u ' v ' is the time average of the product of the fluctuating components u ' and v ' ,
which is not zero, but a negative quantity. If v ' is positive, the fluid particles with a
certain velocity u travel upward to a y-plane where the velocity u is more, these will
tend to slow down the particles in that plane giving rise to a negative velocity component
u ' . And if v ' is negative, it will tend to accelerate the flow, giving rise to a positive u ' .
So a positive v ' is associated with a negative u ' and vice versa, so that the time-average
of the product u ' v ' is not zero, but a negative quantity. The turbulent shear stress as
defined by Eq. (8.75) is thus positive and has the same sign as the corresponding laminar
shearing stress
61
Convection
du du
l    v . . . (8.76)
dy dy
It may be noted that the laminar shearing stress  1 is a true stress, whereas the apparent
turbulent shearing stress or Reynolds stress  t is a concept introduced to account for the
effects of momentum transfer due to turbulent fluctuations.
We can express the total shear stress in turbulent flow as
viscous force
  + turbulent momentum flux . . . (8.77)
area
du
To relate the turbulent momentum flux with the time-average velocity gradient ,
dy
Prandtl postulated that the fluctuations of macroscopic fluid particles in turbulent flow
are, on the average, similar to the motion of molecules in a gas, i.e. they travel a distance
l perpendicular to u before coming to rest n another y-plane as shown in Figure. 8.7.
This distance l is known as Prandtl’s mixing length and qualitatively corresponds to the
mean free path of a gas molecule. Prandtl further argued that the fluid particles retain
their identity and physical properties during the cross motion and that the turbulent
fluctuations arise because of the difference of the time-mean properties between y planes
distance l apart. If a fluid particle travels from a layer at y plane to a layer at (y + 1) plane,
du
u' 1 . . . (8.78)
dy
The turbulent shearing stress is then
du du
 t    u ' v '    v '1   M . . . (8.79)
dy dy

 
where  M  v ' 1 is called the eddy viscosity or the turbulent exchange coefficient for
momentum. The eddy viscosity  M is analogous to the kinematic viscosity  . But v is a
physical property, whereas  M is not, and it depends on the dynamics of flow.
Total shearing stress,
du du du
   1   t  v   M   v   M  . . . (8.80)
dy dy dy
For turbulent flow,
du
 M  v and    M . . . (8.81)
dy
For laminar flow,
du
 M  0 and   v . . . (8.82)
dy
For buffer layer (transition zone),
du
   v   M  . . . (8.83)
dy
The transfer of energy as heat in turbulent flow can be visualized in a similar fashion. Let
us consider a two-dimensional time-mean temperature distribution as shown in Figure
8.9. The fluctuating velocity components transport fluid particles and the energy stored in
them across a plane normal to the y-direction.
62
 Boundary Layer
l
dT Formation
dy

+qt
y+ +’

l
T (y )

+T’
l

y
dT
l
dy

Figure 8.9 : Mixing Length for Energy Transfer in Turbulent Flow

Instantaneous rate of energy transfer per unit area at any point in the y-direction.

  v  c p T
'

The time average of the instantaneous rate of heat transfer per unit area is equal to,
turbulent rate of heat transfer Qt .

t*

 v   T  T dt  c
Qt 1 ' '
 * p p v 'T ' . . . (8.84)
A t 0

where T T T' . . . (8.85)

Using Prandtl’s concept of mixing length, we can relate the temperature fluctuation to the
time-mean temperature gradient by the equation

dT
T' 1 . . . (8.86)
dy

When a fluid particle migrates from the y plane to the (y ± 1) plane, the resulting
temperature fluctuation is caused by the difference between the time-mean temperatures
of the two planes. Assuming that the transport mechanisms of temperature (or energy)
and velocity (or momentum) are similar, the mixing lengths in Eqs. (8.78) and (8.86) are
equal. The product v 'T ' in Eq. (8.84) however, is positive, because a positive v ' is
accompanied by a positive T ' , and vice versa. Combining Eqs. (8.84) and (8.86), the
turbulent rate of heat transfer per unit area

Qt dT
  c p v ' 1 . . . (8.87)
A dy

where the minus sign is a consequence of the second law of thermodynamics, with heat
dissipation from the system. To express the turbulent heat flux in a form analogous to the
Fourier conduction equation, we define  H ' a quantity called eddy diffusivity of heat or
the turbulent exchange coefficient for temperature, by the equation e H  v ' 1 . Therefore,

Qt dT
  c p  H . . . (8.88)
A dy
The total rate of heat transfer per unit area
Qt
 (Molecular conduction)/area + (Turbulent transfer through eddies)/area
A
63
Convection
dT dT
  c p  c p  H
dy dy
Qt dT
or    c p    H  . . . (8.89)
A dy
k
where   . The contribution to the heat transfer by molecular conduction is
c p
proportional to  , and turbulent contribution is proportional to  H . For all fluids except
liquid metals,  H >>  in turbulent flow. For laminar flow,
H  0 . . . (8.90)

Q dT dT
and  c p   k . . . (8.91)
A dy dy

Q dT
In transition zone,  c p     H  . Prandtl number was defined as the ratio of
A dy
v
two transport properties, , those of momentum and energy. Similarly, the ratio of the


turbulent eddy viscosity to the eddy diffusivity, M , is called the turbulent Prandtl
H
number, Prt . Therefore,

M
Prt  . . . (8.92)
H

According to Prandtl’s mixing length theory, since  M   H  v ' 1 , Prt is unity. For
Prt = 1, the turbulent heat flux can be related to the turbulent shear stress. By combining
Eqs. (8.79) and (8.88),

Qt  c p  H d T / dy
 . . . (8.93)
A t  M d u / dy
Qt dT
or   t c p . . . (8.94)
A du
This relationship was first derived in 1874 by the British physicist Osborne Reynolds and
is called Reynolds analogy. It is a good approximation whenever the flow is turbulent
and can be applied in turbulent boundary layers as well as to turbulent flow in pipes and
ducts. This analogy, however, does not hold good in the viscous sublayer where the flow
is laminar.
SAQ 4
(a) Define eddy viscosity. How is it different from from kinematic viscosity?
(b) Define eddy diffusivity. How is it related to the thermal diffusivity?
(c) Explain Prandtl mixing length concept for turbulent flow over a flat plate.

8.6 REYNOLDS ANALOGY FOR TURBULENT FLOW OVER

A FLAT PLATE
To derive a relation between heat transfer and friction coefficients in flow over a flat
64 plate for a fluid having Pr = 1, we recall that the laminar shear stress
 du Boundary Layer
  . . . (8.95) Formation
dy
and the rate of heat transfer per unit area across any plane normal to the y-direction is
dT
q"  k . . . (8.96)
dy
Combining these two equations, we obtain
k dT
q "   . . . (8.97)
 du
k c p
In Eqs. (8.94) and (8.97), we observe that if  c p or  1 , i.e. Pr = 1, the same
 k
equation of heat flow holds good in the laminar and turbulent layers. To determine the
rate of heat transfer from a flat plate to a fluid with Pr = 1 flowing over it in turbulent
k
flow, we replace by c p and separate the variables in Eq. (8.97). Assuming that q "

and  are constant, we get

q w"
du   dT . . . (8.98)
 wc p

where the subscript w indicates that q " and  are taken at the wall of the plate.
Integrating Eq. (8.98) between the limits u = 0 when T  Tw and u  u  when T  T
gives
q w"
u   Tw  T . . . (8.99)
 wc p
Now, by definition, the local heat transfer coefficient hcx and local friction coefficient
C f x are

q w"
hcx  . . . (8.100)
Tw  T

u 2
and  wx  C f . . . (8.101)
x
2
hcx 2u 
1 . . . (8.102)
C f x u 2 c p

hcx Cf
or,  x  St x . . . (8.103)
c p u  2

where St x  the local Stanton number

Nu x
St x 
Re x
hcx
Pr 
 c p u

65
Convection Eq. (8.103) is satisfactory for gases in which Pr = 1. Colburn [12] has shown that this
equation can also be used for fluids having 0.6 < Pr < 50 if it is modified in accordance
with
C fx
St x Pr 2 / 3  . . . (8.104)
2
where the subscript x denotes the distance from the leading edge of the plate. This
expression is referred to as the Reynolds-Colburn analogy, and St x Pr 2 / 3 is called
Colburn’s j-factor.
To apply the analogy between the heat transfer and momentum transfer, in practice it is
necessary to know the friction coefficient C f x . For turbulent flow over a flat plate the
empirical equation for the local friction coefficient
1 / 5
u x 0.0576
C fx  0.0576    . . . (8.105)
 v  Re x 0.2
is in good agreement with experimental results for Reynolds number varying between
5 × 10 5 and 5 × 10 7 .
If the turbulent boundary layer is assumed to start at the leading edge, the average friction
coefficient over a plane surface of length L is
L 1 / 2
1 u L 0.072
Cf 
L0 C f x dx  0.072  
 v 

Re L 0.2
. . . (8.106)

Again, from Eq. (8.104), for turbulent flow

Nu x Cf 0.0576
St x Pr 2 / 3  Pr 2 / 3  x  = 0.0288 Re x 0.2 . . . (8.107)
2  Re x 
0 .2
Re x Pr 2

Nu x  0.0288 Re 0x.8 Pr 1 / 3 . . . (8.108)

It is the local Nusselt number at any value of x larger then x c , since between x = 0 to
x = x c the flow is laminar. The average heat transfer coefficient in turbulent flow over a
plane surface of length L is
L L 0 .8
1 1 k u 
hm  hc   hcx dx   0.0288 Pr 1 / 3    x 0.8 dx . . . (8.109)
L0 L0x  v 
On integrating the non-dimensionalising

Nu L  0.036 Re 0L.8 Pr 1 / 3 . . . (8.110)

This is valid only when L >> x c .

To consider the mixed boundary layer in a flat plate with laminar flow from x = 0 to
x = x c and turbulent flow from x = x c to x = L, as shown in Figure. 8.10.

L 0 .8
1  u  k 0 .8 1 / 3
hm 
L  xc x 0.0288 v  x
x Pr dx . . . (8.111)
c

66
 h Boundary Layer
Formation
Cf

hx or Cf, x

x

T

Laminar Transition Turbulent

T

x xc

Figure 8.10 : Variation of Local Friction and Heat Transfer Coefficients Cfx and hx

hc L  x c 
k

 0.036 Re 0L.08  Re 0x.c8 Pr 1 / 3 

 0.036 Re 0L.8  36,239 Pr 1 / 3  . . . (8.112)

If Eq. (8.70) is used between x = 0 and x = x c and equation 8.108 between x = x c and
x = L for the integrating of hcx it yields, with Re xc  500,000.

1  c 
x L
h
L  0
hlam . dx  x turb. 
h dx . . . (8.113)
c 
0 .5
u x hlam. x
Putting Nu iam.  0.332   Pr 1 / 3  . . . (8.114)
 v  k
0 .8
u x hturb. x
and Nu turb.  0.0288   Pr 1 / 3  . . . (8.115)
 v  k

k 
0 .5 x c 0 .8 L
u  u 
x x
1 / 2  0 .2
h  0.332   dx  0.288   dx  Pr 1 / 3 . . . (8.116)
L   v  0  v  xc 

 
Nu L  0.664 Re1x/c 2  0.036 Re L  Re 0x.c8 Pr 1 / 3
0 .8
 . . . (8.117)

or, 
Nu L  0.036 Re 0L.8  835 Pr 1 / 3  . . . (8.118)

Adding the laminar friction drag between x = 0 and x = x c to the turbulent drag between
x = x c and x = L gives per unit width,

0.072 Re 1 / 5
L L  0.072 Re x1c / 5 x c  1.33 Re x1c / 2 x c 
Cf  . . . (8.119)
L
For a critical Reynolds number, Re xc  500,000, this reduces to

 0.0464 x c 
C f  0.072 Re L1 / 5  
 L  67
Convection 0.072 0.0464  0.072  500,000
= 
Re 0L.2 Re L
0.072 1670
  . . . (8.120)
Re 0L.2 Re L
SAQ 5
(a) Explain Reynolds analogy. Is there any restriction on its use?
(b) What do you mean by Colburn’s j  factor?
(c) What is Reynolds-Colburn analogy?

8.7 CONSTANT HEAT FLUX BOUNDARY CONDITION

The analysis given in Sub-section 8.3 has considered the laminar heat transfer from an
isothermal surface. In many practical problems the surface heat flux is essentially
constant, and the objective is to find the distribution of the plate-surface temperature for
given fluid-flow conditions. For the constant heat flux case it can be shown that the local
Nusselt number is given by
Nu x  0.453 Re1x/ 2 Pr 1 / 3 . . . (8.121)
which may be expressed in terms of wall heat flux and temperature difference as
qw x
Nu x  . . . (8.122)
k Tw  T 
The average temperature difference along the plate, for the constant heat flux condition,
may be obtained from
L L
1 1 qw x
 
L o L 0 kNu x
Tw  T  T w  T dx  dx

qw L / k
 . . . (8.123)
0.6795 Re 1L/ 2 Pr 1 / 3
3
qw  hx  LTw  T  . . . (8.124)
2
where q w is the wall heat flux in W/m 2 .

8.7.1 Other Relations

Eq. (8.35) is applicable to fluids having Prandtl numbers between 0.6 and 50. It would
not apply to fluids with very low Prandtl numbers like liquid metals or to high Prandtl
number fluids like heavy oils or silicones. For a very wide range of Prandtl numbers,
Churchill and Ozoe [5] have corrected a large amount of data to give the following
relation for laminar flow on an isothermal plate :
0.3387 Re1x/ 2 Pr 1 / 3
Nu x  1/ 4
. . . (8.125)
  0.0468  2 / 3 
1    
  Pr  
for Re x Pr > 100.

For the constant heat flux case, 0.3387 is changed to 0.4637 and 0.0468 is changed to
0.0207. The properties are evaluated at the film temperature.
68
8.7.2 Boundary Layer Thickness in Turbulent Flow Boundary Layer
Formation
As shown in Figure 8.11(a) the turbulent boundary layer the shape of the velocity profile
is much more curved than in the laminar layer.
u
Transition
zone

u tu

b
ub 
b
xc
tw

(a) (b)
Figure 8.11 : Laminar Sublayer b in a Turbulent Boundary Layer tu

The measured velocity profile agrees satisfactorily with the equation proposed by
Prandtl :
1/ 7
 y
u  u   . . . (8.126)
 
This equation does not hold good near the wall. The velocity gradient
du 1 u 
 . . . (8.127)
dy 7  1 / 7 y 6 / 7
du
and at the wall where y = 0,   . An infinite value of shear stress at the wall is
dy
impossible. In fact the turbulence always dies down near the wall, where a laminar
sublayer exists and where the velocity increases linearly with the distance y as shown in
Figure. 8.11(b). Outside this sublayer Eq. (8.126) holds true. For not too large Reynolds
numbers and smooth surfaces, Blasius equation holds
1/ 4
 v 
 w  0.0228 u  2

 . . . (8.128)
 u  
The momentum integral equation, Eq. (8.41),

d
u u   u dy   w
dx 0
. . . (8.129)

u
Substituting u  y1/ 7 . . . (8.130)
 1/ 7

 
 y   y
1/ 7 1/ 7 
  y 1 / 7   y 1 / 7
Integrand, I    u   u    u    dy  u   1      dy
2


0               
0 

   y 1 / 7 
 y
2/7

 u     dy     dy 
2

 0    0
 

 1  8/7 1  9/7 
 u 2  1 / 7  2/7 
 8/7  9 / 7 
69
Convection
 7 7  7 u  
2
 u     2
 . . . (8.131)
8 9 72
1/ 4
d  7 u 2    v 
Substituting,     w  0.0228 u 2   . . . (8.132)
dx  72   u  
1/ 4
d 72  0.0228  v 
   . . . (8.133)
dx 7  u  
 x 1/ 4
 v 
 d   0.235 
1/ 4
dx . . . (8.134)
0 0  u 
1/ 4
4 5/ 4  v 
  0.294  x . . . (8.135)
5  u 
1/ 5 1/ 5
 u   v 
  0.294  4/5
  x  4/5
 0.376  x . . . (8.136)
 u   u x 
 0.376
 . . . (8.137)
x Re x 1 / 5
The displacement thickness

 u

 
  y 1 / 7  
1

   1 
*
dy   1     dy   dy  1 / 7 y
1/ 7
dy
0
u      
0  0  0

1 7 8/7 
    . . . (8.138)
 1/ 7 8 8
If the laminar and the turbulent boundary layers are calculated for the critical distance x c ,
it can be seen that the turbulent layer is thicker. In reality an instantaneous increase in
boundary layer thickness is not possible. The transition from the laminar to the turbulent
boundary layer takes place in a transition zone as is indicated in Figure 8.12.

Laminar boundary Transition Turbulent boundary

layer region layer
u
Turbulent
u layer
Buffer layer

x
xc Laminar
sublayer
Boundary
layer thickness, 

Figure 8.12 : Transition from Laminar to Turbulent Boundary Layer

For heat transfer calculations the thickness  b of the laminar sublayer is needed. For this
we must first calculate the velocity u b at the border between the turbulent layer and the
laminar sublayer. The linear velocity increase in the sublayer is derived from the shear
stress on the wall.
du u
w    . . . (8.139)
dy y
70
 1/ 4 Boundary Layer
 v  u Formation
Now,  w  0.028 u  2

  . . . (8.140)
 u   y
1/ 4
 v  y
u  0.0228 u  2

 . . . (8.141)
 u   
When y   b , u  u b
1/ 4 1/ 4
u 2  v  u 2  v 3  b
u b  0.0228     b  0.0228    . . . (8.142)
  u     u  
1/ 4 1/ 4
b 1 v  u  1 ub  v3 
   ub   3 3  . . . (8.143)
 0.0228 u 2  v 3  0.0228 u   u  
1/ 7
u  y
Now,   . . . (8.144)
u   
When y   b , u  u b. . . . (8.145)
1/ 7
ub   b 
  . . . (8.146)
u   
Substituting in Eq. (8.138),
3/ 4 7
b 1 ub  v  u 
     b  . . . (8.147)
 0.0228 u   u    u 
1/ 6 1/ 8
ub  1   v 
    . . . (8.148)
u   0.0228   u  
ub 1.869
or,  . . . (8.149)
u  Re  1 / 8
 0.376
Since  . . . (8.150)
x Re x 1 / 5
1 / 8 1 / 8
ub u   u 0.376 x 
 1.869    1.869  

u  v   v u  x / v 
1/ 5

1 / 8 1 / 10
 u4/5x4/5  u x
 2.11   2.11  
 v 
4/5
 v 
ub 2.11
 . . . (8.151)
u  Re x 0.1
7
 b  ub 
   
2.11
7

191
. . . (8.152)
  u  Re x  Re 0x.7
0 . 1 7

b 191
 . . . (8.153)
 Re x 0.7

71
Convection 1/ 4
 v u  x / v 1 / 5 
1/ 4
ub  v 
 w   0.0228 u  2

  0.0228 u   
b  u   u  0.376 x 
0.0296 u 2
 . . . (8.154)
Re x 0.2
At any given value of x, a turbulent boundary layer increase at a faster rate than a laminar
boundary layer (  tu x 4 / 5 and  lam x 1 / 2 ). Despite its greater thickness, the turbulent
boundary layer offers less resistance to heat flow than a laminar layer because the
turbulent eddies produce continuous mixing between the warmer and cooler fluids on a
macroscopic scale. These eddies diminish in intensity in the buffer layer and hardly
penetrate the laminar sublayer.
Prandtl divided the flow field into a laminar and a turbulent layer, but neglected the
buffer layer in his analysis and obtained
Nu x C fx / 2
St x   . . . (8.155)
Re x Pr 1  2.1Re x0.1 Pr 1 
C fx
From Pr = 1, St x  , which is Reynolds analogy. The second term in the
2
denominator is a measure of the thermal resistance in the laminar sublayer. This part of
the total thermal resistance increases as the Prandtl number becomes larger.
Prandtl’s analysis was later refined by von Karman who divided the flow field into three
zones : a laminar sublayer adjacent to the surface in which the eddy diffusivity  H is
zero, and heat flows only by conduction. Next to it is a buffer layer in which both
conduction and convection contribute to the heat transfer mechanism, i.e.  and  H are
of the same order of magnitude. Finally, a turbulent region in which conduction is
negligible compared to convection, and the Reynolds analogy applies. He used
experimental data for the velocity distribution and the shear stress to evaluate  H and
assumed  M   H . Also assuming the physical properties of the fluid to be independent
of temperature, he determined the thermal resistances in each of the three zones. The von
Karman analogy gives the following equation :
0.0296 Re 0.8 x Pr
Nu x  . . . (8.156)
 5 
1  0.86 Re x 0.1  Pr  1   In 1  Pr  1 
 6 
for 5  10 5 < Re x < 10 7 .

8.9 FORCED CONVECTION INSIDE TUBES AND DUCTS

Let us consider a fluid entering a circular tube at a uniform velocity. The fluid particles in
the layer in contact with the surface of the tube will come to a complete stop. The layer
will cause the fluid particles in the adjacent layers to slow down gradually as a result of
friction. To make up for this velocity reduction, the velocity of the fluid at the mid-
section of the tube will have to increase to keep the mass flow rate through the tube
constant. As a result, velocity boundary layer is as shown in Figures 8.13(a), (b) develops
along the tube. The thickness of this boundary layer increases in the flow direction until
the boundary layer reaches the tube centre and thus fills the entire tube. The region from
the tube inlet to the point at which the boundary layers merge at the centerline is called
the hydrodynamic entry region, and the length of this region is called the hydrodynamic
entry length Le h . The region beyond this entry region where the velocity profile is fully
developed and remains unchanged is called the hydrodynamically developed region. The

72
velocity in this region is parabolic for laminar flow, and somewhere flatter in turbulent Boundary Layer
Formation
flow.
Velocity boundary layer Velocity profile

x
Hydrodynamically
Hydrodynamic entry region developed region

(a)

Laminar
Sublayer
u
Turbulent
core

(b)
Figure 8.13 : Development of Velocity Boundary Layer for Laminar Flow in
(a) a Tube and (b) Velocity Profile in a Turbulent Flow

Now, let us consider that a fluid at a uniform temperature enters a circular tube with its
wall at a different temperature. The fluid particles in the layer in contact with the surface
of the tube will assume the tube surface or wall temperature Tw . This will initiate
convection heat transfer in the tube, and the development of the thermal boundary layer
as shown in Figure 8.14 along the tube.
Thermal profile

T1 Tw

x
Thermally
Thermal Developed region
Entry region
(a)

Surface condition qs”

Tw > T(r,0)
y=r0

t r0
r

t

T(r,0)
T(r,0) Tw T(r,0) Tw T(r,0) T(r)

Thermal entrance region Fully developed region

x Xfd, t
(b)
Figure 8.14 : Development of Thermal Boundary Layer in a Tube
(a) when T > Tw and (b) when Tw > T
73
Convection The thickness of this boundary layer also increases in the flow direction until the
boundary layer reaches the tube centre and thus fills the entire tube. The region of flow
over which the thermal boundary layer develops and reaches the tube centre is called the
thermal entry region, and the length of this region is called the thermal entry length Le t .
As shown in Figure 8.15, the region beyond the thermal entry region in which the
temperature profile remains unchanged is called the thermally developed region. The
T  Tw 
dimensionless temperature profile does not also change upstream of Le t .
Tc  Tw 

U/U U/U U/U Velocity profile

x
-hydrodynamic boundary layer

Temperature profile
Ti Tb Ti Tb Ti Tb for fluid being
Cooled (T = 0)

T
t-thermal boundary layer

hcx
hc

1.0

x/D

Figure 8.15 : Velocity and Temperature Profiles and Variation of Local Heat Transfer Coefficient near
the Inlet for a Fluid being Cooled in a Laminar Flow through an Isothermal Tube

The region in which the flow is both hydrodynamically and thermally developed is called
the fully developed region. The shape of the fully developed temperature profile T (r, x)
differs according to whether a uniform surface temperature Tw  or a uniform heat flux is
maintained. For both surface conditions, however, the amount by which fluid
temperatures exceed the entrance temperature increases with increasing x.
The hydrodynamic and thermal entry lengths in laminar flow are given approximately as

 Le  h  0.05 Re d D . . . (8.157)

Le t  0.05 Re d Pr D . . . (8.158)

In turbulent flow the hydrodynamic and thermal entry lengths are independent of Re d
and Pr, and are generally taken to be

Le h  Le t  10 D . . . (8.159)

The final shapes of the velocity and temperature profiles depend on whether the fully
developed flow is laminar or turbulent. Figures 8.15 and 8.16 illustrate qualitatively the
growth of boundary layers as well as the variations in the local heat transfer coefficient
near the entrance of a tube for laminar and turbulent flow conditions, respectively. The
74
 heat transfer coefficient is largest near the entrance and decreases long the duct until both Boundary Layer
Formation
the velocity and temperature profiles for the fully developed flow have been established.
If Re d for the fully developed flow is below 2100, the entrance effects may be
appreciable for a length as much as 100 D from the entrance.
q q q

Growth of
Boundary layers

q q q

Variation of
velocity
distribution

hcx
hc Turbulent flow
behaviour
Laminar flow
behaviour
Laminar Turbulent boundary Fully established
Boundary layer layer velocity distribution

x/D

Figure 8.16 : Velocity Distribution and Variation of Local Heat Transfer Coefficient
near the Entrance of a Uniformly Heated Tube for a Fluid in Turbulent Flow

For a given fluid the Nussalt number depends primarily on the flow conditions which can
be characterized by the Reynolds number, Re d . For flow on long conduits the
characteristic length in Reynolds number as well as in Nusselt number is the hydraulic
diameter D H and the velocity to be used is the mean over the flow cross-sectional
area, u m ,

u m DH  u m DH
or Re d   . . . (8.160)
 v

h c DH
and Nu d  . . . (8.161)
k
4A
where D H  ,
P
A being the flow cross-sectional area and P the wetted perimeter Figure 8.17(a). For a
 2
circular tube or a pipe, A D P  D, so D H  D, the inside tube diameter. For an
4

annulus formed between two concentric tubes Figure 8.17(b), A  D22  D12  and
4
P   D1  D2  so that

D H  D2  D1 . . . (8.162)

75
Convection

Flow cross-sectional
area

Wetted perimeter

(a)

D1 D2

(b)
Figure 8.17 : Hydraulic Diameter for (a) Triangular Cross Section and (b) Annulus

For a rectangular conduit of sides a and b,

4ab 2ab
DH   . . . (8.163)
2a  b a  b

In engineering practice the Nusselt number for flow in conduits is usually evaluated from
empirical equations based on experimental results. From a dimensional analysis, the
experimental results obtained in forced convection heat transfer experiments in long
conduits can be correlated by an equation of the form

Nud    Red    Pr  . . . (8.164)

where the symbols  and  denote functions of the Reynolds number and Prandtl
number, respectively. In long ducts, where the entrance effects are not important, the
flow is laminar when Re d  2100 . In the range 2100  Re d  10,000, a transition
from laminar to turbulent flow takes place. The flow in this region is transitional. At
Re d > 10,000, the flow becomes fully turbulent.

In laminar flow through a duct, just as in laminar flow over a plate, there is no mixing of
warmer and colder fluid particles by eddy motion, and the heat transfer takes place solely
by conduction. Since all fluids except liquid metals have small thermal conductivities, the
heat transfer coefficients in laminar flow are relatively small. In transitional flow a
certain amount of mixing occurs through eddies that carry warmer fluid into cooler
regions and vice versa. Since the mixing motion, though on a small scale, accelerates the
transfer of heat considerably, a marked increase in heat transfer of coefficient occurs
above Re d  2100, as can be seen in Figure 8.18 where experimentally measured values
of the average Nusselt number for atmospheric air flowing through a long heated tube are
plotted as a function of the Reynolds number. Since the Prandtl number for air does not
vary appreciably, Eq. (8.164) reduces to Nu    Re  and the curve in the Figure (8.18)
shows the dependence of Nud on the flow conditions. In the laminar regime we find that
Nud remains small increasing from about 2.3 at Re d  200 to 5.0 at Re d  2100 . At
Re d > 2100, Nud begins to increase rapidly until Re d  8000 . As Re d is further
increased, Nud continues to increase, but at a slower rate.

76
 200 
 Boundary Layer
100  
Formation
 
50
 
NuD  ReD0.8
20 
00
hcD 
NuD  
k 10  Turbulent
Laminar  Transitional
5.0 00   
  
 

NuD  ReD0.3
2.0 00
1.0
100 200 500 1000 2000 5000 10,000 20,000 50,000

ReD = puD/
00
Figure 8.18 : Nusselt Number Varying with Reynolds Number for Air Flowing in
a Long Heated Tube at Uniform Wall Temperature

8.10 ANALYSIS OF LAMINAR FORCED CONVECTION IN

A LONG TUBE
Let us consider laminar flow through a tube under fully developed conditions with
constant heat flux at the wall. In the fluid element (Figure 8.19), the pressure is uniform
over the cross-section, and the pressure forces are balanced by the viscous shear forces
acting over the surface :
 du 
 p   p  dp   r 2   2 rdx     2 rdx . . . (8.165)
 dr 

du
u(r) (2 rdx)   (2 rdx)
dr
00
r
0 x
pr2
(p+dp) r 2
00
dx R
00

Figure 8.19 : Force Balance on a Cylindrical Fluid Element inside a Tube of Radius R

1  dp 
du    r dr . . . (8.166)
2   dx 
dp
where is the axial pressure gradient. The radial distribution of the axial velocity is
dx
then
1  dp  2
u r    r  C . . . (8.167)
4   dx 
where C is a constant of integration. When r = R, u = 0
1  dp  2
C  R . . . (8.168)
4   dx 

r 2  R 2 dp
Substituting u r   . . . (8.169)
4  dx
The maximum velocity occurs at the centre where r = 0.
77
Convection
R 2 dp
u max   u0 . . . (8.170)
4  dx
There the velocity distribution in dimensionless form becomes
2
u r u
 1    . . . (8.171)
u max R u0
Thus the velocity distribution in fully developed laminar flow is parabolic. The mean
velocity of fluid, um, is
R R
1  dp 
 urdr  4 r 
2
 R 2  rdr
 dx 
um  0
R
 0

R2 / 2
 rdr
0

1 dp  2 R 2 R 4  R 2 dp
  R .     . . . (8.172)
2  R 2 dx  2 4  8  dx

R 2 dp
u0  fluid velocity at the centreline  
4  dx
u0  2u m . . . (8.173)

To obtain the pressure drop of the fluid in the tube of length L, a force balance gives
p R 2  2 R w L . . . (8.174)

2 w L
p  . . . (8.175)
R
The pressure drop can also be related in the form

fL U m2
p  . . . (8.176)
D 2
where f is the Darcy friction factor.
2 w L fL  um2
 . . . (8.177)
R D 2
f um2
or  w   um  C f
2
. . . (8.178)
8 2
where C f is the Fanning friction coefficient

f
Cf  . . . (8.179)
4
dp p
From Eq. (8.172), putting 
dx L
R 2 p
um   . . . (8.180)
8 L
4 32  Lum
p  8 Lu m 2
 . . . (8.181)
D D2
78
 32  Lum Boundary Layer
p  p1  p2  . . . (8.182) Formation
D2
This is known as Hagen-Poiseuille equation for laminar flow. If a fluid flows through a
 m 
capillary tube of length L and diameter D, and the mass flow rate  um   and the
 A
pressure drop p are measured, the viscosity of the fluid  can be estimated from the
above equation. Using Eq. (8.176), we get

32  Lum fL  um2
p   . . . (8.183)
D2 D 2
64 64
f   . . . (8.184)
 um D Red

If the volumetric flow rate of the fluid is V  mv  and p is the pressure drop, then the
pumping power

V
Pp  p . . . (8.185)
p

where  p is the pump efficiency.

8.11 HEAT TRANSFER COEFFICIENT FOR LAMINAR

FLOW IN A TUBE
There are two boundary conditions in which heat transfer coefficient can be determined :
(a) Constant Heat Flux
(b) Constant Wall Temperature
Other analytical methods for arbitrarily varying temperature or arbitrarily varying heat
flux are quite complex.
Constant Heat Flux
Let us consider the control volume (Figure (8.20)) for laminar flow through a tube
where heat is transferred by conduction into and out of the element in a radial
direction.

Dqr + dr

dqr
Dqc,out
r  Tdx 
 (2 rdr )pcp u(r ) T(x) 
Dqc,in = dr  
(2rdr)pcpu(r)T(x) 00 dx  x 
00

Tube
r = rs

Figure 8.20 : Control Volume for Energy Analysis in Flow through a Pipe

79
Convection Rate of heat conduction into the element
dT
dQr   k 2rdx . . . (8.186)
dr
Rate of heat conduction out of the element

 T  2T 
dQr dr  k 2  r  dr  dx   2 dr  . . . (8.187)
 r r 
Heat carried away by the fluid
T
dQc  u 2rc p dx . . . (8.188)
x
By energy balance,
dQr  dQr  dr  dQc . . . (8.189)

T  2T T
 k 2rdx  k 2rdx 2  k 2rdx dr
r r x
 2T  2T
 k 2drdx  k 2drdx dr
x 2 r 2
T
 u 2rdrc p dx . . . (8.190)
x
Neglecting the last term of the L.H.S. of the equation and simplifying,

 T  2T  T
k   r 2   uc p r . . . (8.191)
 r r  x

1   T  1 T
or, r  . . . (8.192)
ur r  r   x
Since the heat flux over the surface is uniform, the fluid temperature increases
linearly with distance x, and so,
T
C . . . (8.193)
x
Eq. (8.192) then becomes an ordinary differential equation with r as the only
variable.
T
At r  0,  0 (at the centre)
r
 T 
At r  R, k    q w"  constant
 r  r  R
From Eq. (8.192),

  T   r 2  r T
 r   u 
0 1  
2 
. . . (8.194)
r  r   R   x

T u 0 T  r 2 r4 
r      C1 . . . (8.195)
r  x  2 4 R 2 
80
 Boundary Layer
T u 0 T  r r3  C1 Formation
     . . . (8.196)
r  x  2 4 R 2  r

u 0 T  r2 r4 
T r , z       C1 In r  C 2 . . . (8.197)
 x  4 16 R
2

T
At r = 0,  0,  C1  0
r
r  0, T  T0 , C 2  T0 ,

The centerline temperature

u 0 T R 2  r  1 r  
2 4

T  T0        . . . (8.198)
 x 4  R  4  R  

This is the temperature distribution of the fluid in the radial direction. The average
bulk temperature of the fluid Tb  is the temperature if the fluid were well mixed
adiabatically so that there is no radial variation of temperature at any cross-section.
It is also called the mixing cup temperature. Therefore,
R R

 u 2rdrc p T  Turdr
Tb  0
R
 0
R
. . . (8.199)

 u 2rdrc
0
p  urdr
0

At r  R, T  Tw . Therefore, Eq. (8.192) becomes

u 0 T R 2 3 3 u 0 R 2 T
Tw  T0   T0  . . . (8.200)
 x 4 5 16  x
Substituting Eqs. (8.198) and (8.171) for T and u, respectively in Eq. (8.199), we
obtain

r 2  u 0 T R 2  r 2 1 r 2 
R
R

 Turdr 0 0  R 2  0  x . 4  R 2  4 R 4  rdr
u  1   T 
Tb  R
0
  
R
 r 
2

0 urdr 0 u 0 1  R 2 rdr

r 2  
R
 u T  r 2 r4
2u m  1  2  T0  0 .   rdr
0 R   x  4 16 R 2 

 R2 1 R4 
2u m   2 
 2 R 4 

4   
R
r 3  u 0 T  r 3 r5 r5 r7
R 2 0  
 T
 0  r    .    dr
R 2   x  4 16 R 2 4 R 2 16 R 4 

4   R2 1 R 4  u 0 T  R 4 R6 R6 R8 
 T 
 0  
  .
    
R2   2 R 4   x  16 96 R
2 2
24 R 2 128 R 4 
81
Convection
4  R 2 u 0 T 4 7 
  T0  R 
R2  4  x 384 

7 u 0 R 2 T
 T0  . . . (8.201)
96  x
Rate of heat transfer

 T 
Q  h c ATw  Tb   kA  . . . (8.202)
 r  r  R
 T 
k 
 r  r  R
hc  . . . (8.203)
Tw  Tb
From Eq. (8.198),

 T  u T R 2  1 1  u T R
   0  2 2R  4R 3   0 . . . (8.204)
 r  r  R  x 4  R   x 4
4
4R
Substituting Eqs. (8.204), (8.201) and (8.200) in Eq. (8.203)
u 0 T R u 0 T R
k k
h c   x 4   x 4
Tw  Tb 3 u 0 R 2 T 7 u 0 R 2 T
T0   T0 
16  x 96  x
k 48 k  2 48 k
   . . . (8.205)
 11  11 D 11 D
 R
 24 
hD 48
Nu d    4.364 . . . (8.206)
k 11
(for q w" = constant).

For constant heat flux boundary condition,

Q
 q c  constant . . . (8.207)
A
 2
q c Ddx  mc p dTb  D u m c p dTb . . . (8.208)
4
dTb q c D 4q c 1
  .  constant . . . (8.209)
dx mc p D c p u m

Bulk fluid temperature Tb varies linearly with x. Also,

k
q c  hc Tw  Tb   Nu d Tw  Tb  . . . (8.210)
D
qc D
Tw  Tb   constant . . . (8.211)
4.364k
In the fully developed laminar forced convection heat transfer, Tw  Tb  also
remains constant with x, as shown in Figure 8.21(a).
82
 Entrance Boundary Layer
Entrance
region developed Formation
region
T

Surface temperature, Tw(x)

(Tw – Tb)

Bulk temperature, Tb(x)

O X
Distance from entrance

(a)
T
Tw(x)

T-Tb Tb(x)

T in

O X
Distance from entrance

(b)
Figure 8.21 : Variation of Average Bulk Temperature of a Fluid in a Pipe for
(a) Constant Heat Flux and (b) Constant Wall Temperature

Uniform Wall Temperature

When the tube wall temperature is constant, the analysis is more difficult because
the temperature difference between the wall and the bulk fluid, Tw  Tb , varies
Tb
along the tube, i.e.  f  x  , as shown in Figure 8.21(b). Eq. (8.192) can be
x
solved subject to the boundary condition that at r  R , T  x, R   constant, which
needs an interactive procedure. Here too, the Nusselt number is a constant

Nu d  3.66 ( Tw = constant) . . . (8.212)

The energy balance for a differential length dx gives

dQc  mc p dTb  q w" pdx . . . (8.213)

where p is the perimeter of the duct and q w" is the wall heat flux.

dTb q w" p p
  hc Tw  Tb  . . . (8.214)
dx mc p mc p

dTb T  T 
Since  d b w for a constant wall temperature,
dx dx

d Tb  Tw  p
 hc Tw  Tb  . . . (8.215)
dx mc p
83
Convection Putting T  Tw  Tb ,

d T  p
  hc T . . . (8.216)
dx mc p

d T 
Te L
p
T T   mc p 0 hc dx . . . (8.217)
i

Te pL
In  hc . . . (8.218)
Ti mc p
L
1
Where h c   hc dx
L0
Rearranging Eq. (8.218),

Te  h c pL 
 exp   . . . (8.219)
Ti  mc 
 p 

The rate of heat transfer by convection to or from a fluid through a duct with
Tw = constant can be written as

Qc  mc p Tw  Tb ,i   Tw  Tb ,e   mc p Ti  Te  . . . (8.220)

Substituting mc p from Eq. (8.219),

hc pl T  Ti
Qc   Ti  Te   h c Aw e
T T
In e In e
Ti Ti

 h c Aw T1m . . . (8.221)

where T1m is the log mean temperature difference (LMTD).

Hausen [18] recommended the following relation for the average convection
coefficient in laminar flow through ducts with uniform surface temperature
0.14
0.0668 Re DH . Pr .D / L  b 
Nu d  3.66    . . . (8.222)

1  0.045 Re DH . Pr .D / L  0.66
 w 
A relatively simple empirical equation suggested by Sieder and Tate [6] has been
widely used to correlate experimental results for liquids in tubes and can be written
in the form
0.33 0.14
 Re DH . Pr .D H   b 
Nu DH  1.86 
   . . . (8.223)
 L   w 
where all the properties in Eqs. (8.222) and (8.223) are based on the bulk
0.14
 
temperature and the empirical correlation factor  b  is introduced to account
 w 
for the temperature variation on the physical properties.
An additional complication in the determination of a heat transfer coefficient in
laminar flow arises when the buoyancy forces are of the same order of magnitude
as the external forces due to the forced convection. Such a condition may arise in
oil coolers when low flow velocities are used. Also, in the cooling of the rotor
84
blades of gas turbines, the natural convection effects may be so large that their Boundary Layer
Formation
effect on the velocity pattern cannot be neglected even in high-velocity flow.
When the buoyancy forces are in the same direction as the external forces, they
increase the rate of heat transfer. When the external and buoyancy forces act in
opposite directions, the heat transfer is reduced. In practice, natural convection
effects are hardly ever significant in turbulent flow.
106
Forced convection turbulent flow
Mixed convection turbulent flow
105
NuD = 4.69 ReD0.27 pr0.21 GrD0.07 (DIL)0.36

ReD 104 Laminar turbulent transition

Forced 103
convection
laminar flow
102
Mixed convection
laminar flow
10
Natural convection
1
102 103 104 105 106 107 108
GrDpr D
L
(a)
10 6
Forced convection turbulent flow

10 5 Mixed convection turbulent flow

Laminar turbulent transition

104
ReD

103
Natural convection
102
Mixed convection
Laminar flow
10 Forced convection
Laminar flow
1
102 103 104 105 106 107 108
GrDpr D
L
(b)
Figure 8.22 : Forced, Natural and Mixed Convection Regimes for
(a) Horizontal Pipe Flow and (b) Vertical Pipe Flow

The influence of natural convection on the heat transfer to fluids in horizontal

isothermal tubes was investigated by Depaw and August [7] who correlated the
data by the equation
0.14
  0.88 1/ 3
Nud  1.75  b 
 d 
Gz  0.12 GzGr 1/ 3 Pr 0.36  

. . . (8.224)
 w 
where Gz is the Graetz number, defined by
 D
Gz  Re d Pr . . . (8.225)
4 L
g  D 3
and Grd 
v2
is the Grashoff number.
It is valid in the range 25 < Gz < 700, 5 < Pr < 400 and 250 < Grd < 105. Physical
properties except for  w are to be evaluated at the average bulk temperature.
85
Convection
8.12 VELOCITY DISTRIBUTION IN TURBULENT FLOW
THROUGH A PIPE
Velocity distribution in turbulent flow has been investigated extensively because of its
importance in practice, but no fundamental theory is yet available to determine this
velocity distribution rigorously by purely theoretical approaches. Therefore, empirical
and semi-empirical relations are used to correlate the velocity field in turbulent flow.
Nikuradse was an early investigator who presented a careful measurement of velocity
distribution in turbulent flow through a smooth pipe. Later more experiments were
conducted to develop empirical relations to fit the velocity distribution. We would
discuss the flow field into three distinct layers :
(a) a very thin layer at the wall where viscous shear stress dominates, called the
viscous sublayer,
(b) a buffer layer, where viscous and turbulent shear stresses are equally
important, and
(c) a turbulent layer, in which turbulent shear stress dominates.
In the study of velocity distribution for turbulent flow, the following two dimensionless
quantities are introduced:
u
u   dimensionless velocity
 w /  1 / 2
1/ 2
 y  
y   w   dimensionless distance
v  
where  w is the shear stress at the wall and u is the velocity component parallel to the
wall surface.
Experiments have shown that the viscous sublayer is maintained in the region y  < 5.
u
The shear stress is  w   and integrating it with u = 0 at y = 0, we get u   y  for
y
viscous sublayer, 0 < y  < 5.

The buffer layer is considered to extend from y  = 5 to y  = 30 and a logarithmic

velocity distribution law u   A In y  + B is assumed, where constants A and B are
determined from the condition that u  be equal to that of the laminar sublayer and of the
turbulent layer at y  = 5 and y  = 30, respectively. The resulting velocity distribution
becomes
u  = 5.0 In y   3.05 for buffer layer 5  y   30

The region y  > 30 is considered to be the turbulent layer. By utilizing the

mixing-length concept and assuming that the mixing length varies linearly with the
distance from the wall, 1  y , it can be shown that the velocity distribution in turbulent
layer has a logarithmic profile in the form
1
u  In y   C . . . (8.226)

where  is called the universal constant. For smooth pipes, C = 5.5, and the velocity
distribution in the turbulent layer becomes
u  = 2.5 In y  + 5.5 for turbulent layer y  > 30 . . . (8.227)
86
Figure 8.23 shows a correlation of the velocity distribution law as obtained above for the Boundary Layer
Formation
three layers along with Nikuradse’s measured experimental data. It is often termed as
“universal velocity profile”.
35

sublayer

Turbulent
Laminar

Buffer
layer
30

layer
25

20
u+ = 2.5 ln y+ + 5.5
 15
u  u / 0 / 

10
u+ = y+ u+ = 5.0 ln y+ - 3.05
5

0
1 10 102 103 104 105

 y 0
y 
 
Figure 8.23 : Logarithms Velocity Distribution Law and Nikuradse’s Experimental Data for
Turbulent Flow inside Smooth Pipe

8.13 ANALOGY BETWEEN HEAT AND MOMENTUM

TRANSFER IN TURBULENT FLOW
The analysis of heat transfer for turbulent flow is much more involved than that for
laminar flow. Based on the analysis of laminar flow along a flat plate, we developed a
relation between the heat transfer coefficient and the local drag coefficient C f x as

St x Pr 2 / 3  C f x / 2 . . . (8.228)

Nu x hx
where St x   . . . (8.229)
Re x Pr c p u 

w
and C fx  . . . (8.230)
1 2
u 
2
For turbulent flow inside a circular pipe a similar expression can be found out. Making a
force balance,

 2 fL u m2  2
 wDL  p D  D . . . (8.231)
4 D 2 4
f
w  u m2 . . . (8.232)
8
where f is the Darcy friction factor. Heat transfer rate at the wall
Q dT
qw   k . . . (8.233)
A dy
and the shear stress
du
w   . . . (8.234)
dy
87
Convection Dividing the two equations
Q k dT
 . . . (8.235)
A w  du
cp
For fluids having Pr = 1, i.e. 1
k
Q dT
 c p . . . (8.236)
A w du
u T
Q m b

A 0
du   c p w  dT
 . . . (8.237)
Tw

Qu m
 c p w Tw  Tb  . . . (8.238)
A
Q c p w
 hc  . . . (8.239)
ATw  Tb  um
Substituting  w from Eq. (8.232),

cp u m2
hc  f . . . (8.240)
um 8
Nu d hc f
St d    . . . (8.241)
Re d Pr c p u m 8
This is known as Reynolds analogy for momentum and heat transfer. Reynolds first
assumed that the entire flow consists of a single zone of highly turbulent region. He
neglected the presence of the viscous sublayer and the buffer layer. In such a turbulent
core, the molecular diffusivity of heat  and that of momentum v are negligible compared
with turbulent diffusivities, i.e.
v <<  M and  <<  H

Then, the equations

du
   v   M  . . . (8.242)
dy
Q dT
and     H  . . . (8.243)
Ac p dy
where y is the distance measured from the tube wall, become
 du
 M . . . (8.244)
 dy
qw dT
and   H . . . (8.245)
c p dy

Assuming the turbulent diffusivities to be equal,  M   H , i.e. Prt , turbulent Prandtl

number = 1, we obtain by division,
qw dT
 . . . (8.246)
c p du
88
On integration, we obtain the same Eq. (8.239). The Eq. (8.241), called Reynolds analogy, Boundary Layer
Formation
is valid for both laminar and turbulent flow in a pipe for Pr = 1 or Prt  1 . Prandtl
assumed that the flow field consisted of two layers, a viscous sublayer where the
molecular diffusivities are dominant, i.e.
 M << v and  H << 
and a turbulent core where the turbulent diffusivities are dominant, i.e.
v <<  M and  <<  H and  M =  H
These assumptions are utilized to simplify Eq. (8.243) for each layer, the equations are
integrated, and the definitions of the friction factor and the heat transfer coefficient are
introduced. The following result is obtained.
hc f 1
St d   . . . (8.247)
c p u m 8 1  5 f / 81 / 2 Pr  1
These relationship is known as the Prandtl analogy for momentum and heat transfer for
fully developed turbulent flow in a pipe. We note that for Pr = 1, the Prandtl analogy
reduces to the Reynolds analogy.
Von Karman extended Prandtl’s analogy by separating the flow field into three distinct
layers: a viscous sublayer, a buffer layer and a turbulent core. He made assumptions
about the relative magnitudes of the molecular and turbulent diffusivities of heat and
momentum in the three layers and obtained the following result :
hc f 1
St d   . . . (8.248)
c p u m 8 1  5 f / 8 Pr  1  In5 Pr  1 / 6
1/ 2

which is known as von Karman analogy for momentum and heat transfer for fully
developed turbulent flow in a pipe.
According to experimental data for fluids flowing in smooth pipes in the range of
Reynolds numbers from 10,000 to 1,000,000 the friction factor is given by the empirical
relation
f  0.184 Re d0.2 . . . (8.249)

Using this relation, Reynolds analogy becomes

Nu d f 0.184 Re d0.2
St d   
Re d Pr 8 8

= 0.023 Re d 0.2 . . . (8.250)

Since Pr = 1,

Nu d  0.023 Re 0d.8 . . . (8.251)

0 . 8
0 .8  0 .2 
or, hc  0.023u D m k   . . . (8.252)

1
In fully developed turbulent flow, h c  u m0.8 and h c  . For a given flow rate, an
D 0 .2
increase in the tube diameter D reduces the velocity and hence h c . The use of small
tubes and high velocities result in high h c . But large velocities require more pumping
power. In the design of heat exchanger equipment, it is necessary to strike a balance
between the gain in heat transfer rates and the increase in pumping requirements.
89
 Convection Figure 8.24 shows the effect of surface roughness on the friction factor (Moody’s chart).
We observe that the friction factor increases appreciably with the relative roughness,
defined as the ratio of the average asperity height  to the diameter D. Roughening the
surface would increase the friction factor and hence the heat transfer coefficient. In
Figure 8.25, Stanton number is plotted against the Reynolds number for various values of

the roughness ratio, for the artificially roughened tubes with sand grains.
D
0.1
0.09
0.08 0.05
0.07 0.04
0.03
0.05
0.02

Relative Roughness
0.04 0.015
0.01
0.008
Friction Factor

0.03
0.005
0.025 0.004
Equation 6.5
0.002
0.02 Laminar flow
0.001
64 0.0008
R=
ReD 0.0006
0.015 0.0004
0.0002
Laminar Transition
0.01 flow zone 0.0001
0.000.005
0.009
0.008 0.000.01
103 2 3 4 5 6 7 8 9104 2 3 4 5 6 7 8 9105 2 3 4 5 6 7 8 9106 2 3 4 5 6 7 8 9 107 2 3 4 5 6 7 8 9108

Figure 8.24 : Moody’s Diagram: Friction Factor varying with Reynolds Number for Laminar and
Turbulent Flow in Tubes with Various Surface Roughnesses

At small Re d , St d has the same value for rough and smooth surfaces. For each value of

, St d reaches a maximum, and with a further increase in Re d , it begins to decrease.
D
4  10-3 0.08
/ D = 0.04
3
0.002
St = NuD/ReDPr

2
0.01
/ D = 0.04
/ D = 0.001
10-3 0.0005
8

6 Smooth pipe
4  10-4
5  103 8  104 2 4 6 8 105 2 4 6 8  105

ReD
ε
Figure 8.25 : Heat Transfer in Artificially Roughened Tubes, St d vs Re d for various values of
D

SAQ 6
(a) What are the generally accepted values of critical Reynolds numbers for
(i) flow over a flat plate,
(ii) flow over a circular tube, and
(iii) flow in a tube?
(b) How is the friction factor for flow in a tube related to the pressure drop?
90
 (c) What will be the relative magnitude of heat flux In case of forced Boundary Layer
Formation
convection (laminar flow) in a tube for
(i) at inlet to the tube, and
(ii) near exit of the tube?

8.14 EMPIRICAL CORRELATIONS

The phenomena of turbulent forced convection are so complex that empirical correlations
are used in practice in engineering design.
(a) The Dittus-Boelter equation, given below, extends the Reynolds analogy to
fluids with Prandtl numbers between 0.7 and 160 by multiplying the right
hand side of Eq. (8.251) by a correction factor of the form Pr n :

hc D
Nu d   0.023 Re 0d.8 Pr n . . . (8.253)
k
where n = 0.4 for heating Tw Tb  and n = 0.3 for cooling Tw Tb  . It is
valid within ± 20% for uniform wall temperature as well as uniform heat
flux conditions within the following ranges of parameters :

6000 < Re d < 10 7

0.5 < Pr < 120

L
> 60
D
It should be used only for situations with moderate temperature differences
Tw  Tb  , since variations in physical properties due to temperature
gradient at a given cross-section are not taken into account by the correlation.
(b) For situations in which significant property variations exist due to a large
temperature difference Tw  Tb  , a correlation developed by Sieder and
Tate [6] is recommended:
0.14
 b 
Nu d  0.027 Re 0 .8
d Pr 1/ 8
  . . . (8.254)
 w 
where all properties except  w are evaluated at the bulk temperature. The
viscosity  w is evaluated at the wall temperature. The correlation is valid
both for uniform wall temperature and uniform heat flux in the following
range of conditions :

6000 < Re d < 10 7

0.7 < Pr < 10,000

L
> 60
D
To account for the variation in physical properties due to the temperature gradient in the
flow direction, the wall and bulk temperatures should be the mean values between the
inlet and outlet of the duct. For noncircular ducts, hydraulic diameter D H should be used.
(a) A similar correlation, but restricted to gases was suggested by Kays and
London [8] for long ducts 91
Convection n
T 
Nu d  C Re 0 .8
d Pr  b
0 .3
 . . . (8.255)
 Tw 
where all properties are based on the bulk temperature Tb . The constant C
and exponent n are
C  0.02 for uniform wall temperature Tw
C  0.021 for uniform heat flux, qw"
n  0.575 for heating
n  0.150 for cooling
(b) Petukhov [9] developed a more accurate expression for fully developed
turbulent flow in smooth tubes :

 f / 8  Re d Pr
n
 b 
Nu d    . . . (8.256)
1.07  12.7  f / 8  Pr 2 / 3  1    w 
1/ 2

where n = 0.11 for Tw  Tb , n  0.25 for Tw Tb and n = 0 for constant

Tw  Tb 
heat flux or for gases. All properties are evaluated at T f  except
2
for b and  w . The friction factor may be obtained from Moody’s chart or
form
f  1.82 log10 , Re d  1.64 
2
. . . (8.257)
(c) Hausen [10] presents the following empirical relation for fully developed
laminar flow in tubes at constant wall temperature (as stated earlier)
d
0.0668   Re d Pr
Nu d  3.66  L . . . (8.258)
2/3
 d  
1  0.04   Re d Pr 
 L  
The Nusselt number approaches a constant value of 3.66 when the tube is
sufficiently long.
Example 8.1
Assume a linear temperature profile T  C  dy for flow of a fluid over a flat plate.
(a) Apply the appropriate boundary condition and express T in terms of
 t , Tw and T .
(b) Assume a linear velocity profile u  a  by .

Obtain an expression for as a function of Prandtl number.
t
(c) Obtain an expression for Nu x .

Solution
Writing the energy balance equation for the control volume (Figure 8.26),

92
 Boundary Layer
Formation
 ' 
Cp dx
x 0 T udy 
C,V

lt(x) t   ' 
l Cp  Tu dy   Cp  Tu dy dx
Cp d  Tu dy 0 x  0 
0

dx

 T 
k   dx1
 y y 0

Figure 8.26

  1  
1
T 
 c p   Tudy  dx   c p dx  T udy  k   dx  0
 x 0  x 0  y  y 0

Since for y   t , the integrand is zero. Therefore, the energy equation can be
written as

 t  T 
 
x 0
T  T udy     . . . (1)
 y  y 0
The temperature profile is given to be
T  C  dy

 T 
  d
 y  y 0

At y  0, T  Tw  C

At y   t , T  T

T  Tw  d  t

d  T  Tw  /  t
The temperature distribution is given by
T  Tw
T  Tw  y
t
T  Tw y
or,  . . . (2)
T  Tw  t

1  T  1
  
T  Tw  y  y 0 t

 T  T  Tw
  
 y  y 0 t
93
Convection Substituting in Eq. (1)

 t u T T
u  T  Tw   T  Tw  dy    w
x 0 u t

 t T  Tw  u 
u  1   dy 
x 0  T  Tw  u t
. . . (3)

The velocity profile has been given to be

u  a  by
At y  0, u  0,  a  0

y   , u  u   b

u
 b

u
 u y

u y
or,  . . . (4)
u 
Substituting Eqs. (2) and (3),

 t y y 
u  1 
x 0   t
 dy 
 t
t
Let   , so that  t  


 t y  y 
u  1  
x 0   t   t
dy 
t

 t   y  y2  
or u  
x 0   t
 2  dy 
t  t

    t2   t3  
or u   
x   t 2  t2 3   t

u d 
6 dx

 2 


u  2 d d  
or     2 
6  dx dx  

u  3 d d 
or    2 2 2   . . . (5)
6  dx dx 
From the x-momentum equation, we obtain

  u 
   u  u  udy   w    
x 0  y  y 0
94
  Boundary Layer
  u  u u
u   1   dy  
Formation
2
or 
x 0  u   u  

d  y y2  
or  u    2  dy 
dx 0     

d  1 2 1 3  
or  u    
dx   2  2 3  
 u d  
or 
6 dx 
d 6 6v
   . . . (6)
dx u  u 
 
6v
0  d  u 0 dx
 2 6vx
or 
2 u

 2 12v 12
 
x 2 u x Re x
 3.464
 . . . (7)
x  Re x 1/ 2

Substituting Eq. (6) in Eq. (5),

u  3 6v 12vx d  
  2 2  
6  u u dx 
d 1
 3  4 2 x  . . . (8)
dx Pr
4 d 3 1
or 3 x 
3 dx Pr
4 dy 1
Putting  3  y, y  x 
3 dx Pr
Particular Integral
1
y
Pr
Complementary Function
4 dy
y x 0
3 dx
dy 3 dx
or 
y 4 x
y  Cx 3/ 4
1
  3  Cx 3 / 4  . . . (9)
Pr 95
Convection Let the portion x0 of the plate be unheated (Figure 8.27) so that the thermal
boundary layer starts from x  x0 .

At x  x0 ,  t  0,   0
(x)

t(x)

Tw

x
X0
Figure 8.27

Substituting in Eq. (9)

1
0  Cx03/ 4 
Pr
x03/ 4
C
Pr
x03/ 4 3/ 4 1
or 3  x 
Pr Pr

1  x  
3/ 4

 1   0  
Pr   x  
1/ 3
  x0 3/ 4 
  Pr 1/ 3
1    
  x  
If x0  unheated portion = 0

  Pr 1/ 3
t
or  Pr 1/ 3


 Pr1/ 3
t
Q  T  T  Tw k
 k    k  T w  T  
A  y  y 0 t t

k  Re x 
1/ 2
k k
hc   
 t  Pr 3.464 x  Pr 
1/ 3 1/ 3

hc x
Nu x   0.288 Re1/x 2 Pr1/ 3
k
Example 8.2
Air at a temperature of T flows over a flat plate with a free stream velocity of u .
The plate is maintained at a constant temperature of Tw . The velocity u and
temperature T of air at any location are given by
96
 2 Boundary Layer
u y T  Tw  y  y Formation
 sin and  2    
u 2 T  Tw  t   t 
where y is the distance measured from the plate along its normal, and  and  t
are the hydrodynamic and thermal boundary layer thickness, respectively. Find the
ratio of heat transfer coefficient to shear stress at the plate surface using the
following data

u  10 m/s,  Pr 1/ 3, T w  200 o C, T  50o C,
t
 air   2.5  10  5 kg/ms, kair   0.04 W/mk,

c p  air   1.0 kj/kgK.

Solution
Shear stress at the wall
 u 
w    
 y  y 0
Velocity profile
u  y 
 sin  
u 2
 u    y   u 
   u  cos  
 y  y 0  2   y 0 2 2

 u
 w   . . . (1)
2
Temperature distribution
2
T  Tw  y  y
 2    
T  Tw  t   t 
1  T  2
  
T  Tw  y  y 0 t

 T  2 T  Tw 
  
 y  y 0 t

 T  2 Tw  T 
h Tw  T   k   k
 y  y 0 t

2k
h . . . (2)
t
From Eqs. (1) and (2),
h 2k 2 4k
  Pr1/ 3
w  t  u  u
h
This is the desired expression for .
w
97
Convection  c p 2.5 10 5 1
Pr    0.625
k 0.04 10 3

4  0.04   0.625 
1/ 3
h W ms s Nm

w 5
2.5 10 10 mk kg m sW

= 174.18ms 1K 1 .
Example 8.3
A rectangular plate is 120 cm long in the direction of flow and 200 cm wide. The
plate is maintained at 80oC when placed in nitrogen that has a velocity of 2.5 m/s
and a temperature of 0oC. Determine
(a) the average heat transfer coefficient, and
(b) the total heat transfer from the plate. The properties of nitrogen at
40oC are   10142 kg/m 3 , c p  1.04 kJ/kg.k,
v  15.63  10  6 m 2/s and k  0.0262 W/m K.
Solution
u xc 2.5  xc
Re xc  500, 000  
v 15.63 10 6
5 105 15.63 10 6
xc  m  2 1.563
2.5
= 3.126 m = 312.6 cm
Since the plate length is 120 cm in flow direction, laminar flow persists in the
entire length of the plate, for which

Num  0.664 Re1/L 2 Pr1/ 3

u L 2.5 1.2
Where Re L    191.938.6
v 15.63 10 6

 Re L 
1/ 2
 438.1

cp  1.04 15.63 10 6 1.142

Pr  
k 0.0262 10 6
 708.53  10 3  0.708

 Pr 
1/ 3
 0.89

hm L
Num  0.664  438.1 0.89  258.9 
k
258.9  0.0262
hm   5.653 W/m 2K
1.2
Rate of heat transfer from the plate

Q  hm  Lb Tw  T   5.653 1.2 2.0 80  0 

= 1085.4 W = 10.85 kW
98
Example 8.4 Boundary Layer
Formation
Water flows over a flat plate measuring 1 m  1 m with a velocity of 2 m/s. the
plate is at a uniform temperature of 90oC and the water temperature is 10oC.
Estimate the length of plate over which the flow is laminar and the rate of heat
transfer from the entire plate. The properties of water at 50oC are
  988.1 kg/m 3 , v  0.556  10 6 m 2/s, Pr  3.54 and k  0.648 W/m K.
Solution
u xc
Re c  500, 000 
v
500, 000  0.556 10 6
xc   0.139 m
2
The length of plate up to which the flow is laminar is 0.139 m.
Laminar Part

Nuc  0.664  Rec  Pr 1/ 3  0.664 500, 000  3.54 

1/ 2 1/ 2 1/ 5

 0.664  707.11 1.523

hm xc
 715.08
k
715.08  0.648
hm   3333.6 W/m 2K
0.139
Qla min ar  hm A Tw  T 

 3333.6  0.139  1 90  10   37069.6 W

 37.07 kW
Turbulent Part
hm  L  xc 
k

 0.036 Re 0.8 0.8
L  Re c Pr 1/ 3 
u L 2 110 6
Re L    3.597 10 6
v 0.556
 Re L   3,597, 000 
0.8 0.8
 175692

 Rec   500, 000 

0.8 0.8
 36239

 Pr   3.54 
1/ 3 1/ 3
 1.523

hm  l  0.139 
 0.036 175, 692  36, 239  1.523
0.648
 7645.93
hm  5754.4 W/m 2K
Qturb  5754.4  0.861 1 80  396365 W
 396.37 kW
Qtotal  Q1am .  Qturb .  37.07  396.37
 433.44 kW. 99
Convection Example 8.5
It was found during a test in which water flowed with a velocity of 2.44 m/s
through a tube (2.54 cm inner diameter and 6.08 m long), that the heat lost due to
friction was 1.22 m of water. Estimate the surface heat transfer coefficient based
on Reynolds analogy. Take   998 kg/m 3 and c p  4.187 kJ/kgK.

Solution
p  h  g  1.22 m  998 kg/m 3  9.81 m/s 2  10 3
 11.944 kPa

fL  um2
p 
D 2

f  6.08 998   2.44 

2

 2
.  11,944 N/m 2
2.54 10 2
11,944  2.54 10 2  2
 f   0.0168
6.08  998   2.44 
2

By Reynolds analogy,
h f
Std  
 c p um 8
0.0168
 h  998  4.18  2.44
8
 21.38 kW/m 2K
Example 8.6
Air at atmospheric pressure and 100oC enters a 2 m long tube (4 cm diameter) with
a velocity of 9 m/s. A 1 kW electric heater is wound on the outer surface of the
tube. Find
(a) the mass flow rate of air,
(b) the exit temperature of air, and
(c) the wall temperature at outlet.
Assume that the rate of heat absorption by air per unit area is uniform throughout
the length of the tube. Take for air, R  0.287 kJ/kg K and c p  1.005 kJ/kg K.

Solution
Density of air at 100oC,
p 101.325
 
RT 0.287  373
 0.946 kg/m 3
Mass flow rate of air, m   Au m

 2
 0.946   4  10 4  9
4
 0.0107 kg/s
100
 Now, Q  mc p Te  Ti   1 kW Boundary Layer
Formation

1
 Te  100   193 o C
0.0107 1.005
Ti  Te 293
Mean air temperature, Tm    146.5o C
2 2
At 146.5oC, the properties of air are,

  0.84 kg/m 3 , Pr  0.683, k  0.026 W/m K

and v  28.8  10 6 m 2 /s.

um D 9  4 10 2
Re d    12,500
v 28.8 10 6
Using Dittus-Boelter equation

Nud  0.023  Red  Pr 

0.8 0.4

 0.023 12,500  0.683 

0.8 0.4

 0.023  1894.6  0.859  37.43

37.43  0.026
h  24.33 W/m 2 K
0.04

Now, 
Q  hA Twe  Te 
where Twe is the exit wall temperature and Te is the exit air temperature. Since
Q
heat flux is uniform,
A


1 kW  24.33 10 3 kW/m 2K    4 10 2  2 m 2  Twe  193 K 
10 4
Twe   193  356.5 0 C
2.433    8
The exit wall temperature is 356.5oC.
Example 8.7
Lubricating oil at a temperature of 60oC enters a 1 cm diameter tube with a
velocity 3.5 m/s. The tube surface is maintained at 30oC. Calculate the tube length
required to cool the oil to 45oC. Assume that the oil has the following average
properties for the temperature range of this problem :

  865 kg/m 3 , k  0.14 W/mK, c p  1.78 kJ/kgK and v  9  10 6 m 2 /s.

Solution
um D 3.5  0.01
Re d    3888.9
v 9 106
cp 1.78  9 10 6  865
Pr    98.98
k 0.14 10 3 101
Convection Using Dittus-Boelter equation,
Nud  0.023  Red  Pr 
0.8 0.3

 0.023  3888.9  98.98 

0.8 0.3

 0.023  744.5  3.97  67.98

67.98  0.14
h  951.72 W/m 2K
0.01
For isothermal tube surface,
h DL T T
 1n w i
mc p Tw  Te

  0.01  3.5  0.238 kg/s
2
where m   Au m  865 
4
0.238 1.78
L  14.18 m
0.951    0.01
Example 8.8
Air at 20oC and 1 atm flows over a flat plate at 40 m/s. The plate is 80 cm long and
is maintained at 60oC. Assuming unit depth in z-direction, calculate the heat
transfer rate from the plate. Properties of air at 40oC are : Pr = 0.7,
k = 0.02723 W/m K, c p  1.007 kJ/kg K and   1.906  10 5 kg/ms.
Solution
20  60
Tf   40o C = 313 K
2
p 101.325
   1.128 kg/m 3
RT 0.287  313
 u L 1.128  40  0.80
Re L     1.89 10 6
 1.906 10 5
The boundary layer is turbulent, since Re L > 5 105

hL
Nu L 
k

 0.036 Re 0.8
L  871 Pr
1/ 3


  0.036 1.89  10 6   871 0.7 
0.8 1/ 3
 
 2908.83  0.89  2588.86
2588.86  0.02723
h  88.12 W/m 2 K
0.8
Q  hA Tw  T   88.12   0.8 160  20 
 2820 W = 2.82 kW
Example 8.9
Air at 20oC and 1 atm flows across a sphere of 15 mm diameter at a velocity of
5 m/s. A small heater inside the sphere maintains the surface temperature at 77oC.
Estimate the rate of heat transfer from the sphere. Properties of air at 27oC or
300 K are :    1.8462  10 5 kg/ms, k  0.02624 W/m K,
v  15.69  10 6 m 2 /s and Pr = 0.708 At Tw  77 o C  350 K,
 w  2.075  10 5 kg/ms.
102
Solution Boundary Layer
Formation
u D 5  0.015
Re d    4780
v 15.69 10 6
1/ 4
  
Now, Nu  2  0.4 Re  1/ 2
d  0.06 Re 2/3
d Pr 0.4
 
 w 
Where properties are evaluated at free-stream temperature,
1/ 4
 1.8462 
Nu  2  0.4  4780   0.06 4780   0.708 
1/ 2 1/ 3 0.4

   
 2.075 

 2   27.66  17.07  0.87  0.97  39.75

0.02624
h  39.75   69.54W / m 2K
0.015

 0.015 
2

Q  hA Tw  T   69.54  4   77  27 

4
= 2.46 W
Example 8.10
In a power plant feedwater is flowing through a duct of rectangular cross-section
8 cm  4 cm and the wall temperature is maintained at 170oC throughout. The
feedwater flows at a rate of 300 kg/min, enters at a temperature of 20oC and is
heated to 150oC. Compare the heat transfer coefficients obtained using
(a) Dittus-Boelter equation, and
(b) Sieder-Tate equation and estimate the required length of the duct.

Properties of water at 105oC are Pr = 1.64,   265  10 5 kg/ms,

c p  4.226 kJ/kg K, k  683  10 6 kW/m K and
 w at 170o C  158  10 6 kg/ms.
Solution

Dh  Hydraulic diameter of the duct

4 A 4 8  4
   5.333
P 2 8  4 

150  20
T1m   64.52o C
150
1n
20
Properties values of water were therefore selected at 170  64.52  105 o C.

 uD mDh 300 5.333 10 2

Re d    
 A 60 32 10 4  265 10 6
= 314,400
Pr = 1.64
103
Convection Dittus-Boelter equation

Nud  0.023  Red  Pr 

0.8 0.4

 0.023   314, 400  1.64 

0.8 0.4

 0.023  2.5  10 4  1.219  700.8

700.8  683 10 6
h  8.975 kW/m2 K.
5.333 10 2
Sider-Tate equation
0.14
0.8 0.33   
Nud  0.027 Re Pr d  
 w 
0.14
4 265 10 6 
 0.027  2.5  10  1.179  6 
 158 10 
= 855.7

855.7  683 10 6

h  10.96 kW/m 2K
5.333 10 2
The latter is probably more accurate as it takes into account viscosity effects
varying with temperature. Taking the latter value of h, i.e. 10.96 kW/ m 2 K, the
duct length has been calculated.

Q  mc p  T  water  5 4.226  150  20 

= 2747 kW
Q  hAT1m

2747 = 10.96  2  0.4  0.08  L  64.52

L  16.19 m.
Exercise 8.1
(a) Show by order-of-magnitude analysis for flow over a plane surface
 1

x  Re x 1/ 2

(b) By Reynolds-Colburn analogy show that for turbulent flow over a plane
surface the local Nusselt number is Nu x  0.0288 Re x0.8 Pr 1/ 3 and the local
drag coefficient is
0.0576
C fx 
 Re x 
0.2

(c) Show that for laminar flow from x = 0 to xc and for turbulent flow from

x  xc to x  L over a flat plate Nu L  0.036 Re L0.8  835 Pr1/ 3 
0.072 1670
and Cf  
Re 0.2
L Re L
104
 (d) For constant heat flux boundary condition for laminar flow over a flat plate Boundary Layer
Formation
Nu x  0.453Re1/x 2 Pr1/ 3
show that the average temperature difference along the plate is
qw L / k
Tw  T 
0.6795 Re1/L 2 Pr 1/ 3
where the qw is the heat flux in W/m2.

Exercise 8.2
(a) Taking the velocity profile in the turbulent flow over a plane surface as
1/ 7
 y
u  u  
 
and wall shear stress as
1/ 4
 v 
 w  0.0228 u  
2

 u  
show that the boundary layer thickness is given by
 0.376

x  Re x 1/ 5

(b) Show that the thickness of laminar sublayer, where the velocity varies
linearly, is

 b /   191/  Re x 
0.7

and the wall shear stress is

0.0296  u2
w 
 Re x 
0.2

(c) Show that for laminar flow through a tube the Fanning friction coefficient
C f is equal to

Cf  f / 4
(d) Calculate the heat transfer coefficient for water flowing through a 25 mm
diameter tube at the rate of 1.5 kg/s, when the mean bulk temperature is
40oC. For turbulent flow of a liquid take
Nud  0.0243Red0.8 Pr 0.4
where all properties are evaluated at the mean bulk temperature.
Exercise 8.3
(a) The crank case of an automobile is 0.6 m long, 0.2 m wide and 0.1 m deep.
Assuming the surface temperature of the crank case is 350 K, estimate the
rate of heat transfer from the crank case to atmospheric air at 276 K at a road
speed of 30 m/s. Assuming that the vibration of the engine and the chassis
induce the transition from laminar to turbulent flow so near to the leading
edge that, for practical purposes, the boundary layer is turbulent over the
entire surface. Neglect radiation and use for the front and rear surface the
same average heat transfer coefficient as for the bottom and sides.
(b) Used engine oil is required to be recycled in a system in which engine oil
flows through a 1 cm inner diameter 0.02 cm wall copper tube at the rate of 105
Convection 0.05 kg/s. The oil enters at 35oC and is to be heated to 45oC by atmospheric
pressure steam condensing on the outside. Calculate the length of the tube
required. Properties of engine oil at 40oC are c p  1964 J/kg K,
  876 kg/m 3 , k  0.144 W/m K,   0.210 N s/m2 and Pr = 2870.
(c) A rectangular plate is 120 cm long in the direction of flow and 200 cm wide.
The plate is maintained at 80oC when placed in nitrogen that has a velocity
of 2.5 m/s and a temperature of 0oC. Determine
(i) the average friction coefficient,
(ii) the viscous drag exerted on the plate,
(iii) the average heat transfer coefficient, and
(iv) the total heat transfer rate from the plate.
(d) A fluid at 27oC flows with a velocity of 10 m/s across a 5 cm outer diameter
tube whose surface is kept at a uniform temperature of 120oC. Determine the
average heat transfer coefficients and the heat transfer rates per metre
length of the tube for
(i) air at atmospheric pressure,
(ii) water, and
(iii) ethylene glycol.

Exercise 8.4
(a) Air at 27oC flows with a free-stream velocity of 40 m/s along a flat plate
2 m long. Calculate the boundary layer thickness at the end of the plate for
air at
(i) 1/2 atm,
(ii) 1 atm, and
(iii) 2 atm.

(b) Engine oil (   868 kg/m 3 , v  0.75  10 4 m 2 /s ) flows with a mean

velocity of 0.15 m/s inside a circular tube having an inside diameter of
2.5 cm. Calculate the friction factor and the pressure drop over the length
100 m of the tube.
(c) Water entering at 10oC is to be heated to 40oC in a 20 mm inner diameter at
a mass flow rate of 0.01 kg/s. The outside of the tube is wrapped with an
insulated heating element that produces a uniform flux of 5 kW/m2 over the
surface. Neglecting any entrance effects, determine
(i) the Reynolds number,
(ii) the heat transfer coefficient,
(iii) the length of tube needed for 30oC temperature rise,
(iv) the inner surface temperature,
(v) the friction factor,
(vi) the pressure drop, and
(vii) the pumping power required if the pump is 50% efficient.
(d) A flat plate 100 cm wide and 150 cm high is to be maintained at 90oC in air
with a free-stream temperature of 10oC. Determine the velocity at which the
106
 air must flow over the flat plate along 150 cm side so that the rate of energy Boundary Layer
Formation
dissipation from the plate is 3.75 kW. Properties of air at 50oC are
  1.09 kg/m3, k = 0.028 W/m K, c p  1.007 kJ/kg K and
  2.03  10 5 kg/ms.

8.15 SUMMARY
We have discussed here about the heat transfer and hydrodynamics for flow over a flat
plate, flow across a tube banks and flow through a tube. Concept of boundary layer is
introduced to deal with the effect of viscosity on fluid flow and heat transfer. Analytical
formulations are done but you can recognize the complicacies associated with the
solutions. Approximate methods are introduced for quick solution as the expense of
accuracy. Turbulent flow concept is introduced. After studying this unit you will be able
to formulate and solve problems for convective heat transfer.

8.16 KEY WORDS

Laminar Flow : Tangents drawn at a flow lines are parallel.
Turbulent Flow : Tangents drawn at a flow lines are not parallel,
cross flows will be present.
Boundary Layer : Locus of all the points differentiating the viscous
and potential flow.
Hydrodynamic Boundary Layer : Related to flow parameters such as velocity.
Thermal Boundary Layer : Related to temperature of surface and fluid.
Hydraulic Diameter : It is an equivalent diameter when flow cross
section is other than circular.
Constant Heat Flux : Heat flux (heat flow per unit area normal to the
flow direction) being constant in spatial directions.
Internal Flow : Flow though a passage like tube or pipe.
External Flow : Flow over a body or surface.

8.17 ANSWERS TO SAQs

Please refer the relevant preceding text in this unit for answers to SAQs.

REFERENCES
H. Schlichiting, (1968), Boundary Layer Theory, 6th Edition, J. Kesitn, Trans,
McGraw-Hill, New York.
Frank Kreith and Mark S. Bohn, (1997), Principles of Heat Transfer, 5th, PWS
Publishing Co, Boston.
E. R. G. Eckert and R. M. Drake, (1959), Heat and Mass Transfer, McGraw-Hill,
Kogakusha.
J. P. Holman, (2002), Heat Transfer, 9th Edition, Tata McGraw Hill, New Delhi.
E. Achenbach, (1989), Heat Transfer from a Staggered Tube Bundle in Cross-flow at
High Reynolds Numbers, International Journal of Heat and Mass Transfer, Volume 32,
pp. 271-280.

107
Convection E. N. Sieder and C. E. Tate, (1936), Heat Transfer and Pressure Drop of Liquids in
Tubes, Ind. Eng. Chem., Volume 28, pp. 1429.
C. A. Depew and S. E. August, (1971), Heat Transfer due to Combined Free and Forced
Convection in a Horizontal and Isothermal Tube, Trans. ASME Ser., C, J. Heat Transfer,
Volume 93, pp. 380-384.
W. M. Kays and A. L. London, (1984), Compact Heat Exchangers, 3rd Edition,
McGraw-Hill, New York.
B. S. Petukhov, (1970), Heat Transfer and Friction in Turbulent Pipe Flow with Variable
Physical Properties, Advances in Heat Transfer, Academic Press, New York, pp. 504-
564.
H. Hausen, (1983), Heat Transfer in Counter Flow, Parallel Flow and Cross Flow,
McGraw-Hill, New York.
M. N. Ozisik, (1985), Heat Transfer A Basic Approach, McGraw-Hill International
Edition.
F. P. Incropera and D. P. Dewitt, (2004), Fundamentals of Heat and Mass Transfer,
5th Edition, John Wley and Sons.
P. K. Nag, (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.

CONVECTION
Objective of the present block is to discus about the Heat Transfer by Convection. Both
the units, i.e. Units 7 and 8 are devoted to discuss the physical and mathematical basis for
the understanding of convective transport and to reveal various heat transfer correlations.
In engineering applications, the pressure drop or drag force associated with the flow
inside ducts or over bodies is also of interest. Therefore, appropriate correlations are
presented to predict the pressure drop or drag force in flow.
The analysis of convection is complicated because the fluid motion is affected by
pressure drop, drag force and heat transfer. The literature of convective heat transfer is
overwhelming and ever growing. In recent years, with the availability of high speed,
large capacity digital computers, great advances have been made in the analysis of very
complicated heat transfer problems. However, a large number of simple engineering
problems can be handled with the use of standard heat transfer correlations. To achieve
this objective, basic concepts associated with the flow over a body, flow inside a duct and
turbulence are discussed. Effect of fluid viscosity is dealt in particular for flow and heat
transfer problems with the help of boundary layer theory.

108
Indira Gandhi MRW-002
National Open University
School of Engineering & Technology HEAT TRANSFER

BLOCK 4

Radiation
 Indira Gandhi
National Open University
MRW – 002
School of Engineering & Technology Heat Transfer

Block

4
RADIATION
UNIT 9
Radiation Principles 5

UNIT 10
Radiation Exchange 43
RADIATION

Preceeding blocks where devoted to heat transfer by conduction and convection.

In the present block, the third mode of heat transfer, I.e. radiation has been-
considered. Thermal radiation is that electromagnetic radiation emitted by a body
as a result of its temperature. Solution of radiation problems are very complicated
as large number of variables are associated with it. In dealing with such problems
you will notice that radiation is dependent on spatial variables (x,y,z), spectral
properties (λ) as well as polar (θ) and azimuthal angle (ϕ). Treatment becomes
still complicated if the intervening medium is participating in nature.

In Unit 9, we will consider the principles of radiation. Radiation intensity and

emissive power are discussed. Concept of blackbody and laws of radiation are
discussed. Radiative transfer rate depends on the surface properties. This unit will
give you more insight on thermal radiation.

Unit 10 discusses estimation of thermal radiation with the help of view factor.
View factor algebra is more elaborately given so that you can estimate the
radiative transfer for different geometries and orientations. An electrical analogy
method is provided. Based on this unit, you will be able to solve surface radiation
problems. Radiation with participating medium is more involved. Hence the same
is not considered in these units.
 Radiation Principles
UNIT 9 RADIATION PRINCIPLES
Structure
9.1 Introduction
Objectives
9.2 Radiation Intensity
9.3 Intensity Related to Emission
9.4 Relation to Irradiation
9.5 Relation to Radiosity
9.6 Blackbody Radiation
9.6.1 Laboratory Blackbody
9.6.2 Spectral Energy Distribution of Blackbody
9.7 Planck’s Law
9.8 Wien’s Displacement Law
9.9 Planck’s Law in Dimensionless Form
9.10 The Stefan-Boltzmann Law
9.11 Surface Emission
9.12 Surface Absorption, Reflection and Transmission
9.12.1 Absorptivity
9.12.2 Reflectivity
9.12.3 Transmissivity
9.12.4 Special Considerations
9.13 Kirchhoff’s Law
9.14 The Gray Surface
9.15 Summary
9.16 Key Words
9.17 Answers to SAQs

9.1 INTRODUCTION
Thermal radiation, commonly known as radiation heat transfer, is distinctly different
from conduction and convection. Both the conduction and convective heat transfer rates
are linearly proportional to the temperature differences. However, thermal radiation heat
transfer is proportional to the differences of the individual absolute temperatures of the
bodies each raised to the fourth power. Thus, it is evident that the importance of thermal
radiation becomes intensified at high absolute temperature levels. Consequently,
radiation contributes substantially in combustion applications such as fires, furnaces, IC
engines, gas turbines, nuclear reactors, etc.
Objectives
After studying this unit, you should be able to
 understand the mechanism of heat transfer by radiation,
 understand parameters such as intensity, emission,
 interprete a blackbody and its role as a standard for radiation comparison
with other bodies, and
 estimate radiative heat transfer by different laws.
5
Radiation
9.2 RADIATION INTENSITY
Radiation incident on a surface may come from different directions. Response of the
surface to this radiation depends on the direction. Such directional effects may be treated
by introducing the concept of radiation intensity.
Consider the emission in a particular direction from an elemental area dA1 (Figure9.1).
Emitted Radiation

n
dAn

dA
dω
1

Figure 9.1 : Emission of Radiation from a Differential Area dA1 into a Solid Angle d
Subtended by dAn at a Point on dA1

Because of the three dimensional nature of radiation, we will be using the spherical
coordinate system as shown in Figure 9.2 to locate the direction of the ray. This direction
may be specified in terms of the zenith angle  and azimuthal angle .
z

(r,,)

r



x

Figure 9.2 : The Spherical Coordinate System

A differentially small surface area dAn, through which this radiation passes, subtends a
solid angle d when viewed from a point on dA1. From Figure 9.3(a) we see that the
differential plane angle d is defined by a region between the rays of a circle to the
radius r of the circle. Similarly, from Figure 9.3(b) the differential solid angle d is
defined by a region between the rays of a sphere and is measured as the ratio of the
element of area dAn on the sphere to the square of the sphere’s radius.

dAn
Accordingly, d  . . . (9.1)
r2

The area dAn is normal to the (, ) direction, and as shown in Figure 9.4(a), it may be
represented as dAn  r 2 sin  d  d  for a spherical surface.

6
 r r Radiation Principles

dl

+ 

dl
d 
r
(a) dAn (b)
d 
r2 dAn
Figure 9.3 : Definition of (a) Plane and (b) Solid Angle

dAn = r2 sin θ d θ d  dA1

rd θ r
r sinθ
r sinθd
θ r

dAn
dθ dω ≡
dA1 r2

d

(a)
n

dA1 cos θ

dA
1
(b)
Figure 9.4 : (a) The Solid Angle Subtended by dAn at a Point on dA1 in the
Spherical Coordinate System and (b) The Projection of dA1 Normal to the Direction of Radiation

Accordingly, d   sin  d  d  . . . (9.2)

whereas the plane angle d has the unit of radians (rad), the unit of the solid angle is the
steradian (sr).
Now, consider the rate of emission from dA1 and received by the area dAn (Figure 9.1).
This quantity may be expressed in terms of spectral intensity I, e of the emitted radiation.
We formally define I, e as the rate at which radiant energy is emitted at wavelength  in
the (, ) direction, per unit area of the emitted surface normal to its direction, per unit
solid angle about this direction, and per unit wavelength interval d about . Note that
the area used to define the intensity is the component of dA1 perpendicular to the
direction of the radiation. From Figure 9.4(b), it is clear that the projected area is equal to
dA1 cos  . In effect it is how dA1 will appear to an observer situated on dAn. The spectral
intensity as per definition is 7
Radiation dq
I  , e (, , )  . . . (9.3)
dA1 cos  . d  . d 

 dq 
where    dq is the rate at which radiation of wavelength  leaves dA1 and passes
 d 
through dAn. Rearranging Eq. (9.3), it follows that
dq  I  , e ( , , ) dA1 cos  . d  . . . (9.4)

where dq has the units of W/m. This important expression allows us to compute the
rate at which radiation emitted by a surface propagates into the region of space defined
by the solid angle d about the (, ) direction. However, to compute this rate, the
spectral intensity I, e of the emitted radiation must be known. From Eq. (9.4) and
Eq. (9.2), we may associate the spectral radiation flux with dA1 as
dq  I  , e ( , , ) cos  sin  . d  . d  . . . (9.5)

If the spectral and directional distributions of I, e are known, i.e. I  , e (, , ) , the heat
flux associated with emission into finite solid angle or over any finite wavelength interval
may be determined by integrating Eq. 9.5. For example, spectral heat flux associated with
emission into hypothetical hemisphere above dA1 as shown in Figure 9.5 is

2 2
q ()    I  , e (, , ) cos  sin d  d  . . . (9.6)
0 0 n

0    /2

d dAn

dA1

d

02

Figure 9.5 : Emission from a Differential Element of Area dA1 into a Hypothetical
Hemisphere Centered at a Point on dA1

The solid angle associated with the entire hemisphere may be obtained by integrating

Eq. (9.2) over the limits  = 0 to  = 2 and  = 0 to   .
2
 
2 2 2
Hence,  d    sin  d  d   2  sin  d   2 sr . . . (9.7)
h 0 0 0

where the subscript h refers to integration over the hemisphere. The total heat flux
associated with emission in all directions and at all wavelengths is then

8
  Radiation Principles
q   q ( ) d  . . . (9.8)
0

9.3 INTENSITY RELATED TO EMISSION

The radiation intensity is related to several important radiation fluxes. It was mentioned
in Unit 1 that radiation occurs from any surface that is at a finite temperature. The
concept of emissive power is introduced to quantify the amount of radiation emitted per
unit surface area.
The spectral, hemispherical emissive power E (W/m 2 . μm) is defined as the rate at
which radiation of wavelength  is emitted in all directions from a surface per unit
wavelength d about  and per unit surface area. Accordingly, it is related to the spectral
intensity of emitted radiation by the expression

2 2
E ()    I  , e (, , ) cos  sin  d  d  . . . (9.9)
0 0

It may be noted that E is a flux based on the actual surface area, whereas I, e is based on
the projected area. The cos  appearing in this integrand is a consequence of this
difference.
The total, hemispherical emissive power, E (W/m2), is the rate at which radiation is
emitted per unit area at all possible wavelengths and in all possible directions.

Accordingly, E  E ( ) d  . . . (9.10)
0

or, from Eq. (9.9)


 2 2
E    I  , e ( ,  ,  ) cos  sin  d  d  d  . . . (9.11)
0 0 0

As emissivity indicates emission in all directions, the adjective hemispherical is

redundant and is often dropped. Hence, emissive power can be spectral or total emissive
power.
The directional distribution of surface emission varies according to the nature of the
surface. However, there is a special case that provides a reasonable approximation
applicable to many surfaces. We speak of a diffuse emitter as a surface for which the
intensity of emitted radiation is independent of direction, in which case
I  , e (, , )  I  , e () . Removing I, e from the integrand of Eq. (9.9) and performing
the integration, it follows that
E ()   I  , e () . . . (9.12)
Similarly, from Eq. (9.11)
E   Ie . . . (9.13)
where Ie is the total intensity of the emitted radiation.

9.4 RELATION TO IRRADIATION

In earlier sections we have discussed about the radiation emitted by a surface. Similar
concept may be applied to incident radiation (Figure 9.6).
Such radiation may originate from emission and reflection occurring at other surfaces and
will have spectral and directional distributions determined by the spectral intensity
9
Radiation I  , i (, , ) . This quantity is defined as the rate at which radiant energy of wavelength
 is incident from the (, ) direction, per unit area of the intercepting surface normal to
this direction, per unit solid angle about this direction, and per unit wavelength interval
d about . Accordingly,

2 2
G ()    I  , i (, , ) cos  sin  d  d  . . . (9.14)
0 0

where sin  d  d  is the unit solid angle. The cos  factor originates because G is a
flux based on the actual surface area, where I, i is defined in terms of the projected area.
If the total irradiation G (W/m2) represents the rate at which radiation is incident per unit
area from all directions at all wavelengths, it follows that

G  G ( ) d  . . . (9.15)
0

or, from Eq. (9.14)


 2 2
G    I  , i ( ,  ,  ) cos  sin  d  d  d  . . . (9.16)
0 0 0

Incident Radiation, I, i

n

d
dA1

Figure 9.6 : Directional Nature of Incident Radiation

If the incident radiation is diffuse, I, i is independent of (, ) and it follows that
G ()   I  , i () . . . (9.17)

and G   Ii . . . (9.18)

9.5 RELATION TO RADIOSITY

The last radiative flux of interest, termed radiosity accounts for all the radiant energy
leaving a surface. Since radiation includes the reflected portion of the irradiation, as well
as direct emission (Figure 9.7), the radiosity is generally different from the emissive
power.
The spectral radiosity J (W/m2.m) represents the rate at which radiation of wavelength
 leaving a unit area of the surface, per unit wavelength interval d about . Since it
accounts for radiation leaving in all directions, it is related to the intensity associated with
emission and reflection, I  , e  r (, , ) , by an expression of the form
10
  Radiation Principles
2 2
J  ( )    I  , e  r (, , ) cos  sin  d  d  . . . (9.19)
0 0 Radiosity

Emission
Irradiation
Reflected
Portion of
Irradiation

Figure 9.7 : Surface Radiosity

Hence the total radiosity J (W/m2) associated with the entire spectrum is

J   J  ( ) d  . . . (9.20)
0


 2 2
or J     I  , e  r ( ,  ,  ) cos  sin  d  d  d  . . . (9.21)
0 0 0

If the surface is both a diffuse reflector and diffuse emitter, I  , e  r is independent of

(, ). And it follows that
J  ( )   I  , e  r . . . (9.22)

and J   Ie  r . . . (9.23)

It may be noted that the radiation flux in case of radiosity is based on the actual surface
area while the intensity is based on the projected area.
SAQ 1
(a) What do you mean by radiation intensity?
(b) Explain the following terms :
(i) radiosity,
(ii) emission,
(iii) spectral intensity.

9.6 BLACKBODY RADIATION

The blackbody is an ideal surface having the following properties :
(a) A blackbody absorbs all the incident radiation, regardless of wavelength and
direction.
(b) For a prescribed temperature and wavelength, no surface can emit more
energy than a blackbody.
(c) Although the radiation emitted by a blackbody is a function of wavelength
and temperature, it is independent of direction. That is, the blackbody is a
diffuse emitter. 11
Radiation As the perfect emitter and absorber, the blackbody serves as the standard against which
radiative properties of actual surfaces may be compared.
9.6.1 Laboratory Blackbody
As shown in Figure 9.8, a blackbody may be made in the laboratory from a hollow
enclosure or cavity with a small hole, the walls being kept at uniform temperature. A ray
of radiant energy entering through the hole will be partially absorbed and reflected many
times before it is almost fully absorbed.
Hole Incident
Fourth Reflection Ray Second Reflection and
And Partial Absorption Partial Absorption

Isothermal
Enclosure

Third Reflection and

Partial Absorption

First Reflection and

Partial Absorption
Figure 9.8 : Schematic Diagram of a Blackbody Cavity

The energy emitted from the portion of the surface in the cavity is Eb. After one
reflection, this becomes Eb, after two reflections, 2Eb, etc. Let us imagine the
radiation leaving the hole of this cavity to be composed of rays which have been directly
radiated, reflected once, twice, etc. Then the energy emitted from the hole is

E   Eb   Eb   2  Eb  . . . . . . (9.24)

or, E   Eb (1     2  . . .)

1 1
  Eb   Eb  Eb . . . (9.25)
1 

The energy streaming out from the hole or cavity is black body radiation.
9.6.2 Spectral Energy Distribution of Blackbody
The characteristics of thermal radiation from a solid body that when it is dispersed by
being passed through a prism, a continuous spectrum is formed with an energy
distribution for a black body. The area under each constant temperature curve is the total
rate of energy emission per unit area given by

Eb   Eb d  . . . (9.26)
0

where Eb is the monochromatic emissive power of a blackbody.

SAQ 2
(a) What is a blackbody? What are the characteristics of a black body?

12 (b) What is monochromatic emissive power?

 Radiation Principles
9.7 PLANCK’S LAW
The spectral distribution of blackbody emission was first determined by Planck. It is of
the form

2hc02
I  , b (, T )  . . . (9.27)
  hc  
 5  exp  0   1
  kT  

where h  6.6256  10  34 J.s and k  1.3805  10  23 J/K are the universal Planck and
Boltzmann constants, respectively.

c0  2.998  108 m/s is the speed of light in vacuum.

T is the absolute temperature of the blackbody (K)
Since the blackbody is a diffuse emitter, it follows from Eq. (9.12) that its spectral
emissive power is of the form
C1
E , b (, T )   I  , b (, T )  . . . (9.28)
5 C  
  exp  2   1
  T  
h c0
where C1  2 h c02  3.742  108 W.μm4 /m2 , C2   1.439  10 4 μm.K . Eq. (9.28)
k
is known as Planck distribution as shown in the plot (Figure 9.9).
109
Visible Spectral Region
108
max T = 2898  m.K
107
Solar Radiation
Spectral Emissive Power Eb (W/m4  m)

106
5800 K
105
2000 K
104
1000 K
103

102
800 K
101
300 K
100

10 1
100 K
10 2
50 K
10 3

10 4
0.1 0.2 0.4 0.6 1 2 4 6 10 20 40 60 100

Wavelength ( m)
Figure 9.9 : Spectral Blackbody Emissive Power

Following are the important features of the Planck distribution :

(a) The emitted radiation varies continuously with wavelength.
(b) At any wavelength the magnitude of the emitted radiation increases with
increasing temperature.
(c) The spectral region in which the radiation is concentrated depends on
temperature, with comparatively more radiation appearing at shorter
wavelengths as the temperature increases.
13
Radiation (d) A significant fraction of radiation emitted by the sun, which may be
approximated as a blackbody at 5800 K, is in the visible region of the
spectrum. In contrast, for T < 800 K, emission is predominantly in the
infrared region of the spectrum and is not visible to the eye.

9.8 WIEN’S DISPLACEMENT LAW

It is observed from the Figure 9.8 that the blackbody spectral distribution has a maximum
and corresponding wavelength max depends on the temperature. Differentiating
Eq. (9.28) with respect to  and setting the result equal to zero, we obtain
max T  C3 . . . (9.29)

where C3 = 2897.8 m.K. Eq. (9.29) is known as Wien’s displacement law and the locus
of the points described by this law is plotted as the dashed line of Figure 9.8.
From Figure 9.8, it is observed that the maximum spectral emissive power is displaced to
shorter wavelengths with increasing temperature.

C1   5
Eb  |max  C2
. . . (9.30(a))
e T
 1

or, Eb  |max  C3 T 5 W/m 2 . . . (9.30(b))

This wavelength is in the middle of visble spectrum ( = 0.5 m) for solar radiation,
since the sun emits approximately as a blackbody at 5800 K. For a blackbody at 1000 K,
peak emission occurs at 2.9 m, with some of the emitted radiation appearing visible as
red light. With increasing temperature, shorter wavelengths become more prominent,
until eventually significant emission occurs over the entire visible spectrum. For example,
a tungsten filament lamp operating at 2900 K (max = 1 m) emits white light, although
most of the emission remains in the infrared region.

9.9 PLANCK’S LAW IN DIMENSIONLESS FORM

Planck’s law can be expressed in dimensionless form by using Eqs. (9.28) and (9.30(b)).
Eb  C1 1
Hence,  5
. C2
 f ( T ) . . . (9.31)
Eb  |max C3 ( T )
T
e 1
From Wien’s displacement law
 max T  C3  2897.8 . . . (9.32)

Eb    
Hence,   . . . (9.33)
Eb  |max  max 
Figure 9.10 is obtained by plotting Eqs. (9.31) and (9.33).

14
 10
8 Radiation Principles
6
4

3
2

0.1
8
6
Eb/(Eb)max

4
3
2

0.01
8
6

4
3
2

0.001
8 0.1 2 3 4 6 8 1.0 2 3 4

T, cm C 0

0.4 0.6 1.0 2 3 4 6 8 10

/max

Figure 9.10 : Graphical Representation of Planck’s law in Dimensionless Form

The maximum of this relationship corresponds to the values of
Eb 
1 . . . (9.34)
Eb  |max


and 1 . . . (9.35)
 max
 

 Eb d   Eb d 
0 0
f  
 . . . (9.36)
T 4
 Eb d 
0

This is also equal to the ratio of area A1 to the total area under the curve at T
(Figure 9.11).
Similarly, for the wavelength range between 1 and 2 and, the fraction of radiation at
temperature T, as shown in Figure 9.11, would be
2

 Eb d 
 2 1 
1 1 Area A2
f     E b d    E b d    . . . (9.37)

 T 4  0  Total Area


0
Eb d 
0

 E (0  T ) 
Table 9.1 gives the values of fractional area  b  for various values of T.
 T 4 
This table will help in computing the fraction of radiant energy falling in a certain
wavelength range.
15
Radiation

Area A1

Eb Eb

Area A2
T

0 d  1 2


Figure 9.11 : Fraction of Radiation in a Range of Wavelength Expressed as Area Ratio

Table. 9.1 : Blackbody Radiation Functions

Eb (0  T ) Eb (0  T )
 T ( mK  10 3 )  T ( mK  10 3 )
T 4 T 4

0.2 0.341796  10– 26 4.6 0.579316

0.4 0.186468  10 – 11 4.8 0.607597
0.6 0.929299  10– 7 5.0 0.633786
0.8 0.164351  10 –4 5.2 0.658011
1.0 0.320780  10– 3 5.4 0.680402
1.2 0.213431  10 –2 5.6 0.701090
1.4 0.779084  10– 2 5.8 0.720203
1.6 0.197204  10 –1 6.0 0.737864
1.8 0.393449  10– 1 6.2 0.754187
2.0 0.667347  10 –1 6.4 0.769282
2.2 0.100897 6.6 0.783248
2.4 0.140268 6.8 0.796180
2.6 0.183135 7.0 0.808160
2.8 0.227908 8.0 0.856344
3.0 0.273252 9.0 0.90090
3.2 0.318124 10.0 0.914263
3.4 0.361760 12.0 0.945167
3.6 0.403633 15.0 0.969056
3.8 0.443411 20.0 0.985683
4.0 0.480907 40.0 0.998057
4.2 0.516046 50.0 0.999045
4.4 0.548830 100.0 1.000000
16
 Radiation Principles
9.10 THE STEFAN-BOLTZMANN LAW
Substituting the Planck distribution Eq. (9.28) into Eq. (9.10), the total emissive power of
a blackbody Eb may be expressed as

C1
Eb    C  
d . . . (9.38)
5
0   exp  2   1
  T  
1
Substituting  1  x , or    2 d   dx , or d    dx , and when  = , x = 0 and
x2
 = 0, x = , we have
0
C1 x 5  1 
Eb   C2 x   2  dx
 x 

e T 1
1
  C2 x 
3
 C1  x  e T  1
 
dx
0  
 C2 x  2C2 x  3C2 x  4C2 x

 C1  x 3 (e T e T e T e T  . . .) dx
0

  C2 i x
3
 C1  x e T dx i = 2, 2, 3, . . . . . . (9.39)
0


n!
Since,  x n e  ax dx  , considering only first 3 terms,
0 a n 1

3! 6C1 T 4
Eb  C1 3 1

 C2 i  C24 i 4
 T 
 

6C1 T 4  1 1 1 
 4  4  4  4  . . .
C2  1 2 3 

6C1 T 4 4
 . . . . (9.40)
C24 90

Substituting the values of C1, C2 and , we get

6  3.74  10  6   4
Eb  2 4
. T 4  5.67  10 8 T 4   T 4 . . . (9.41)
(1.438  10 )  90

where   5.670  10  8 W/m 2 .K 4 is the Stefan-Boltzmann constant. Eq. (9.41) is

known as Stefan-Boltzmann law and it enables calculation of the amount of radiation
emitted in all directions and over all wavelengths simply from the knowledge of the
temperature of the blackbody. Because this emission is diffuse, it follows from Eq. (9.13)
that the total intensity associated with blackbody emission is
Eb
Ib  . . . (9.42)

Blackbody radiation in a certain range of wavelength.
17
Radiation The fraction of total radiation f from a blackbody in the wavelength range of 0 –  at a
certain temperature T is given by Eq. (9.36).
From Planck’s law

C1   5
Eb  C2
. . . (9.43)
e T 1

Eb C1 ( T )  5
or,  C2
. . . (9.44)
T5
e T
 1
Eb
Therefore, is a function of T.
T5

Again, Eb   Eb d    T 4 . . . (9.45)
0

 
Eb Eb
   T 5
d   T5
d ( T ) . . . (9.46)
0 0

Eb C1 (T )  5
Again,  C2
. . . (9.46(a))
T5
e T
 1

C1 (T )  5
   C2
d ( T ) . . . (9.47)
0
e T
 1
Eb
Figure 9.12 shows the plot of vs T. At T = 1 K, the curve represents Eb vs . The
T5
area under the curve is equal to the Stefan-Boltzmann constant.

Total Energy Found

0.2898 cm K Below T
Eb \ T5

Eb

T5

Total Area
=
at T = 1 K

0 d (T)
T (mK)

Figure 9.12 : Total Energy of Blackbody Radiation below T

18
The fractional energy in the range of 1T and 2T is given by Radiation Principles

2 2

 Eb d   Eb d 
1 1
f  
 . . . (9.48)
T 4
 Eb d 
0

 2T
1 Eb
or, f 
  T5
(T ) . . . (9.49)
1T

Which also represents the area ratio which can be read from the scale at the top of
Figure 9.11.
SAQ 3
(a) What is Stefan-Boltzmann law? How is Stefan-Boltzmann law derived from
Planck’s law of thermal radiation?
(b) Derive Wien’s displacement law from Planck’s equation.
(c) What is the value of Stefan-Boltzmann constant?

9.11 SURFACE EMISSION

In the preceding sections surface emission for the blackbody was considered. For any real
surface, emission will be never more than a blackbody at the same temperature. The term
emissivity is used to compare the emission characteristics of a real surface with respect to
a blackbody.
Emissivity is defined as ratio of the radiation emitted by the real surface to the radiation
emitted by a blackbody at the same temperature.
Spectral directional emissivity  ,  ( , , , T ) of a surface at temperature T is defined
as the ratio of the intensity of radiation emitted at the wavelength  and in the direction
(, ) to the intensity of the radiation emitted by a blackbody at the same values of T
and .
I  , e (, , , T )
Hence,  ,  ( , , , T )  . . . (9.50)
I  , b (, T )

Similarly, total directional emissivity  may be defined as

I e (, , T )
 (, , T )  . . . (9.51)
I b (T )
For most engineering calculations, it is desired to work with surface properties that
represent directional averages. A spectral, hemispherical emissivity is therefore
defined as
E (, T )
 (, T )  . . . (9.52)
E , b (, T )

It may be related to the directional emissivity ,  as follows :

19
Radiation 
2 2

  I  , e (, , , T ) cos  sin  d  d 

0 0
  ( , T )  
. . . (9.53)
2 2

  I  , b (, T ) cos  sin  d  d 

0 0

In constrast to Eq. (9.9), the temperature dependence of emission is now acknowledged.

From Eq. (9.50) and the fact that I, b is independent of  and , it follows that

2 2

   ,  ( , , , T ) cos  sin  d  d 
0 0
  ( , T )  
. . . (9.54)
2 2

  cos  sin  d  d 
0 0

Assuming ,  to be independent of , which is a reasonable assumption for most

surfaces and evaluating the denominator, we obtain

2
 ,  ( , T )  2   ,  ( , , T ) cos  sin  d  . . . (9.55)
0

The total, hemispherical emissivity, which represents an average over all possible
directions and wavelengths, is defined as
E (T )
 (T )  . . . (9.56)
Eb (T )
Emissivity depends strongly on the nature of the surface, which can be influenced by the
method of fabrication, thermal cycling and chemical reaction with the environment.
Some of the important points regarding emissivity are :
(a) The emissivity of metallic surface is generally small, achieving values as
low as 0.02 for highly polished gold and silver.
(b) The presence of oxide layers may significantly increase the emissivity of
metallic surfaces. Contrast the value of 0.10 for lightly oxidized stainless
steel with the value of approximately 0.50 for the heavily oxidized form.
(c) The emissivity of nonconductors is comparatively large, generally
exceeding 0.6.
(d) Emissivity of conductors increases with increase in temperature; however,
depending on the specific material, the emissivity of nonconductors may
either increase or decrease with increasing temperature.

9.12 SURFACE ABSORPTION, REFLECTION AND

TRANSMISSION

We defined the spectral radiation G (W/m2.m) as the rate at which radiation of

wavelength  is incident on a surface per unit area of the surface and per unit wavelength
interval d about . It may be incident from all possible directions, and it may originate
from several different sources. The total irradiation G (W/m2) encompasses all spectral
contributions and may be evaluated from Eq. (9.15). In this section we consider the
processes resulting from interception of this radiation by a solid or liquid medium.

20
In the most general situation the irradiation interacts with a semitransparent medium, Radiation Principles
such as a layer of water or a glass plate. Spectral component of the irradiation is shown in
Figure 9.13.
Irradiation
Reflection G
G,ref

G = G,abs + G.ref + G,tr

Semitransparent Absorption
Medium G,abs

Transmission
G,tr

Figure 9.13 : Absorption, Reflection and Transmission Processes

Associated with a Semitransparent Medium

Part of the radiation is reflected, absorbed and transmitted. From radiation balance on the
medium, it follows that
G  G , ref  G , abs  G , tr . . . (9.57)

In general, determination of these components is complex, depending on the upper and

lower surface conditions, the wavelength of the radiation, and the composition and
thickness of the medium. Moreover, conditions may be strongly influenced by volumetric
effects occurring within the medium.
In simpler situation pertaining to most engineering applications, the medium may be
opaque to the incident radiation. In this case, G, tr = 0 and the remaining absorption and
reflection processes may be treated as surface phenomena. That is, they are controlled by
processes occurring within a fraction of a micrometer from the irradiated surface. It is
therefore appropriate to speak of irradiation being absorbed and reflected by the surface,
with the relative magnitudes of G, abs and G, ref depending on  and the nature of the
surface material. There is no net effect of reflection process on the medium, while
absorption has the effect of increasing the internal thermal energy of the medium.
It is interesting to note that the surface absorption and reflection are responsible for our
perception for colour. Unless the surface is at a high temperature (Ts  1000 K), such that
it is incandescent, colour is no way due to emission, which is concentrated in the IR
region, hence imperceptible to the eyes. Colour is instead due to selective reflection and
absorption of the visible portion of the irradiation that is incident from the sun or an
artificial source of light. A shirt is ‘red’ because it contains a pigment that preferentially
absorbs the blue, green and yellow components of the incident light. Hence the relative
contributions of these components to the reflected light, which is seen, is diminished, and
the red component is dominant. Similarly, a leaf is green because its cells contain
chlorophyll, a pigment that shows strong absorption in the blue and red and preferentially
reflects green light. A surface appears ‘black’ if it absorbs all the incident visible
radiation, and it is ‘white’ if it reflects all the visible light. However, we must be careful
how we interpret such visual effects. For a prescribed irradiation, the ‘colour’ of a surface
may not indicate its overall capacity as an absorber or reflector, since much of the
irradiation may be in the IR region. A ‘white’ surface such as snow, for example, is
highly reflective to visible radiation but strongly absorbs IR radiation, thereby
approximating blackbody behaviour at long wavelengths.
21
Radiation 9.12.1 Absorptivity
The absorptivity is a property that determines the fraction of the irradiation absorbed by a
surface. Determination of absorptivity is complicated because of its spectral and
directional nature.
The spectral, directional absorptivity   ,  ( , , ) of a surface is defined as a fraction
of the spectral intensity in the direction (, ) that is absorbed by the surface. Hence
I  , iabs (, , )
  ,  ( , , )  . . . (9.58)
I  , i (, , )

It may be noted that the effect of surface temperature has a relatively small effect on the
spectral, directional absorptivity. Hence effect of surface temperature is not considered in
Eq. (9.58).
It is implicit in the foregoing result that surfaces may exhibit selective absorption with
respect to the wavelength and direction of the incident radiation. For most engineering
calculations, however, it is desirable to work with surface properties that represent
directional averages. We, therefore, define a spectral hemispherical absorptivity  () as
G , abs ()
  ( )  . . . (9.59)
G ()

From Eqs. (9.14) and (9.58), we can write


2 2

    ,  ( , , ) I , i ( , , ) cos  sin  d  d 
0 0
  ( )  
. . . (9.60)
2 2

  I  , i (, , ) cos  sin  d  d 

0 0

Hence,  depends on the directional distribution of the incident radiation, as well as on

the wavelength of the radiation and the nature of the absorbing surface. If the incident
radiation is diffusely distributed and ,  is independent of , Eq. (9.60) reduced to

2
  ( )  2    ,  (, ) cos  sin  d . . . (9.61)
0

The total, hemispherical emissivity, , represents an integrated average over both

direction and wavelength. It is defined as the fraction of the total irradiation absorbed by
a surface
Gabs
 . . . (9.62)
G
From Eqs. (9.15) and (9.59), it may be expressed as


   ( ) G ( ) d 
0
 
. . . (9.63)
 G () d 
0

Accordingly, depends on the spectral distribution of the incident radiation as well as on

its directional distribution and the nature of the absorbing surface.

22
9.12.2 Reflectivity Radiation Principles

The reflectivity is a property that determines the fraction of the incident radiation
reflected by a surface. Reflectivity is bidirectional in nature and because of this nature, its
specific definition may take several different forms. Bidirectional nature of reflectivity
indicates that it depends on the direction of incidence as well as the direction of reflection.
We shall avoid this complication by working exclusively with a reflectivity that
represents an integrated average over the hemisphere associated with the reflected
radiation and provides no information concerning the directional distribution of this
radiation.
Spectral, Directional Reflectivity  ,  ( , ,  )

Spectral, directional reflectivity  ,  ( , , ) of a surface is defined as the

fraction of the spectral intensity incident in the direction of (, ), which is
reflected by the surface.
I  , i , ref (, , )
Hence,  ,  ( , , )  . . . (9.64)
I  , i (, , )

Spectral, Hemispherical Reflectivity  ()

Spectral, hemispherical reflectivity  () is defined as the fraction of the spectral
irradiation that is reflected by the surface.
G , ref ()
Hence,   ( )  . . . (9.65)
G ()

Which is equivalent to

2 2

   ,  ( , , ) I , i ( , , ) cos  sin  d  d 
0 0
  ( )  
. . . (9.66)
2 2

  I  , i (, , ) cos  sin  d  d 

0 0

Total, Hemispherical Reflectivity 

The total, hemispherical reflectivity  is defined as
Gref
 . . . (9.67)
G
Surfaces may be idealized as diffuse or specular, according to the manner in which
they reflect radiation (Figure 9.14).
Incident
Ray
Reflected
Radiation of
Uniform intensity

(a)
23
Radiation 1 = 2

Reflected
Incident ray
ray

2
1

(b)
Figure 9.14 : (a) Diffuse and (b) Specular Reflection

Diffuse reflection occurs if, regardless of the direction of the incident radiation, the
intensity of the reflected radiation is independent of the reflection angle. In
contrast, if the angle of incident is equal to the angle of reflection, the reflection is
called specular reflection. No surface is perfectly diffuse or specular. Specular
condition is approximated for the mirror like polished surfaces. Rough surfaces are
diffuse type. Assumption of diffuse reflection is reasonable for most engineering
applications.
9.12.3 Transmissivity
Hemispherical transmissivity of a surface is defined as
G , tr ()
  . . . (9.68)
G ()

Gtr
and  . . . (9.69)
G
The total transmissivity  is related to the spectral component  by
 

 G , tr () d    G () d 
0 0
 
 
. . . (9.70)
 G () d   G () d 
0 0

SAQ 4
(a) What do you mean by surface property? How does radiation varies with
surface properties?
(b) Define the following :
(i) Absorptivity
(ii) Reflectivity
(iii) Transmissivity.
(c) How does surface property depend on wavelength of radiation?
9.12.4 Special Considerations
From radiation balance in a semitransparent medium, we get
       1 . . . (9.71)
For properties that are averaged over the entire spectrum, it also follows that
   1 . . . (9.72)
24
If the medium is opaque,     0 , hence, Radiation Principles

     1 . . . (9.73)
and    1 . . . (9.74)
Hence, knowledge of one property implies knowledge of the other.

9.13 KIRCHHOFF’S LAW

A small body of surface area A1 is placed in a hollow evacuated space kept at a constant
uniform temperature T as shown in Figure 9.15. After sometime the body at steady state
will attain the same temperature as that of the interior of the space and thereafter will
radiate as much energy as it receives.

TS
G
G = Eb (TS)
A2
A1 E2

A3
E3

E1

Figure 9.15 : Radiative Exchange in an Isothermal Enclosure

Let I = The radiant energy falling upon the body per unit time per unit surface area,
E1 = Total emissive power of the body, and
1 = Absorptivity of the body.
At steady state,
Energy absorbed = Energy emitted
or, I A1 1  E1 A1 . . . (9.75)
E1
or, I  . . . (9.76)
1
If this body is replaced by a second body of the same shape and surface area, but of
different material at exactly the same location at steady state
I A1  2  E2 A1 . . . (9.77)
E2
or, I  . . . (9.78)
B
Similarly, if the second body is replaced at the same location by a black body of the same
shape and surface area,
I A1 b  Eb A1 . . . (9.79)
Eb
I  . . . (9.80)
b

Since, I is the same on each body,

E1 E2 E
  b  Eb . . . (9.81)
1  2  b
25
Radiation (since b = 1)
E1 E
 1  , 2  2 . . . (9.82)
Eb Eb

Therefore, for any body

E
 . . . (9.83)
Eb

E
But the ratio is defined as emissivity.
Eb

Hence,  . . . (9.84)

This is Kirchhoff’s law which states that the emissivity of the surface of a body is equal
to its aborptivity when the body is in thermal equilibrium with its surroundings.

9.14 THE GRAY SURFACE

A gray body is one the monochromatic emissivity of which has the same value at all
wavelengths. For a gray body,  is independent of  so that
        . . . (9.85)

Eq. (9.85) holds good even though the temperatures of the incident radiation and the
receiving surface are not the same. For the purpose of heat transfer calculations, most real
surfaces are considered to be gray. In general, however, the emissivity of real surfaces
varies with wavelength. If the variation of monochromatic emissivity  with  is known,
the emissive power of real body can be found by plotting the products
 Eb vs  (Figures 9.16 (a)-(b)).

E

Black body,  = 1
1

Gray Surface,  = 1 Constant

Real Surface,  

T = Const.
0
0 
(a)

26
 E Black body, Eb Radiation Principles

E b
Gray Surface
E =  Eb

E

Eg

Gray Surface
E =  Eb
T = Const. (b)

Figure 9.16 : Variation (a)  with  and (b) E with  

An average emissivity of a real body can then be obtained from

 

E
 E d    Eb d 
0 0
  
 
. . . (9.86)
Eb
 Eb d   Eb d 
0 0

E E
since,   . If the real body is gray body, then the ratio  is constant for all
Eb Eb
wavelengths.
The toal average absorptivity can similarly be obtained from the distribution of
monochromatic absorptivity .


   Eb d  
Eb

0
Thus,  
 d ( T ) . . . (9.87)
T 5

0
Eb
0

If a plot of  with  is known,  can be obtained from the above equation.

SAQ 5
Explain Kirchhoff’s law. What do you mean by the statement “A perfect absorber
of radiant energy is also a perfect emitter”?
Example 9.1
A small surface area A1 = 10– 4 m2 is known to emit diffusely and from
measurements the total intensity associated with emission in the normal direction
is In = 7500 W/m2.sr.

27
 A3

Radiation

A2 A4

0.6 m

0.6 m

0.6 m

60 45

A1
Figure 9.17

Radiation emitted from the surface is intercepted by three other surfaces of area
A2  A3  A4  10 4 m 2 , which are 0.6 m from A1 and are oriented as shown in
Figure. What is the intensity associated with emission in each of the three
directions? What are the solid angles subtended by the three surfaces when viewed
from A1? What is the rate at which radiation emitted by A1 is intercepted by the
three surfaces?
Solution
Assumptions
(a) Surface A1 emits diffusely.
(b) A1, A2, A3 and A4 may be approximated as differential surfaces,
 Aj 
    1. A3
 r j2 
 
r3 = 0.6 m

A2 1n = 7500 w/m2 , sr A4

2 = 300

r4 = 0.6 m
A2 r2 = 0.6 m
A2, n = A2 Cos 2 1= 600 1= 450
n
2 = 30o

A1

2-1

Figure 9.18

Intensity of the emitted radiation, I = 7500 W/m2.sr for each of the three directions.
Treating A2, A3 and A4 as differential surface areas, the soild angles may be
computed as
dAn
d 
r2
where dAn is the projection of the surface normal to the direction of the radiation.
dAn, j  dA j  cos  j

where j is the angle between the surface normal and the direction of the radiation.
28
 The solid angle subtended by surface A2 with respect to A1 is Radiation Principles

A2 cos 2 10 4 m 2  cos 30o

2 1    2.4  10  4 sr
r2 (0.6 m)2

A3 cos 3 10 4 m 2  cos 0o

Similarly, 3 1    2.78  10  4 sr
r2 (0.6 m)2

A4 cos 4 10 4 m 2  cos 0o

4 1  2
 2
 2.78  10  4 sr
r (0.6 m)
Total radiation
q1  j  I  A1 cos 1   j 1

where 1 is the angle between the normal to surface 1 and the direction of the
radiation. Hence,
q1  2  7500 W/m 2 .sr  (10  4 m 2  cos 60 o )  2.4  10  4 sr  9  10  5 W

q1  3  7500 W/m 2 .sr  (10  4 m 2  cos 0 o )  2.78  10  4 sr  2.085  10  4 W

q1  4  7500 W/m 2 .sr  (10  4 m 2  cos 45o )  2.78  10  4 sr  1.47  10  5 W

Example 9.2
The spectral distribution of surface irradiation is as follows :

2000
G (W / m2.  m)

1500

1000

500

0
0 5 10 15 20 25 30

 ( m)
Figure 9.19

What is the total irradiation?

Solution
The total irradiation is evaluated by breaking into parts. That is
5m 25m 30m 
G  G d    G d    G d    G d 
0 5 m 25 m 30 m

1
or, G  (2000 W/m 2 .μm)  (5  0) m  (2000 W/m 2 .μm) (25  5) m
2
1
  (2000 W/m 2 .μm)  (30  25)  m  0
2

 (5000  40,000  5000) W/m 2

29
Radiation
or G  50,000 W/m 2
Example 9.3
Consider a large isothermal enclosure that is maintained at a uniform temperature
of 1800 K. Calculate the emissive power of the radiation that emerges from a small
aperture on the enclosed surface. What is the wavelength 1 below which 10% of
concentration of emission is concentrated? What is the wavelength 2 above which
10% of the radiation is concentrated? Determine the maximum spectral emissive
power and the wavelength at which this emission occurs. What is the irradiation
incident on a small object placed inside the enclosure?
Solution
(a) Emission from the aperture of any isothermal enclosure will have the
characteristics of blackbody radiation. Hence,

E  Eb (T )  T 4  5.67  10 8 W/m 2K 4  (1800 K ) 4

Hence, E  5.95  105 W/m 2

E = Eb (T)

G = Eb (T)

E, b (T) 10%

Small
Object 10%

Enclosure
T = 1800 K 1 max 2

Figure 9.20

(b) The wavelength 1 corresponds to the upper limit of the spectral band
(0  1) containing 10% of the emitted radiation. With F (0  1) and
1 T  2200 m .
2200
Hence, 1   1.22  m
1800
The wavelength 2 corresponds to the lower limit of the spectral band
(2  ) containing 10% of the emitted radiation.
Now F ( 2  )  1  F (0   2 )  0.1
or F (0   2 )  0.9
 2 T  9382 m.K . Hence  2  5.21 m
(c) From Wien’s law
 max T  2898 m.K
  max  1.61 m
The spectral emissive power may be computed by

I 1 b (1.61 m, T )  0.722  10  4 T 5

Hence, I 1 b (1.61 m, 1800 K )  0.722  10  4 (1/ μm.K.sr)

5
 5.67  10  8 W/m 2K 4 1800 K 
30
 Radiation Principles
= 0.773  10 5 W/m 2 .sr.μm
Since emission is diffuse

E , b   I 1 b  2.43  105 W/m 2 .μm

(d) Irradiation of any small object inside the enclosure may be approximated as
being equal to emission from a blackbody at the enclosure surface
temperature. Hence, G = Eb (T), in which case

G  5.95  105 W/m 2

Example 9.4
A surface emits as a blackbody at 2000 K. What is the rate per unit area (W/m2) at
which it emits radiation in directions corresponding to 0o    60o and in the
wavelength interval 3 m    5 m ?
Solution
The desired emission inferred from the following equation, with the limit of
integration restricted as follows:

5m 2  3
E     I  , b cos  sin  d  d  d 
3m 0 0

 = 600

Black Body at 2000 K

Figure 9.21

Since blackbody emits diffusely

  
5  2 3 
E   I , b 
0
 cos  sin  d  d   d 

3  0 
 
5
 0.75   I  , b d 
3

5
E , b
 0.75 Eb  Eb
d
3

 0.75 Eb [ F (0  5)  F (0  3)]
1 T  3 μm × 2000 K = 6000 μm.K : F (0  3)  0.737
 2 T  5 μm × 2000 K = 10,000 μm.K : F (0  5)  0.914 31
Radiation Hence, E  0.75 × (0.914  0.737)  E b = 0.13275 E b

 0.13275 × 5.67  10  8  (2000) 4 W/m 2

 1.2 × 105 W/m 2

Example 9.5
A diffuse surface at 2000 K has the spectral, hemispherical emissivity shown as
follows. Determine the total, hemispherical emissivity and the total emissive
power. At what wavelength will the spectral emissive power be a maximum?
1.0
2
0.8
T = 2000 k
1 = 3  m
2 = 5  m

1
0.4
 = ()

0 3 5
 = ( m)

Figure 9.22
Solution
(a) The total, hemispherical emissivity is given by
3 5


 E , b d 
1  E, b d  2  E, b d 

0 3
  
0
E Eb Eb

or   1 F(0  3m)  2 F(0  5m)  F(0  3m) 

From Table 9.1 we obtain

1 T  3 m  2000  6000 m.K : F(0 3 m)  0.737

 2 T  5 m  2000  10000 m.K : F(0 5 m)  0.914

Hence,   0.4  0.737  0.8 [0.914  0.737]  0.4364

(b) Total emissive power,
E   Eb   T 4

E  0.4364 (5.67  10  8 w/m 2.k 4 ) (2000 K) 4

E  396 kw/m 2
(c) From Wien’s law,
2898 μm.K
 max  = 1.449 μm
2000 K
The spectral emissive power at this wavelength may be obtained by,
E ( max , T )  E ( max ) E , b ( max , T )

or, since the surface is a diffusive emitter,

32
 E ( max , T )    ( max ) I  , b ( max , T ) Radiation Principles

I  , b ( max , T )
   ( max ) 5
 T 5
T

E (1.449 μm, 2000 K)    0.4  0.722  10  4 (1/μm.k.sr)

5.67  10  8 w/m 2 .K 4 × (2000 K) 5

= 164.62 kw/m2.m
Since  = 0.4 from  = 0 to  = 3 m, the foregoing result provides the
maximum spectral emissive power for region  < 3 m. However, with the
change in  that occurs at  = 3 m, the value of E at  = 3 m may be
larger than that for  = 1.449 m. To determine whether this is, in fact, the
case, we compute
I  , b (1 , T )
E  (1 , T )    (1 ) 5
 T 5
T
 I  , b (1 , T )  4 1
where, for 1 T  6000 μm.k,  5   0.27  10 (μmk.sr)
 T 

Hence, E (3 μm, 2000 K)    0.8  0.27  10  4 (1/μm.k.sr)

5.67  10  8 w/m 2 .K 4 × (2000 K) 5

E (3 μm, 2000 K)  123.122 kw/m 2  E (1.449, 2000 K)

and peak emission occurs at    max  1.449 m

Example 9.6
The spectral, hemispherical absorptivity of an opaque surface and the spectral
irradiation at the surface are as shown

500
1.0
G (w / m2. m)
T

0.2
0
0
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16

 = ( m)
 = ( m)
Figure 9.23

How does the spectral, hemispherical reflectivity vary with wavelength? What is
the total, hemispherical absorptivity of the surface? If the surface is initially at
450k and has a total hemispherical absorptivity of 0.85, how will its temperature
change upon exposure to the radiation?
Solution
It is assumed that the surface is opaque, negligible surface convection and back of
the surface is insulated.

33
Radiation E =  Eb G

Ts = 800 K,  = 0.85

Figure 9.24


Gabs
   G d 
0
Absorbitivity,    
G
 G d 
0
Subdividing the integral into parts,
6 8 16
0.2  G d   500   d   1.0  G d 
2 6 8
 6 12 16

 G d    G d    G d 
2 6 12

1  1 
 0.2   500 (6  2)  500 0.2 (8  6)  (1  0.2) 2 (8  6) 
2  
 1 
 1  500 (12  8)  1   500 (16  12) 
 2 
 1  1 
   500 (6  2)  500 (12  6)    500 (16  12)
 2  2 
Gabs (200  600  3000) w/m 2
Hence,  
G (1000  3000  1000) w/m 2
3800 w/m 2
  0.76
5000 w/m 2
Neglecting convection effects, the net heat flux to the surface is
  G  E  G   T 4
qnet

Hence,   0.76 (5000)  0.85  5.67  10  8 (450) 4

qnet
  3800  1976.3  1823.7 w/m 2
qnet
  0 , the surface temperature will increase with time.
Since, qnet
Example 9.7
A small, solid metallic sphere has an opaque, diffusive coating for which  = 0.8
for   5 m and  > 0.1 m for  > 5 m. The sphere, which is initially at a
uniform temperature of 350 k, is inserted into a large furnace whose walls are at
1300k. Determine the total, hemispherical absorptivity and emissivity of the
coating for the initial condition and for the final, steady-state condition.
34
Solution Radiation Principles

(a) Total hemispherical absorptivity is



   ( ) G ( ) d 
0
 

 G () d 
0

or, with G  E , b (T f )  E , b ( , 1300 K)

   ( ) E , b (, 1300 K) d 
0

Eb (1300 K)
1

 E , b (, 1300 K) d 
0
Hence,     ,1
Eb (1300 K)


 E , b (, 1300 K) d 
1
 , 2
Eb (1300 K)

or     ,1 F(0  1 )    , 2 [1  F0  1 ]
1.0
G  1
0.8



  2
0.1
0
Sphere of mass, m, area, 0 5
Furnace
Enclosure, Tf b As, temperature, Ts, and
Specific heat, G  = ( m)

Figure 9.25

From Table 9.1,

1 T f  5  1300  6500 m.k : F(0  1)  0.776

Hence,   0.8  0.776  0.1 (1  0.776)  0.647

The total, hemispherical emissivity ;



  E , b ( , Ts ) d 
0
 (Ts ) 
Eb (Ts )

Since the surface is diffuse,     and it follows that

1

 E , b (, 350 k) d 
0
    ,1
Eb (350 k)
35
Radiation 

 E , b (, 350 k) d 
1
 , 2
Eb (350 k)

or,     ,1 F(0  1 )    , 2 [1  F(0  1 ) ]

From Table 9.1,

1 Ts  5  350  1750 m.k : F(0  1)  0.024

Hence,   0.8  0.024  0.1 (1  0.024)  0.1168

(b) Because the spectral characteristics of the coating and the furnace
temperature remain fixed, there is no change in the value of  with
increasing time. However, as Ts increases with time, the value of  will
change. After a sufficiently long time, Ts = Tf and    (  0.647) .

Example 9.8
A flat-plate solar collector with no cover plate has a selective absorber surface of
emissivity 0.15 and solar absorptivity 0.9. At a given time of day the absorber
surface temperature Ts is 150oC when the solar irradiation is 800 W/m2, the
effective sky temperature is – 12oC, and the ambient air temperature T is 12oC.
Assume that the heat transfer convection coefficient for the calm day conditions
can be estimated from
1
h  0.22 (Ts  T ) 3 W/m 2.K

Calculate the useful heat removed rate (w/m2) from the collector for these
conditions. What is the corresponding efficiency of the collector?
Solution
(a) Performing an energy balance on the absorber,
Ein  Eout  0

Or, per unit surface area,

  E  qu  0
 s Gs   sky Gsky  qconv

4
Gsky   Tsky

Since the sky radiation is connected in approximately the same spectral

region as that of surface emission, it is reasonable to assume that
sky    0.15

4
  h (Ts  T )  0.22 (Ts  T
with qconv )3 and E    Ts4 it follows that
4
4
qu   s Gs    Tsky  0.22 (Ts  T  )3    Ts4

4
qu   s Gs  0.22 (Ts  T ) 3    (Ts 4  T sky
4
)

4
qu  0.9  800  0.22 (150  25) 3  0.15  5.67  10  8 (423 4  261 4)

36
 Radiation Principles
qu  (720  137.5  232.8)  349.7 w/m 2

Tsky = -120C

Gsky

Air
T = 250C

E = 0.15 h = 0.22 (Ts - T)1/3 w/m2K

s = 0.9
sGs skyGsk
qconvv E
y
Ts = 150 C
0

qu
Figure 9.26

(b) The collector efficiency, defined as the fraction of the solar irradiation
extracted as useful energy, is then
qu 349.7
   0.437
Gs 800

Exercise 9.1
(a) Consider a large isothermal enclosure that is maintained at a uniform
temperature of 1800 K. Calculate the emissive power of the radiation that
emerges from a small aperature on the enclosure surface. What is the
wavelength 1 below which 10% of the emission is concentrated? What is
the wavelength 2 above which 10% of the emission is concentrated?
Determine the maximum spectral emissive power and the wavelength at
which this emission occurs. What is the irradiation incident on a small
object placed inside the enclosure?
(b) Consider a small surface of area A1 = 10– 4 m2, which emits diffusely with a
total, hemispherical emissive power of E1 = 5  104 W/m2.

n2
2 = 300
n1 A2

1 = 600
r2 = 0.5 m

A1
Figure 9.27

(i) At what rate is this emission intercepted by a small surface of area

A2 = 10– 4 m2, which is oriented as shown?
(ii) What is the irradiation G2 on A2?
(iii) For zenith angles of 2 = 0.30 and 60o, plot G2 as a function of the
separation distance for 0.25  r2  1.0 m. 37
Radiation (c) Estimate the wavelength corresponding to maximum emission from each of
the following surfaces :
(i) the sun,
(ii) a tungsten filament at 2500 K,
(iii) a heated metal at 1500 K,
(iv) human skin at 305 K, and
(v) cryogenically cooled metal surface at 60 K.
Exercise 9.2
(a) An enclosure has an inside area of 100 m2, and its inside surface is black and
is maintained at a constant temperature. A small opening in the enclosure
has an area of 0.02 m2. The radiant power emitted from this opening is 70 W.
What is the temperature of the interior enclosure wall? If the interior surface
is maintained at this temperature, but is now polished, what will be the value
of the radiant power emitted from the opening?
(b) The spectral hemispherical emissivity of tungsten may be approximated by
the distribution depicted below. Consider a cylindrical tungsten filament that
is of diameter D = 0.8 mm and length L = 20 mm. The filament is enclosed
in an evacuated bulb and is heated by an electric current to a steady state
temperature of 2900 K.

0.6  = 0.45

0.4


0.2  = 0.10

0
0 2 4

 ( m)

Figure 9.28

(i) What is the total hemispherical emissivity when the filament

temperature is 2900 K?
(ii) Assuming the surroundings are at 300 K, what is the initial rate of
cooling of the filament when the current is switched off?
(iii) Generate a plot of the emissivity as a function of filament temperature
for 1300  T  2900 K
(iv) Estimate the time required for the filament to cool from 2900 K to
1300 K.
(c) A small opaque, diffuse object at Ts = 400 K is suspended in a large furnace
whose interior walls are at Tf = 2000 K. The walls are diffuse and gray and
have emissivity of 0.20. The spectral, hemispherical emissivity for the
surface of the small object is given below

38
 Radiation Principles
0.7

0.5



0
0 1 3

 ( m)

Figure 9.29
(i) Determine the total emissivity and absoptivity of the surface.
(ii) Evaluate the reflected radiant flux and the net radiative flux to the
surface.
(iii) What is the spectral emissive power at  = 2 m?
(iv) What is the wavelength 1/2 for which one-half of the total radiation
emitted by the surface is in the spectral region   1/2.
Exercise 9.3
(a) A large body of non-luminous gas at a temperature of 1200 K has emission
bands between 2.5 and 3.5 m and between 5 and 8 m. The effective
emissivity in the first band is 0.8 and in the second 0.6. Determine the
emissive power of this gas.
(b) A small disk 5 mm in diameter is positioned at the centre of an isothermal,
hemispherical enclosure. The disk is diffuse and gray with an emissivity of
0.7 and is maintained at 900 K. The hemispherical enclosure, maintained at
300 K has a radius of 100 mm and emissivity of 0.85. Calculate the radiant
power leaving an aperture of diameter 2 mm located on the enclosure as
shown.
(c) The spectral transmissivity of plain and tinted glass can be approximated as
follows :
Plain glass  = 0.9 0.3    2.5 m
Tinted glass  = 0.9 0.5    1.5 m
Outside the specified wavelength ranges, the spectral transmissivity is zero
for both the glasses. Compare the solar energy that could be transmitted
through the glasses. With solar irradiation on the glasses, compare the
visible radiant energy that could be transmitted.

9.15 SUMMARY
In the present unit concept of thermal radiation has been given. Radiation intensity and
emission, irradiation, etc. are defined. Radiation behaviour of a real surface depends on
properties of the surface. Comparison of radiation emitted by a real surface to a standard
surface is given elaborately. Blackbody concept is given for this purpose. Surface
properties on radiative behaviour are presented. Many a times such properties are
wavelength dependent. Hence, spectral properties are given in appropriate places. Laws
governing radiation are also discussed.

39
Radiation
9.16 KEY WORDS
Spectral : Wavelength dependent.
Intensity of Radiation : A concept to treat the directional effects of surface
radiation.
Irradiation : Radiations from all directions incident on a surface.
Specular : Angle of incidence is equal to angle of reflection.
Diffuse : Incident radiation is distributed uniformly after
reflection.
Blackbody : Ideal surface, perfect emitter and absorber.
Absorptivity : Absorption property of a surface/body.

9.17 ANSWERS TO SAQs

Refer the preceding text for all the Answers to SAQs.

40
 Radiation Principles
REFERENCES
J. P. Holman (2002), Heat Transfer, 9th Edition, Tata McGraw-Hill, New Delhi.
M. N. Ozisik (1985), Heat Transfer – A Basic Approach, McGraw-Hill International
Edition.
F. P. Incropera and D. P. Dewitt (2004), Fundamentals of Heat and Mass Transfer,
5th Edition, John Wiley and Sons.
P. K. Nag (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.

41
 Radiation Exchange
UNIT 10 RADIATION EXCHANGE
Structure
10.1 Introduction
Objectives
10.2 The View Factor
10.2.1 View Factor Integral
10.2.2 View Factor Relations
10.3 Blackbody Radiation Exchange
10.4 Radiation Exchange between Diffuse, Gray Surfaces in an Enclosure
10.5 Net Radiation Exchange at a Surface
10.6 Radiation Exchange between Surfaces
10.7 The Two Surface Enclosure
10.8 Radiation Shield
10.9 The Reradiating Surface
10.10 Multimode Heat Transfer
10.11 Crossed String Method
10.12 Summary
10.13 Key Words
10.14 Answers to SAQs

10.1 INTRODUCTION
In this unit we will consider the problem of radiative exchange between two or more
surfaces. This exchange depends strongly on the surface geometries and orientations, as
well as on their radiative properties and temperatures. We will assume that the surfaces
are separated by a nonparticipating medium, since such a medium does not emit, absorb
or scatter the radiation. It has no effects on the transfer of radiation between surfaces. A
vacuum meets these requirements exactly, and most gases meet them to an excellent
approximation.
Objectives
After studying this unit, you should be able to
 understand method of estimation of radiative heat transfer between surfaces,
 apply the concept of view factor to find the radiative heat transfer between
different geometric surfaces,
 apply view factor algebra for calculation of radiative heat transfer,
 utilise net radiation method for radiative exchange between surfaces, and
 understand the concept of radiation shield and reduction of radiation mode
of heat transfer in some applications.

10.2 THE VIEW FACTOR

To compute radiation exchange between any two surfaces, we must first introduce the
concept of a view factor (also called a configuration or shape factor).
43
Radiation 10.2.1 The View Factor Integral
The view factor Fij is defined as the fraction of the radiation leaving surface i that is
intercepted by surface j. To develop a general expression for Fij, we consider the
arbitrarily oriented surfaces Ai and Aj of Figure 10.1.
dAj

nj
j

AjTj dAj cos j

R

ni
ni dwj
i

dAi

dAi

AiTi
Figure 10.1 : View Factor Associated with Radiation Exchange between
Elemental Surfaces of Area dAi and dAj
Elemental areas on each surface, dAi and dAj are connected by a line of length R, which
forms the polar angles i and j, respectively, with the surface normals ni and nj. The
values of R, i and j, vary with the position of the elemental areas on Ai and Aj.
From the definition of the radiation intensity, and Eq. (9.4), the rate at which radiation
leaves dAi and is intercepted by dAj may be expressed as :
dqi  j  I i cos i d Ai d  j  i . . . (10.1)
where Ii is the intensity of the radiation leaving surface i and d  j  i is the solid angles
(cos  j d A j )
subtended by dAj when viewed from dAi. With d  j  i  from Eq. (9.1), it
R2
follows that
cos i cos  j
dqi  j  I i dAi dA j . . . (10.2)
R2
Assuming that surface i emits and reflects diffusely and substituting from Eq. (9.23), we
then obtain
cos i cos  j
dqi  j  J i dAi dA j . . . (10.3)
 R2
The total rate at which radiation leaves surface i and is intercepted by j may then be
obtained by integrating over the two surfaces. That is,
cos i cos  j
qi  j  J i    R2
dAi dA j . . . (10.4)
Ai Aj

where it is assumed that the radiosity Ji is uniform over the surface Ai. From the
definition of the view factor as the fraction of the radiation that leaves Ai and is
intercepted by Aj,
qi  j
Fij  . . . (10.5)
Ai J i
44
It follows that, Radiation Exchange

1 cos i cos  j
Fij 
Ai    R2
dAi dA j . . . (10.6)
Ai Aj

Similarly, the view factor Fji is defined as the fraction of the radiation that leaves Aj and
is intercepted by Ai. The same development then yields
1 cos i cos  j
F ji 
Aj    R2
dAi dA j . . . (10.7)
Ai Aj

Either Eq. (10.6) or Eq. (10.7) may be used to determine the view factor associated with
any two surfaces that are diffuse emitters and reflectors and have uniform radiosity.
10.2.2 View Factor Relations
An important view factor relation is suggested by Eqs. (10.6) and (10.7). In particular,
equating the integrals appearing in these equations, it follows that
Ai Fij  A j F ji . . . (10.8)

This expression, termed the reciprocity relation, is useful in determining one view factor
from knowledge of the other.
TN

JN
T1 J1
T1

J1

J2

T2
Figure 10.2 : Radiation Exchange in an Enclosure

Another important view factor relation to the surfaces of an enclosure as shown in

Figure 10.2. From the definition of the view factor, the summation rule may be applied to
each of the N surfaces in the enclosure. The rule follows from the conservation
requirements that all radiation leaving surface i must be intercepted by the enclosure
surfaces. The term Fij appearing in this summation represents the fraction of the radiation
that leaves surface i and is directly intercepted by i. If the surface is concave, it sees
itself and Fij is nonzero. However, for a plane or convex surface, Fij = 0.
N
 Fij  1 . . . (10.9)
j 1

To calculate radiation exchange in an enclosure of N surfaces, a total of view factors are

needed. This requirement becomes evident when the view factors are arranged in the
matrix form.

 F11  F1 N 
 
     . . . (10.10)
F 
 N 1  FNN 
45
Radiation However, all the view factors need not be calculated directly. A total of N view factor
may be obtained from the N equations associated with application of the summation rule,
N ( N  1)
Eq. (10.9), to each of the surface in the enclosure. In addition, view factors
2
N ( N  1)
may be obtained from applications of the reciprocity relation, Eq. (10.8),
2
 N 2  N  N ( N  1)  N ( N  1)
which are possible for the enclosure. Accordingly, only   
 2  2
view factors need be determined directly. For example, in a three-surface enclosure this
3 (3  1)
required correspond to only  3 view factors. The remaining six view factors
2
may be obtained by solving the six equations that results from use of Eqs. (10.8)
and (10.9).
Table 10.1 : View Factors for Two-dimensional Geometries
Geometry Relation

Parallel Plates with Midlines

Connected by Perpendicular

W1
1 1
i 2 2
[(Wi  W j )  4]  [(W j  Wi )  4] 2
2
Fij 
2Wi
L
w w
Wi  , Wi 
L L

W2

Inclined Parallel Plates of Equal j

Width and a Common Edge

W

Fij  1  sin  
2


j
Perpendicular Plates with
a Common Edge

1
 2 2
 w  2
1     1    
 wi    wi  
Fij 
2

W1

46
 Radiation Exchange
Three Sided Enclosure

Wj
Wk wi  w j  wk
j Fij 
2 wi
k

Wi

 1
1  2 2
Fij    [C  ( R  1) ] 2
2 
Parallel Cylinders of
Different Radil
1
2 2
 [C  ( R  1) ] 2
rj
ri  1  R   1  
 ( R  1) cos      
 C   C  
j
i  R   1   
 ( R  1) cos 1      
 C   C   

r s
R ,S 
ri ri

C 1 R  S

r
Cylinder and Parallel Rectangle

j
L
r  1 S1 s 
Fij   tan  tan 1 2 
s1  s2 L L

S2

S1

Infinite Plane and Raw of Cylinders

s D 1
 D 2 2
 
Fij  1  1    
  s  
j
1
D 1  s2  D2  2
   tan  
 s  2 
i  D 

47
Radiation To illustrate the foregoing procedure, consider a simple, two-surface enclosure involving
the spherical surfaces as shown in Figure 10.3.

F12 = 1, F11 = 0
A A
F12 = 1 , F22 = 1  1
A2 A2
J1
J2

Figure 10.3 : View Factors for the Enclosure Formed by Two Spheres

Although the enclosure is characterized by N2 = 4 view factors (F11, F12, F21, F22),
( N  1)
N  1 only view factor need be determined directly.
2
In this case such a determination may be made by inspection. In particular, since all
radiation leaving the inner surface must reach the outer surface, it follows that F12 = 1.
The same may not be said of radiation leaving the outer surface, since this surface sees
itself. However, from the reciprocity relation, Eq. (10.8), we obtain

A  A 
F21   1  F12   1  . . . (10.11)
 A2   A2 

From the summation rule, we obtain

F11  F12  1 . . . (10.12)

in which F11 = 0, and

F21  F22  1 . . . (10.13)

A 
in which case F22  1   1  . . . (10.14)
 A2 

For more complicated geometries, the view factor may be determined by solving the
double integral of Eq. (10.6). Such solutions have been obtained for many different
surface arrangements and are available in equation, graphical and tabular form. Results
for several common geometries are presented in Tables 10.1 and 10.2 and Figures 10.4
to 10.6.
The configurations of Table 10.1 are assumed to be infinitely long (in a direction
perpendicular to the page) and are hence two-dimensional. The configuration of
Table 10.2 and Figures10.4 to 10.6 is three-dimensional. It is useful to note that the
results of Figures 10.4 to 10.6 may be used to determine other view factors. For example,
the view factor for an end surface of a cylinder (or a truncated cone) relative to the lateral
surface may be obtained by using the results of Figures 10.4 and 10.6 may be used to
obtain other useful results if two additional view factor relations are developed.

48
 Table 10.2 Radiation Exchange

Geometry Relation

Aligned Parallel Rectangles X Y

X  ,Y 
L L

 1
2   (1  X 2 ) (1  Y 2 )  2

Fij  ln  
XY   1  X 2  Y 2 

i
L

j 1
Y X
2 1
X  X (1  Y ) 2 tan
1
(1  Y 2 ) 2

1
2 1 Y
 Y (1  X ) 2 tan
1
(1  X 2 ) 2

 X tan 1 X  Y tan 1 Y 
Coaxial Parallel Disks r r
Ri  . R j 
L L

rj
j 1  R12
S 1
ri L R12

i
 1
  2 2 
1 r
Fij  S  S 2  4    
2    ri   

 
 

Perpendicular Rectangles with a Common Edge Z Y

H  ,W 
X X

1  1 1  H tan 1 1 
Fij  W tan 
W  W H

1
2 2 1 1
 (H  W ) 2 tan
1
Z
J (H 2  W 2) 2

Y X 1


1  (1  W 2 ) (1  H 2 )  W 2 (1  W 2  H 2 )  2
 ln   
4  1  W2  H2  (1  W 2 ) (W 2  H 2 ) 


1
 H 2 (1  H 2  W 2 )  2 
  
 (1  H 2 ) ( H 2  H 2 )  


49
 1.0
Radiation 
4
0.7 10
i 2
0.5 L 1.0
0.4 j
Y 0.6
0.3 X
0.4
0.2

Fij
0.2
0.1

0.07
0.05
0.04 Y/ L = 0.1

0.03

0.02

0.01
0.1 0.2 0.3 0.5 1.0 2 3 4 5 10 20
X/L
Figure 10.4 : View Factor for Aligned Parallel Rectangles
1.0
8 rj
i
6 ri L
0.8
5 j

4
1.5
0.6 3
1.25
FiJ

Rj / L
=2 1.0
0.4
0.8

0.2
0.6 0.4

0.3
0
0.1 0.2 0.4 0.6 0.8 1 2 4 6 8 10
Z/ri
Figure 10.5 : View Factor For Parallel Flat Circular Plates

0.5 Y/X = Z J
0.02 I
Y
X
0.05
0.4
0.1

0.2
03
Fn

0.4

0.6
0.2
1.0

1.5
2.0 4
0.1
10

0 20
0.1 0.2 0.4 0.6 0.8 1 2 4 6 8 10
Z/X
Figure 10.6 : View Factor for Perpendicular Rectangles with a Common Edge
50
The first relation concerns the additive nature of view factors for a subdivided surface Radiation Exchange
and may be inferred from Figure 10.7.

An
Aj
Ak

A1

n
Aj   Ak
k 1

Figure 10.7 : Method for Finding View Factor by Dividing Areas

Considering radiation from surface i to surface j, which is divided into n components, it

is evident that
n
Fi ( j )   Fik . . . (10.15)
k 1

where the parentheses around a subscript indicate that it is a composite surface, in which
case (j) is equivalent to (1, 2, 3, . . . , k, . . . , n). This expression simply states that
radiation reaching a composite surface is the sum of the radiation reaching its parts.
Although it pertains to subdivision of the receiving surface, it may also be used to obtain
the second view factor relation, which pertains to subdivision of the originating surface.
Multiplying Eq. (10.15) by Ai and applying the reciprocity relation, Eq. (10.8), to each of
the resulting terms, it follows that
n
A j F( j ) i   Ak Fki . . . (10.16)
k 1

n
 Ak Fki
k 1
F( j ) i  n
. . . (10.17)
 Ak
k 1

Eqs. (10.16) and (10.17) may be applied when the originating surface is composed of
several parts.
SAQ 1
(a) What is view factor?
1 cos 1 cos 2
(b) Show that A1 F12 
   r2
dA1 dA2 .
A1 A2

(c) Explain how the shape factor is determined by decomposing one or both the
surfaces into subdivisions.
(d) What is the shape factor with respect to itself if the surface is concave,
convex or flat?

51
Radiation
10.3 BLACKBODY RADIATION EXCHANGE
In general, radiation may leave a surface due to both reflection and emission and on
reaching a second surface, experience reflection as well as absorption. However, matters
are simplified for surfaces that may be approximated as blackbodies, since there is no
reflection. Hence energy only leaves as a result of emission and since radiation is
absorbed.
Consider radiation exchange between two black surfaces of arbitrary shape as shown in
Figure 10.8.

nj
ni

Ji = Ebi Jj = Ebj

Aj Tj

Ai Ti

Figure 10.8 : Radiation Transfer Between Two Surfaces that may be Approximate as Blackbodies

Defining qi  j as the rate at which radiation leaves surface i and is intercepted by

surface j, it follows that
qi  j  ( Ai J i ) Fij . . . (10.18)

or, since radiosity equals emissive power for a black surface ( J i  Ebi ) ,

qi  j  Ai Fij Ebi . . . (10.19)

Similarly, q j  i  A j F ji Ebj . . . (10.20)

The net radiative exchange between two surfaces may then be defined as
qij  qi  j  q j  i . . . (10.21)

From which it follows that

qij  Ai Fij Ebi  A j F ji Ebj . . . (10.22)

or qij  Ai Fij  (Ti 4  T j4 ) . . . (10.23)

Eq. (10.23) provides the net rate at which radiation leaves surface i as a result of its
interaction with j, which is equal to the net rate at which j gains radiation due to its
interaction with i.
The foregoing result may also be used to evaluate the net radiation transfer from any
surface in an enclosure of black surfaces. With N surfaces maintained at different
temperatures, the net transfer of radiation from surface i is due to exchange with the
remaining surface and may be expressed as
N
qi   Ai Fij  (Ti 4  T j4 ) . . . (10.24)
ji

52
SAQ 2 Radiation Exchange

Show that the hemispherical black cavity with flat cover over it emits 50% of
radiation to the surface itself and is absorbed.

10.4 RADIATION EXCHANGE BETWEEN DIFFUSE, GRAY,

SURFACES IN AN ENCLOSURE
Although useful to a point, the foregoing results are limited by the assumption of
blackbody behaviour. The blackbody is of course, an idealized, which, although closely
approximated by some surfaces, is never precisely achieved. A major complication
associated with radiation exchange between non-black surface is due to surface reflection.
In an enclosure, such as that of Figure 10.9, radiation may experience multiple reflections
off all surfaces, with partial absorption occurring at each.

T1A1E1
TjAjEj
Q1 G1
J1

TiAiEi
Figure 10.9 : Radiation Exchange in an Enclosure

Analysing radiation exchange in an enclosure may be simplified by making certain

assumptions. Each surface of enclosure is assumed to be isothermal and to be
characterised by a uniform radiosiy and irradiation. Opaque, diffuse, gray surface
behaviour is also assumed, and the medium within the enclosure is taken to be
nonparticipating. The problem is generally one in which the temperature Ti associated
with each of the surface is known, and the objective is to determine the net radiative heat
flux qi from each surface.

10.5 NET RADIATION EXCHANGE AT A SURFACE

The term qi , which is the net rate at which radiation leaves surfaces i , represents the net
effect of radiation interactions occurring at the surface is as shown in Figure 10.10.
Gi Ai
Ji Ai

qi
Figure 10.10 : Radiative Balance as per Eq. (10.25)

It is the rate at which energy would have to be transferred to the surface by other means
to maintain it at a constant temperature. It is equal to the difference between the surface
radiosity and irradiation and may be expressed as
53
Radiation qi  Ai ( J i  Gi ) . . . (10.25)
From Figure 10.11 and the definition of the radiosity Ji,
J i  Ei  i Gi . . . (10.26)
Gi Ai

i Gi Ai

Ei Ai

i Gi Ai

qi
Figure 10.11 : Radiative Balance according to Eq. (10.27)

It is evident that the net radiative transfer from the surface may also be expressed in
terms of the surface emissive power and the absorbed irradiation
qi  Ai ( Ei  i Gi ) . . . (10.27)

Substituting from Eq. (9.52) and recognizing that i  1  i  1  i for an opaque,

diffuse, gray surface, the radiosity may also be expressed as
J i  i Ebi  (1  i ) Gi . . . (10.28)

Solving for Gi and substituting into Eq. (10.25), it follows that

 J  i Ebi 
qi  Ai  J i  i  . . . (10.29)
 1  i 

Ebi  J i
or qi  . . . (10.30)
(1  i )
i Ai
Eq. (10.30) provides a convenient representation for the net radiative heat transfer rate
from a surface. This transfer, which may be represented by the network element of
Figure 10.12, is associated with the driving potential (Ebi – Ji) and a surface radiative
(1  i )
resistance of the form .
i Ai
Ji

1 - i
i A i

Ebi

q1
Figure 10.12 : Network Element Representing the Net Radiation Transfer from a Surface

Hence, if the emissive power that the surface would have if it were black exceeds its
radiosity, there is net radiation heat transfer from the surface; if the inverse is true, the net
transfer is to the surface.
54
SAQ 3 Radiation Exchange

(a) Explain the electrical analogy for radiaitve heat transfer in a black enclosure.
(b) Draw the equivalent electrical network for radiative flux between four walls
of a black body.

10.6 RADIATION EXCHANGE BETWEEN SURFACES

To use Eq. (10.30), the surface radiosity Ji must be known. To determine this quality; it is
necessary to consider radiation exchange between the surfaces of the enclosure.
The irradiation of surface i can be evaluated from the radiosities of all the surface in the
enclosure. In particular, from the definition of the view factor, it follows that the total rate
at which radiation reaches surface i from all surfaces including i, is
N
Ai Gi   F ji A j J j . . . (10.31)
j 1

or, from the reciprocity relation, Eq. (10.8),

N
Ai Gi   Ai Fij J j . . . (10.32)
j 1

Canceling the area Ai and substituting into Eq. (10.25) for Gi,
 N 
qi  Ai  J i 
  Fij J j 

. . . (10.33)
 j 1 
or, from the summation rule, Eq. (10.9),
 N N 
qi  Ai   Fij J i   Fij J j  . . . (10.34)
 j 1 
 j 1 
N N
Hence, qi   Aij Fij (J i  J j )   qij . . . (10.35)
j 1 j 1

This result equates the net rate of radiation transfer from surface i, qi, to the sum of
components qij related to radiative exchange with the other surfaces. Each components
may be represented by a network element for which (Ji – Jj) (Ai Fij– 1) is a space or
geometrical resistance (Figure 10.13).
Combining Eqs. (10.30) and (10.35), we then obtain

Ebi  J i N Ji  J j
 . . . (10.36)
(1  i ) j  1 ( A F ) 1
i ij
i Ai

As shown in Figure 10.13 this expression represents a radiation balance for the radiosity
node associated with surface i. The rate of radiation transfer (current flow) to i through its
surface resistance must equal the net rate of radiation transfer (current flow) from i to all
other surfaces through the corresponding geometrical resistances.

55
 qi1

Radiation Jl
qi2
J2

qi3
(Ai Fi1)-1 J3

(Ai Fi 2)-1

(Ai Fi 3)-1

Ebi Ji
JN-1
qi qi(N-1)
1 - i (Ai Fi (N-1))-1
i A i

Node Corresponding (Ai FiN)-1

To the Surface I
JN
qiN

Figure 10.13 : Network Representation of Radiative Exchange between Surface i and

the Remaining Surfaces of an Enclosure

Note that Eq. (10.36) is especially useful when the surface temperature Ti (hence, Ebi) is
known. Although this situation is typical, it does not always apply. In particular,
situations may arise for which the net radiation transfer rate at the surface qi, rather than
the temperature Ti, is known. In such cases the preferred form of the radiation balance is
Eq. (10.35), rearranged as
N Ji  J j
qi   ( Ai Fij ) 1
. . . (10.37)
j 1

Use of network representation to solve enclosure radiation problems was first suggested
by Oppenheim. The method provides a useful tool for visualizing radiation exchange in
the enclosure and at least for simple enclosures may be used as the basis for predicting
this exchange. However, a more direct approach simply involves working with
Eqs. (10.36) and (10.37). Eq. (10.36) is written for each surface at which Ti is known, and
Eq. (10.37) is written for each surface at which qi is known. The resulting set of N linear,
algebraic equations is solved for the N unknowns, J1, J2, . . . , JN. With knowledge of the
Ji, Eq. (10.34) may be used to determine the net radiation heat transfer rate qi at each
surface of known Ti or the value of Ti at each surface of known qi.
For any number N of surface in the enclosure, the foregoing problem may readily be
solved by iteration or matrix inversion. For each of the N surfaces Eq. (10.36) or (10.37)
may be rearranged to obtain the following system of N equations :

 a11 J1  . . .  a1i J i  . . . a1 N J N  C1 
 
 ...................... ................... .................... 
 ...................... ................... .................... 
 
 ai1 J1  . . .  aii J I  . . . aiN J N  Ci  . . . (10.38)
 ...................... ................... ..................... 
 
 ...................... ......... ........... ..................... 
 a J  ... a Ni J i  . . . a NN J N  C N 
 N1 1
where the coefficients aij and Ci are known qualities. In matrix form these equations may
be expressed as

 A  J   C  . . . (10.39)

56
 Radiation Exchange
 a11 . . . a1i . . . a1 N   J1   C1 
     
............................  .  . 
............................  .  . 
     
where  A  ai1 . . . aii . . . aiN  J   Ji   C    Ci  . . . (10.40)
............................  .  . 
     
............................  .  . 
a . . . a . . . a    C 
 N1 Ni NN  JN   N
Expressing the unknown radiosities as
 J1  b11 C1  . . .  b1i C i  . . .  b1 N C N 
 
.................................................................. 
.................................................................. 
 
 J i  bi1 C1  . . .  bii Ci  . . .  biN C N  . . . (10.41)
.................................................................. 
 
.................................................................. 
J  b C  . . .  b C  . . .  b C 
 N N1 1 Ni i NN N 

They may be found by obtaining the inverse of [A], [A]– 1 such that

 J    A1 C  . . . (10.42)

b11 . . . b1i . . . b1 N 
 
...............................
...............................
 
where  A1  bi1 . . . bii . . . biN  . . . (10.43)
...............................
 
...............................
b . . . b . . . b 
 N1 Ni NN 

The foregoing matrix inversion may readily be obtained by using any of numerical
computer routines available. The system of equations may also be solved by the
Gauss-Seidel iteration method.
SAQ 4
What do you mean by radiosity and irradiation?

10.7 THE TWO-SURFACE ENCLOSURE

The simplest example of an enclosure is one involving two surfaces that exchange
radiation only with each other. Such a two-surface enclosure is shown schematically in
Figure 10.14(a). Since there are only two surfaces, the net rate of radiation transfer from
surface 1, q1, must equal the net rate of radiation transfer to surface 2,  q2, and both
quantities must equal the net at which radiation is exchanged between 1 and 2.
Accordingly,
q1   q2  q12 . . . (10.44)
The radiation transfer rate may be determined by applying Eq. (10.36) to surface 1 and 2
and solving resulting equations for J1 and J2. The results could then be used with
Eq. (10.30) to determine q1 (or q2). However, in this case the desired result is more
readily obtained by working with the net-work representation of the enclosure shown in
Figure 10.14(b). 57
Radiation

A2 ,T2 ,2

q12

A1 ,T1 ,1

(a)

1 - 1
1 1 - 2
1 A1
Eb1 A1 F12 J2 2 A2 Eb2
J1
q1 - q2

q12
Eb1 – J1 12 – Eb2
q1 = - q2 =
(1 - 1)/ 1 A1 (1 - 2)/ 2A2
(b)
Figure 10.14 : The Two Surface Enclosure (a) Schematic, (b) Network Representation

From Figure 10.14(b) we see that the total resistance to radiation exchange between
surfaces 1 and 2 is comprised of the two surface resistances and the geometrical
resistance. Hence, substituting from Eq. (9.41), the net radiation exchange between
surfaces may be expressed as

 (T14  T24 )
q12  q1   q2  . . . (10.45)
1  1 1 1  2
 
1 A1 A1 F12  2 A2
The foregoing results may be used for any two diffuse gray surfaces that form an
enclosure. Important special cases are summarised in Table 10.3
Table 10.3 : Special Diffuse, Gray, Two-Surface Enclosure
Large (Infinite) Parallel Planes

A1, T1, 1

A1  A2  A A  (T 4  T 4 )
q12  1 2
1 1
 1
F12  1 1 2
A2, T2, 2

Long (Infinite) Concentric Cylinders

r1

r2

A1 r
 1  A1 (T 4  T 4 )
1 2
A2 r2 q12 
1 1   2  r1 
  
1  2  r2 
F12  1

58
 Radiation Exchange
Concentric Spheres

r1
2
A1 r1  A1 (T 4  T 4 )
 1 2
A2 r22 q12 
2
r2 1 1   2  r1 
  
1  2  r2 
F12  1

Small Convex Object in a Large Cavity

A1, T1, 1

A1
0
A2
q12   A1 1 (T14  T24 )
F12  1

A2, T2, 2

10.8 RADIATION SHIELDS

Radiation shields constructed from low emissivity (high reflectivity) materials can be
used to reduce the net radiation transfer between two surfaces. Consider placing a
radiation shield, surface 3, between the two large, parallel planes of Figure 10.15(a).
Radiation Shield

q1 q13 q32 - q2

3, 1
3, 2

A1, T1, 1 A2, T2, 2

A 3 , T3

(a)
Eb1 J1 J3,1 Eb3 J3,2 J2 Eb2

q1
1 - 1 1 1 - 3.1 1 - 3, 2 1 1 - 2
3,1 A3 3,2 A3 2 A2
1 A1 A1 F13 A3 F32
(b)
Figure 10.15 : Radiation Exchange between Large Parallel Planes with a Radiation Shield
(a) Schematic, and (b) Network Representation
59
Radiation Without the radiation shield, the net rate of radiation transfer between surfaces 1 and 2 is
obtained from Table 10.3 (for parallel planes). However, with the radiation shield,
additional; resistances are present, as shown in Figure 10.15(b), and the heat transfer rate
is reduced. Note that the emissivity associated with one side of the shield (3, 1) may
differ from that associated the opposite side (3, 2) and the radiosities will always differ.
Summing the resistances and recognizing that F13 = F32 = 1, it follows that

A1  (T14  T24 )
q12  . . . (10.46)
1 1 1  3,1 1  3, 2
  
1  2 3,1 3, 2

Note that the resistance associated with the radiation shield become very large when the
emissivity 3, 1 and 3, 2 are very small. Eq. (10.46) may be used to determine the net heat
transfer if T1 and T2 are known. From knowledge of q12 and the fact that q12 = q13 = q32,
the value of T3 may then be determined by expressing Equation given in Table 10.3 for
plane parallel plates for q13 or q32.
The foregoing procedure may readily be extended to problems involving multiple
radiation shields. In the special case for which all the emissivities are equal, it may be
shown that, with N shields,
1
( q12 ) N  ( q12 )0 . . . (10.47)
N 1
where (q12)0 is the radiation transfer rate with no shields (N = 0).
SAQ 5
What do you mean by a radiation shield? Where is it used?

10.9 THE RERADIATING SURFACE

The assumption of a reradiating is common to many industrial applications. This
idealized surface is characterised by zero net radiation transfer (qi = 0). It is closely
approached by real surfaces that are well insulated on one side and for which convection
effects may be neglected on the opposite (radiating) side. With qi = 0, it follows from
Eqs. (10.25) and (10.30) that Gi = Ji = Ebi. Hence, if the radiosity of a reradiating surface
is known, its temperature is readily determined. In an enclosure, the equilibrium
temperature of a reradiating surface is determined by its interaction with the other
surfaces, and it is independent of the emissivity of the reradiating surface.
A three-surface enclosure, for which the third surface, surface R, i.e. reradiating, is
shown in Figure 10.16(a), and the corresponding network is shown in Figure 10.16(b).
AR, TR, R

A1, T1, 1 A2, T2, 2

(a)

60
 qR = 0 Radiation Exchange

EbR
1 - R
R AR

1 JR = Eb8
A1 F1R 1
1 - 1
A2 F2R 1 - 2
1 A1 qR2
q1R  2 A2
(b)
q1 Eb1 - q2
J1 1 J2 Eb
A1 F12
Figure 10.16 : A Three-Surface Enclosure with One Surface Reradiating
(a) Schematic, and (b) Network Representation
Surface R is presumed to be well insulated, and convection effects are assumed to be
negligible. Hence, with qR = 0, the net radiation transfer from surface 1 must equal the
net radiation transfer to surface 2. The network is a simple series-parallel arrangement,
and from its analysis it is readily shown that
Eb1  Eb 2
q1   q2  . . . (10.48)
1  1 1 1  2
 1

1 A1  1   1    2 A2
A1 F12    
 A1 F1R   A2 F2 R  
Knowing q1 =  q2 Eq. (10.30) may be applied to surfaces 1 and 2 to determine their
radiosities J1 and J2. Knowing J1, J2, and the geometrical resistances, the radiosity of the
reradiating surface JR may be determined from the radiation balance
J1  J R J  J2
 R 0 . . . (10.49)
 1   1 
   
 A1 F1R   A2 F2 R 
The temperature of the reradiating surface may then be determined from the requirement
that  TR4  J R . Note that the general procedure described in Section 10.6 may be
applied to enclosures with reradiating surfaces. For each surface, it is appropriate to use
Eq. (10.37) with qi = 0.

10.10 MULTIMODE HEAT TRANSFER

Thus far, radiation exchange in an enclosure has been considered under conditions for
which conduction and convection could be neglected. However, in many applications,
convection and/or conduction are comparable to radiation and must be considered in the
heat transfer analysis. Consider the general surface conditions of Figure 10.17(a).
In addition to exchanging energy by radiation with other surfaces of the enclosure, there
may be external heat addition to the surface, as, for example, by electric heating. And
heat transfer from the surface by convection and conduction. From a surface energy
balance, it follows that
qi , ext  qi , rad  qi , conv  qi , cond . . . (10.50)

where qi, rad, the net rate of radiation transfer from the surface, is determined by standard
procedures for an enclosure. Hence, in general, qi, rad may be determined from
Eqs. (10.30) or (10.35), while for special cases such as a two-surface enclosure and a
three-surface enclosure with one reradiating surface, it may be determined from
Eqs. (10.45) and (10.48), respectively.
61
Radiation

Enclosure

qi, rad. qi, conv

qi, ext

qi, cond

(a)
J1

1 - i
qi, rad
i A i

qi, ext qi, conv

Ebi,

qi, cond

(b)
Figure 10.17 : Multimode Heat Transfer from a Surface in an Enclosure
(a) Surface Energy Balance, and (b) Circuit Representation
The surface network element of the radiation circuit is modified according to
Figure 10.17(b), where qi, ext, qi, cond and qi, conv represent current flows to or from the
surface node. Note, however, that while qi, cond and qi, conv are proportional to temperature
difference, qi, rad is proportional to the difference between temperatures raised to the
fourth power. Conditions are simplified if the back of the surface is insulated, in which
case qi, cond = 0. Moreover, if there is no external heating and convection is negligible, the
surface is reradiating.

10.11 CROSSED STRING METHOD

Consider an enclosure (Figure 10.18) consisting of four surfaces. These surfaces are very
long in the direction perpendicular to the plane of the figure. The shape of the surfaces
may be plane, convex or concave. Suppose our interest is to evaluate the view factor F13
between the surfaces A1 and A3. We assume imaginary strings tightly stretched among the
four corners A, B, C and D of the enclosure .
C

L2

B L6
A3
L5 L3

L1
A1 L4

A
Figure 10.18 : Hottel Cross String Method

Let, Li (i = 1, 2, 3, 4, 5, 6) = length of the strings joining the corners.

Sum of the length of the crossed strings = L5 + L6
62
Sum of the length of the uncrossed strings = L1 + L2 Radiation Exchange

Hottel has shown that the view factor F13 is

( L5  L6 )  ( L1  L2 )
L1 F13  . . . (10.51)
2
Eq. (10.51) is useful for determining the view factor between the surfaces of a long
enclosure, such as, a groove, which can be characterized by a 2-D geometry.
SAQ 6
Explain Hottel’s cross string method for estimating shape factor for infinitely long
surfaces. Derive the expression for F12 in terms of areas and lengths of surfaces.
Example 10.1
Consider a diffuse circular disk of diameter D and area Aj and a plane diffuse
surface of area Ai < < Aj. The surfaces are parallel, and Ai is located at a distance L
from the centre of Aj. Obtain an expression for the view factor Fij.
Solution
It is given that surfaces are diffuse and Ai < < Aj.
The desired view factor may be obtained from
1 cos i cos  j
Fij 
Ai A Ai j  R2
dAi dA j

Here, i, j and R are approximately independent of the position on Ai.

dr D

dAj

Aj j

L
R

i

Ai

Figure 10.19 : Schematic Diagram For Example 10.1

cos i cos  j
Hence, Fij  A j  R2
dA j

with i   j  

cos 2 
Fij  A j  R2
dA j

Further, R 2  r 2  L2
L
cos  
R 63
Radiation and dA j  2 r dr
D
2
rdr D2
Hence, Fij  2L2  
0 ( r 2  L2 ) 2 D 2  4 L2
Example 10.2
Determine the view factor F12 and F21 for the following geometries.
(a) Sphere of diameter D inside a cubical box of length L = D.
(b) Diagonal partition within a long square duct.
(c) End and side of a circular tube of equal length and diameter.
Solution
(a) Sphere within a cube
By inspection, F12 = 1
A1  D2 
By reciprocity, F21  F12  2
 1
A2 6L 6
D

A1
A1

L=D A1
L A2
L=D
A2
A2
A3
A3

Figure 10.20 : Geometry For Example 10.2

(b) Partition within a square duct

From summation rule
F11  F12  F13  1

where F11 = 0
By symmetry, F12 = F13
Hence, F12 = 0.5

A1 2L
By reciprocity, F21  F12   0.5  0.71
A2 L

(c) Circular tube

r3 L
From Figure 10.5, with  0.5 and  2, F13  0.17
L r1
From summation rule, F11  F12  F13  1
or, with F11  0, F12  1  F13  0.83

D 2
A
By reciprocity, F21  1 F12  4  0.83  0.21
A2 DL

64
Example 10.3 Radiation Exchange

A brick wall having an emissivity of 0.85 is 6 m wide and 4 m high. It is at a

distance of 4 m from a 500 m  400 m opening in a furnace wall. The centre line
of the opening lies 1 m lower and 1 m left of the centre of the wall. The furnace
temperature is 1500oC and that of the wall is 37oC. Calculate the rate of radiation
heat transfer between the opening and the wall.
Solution
T1 = 173 K
T2 = 310 K
The brick wall is divided into four areas I, II, III and IV as shown in Figure 10.21.
2m 4m
h

i
g
II
III
3m

I IV 1m
b e
a
1m

c d f

1m
5m

Figure 10.21 : Brick Wall Divided in Four Areas

Heat loss from the opening to the wall

(Q1  2 ) net   A1  e F12 (T14  T24 )

where e = 1 . 2 = effective emissivity

The furnace opening is considered to act as a blackbody.

e  1  0.85  0.85

A1  0.5  0.4  0.2 m 2

Rectangle I
D 5
 5
L1 1

D 5
  2.5
L2 2

Using Figure 10.6, F = 0.023

65
Radiation Rectangle II
Similarly for Rectangle II
D 5
  1.67
L1 3

D 5
  2.5
L2 2

F = 0.053
Rectangle III
D 5
  1.67
L1 3

D 5
  1.25
L2 4

F = 0.09
Rectangle IV
D 5
  5.0
L1 1

D 5
  1.25
L2 4

F = 0.036

F12   F  0.023  0.053  0.09  0.036  0.202

 Q12 net  5.67  10  8  0.2  0.85  0.202  (17734  310 4 )

= 19.222 W.
Example 10.4
Two very large parallel plates with emissivities 0.3 and 0.8 exchange radiative
energy. Determine the percentage reduction in radiative energy transfer when a
polished aluminium radiation shield ( = 0.04) is placed between them.
Solution
The radiant heat transfer rate without the shield is given by

Q  (T14  T24 )  (T14  T24 )

   0.279 (T14  T24 )
A 1 1 1 1
 1  1
1  2 0.3 0.8

The radiation network for two infinite parallel plates separated by one radiation
shield is shown in Figure 10.22.

Eb1 J1 J3 Eb3 J 3 J2 Eb2

1 - 1 1 1 - 3 1 - 3 1 1 - 2

1 F13 3 3 F32 2
Figure 10.22 : Radiation Network for Two Infinite Parallel Plates Separated by One Radiation Shield
66
 With shield, the total resistance is Radiation Exchange

1  1 1 1  3 1 1  2
R  2  
1 F13 3 F32 2

1  0.3 1  0.04 1  0.08

  1 2  1
0.3 0.04 0.08
= 52 283
Rate of radiant heat transfer with one shield

Eb1  Eb 2  (T14  T24 )

Q1  2    0.017  (T1 4  T24 )
R 52.283

0.279  0.017
Percentage reduction in heat transfer   100  93.6% .
0.279
Example 10.5
Two infinitely long parallel plates of widths x = 12 cm and y = 6 cm are located at
a distance z = 7 apart as shown in Figure 10.23. Determine the view factor F12.
Solution
Lebel the end points of both the surfaces and draw straight dashed lines between
the end points (Figure 10.23). Using the cross-string method
4

Figure 10.23 : Example 10.5

( L5  L6 )  ( L3  L4 )
F12 
2 L1
where L1 = x = 12 cm
L2 = y = 6 cm
L3 = x = 7 cm
1
2 2 2
L4  (7  6 )  9.22 cm
1
L5  (62  7 2 ) 2  9.22 cm
1
2 2 2
L6  (12  7 )  13.89 cm
(9.22  13.89)  (7  9.22)
F12   0.287
2  12
Example 10.6
(a) Derive an expression for the time required to cool a body of mass m, surface
area A, emissivity  and specific heat Cp from an initial temperature T1 to the
final temperature T2 in a large enclosure, the walls of which are at
67
Radiation temperature Tw. Neglect the convective losses and temperature gradients
inside the body.
(b) A solid copper sphere 0.1 m in diameter is heated to 1000oC and suspended
in a large room, the walls of which are at 30oC. Calculate the time taken by
the sphere to cool to 500oC. Consider only radiaitve energy transfer and
neglect the internal thermal resistance of the sphere. For copper, take
 = 8680 kg/m3, Cp = 0.41 kJ/kgK and  = 0.1.
Solution
(a) Let T denote the temperature of the body at instant t. In time dt, let the
temperature of the body drop by dT. By energy balance,
Energy loss of the body = Energy transfer to the surroundings by radiation
or  mC p dT   A  (T 4  Tw4 ) dt
T2
 A dT
mC p
t  (Tw2  T ) (Tw2  T 2 )
2
T1

1  1 1 

2Tw2
  2 2
 2  dT
2 
 Tw  T Tw  T 

T1
1  1  1 T  1  
T1
1  1
 
2 T 
2Tw  w 
tan  
Tw T 2Tw  
T
 w  T

Tw  T
 dT
 
 2 T2

  T1   T2  
    

1  1
tan  T2   Tw   1 ln (Tw  T1 ) (Tw  T2 ) 
2Tw3  T T  2 (Tw  T1) (Tw  T2 ) 
 1 1  2  
  Tw   Tw  

 
 
mC p 1  1 (T2  Tw ) (T1  Tw )  1 Tw (T1  T2 ) 
t ln  tan
 A  2Tw3  2 T T  Tw2  T1T 2 
 1 1  2  
  Tw   Tw  

(b) T1 = 1273 K
T2 = 773 K
Tw = 303 K

A  4 r 2  4  (0.05) 2  0.0314 m 2

3
m  v  8680       (0.05) 3  4.542 kg
4
4
 293 
Eb3   T34  5.67   2
  418 W/m  0.418 kW/m
2
 100 
1  1  0.5
  1.27 m  2
1 A1 
0.5     1
4
F12 = 0.06

68
 1 1 Radiation Exchange
  21.22 m  2
A1 F12   
   1  0.06
4
F13  1  F12  0.94

2 F23  F23  F23  0.94  1  1.94

where suffix 2 indicates both sides, 2 for the left side and 2 for the right
side of the disc.
1.94
F23   0.97
2
1 4
  1.35 m  2
A1 F13   0.94

1 4
  0.65 m  2
2 A2 F23   (2  0.97)

Algebraic sum of currents at each node is zero.

At Node J1
20.24  J1 Eb 2  J1 0.418  J1
  0
1.27 21.22 1.35
At Node J2
(T2  Tw ) (T1  Tw ) (773  303) (1273  303)
ln  ln
(T1  Tw ) (T2  Tw ) (1273  303) (773  303)

1076  970
 ln  ln 1.409  0.34288
1576  470

Tw (T1  T2 ) 303 (1273  773)

tan 1  tan 1
Tw2  T1 T2 3032  1273  773

 tan 1 0.1408  8.0145  0.1396 rad

4.542  0.41 1
t  11

5.67  10  0.0314  0.1 (12  3033 )

1 
   0.34288  0.1396 
 2 
 5982.2 s = 1 hr 40 min

Exercise 10.1
(a) If the inside surface temperature of a hemispherical cavity of 0.25 m
diameter is 600oC and its emissivity is 0.78, calculate the rate of radiant heat
transfer from the cavity.
(b) A cubical oven has inside sides equal to 0.3 m. One of the faces of the oven
forms the door. If all the other faces are having same emissivity ( - 0.8) and
maintained at 500oC, find the rate of heat loss if the oven door is kept open.
(c) Show that the view factor for two surfaces 1 and 2 connected by a refractory
surface is given by
69
Radiation
A2  A1 F122
f12 
A1  A2  2 A1 F12

What will be its value (a) if A1 = A2 and (b) if the surface 2 does not see the
surface 1?
(d) Show that the radiative flux (Q12)net between two gray surfaces 1 and 2
connected by a non-conducting and reradiating surface is given by

(Q12 ) max   A1 F12 (T14  T24 )

1
where F12 
1  1 A  1 
  1   1   1
 1  f12 A2   2 

A2  A1 F122
and f12 
A1  A2  2 A1 F12

Exercise 10.2
(a) Two large parallel planes having emissivities of 0.4 and 0.7 are maintained
at temperatures 1000 K and 600 K, respectively. Determine the net radiant
heat exchange per unit area between the planes. If a radiation shield having
an emissivity 0.05 on both sides is placed between the two planes, calculate
the temperature of the shield and the heat transfer rate per unit area.
(b) Determine F12 and F21 for the following configurations :
(i) Long duct.
(ii) Small sphere of area A1 under a concentric hemisphere of area
A2 = 2A1.
(iii) Long duct. What is F22 in this case?
(iv) Long inclined plates (Point B is directly above the centre of area A1).
(v) Sphere lying on infinite plane.
(vi) Hemisphere-disk arrangement.
(vii) Long open channel.
(c) Consider the parallel rectangles as shown in figure show that the view factor
F12 can be expressed as

1
F12  [ A14 F14  23  A1 F13  A4 F42 ]
2 A1

(d) Using a suitable method calculate view factor for the figures given below.

70
 Radiation Exchange

A2
A2 A1

A1

900

A1
100 mm
A2 A2
A1

200 mm

A1 A2 Hemisphere,
diameter D

A2
A3 A1 Disk,
diameter D/2

L1
A2

1m A2

L2
A1
D

2m dA1

71
Radiation

A1
b
L
A2

D
A2 c

dA1

dA
A2
1

r2

r2

dA
1

A2

10.14 SUMMARY
In the present unit concept of view factor is brought forth and view factor integral is
derived. Various relations of view factors are also derived. Hottle’s cross string method is
also described. Radiation exchange problems in black and gray enclosures and their
solutions are then taken up. The electrical analogy method is discussed. The radiation
exchange between infinite parallel plates and applications of radiation shields are
discussed.

10.15 KEY WORDS

Cross String Method : A method to find view factor for infinitely long
surface.
Radiation Shield : Means such as a polished surface to reduce
radiation between two surfaces
Radiosity : Total radiant energy leaving a surface per unit
time per unit surface area.

72
 Radiation Exchange
10.16 ANSWERS TO SAQs
Refer the relevant text for all the Answers to SAQs.

REFERENCES
J. P. Holman (2002), Heat Transfer, 9th Edition, Tata McGraw-Hill, New Delhi.
M. N. Ozisik (1985), Heat Transfer : A Basic Approach, McGraw-Hill International
Edition.
F. P. Incropera and D. P. Dewitt (2004), Fundamentals of Heat and Mass Transfer,
5rh Edition, John Wiley and Sons.
P. K. Nag (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.

RADIATION
Preceeding blocks were devoted to heat transfer by conduction and convection. In the
present block, the third mode of heat transfer, i.e. radiation has been considered. Thermal
radiation is that electromagnetic radiation emitted by a body as a result of its temperature.
Solution of radiation problems are very complicated as large number of variables are
associated with it. In dealing with such problems you will notice that radiation is
dependent on spatial variables (x, y, z), spectral properties () as well as polar () and
azimuthal angle (). Treatment becomes still complicated if the intervening medium is
participating in nature.
In Unit 9, we will consider the principles of radiation. Radiation intensity and emissive
power are discussed. Concept of blackbody and laws of radiation are discussed. Radiative
transfer rate depends on the surface properties. This unit will give you more insight on
thermal radiation.
Unit 10 discusses estimation of thermal radiation with the help of view factor. View
factor algebra is more elaborately given so that you can estimate the radiative transfer for
different geometries and orientations. An electrical analogy method is provided. Based
on this unit, you will be able to solve surface radiation problems. Radiation with
participating medium is more involved. Hence the same is not considered in these units.

73
Radiation

74
Indira Gandhi MRW-002
National Open University
School of Engineering & Technology HEAT TRANSFER

BLOCK 5

Conduction
 Indira Gandhi
National Open University
MRW – 002
School of Engineering & Technology Heat Transfer

Block

5
HEAT TRANSFER APPLICATIONS
UNIT 11
Heat Exchangers 5

UNIT 12
Boilers 41
HEAT TRANSFER APPLICATIONS

You have learnt about the heat transfer mechanism in earlier blocks. Both heat
and mass transfer are important from practical engineering aspects. Present block
is devoted to applications of heat transfer in practical engineering fields.

Unit 11 is devoted to heat exchangers. Different types of heat exchangers utilized

in industries are discussed in this unit. Classification of heat exchanger is also
given. Methods to calculate heat transfer rate and sizing and rating of heat
exchangers are discussed elaborately. You will be able to evaluate heat transfer
rate and length of heat transfer tubes for different types of heat exchangers.

In Unit 12, we have discussed about another class of heat exchangers working
with phase change. Boiling and condensation theories are given to understand
such heat exchangers. Different types of boilers and condensers are discussed and
some simple methods to design a condenser is given in this unit.
 Heat Exchangers
UNIT 11 HEAT EXCHANGERS
Structure
11.1 Introduction
Objectives
11.2 Classification of Heat Exchanger
11.2.1 Compact Heat Exchanger
11.2.2 Classification by Construction Type
11.2.3 Classification by Flow Arrangement
11.2.4 Classification by Heat Transfer Mechanism
11.2.5 Classification according to the Type of Fluids
11.3 Temperature Distribution in Heat Exchangers
11.4 The Overall Heat Transfer Coefficient
11.5 Heat Exchanger Analysis
11.5.1 Log Mean Temperature Difference
11.5.2 The Effectiveness – NTU Method
11.6 Summary
11.7 Key Words
11.8 Answers to SAQs

11.1 INTRODUCTION
Heat exchangers are devices that facilitate heat transfer between two or more fluids at
different temperatures. Many types of heat exchangers are developed to meet the demand
of processes. Applications of heat exchangers are wide, such as steam power plants,
chemical processes, building heating, air conditioning, house hold refrigerators, car
radiators, radiators for space vehicles, etc.
Heat transfer, pressure drop analysis, sizing and performance testing, and economic
aspects play important roles in the design of heat exchangers.
Objectives
After studying this unit, you should be able to
 classify different heat exchangers and applications,
 determine the overall heat transfer coefficient,
 find the logarithmic mean temperature difference, and
 apply method of rating and sizing to heat exchangers.

11.2 CLASSIFICATION OF HEAT EXCHANGER

Heat exchangers are classified according to :
(a) compactness,
(b) construction,
(c) flow arrangement, and
(d) heat transfer mechanism.
5
Heat and Mass 11.2.1 Compact Heat Exchanger
Transfer Applications
S
A heat exchanger having surface to volume ratio   more than 700 m2/m3 is known as
V 
compact heat exchanger. High value of compactness reduces the volume for a specific
heat exchanger performance. Compactness becomes importance when one thinks about
the criticality of the process such as aircraft, aerospace vehicles, automobile radiators,
heating, ventilation and air conditioning, etc. Table 11.1 presents some surface densities
of different applications.
Table 11.1 : Compactness of Some Heat Exchangers
Sl. No. Applications Compactness m2/m3
1. Automobile radiator 1100
2. Human lungs 20,000
3. Plane tubular heat exchanger/shell and tube heat 70-500
exchanger
4. Rotary heat exchangers  6500
11.2.2 Classification by Construction Type
Heat exchangers are classified according to their construction features.
Recuperator
In most of the heat exchangers, the fluids are separated by a heat transfer surface
and ideally they do not mix. Such heat exchangers are called direct transfer heat
exchanger or recuperator.
Tabular Heat Exchangers
Tubular exchangers are widely used and they are manufactured in many sizes, flow
arrangements and types. They can accommodate a wide range of operating
pressures and temperatures. The ease of manufacturing and their relatively low
cost have been the principal reason for their wide spread use in engineering
applications. The most common tubular heat exchanger is the shell and tube type
exchanger. 85% of the heat exchangers in the world are the shell and tube type
exchangers. Figure 11.1 illustrates the main features of a shell and tube type heat
exchanger having one fluid flowing inside the tubes and the other flowing outside
the tubes.
Tube-side Shell-side
Header
outlet outlet Baffle

Header
Tubes
Shell
Shell-side Tube-side
Fluid inlet Fluid inlet
Figure 11.1 : Shell and Tube Heat Exchanger

The principal components of this type of heat exchanger are :

(a) tube bundles,
(b) shell,
(c) front header,
6
 (d) rear header, and Heat Exchangers

(e) baffles.
The baffles are used to support the tubes, to direct the fluid flow approximately
normal to the tubes and to increase the turbulence of the shell fluid. There are
various types of baffles and the choice of baffle type, spacing and geometry
depends on the flow rate, allowable shell side pressure drop, tube support
requirement and the flow induced vibration.
Figures 11.2(a) to (c) presents 3 (three) different type of shell and tube type heat
exchangers. Figure 11.2(a) is a fixed head type of shell and tube heat exchanger
(same as shown in Figure 11.1).
Tube-side Shell-side
Header
outlet outlet Baffle

Header
Tubes
Shell
Shell-side Tube-side
Fluid inlet Fluid inlet
(a)

Floating
(b) Head Type

(c) U-type
Figure 11.2 : Different Types of Shell and Tube Heat Exchangers

If the tubes are fixed, during heat transfer operation the tubes gets expanded as
both ends of the tubes are fixed. The tubes may bend due to expansion. To prevent
the bending of the tubes, the floating head arrangements are done (Figure 11.2(b)).
Another improvement in the shell and tube type is the incorporation of U-bend
tubes which will facilitate more turbulence and secondary flow, thus increasing
heat transfer rate (Figure 11.2(c)). Feed water heaters in thermal power plants uses
U-bend type exchangers.
Plate Heat Exchanger
Plate heat exchangers as shown in Figure 11.3 are extensively used in liquid-to-
liquid heat exchange processes. This type of heat exchanger is very popular in
process industries where mixing, evaporation, reaction, distillation, separation
processes are involved. It is one of the compact counter flow heat exchangers.
Q
Effectiveness   act is as high as 0.95. Temperature and pressure are the main
Qmax
7
Heat and Mass drawbacks in case of gasketed units. But with advance technology, better gaskets
Transfer Applications are available and can withstand high temperature and pressure.
Header

Hot
Fluid

Cold
Cold Fluid in
Fluid Out

Hot
Fluid Out

Figure 11.3 : Plate Heat Exchanger

Plate-fin Heat Exchanger

The compactness factor can be significantly improved (upto about 6000 m2/m3) by
using the plate-fin type heat exchanger as shown in Figure 11.4. These types of
heat exchangers are generally used for gas-to-gas applications, but they are used
for low pressure applications not exceeding above 10 atm (1000 kPa). The
maximum operating temperature is limited to about 800oC.

Figure 11.4 : Plate-fin Exchanger

Spiral Plate Heat Exchanger

It is a form of plate heat exchanger usually made of stainless steel. It is often used
in cellulose industries where heat exchanger is subjected to severe fouling and
corrosion. The plates of this type of heat exchanger are very long and thickness of
passage between the plates must be rather small so that after the sheets forming the
upper and lower surfaces are welded together, the unit can be wrapped into spiral
form. That’s why it is called as spiral heat exchanger. The technical features of this
type of heat exchanger are :
(a) Flow rates are relatively low
(b) Pure counterflow heat exchanger
(c) Highly compact (more than 700 m2/ m3)
(d) Can withstand pressure upto 10 bar only
8
 Heat Exchangers

Hot Out

Cold In x

Figure 11.5 : Spiral Heat Exchanger

Regenerator
Heat exchangers in which there is an intermittent flow of heat from hot to cold
fluid via heat storage and heat rejection through the exchanger surface or matrix
are referred to as indirect or storage type heat exchanger or regenerator. The
regenerative type heat exchangers are either static or dynamic.
Static Type Regenerator
(a) No moving parts.
(b) Consists of a porous medium (balls, pebbles, powders, etc.) through which
hot and cold fluid pass alternatively.
(c) A flow switching device regulates the periodic flow of the two fluids.
(d) Compact for use in refrigeration and Stirling Engines.
(e) Non-compact in high temperature (900 – 1500oC) applications.
(f) Low cost and ruggedness are essential for the stationary type.
Storage Type or Regenerative Heat exchanger
Storage type or regenerative heat exchanger is shown in Figure 11.6. In this heat
exchanger energy is stored periodically. Medium is heated or cooled alternatively.
Heating period and cooling period constitute 1 (one) cycle.
Features
(a) Periodic heat transfer-conduction.
(b) Heat transfer fluid can be a liquid, phase changing, non-phase
changing.
(c) Solid storage medium is called matrix.
(d) Matrix may be stationary or rotating
In In

A B

C D

Figure 11.6 : Storage Type/Regenerator Type Heat Exchanger

9
Heat and Mass Classical Applications
Transfer Applications
(a) Gas turbine regenerators : Heating the compressed air by the gas
turbine exhaust before the air goes to the combustor.
(b) Reversed Stirling engine for liquefaction of air-Philips refrigeration
machine.
Dynamic Type Regenerator
(a) Compact in nature, very high surface to volume ratio (more than
6500 m2/m3).
(b) Usually matrix rotates about an axis (i.e. Ljungstrom regenerative air
preheater).
(c) Operating temperature is upto 870oC.
(d) Usually used as gas-to-gas heat exchanger.
Heat Wheels/Rotary Regenerator/Ljungstrom Air Preheater
Features
(a) Annulus is divided into numbers of sectors.
(b) Matrix is filled up in these sectors.
(c) Out of these sectors, few are kept empty so that hot fluid/cold fluid
cannot flow through the empty region.
(d) Matrix rotates at low RPM (3-4 RPM) driven by a motor through
appropriate reduction gear in mesh.
(e) Heat capacity of matrixes are more than that of the gases
m c p |matrix  m c p |gas , hence it is useful for gas-to-gas recovery.

(f) For gas-to-liquid or liquid-to-liquid recovery, it is not useful as

m c p |matrix < m c p |liquid .

Hot
Fluid

Cold
Fluid

Heating Cooling
Zone Zone

Empty

Figure 11.7 : Heat Wheel

10
 Advantages Heat Exchangers

(a) Cheap
(b) Compact (~ 6500 m2/ m3), large surface area, hence high heat transfer
rate.
(c) Can operate upto 870oC with metal matrix
Disadvantages
(a) Large pressure drop through the matrix.
(b) Fouling of surfaces with entrained solid particles causing reduction in
flow passage area leading to increase in pressure drop.
(c) Leakage of one fluid in the duct to the duct where other fluid is
flowing requires extra space to separate the flow passages.
(d) Energy required to rotate the wheel.
Applications
Heat wheels are widely used in thermal power plants to pre-heat air by the
exhaust gases.
Rothemuhle Regenerator
This type of heat exchanger is extensively used in steel plants for heat recovery. In
this heat exchanger, the matrix is stationary and the hood rotates. One half of the
hood is perforated. Air at ambient condition is passed through the central core of
the heat exchanger. On one side of the central pipe also, perforation is provided to
facilitate air movement to the matrix. Hot gas containing the waste heat enters
from the top, passes through the perforated hood to the matrix thereby exchanges
heat through the matrix and goes out at the bottom. Simultaneously, air passes
through the other half of the previously hot matrix and get heated. Two different
positions of this heat exchanger has been shown in Figure 11.8.
Hot Air Out
Hot Air Out
Hot Gas in
Hot Gas in

Rotating Hood Rotating

Hood

Stationary Heat Stationary Heat

Exchanger Matrix Exchanger Matrix
Cooled Cooled
Out gas Out Gas

Cool Air in
Cool Air in
Figure 11.8 : Rothemuhle Heat Exchanger

Matrix Material.
(a) Usually knitted aluminium or stainless steel is used.
(b) For moisture removal from air, hygroscopic material like asbestose
fibre impregnated with LiCl can be used.
Disadvantage
(a) Pressure drop is large.
(b) Subjected to fouling.
11
Heat and Mass 11.2.3 Classification by Flow Arrangement
Transfer Applications
Numerous possibilities exist for flow arrangement in heat exchangers. Some of the flow
arrangements are as follows :
Parallel Flow
The hot and the cold fluid enter at the same end of the heat exchanger, flow
through in the same direction, and leave at the other end, as illustrated in
Figure 11.9.
Cold Out

Hot Hot
In Out

Cold In

Figure 11.9 : Parallel Flow Heat Exchanger

Counter Flow
The hot and cold fluids enter in the opposite ends of the heat exchanger and flow
in opposite directions, as illustrated in Figure 11.10.
Cold In

Hot Hot
In Out

Cold I
Out
Figure 11.10 : Counter Flow Heat Exchanger

Cross Flow
In the cross flow exchanger, the two fluids usually flow at right angles to each
other, as illustrated in Figure 11.11. In the cross flow arrangement, the flow may
be called mixed or unmixed, depending on the design.
Cold In

Hot Hot
In Out

Cold Out

Figure 11.11 : Cross Flow Heat Exchanger

Figure 11.12(a) shows an arrangement in which both the cold and hot fluids flow
through the individual channels formed by corrugation, therefore the fluids are not
free to move in the transverse direction. Then each fluid stream is said to be
unmixed. Figure 11.12(b) illustrates a typical temperature profile for the outlet
temperatures when both fluids are unmixed (Figure 11.12(a)). The inlet
temperatures for both fluids are assumed to be uniform, but the outlet temperatures
exhibit variation transverse to the flow.

12
 Heat Exchangers
T c, out T h, out

Hot Fluid, T h, In
T c, In, Cold Fluid

(a)
T c, out

y
T h, in T h, out
x

T c, in
(b)

Hot
Fluid
Cold
Fluid
inside
Tubes

(c)
Figure 11.12 : Cross Flow Arrangements (a) Both Fluids Unmixed, (b) Temperature Profile when Both
Fluids are Unmixed, and (c) Cold Fluid Unmixed but Hot Fluid Mixed

In the flow arrangements shown in Figure 11.12(c), the cold fluid flows inside the
tubes and is not free to move in the transverse direction. Therefore, the cold fluid is
said to be unmixed. However, the hot fluid stream flows over the tubes and is free
to move in the transverse direction. Therefore, the hot fluid stream is said to be
mixed. The mixing tends to make the fluid temperature uniform in the transverse
direction. Therefore, the exit temperature of a mixed stream exhibits negligible
variation in the crosswise direction.
In general, in a cross flow exchanger, three idealised flow arrangements are
possible :
(a) both fluids are unmixed,
(b) one fluid is mixed and the other fluid is unmixed, and
(c) both fluids are mixed.
The last arrangement is not commonly used.
In a shell and tube exchanger, the presence of large number of baffles serves to
mix the shell side fluid. Hence, temperature of the shell side fluid tends to be
uniform at any cross-section.
13
Heat and Mass Multipass Flow
Transfer Applications
The multipass flow arrangements are frequently used in heat exchanger design,
because multipassing increases the overall effectiveness over the individual
effectiveness. A wide variety of multipass arrangements are possible.
Figure 11.13 illustrates typical arrangements. The heat exchanger in
Figure 11.13(a) is a “one shell pass, two tube pass” arrangement. This is also
called a one-two heat exchanger. Figure 11.13(b) presents a “two shell pass, four
tube pass” arrangement and in Figure 11.13(c) shows a “three shell pass, six tube
pass” arrangement.
Shell-side Fluid

Tube-side Fluid

(a)

Shell-side Fluid

Tube-side Fluid

(b)
Shell-side Fluid

Tube-side
Fluid

(c)

Figure 11. 13 : Multipass Flow Arrangements (a) One Shell Pass, Two Tube Pass,
(b) Two Shell Pass, Four Tube Pass, (c) Three Shell Pass, Six Tube Pass

11.2.4 Classification by Heat Transfer Mechanism

Heat exchangers can be classified based on the heat transfer mechanism, such as :
(a) Single phase forced or free convection (sensible heat transfer).
(b) Phase change (boiling and condensation).
(c) Radiation or combined convection and radiation.
Single phase heat exchangers are already described in the preceeding subsections. In this
type of heat exchangers, heat transfer between fluid streams occur due to exchange of
sensible heat only. Phase change mechanisms and applications to boiler and condenser
are described in Unit 12.

14
Metallic Radiative Recuperator Heat Exchangers

Features
(a) Height upto 50 m
(b) Diameter 0.25-3 m
(c) Natural draught, no need for a fan or blower
(d) Heat transfer is chiefly by radiation
Examples : Metallic recuperators are used in
(a) Steel plants
(b) Glass Melting Furnaces
Waste Gas

Hot Air to Process

Cold Air Intake

Flue Gas

Figure 11.14 : Metallic Recuperator

11.2.5 Classification according to the Type of Fluids

Heat exchangers may be further classified according to the type of fluids, such as
gas-to-gas, gas to liquid and liquid to liquid heat exchanger.
Gas-to-Gas Heat Exchanger
(a) Plate-fin coupled prime surface exchanger
(b) Heat Pipe
(c) Rotary regenerator
(d) Convection recuperator
(e) Liquid heat
Gas-to-Liquid Heat Exchanger
(a) Boiler
(i) Fire tube
(ii) Water tube
(b) Economizer
(c) Fluidized bed heat exchanger
(d) Heat pipe heat exchanger
15
Heat and Mass Liquid-to-Liquid Heat Exchanger
Transfer Applications
(a) Shell and tube
(b) Plate
Gas-to-Gas Heat Exchanger
These types of heat exchangers require special arrangement to have high heat
transfer rate. Gases are having low heat transfer coefficient. Usually heat transfer
coefficients on opposite sides of the heat transfer surface are usually within a
factor of 3 to 4 of each other. The absolute values are generally lower than liquid
to liquid heat exchangers. Hence heat transfer enhancement is essential in such
heat exchangers. Application of fin, surface roughening, inserts are usually
employed. But pressure drop is also high. Leakage of either fluid may be tolerated
upto 4%. Constructions are lighter and less rugged. Various types of gas-to-gas
heat exchangers are :
(a) Plate-fin
(b) primary surface exchanger
(c) rotary regenerator
(d) run around coil
(e) radiation and convective recuperator
(f) heat pipe
Temperature Limitations
The minimum temperature to which the waste gas are cooled is an important
consideration in the design of gas-to-gas waste heat recovery devices, i.e. air
preheater. It is considered good practice to design so that temperature of
combustion gas (waste gas) leaving the heat exchanger does not drop below 300oF
(148.88oC), because if this is not done, corrosion may result as a result of
condensation of sulphurous or sulphuric acid. These substances are formed from
the sulphur present in most fuels reacting with the water vapour from combustion.
Oxidation of carbon steel becomes a serious problem for gas temperature in excess
of 1000 F (537.78oC). Hence, ceramic material is to be used. For example, the air
preheater for steel and blast furnace are ordinarily made of ceramic.
Applications of Gas-to-Gas Recovery Units
(a) Air preheater for steam power plant
(b) Gas fired heater for industrial processes
(c) Gas turbines – overall thermal efficiency may be doubled with gas-to-gas
regenerative type heat exchanger. Used as intercooler between stages and
compressor.
SAQ 1
(a) How do you classify the heat exchangers?
(b) What is a counterflow heat exchanger?
(c) Sketch the schematic diagram of a shell-and-tube heat exchanger and
indicate its various parts.
(d) What is the major difference between a recuperator and a regenerator?
(e) How does a rotary heat exchanger work? Explain with diagram?

16
 Heat Exchangers
11.3 TEMPERATURE DISTRIBUTION IN HEAT
EXCHANGERS
Figure 11.15 presents the variation of temperature for different arrangements along the
flow path (along the length of the heat exchanger).
Condensing
 T1
Hot Fluid
 T1
Temperature

Cold Fluid

Temperature
Cold Fluid  T0
 T0

O Distance from Inlet L

O Distance from Inlet L
(a) Uniform Temperature Difference (b) Uniform Surface Temperature
or Uniform Heat Flux (As in a Condensor)

Hot Fluid
 T0
Temperature

 T1
Boiling

O Distance from Inlet L

(c) Uniform Surface Temperature (As in Boiling)

Hot Fluid
Hot Fluid  T1
Temperature

Temperature

 T0  T1
Cold Fluid
Cold Fluid  T0

O Distance from Inlet L L

O Distance from Inlet
(d) Parallel Flow Heat Exchanger (e) Counter Flow Heat Exchanger
Figure 11.15 : Axial Temperature Distribution in Typical Single-pass Heat Transfer Matrices
Figure 11.15(a) indicates a pure counterflow heat exchanger in which the temperature
rise in the cold fluid is equal to the temperature drop in the hot fluid, thus the temperature
drop T between the hot fluid and the cold fluid is constant throughout. However, in all
other cases (Figures 11.15(b)-(e)), the temperature difference T between the hot and the
cold fluids varies with position along the path of flow.
Figure 11.15(b) corresponds to a situation in which the hot fluid condenses and heat is
transferred to the cold fluid causing its temperature to rise along the path of flow.
In Figure 11.15(c), cold liquid is evaporating and cooling the hot fluid along its path of
flow.
Figure 11.15(d) shows a parallel flow arrangement in which both the fluids flow in the
same direction, with cold fluid experiencing a temperature rise and the hot fluid a
temperature drop. The outlet temperature of the cold fluid cannot exceed that of the hot
fluid. Therefore, the temperature effectiveness of the parallel flow exchangers is limited.
Because of this limitation, generally they are not used for heat recovery.
Figure 11.15(e) presents a counter flow arrangement in which fluids flow in opposite
directions. The exit temperature of the cold fluid can be higher than that of the hot fluid.
Theoretically, the exit temperature of one fluid may approach the inlet temperature of the
other. Therefore, thermal capacity of the counter flow exchanger can be twice that of the
parallel flow heat exchangers. The high heat recovery and temperature effectiveness of
this exchanger makes it preferable to the parallel flow exchanger whenever the design
requirements permit such a choice.
17
Heat and Mass Temperature distributions are more complicated in multi-pass and cross flow
Transfer Applications arrangements. Figure 11.16 shows the temperature distribution in a one shell pass, two
tube pass heat exchanger.
Shell-Side
Fluid in
Tube-Side
Fluid
Out
In

Shell-Side
Fluid Out

Hot Fluid – Shell side

Cold Fluid – Tube side

O L
Distance
Figure 11.16 : Axial Temperature Distribution in One Shell Pass, Two Tube Pass Heat Exchanger
Figure 11.17 shows a typical temperature profile in a cross flow heat exchanger when
both fluids are unmixed. It may be noted that inlet temperatures for both the streams are
uniform but the out let temperatures are non-uniform for both the streams.
Hot Fluid In

Cold Fluid In Cold Fluid Out

Hot Fluid Out

Figure 11.17 : Temperature Distribution in a Cross Flow Heat Exchanger, Both Fluids are Unmixed

11.4 THE OVERALL HEAT TRANSFER COEFFICIENT

In the heat transfer analysis of heat exchangers, various thermal resistances in the path of
heat flow from the hot to the cold fluid are combined into an overall heat transfer
coefficient U.
Consider that the total thermal resistance R to heat flow across a tube, between the inside
and the outside flow, is composed of the following thermal resistances
R = (thermal resistance of inside flow) + (thermal resistance of the tube material)
+ (thermal resistance of outside flow)
and various terms are given by
1 t 1
R   . . . (11.1)
Ai hi k Am Ao ho
where Ao, Ai = outside and inside surface areas of tube respectively, m2
Ao  Ai
Am   logarithmic mean area, m2
 Ao 
ln  
 Ai 
18
ho, hi = Heat transfer coefficients for inside and outside flow respectively, W/(m2.oC), Heat Exchangers

k = Thermal conductivity of tube material, W/(m.oC),

R = Total thermal resistance from inside to outside flow oC/W, and
t = Thickness of tube, m.
The thermal resistance R given by Eq. (11.1) can be expressed as an overall heat transfer
coefficient based on either the inside or the outside surface of the tube. It does not matter
on which area it is based as long as it is specified in its definition. For example, the
overall heat transfer coefficient Uo based on the outside surface of the tube is defined as
1 1
Uo   . . . (11.2)
Ao R  Ao   1   Ao  t  1
      
 Ai   hi   Am  k  ho

1
or, Uo  . . . (11.3)
 Do   1   1   Do  1
       Do ln  
 Di   hi   2k   Di  ho
Ao D D
Since  o ln o
Am 2t Di

Do  Di  2t

and Di and Do are the inside and out side diameter of the tube respectively.
Similarly, the overall heat transfer coefficient Ui based on the inside surface of the tube is
defined as
1 1
Ui   . . . (11.4)
Ai R 1  Ai   t   Ai   1 
    
hi  Am   k   Ao   ho 

1
or Ui  . . . (11.5)
1 1  D   Di   1 
   Di ln  o     
hi  2k   Di   Do   ho 
When the wall thickness is small and its thermal conductivity is high, the tube resistance
can be neglected and Eq. (11.4) reduces to
1
Ui  . . . (11.6)
1 1

hi ho

In heat exchanger applications, the heat transfer surface is fouled with the accumulation
of deposits which in turn introduces additional thermal resistance in the path of heat flow.
The effect of fouling is generally introduced in the form of a fouling factor
F (m2.oC/W).
We now consider heat transfer across a tube, which is fouled by deposit formation on
both the inside and outside surfaces. The thermal resistance R in the path of heat flow for
this case is given by
1 F t F 1
R  i   o  . . . (11.7)
Ai hi Ai k Am Ao Ao ho

where Fi and Fo are the fouling factors at the inside and outer tube surface of the tube.
Then Eq. may be represented in terms of the overall heat transfer coefficient based on the
outside surface of the tube as 19
Heat and Mass 1
Transfer Applications Uo  . . . (11.8)
 Do   1   Do   Do  1
       ln    Fo 
 Di   hi   2k   Di  ho

The values of overall heat transfer coefficients for different types of applications vary
widely. Typical ranges of Uo are given in Table 11.2.
Table 11.2 : Typical Values of Overall Heat Transfer Coefficient Uo
for Different Types of Heat Exchangers

Type of Heat Exchangers Uo (W/m2.oC)

Water to oil exchangers 60-350

Gas-to-gas exchangers 60-600

Air conditioners 350-800

Ammonia condensers 800-1400

Steam condensers 1500-5000

11.4.1 Fouling Factor

Mechanism of fouling is very complicated. No reliable technique is available to predict
the fouling factor. Some recommended values are given in Table 11.3.
Table 11.3 : Fouling Factors

Sl. No. Fluid Fouling Factor (m2.K/W)

1. Sea water 0.000088

2. Treated boiler water 0.000180

3. Fuel oil 0.009000

4. Diesel engine exhaust 0.001800

Category of Fouling
Scaling or Precipitating Fouling
Due to the crystallization from solutions of dissolved substance on to the
heat transfer surface.
Particulate Fouling
Accumulation of finely divided solids suspended in the process fluids on to
the heat transfer surface.
Chemical Reaction Fouling
The deposit formation on to the heat transfer surface by chemical reaction.
Corrosion Fouling
Accumulation of corrosion products on to the heat transfer surface.
Biological Fouling
Attachment of micro-organisms to a heat transfer surface.
Solidification Fouling
Crystallization of a pure liquid or one component from the liquid phase on a
sub-cooled heat transfer surface.

20
SAQ 2 Heat Exchangers

(a) What do you mean by fouling?

(b) How does fouling effect the overall heat transfer coefficient of a heat
exchanger?

11.5 HEAT EXCHANGER ANALYSIS

11.5.1 Log Mean Temperature Difference
To design or to predict the performance of heat exchanger, it is essential to relate the total
transfer rate to quantities such as the inlet and outlet fluid temperatures, the overall heat
transfer coefficient, and the total surface area for heat transfer. Two such relations may
readily be obtained by applying overall energy balances to the hot cold fluids. In
particular, if q is the total rate of heat transfer between the hot and cold fluids and there is
negligible potential and kinetic energy changes, application of the steady flow energy
equation gives
q  m h (ih , i  ih , 0 ) . . . (11.9)

and q  m c (ic , o  ic , i ) . . . (11.10)

where i is the fluid enthalpy. The subscripts h and c refer to the hot and cold fluids, where
i and o designate the fluid inlet and outlet conditions. If the fluids are not undergoing a
phase change and constant specific heats are assumed, these expressions reduce to
q  m h c p , h (Th , i  Th , 0 ) . . . (11.11)

and q  mc c p , c (Tc , o  Tc , i ) . . . (11.12)

where the temperature appearing in the expression refer to the mean fluid temperatures at
the designated locations. Another useful expression may be obtained by relating the total
heat transfer rate q to the temperature difference T between the hot and cold fluids,
where
T  Th  Tc . . . (11.13)
Such an expression would be an extension of Newton's law of cooling, with the overall
heat transfer coefficient Uo used in place of the single convection coefficient h. However,
since T varies with position in the heat exchanger, it is necessary to work with a rate
equation of the form
q  U o A Tm . . . (11.14)
where Tm is an appropriate mean temperature difference.
The Parallel Flow Heat Exchanger
The hot and cold fluid temperature distributions associated with a parallel-flow
heat exchanger are shown in Figure 11.7. The temperature difference T is initially
large but decays rapidly with increasing x, approaching zero asymptotically. It is
important to note that, for such an exchanger, the outlet temperature of the cold
fluid never exceeds that of the hot fluid. In Figure 11.7 the subscripts 1 and 2
designate opposite ends of the heat exchanger. This convection is used for all types
of heat exchangers considered. For parallel flow, it follows that
T1, i , Th, o  T1, 2 , Tc,1 . . . (11.15)

Tc , o  Tc , 2 . . . (11.16)

21
Heat and Mass The form of Tm may be determined by applying an energy balance to differential
Transfer Applications
elements in the hot and cold fluids (Figure 11.18). Each element is of length dx and heat
transfer surface area dA.

Ch Th Th + dTh
dq
dA Heat Transfer

CC Tc Surface Area
TC + dTC

dx

Th,i
Th, Ch
dTh

Th, 
T T dq T2
T1
Tc, 

dTc
Tc,i Tc, Cc

1 x 2

Figure 11.18 : Parallel Flow

The energy balance and the subsequent analysis are subjected to the following
assumptions :
(a) The heat exchanger is insulated from its surroundings, in which case the
only heat exchange is between the hot and cold fluids.
(b) Axial conduction along the tubes is negligible.
(c) Potential and kinetic energy changes are negligible.
(d) The fluid specific heats are constant.
(e) The overall heat transfer coefficient is constant.
The specific heats may of course change as a result of temperature variations, and the
overall heat transfer may change because of variations in fluid properties and flow
condition. However, in many applications such variations are not significant, and it is
reasonable to work with average values of cp, c, cp, h and U for the heat exchanger.
Applying an energy balance to each of the elements of Figure 11.18, it follows that
dq   mh c p , h , dTh  C h dTh . . . (11.17)

and dq   mc C p , c , dTc  Cc dTc . . . (11.18)

where Ch and Cc are the hot and cold fluid heat capacity rates, respectively. These
expressions may be integrated across the heat exchanger to obtain the overall energy
balances given by Eqs. (11.11) and (11.12). The heat transfer across the surface area dA
may also be expressed as
dq  U TdA . . . (11.19)

where T = Th – Tc is the local temperature difference between the hot and cold fluids.
To determine the integrated form of Eq. (11.19), we begin by substituting Eqs. (11.17)
and (11.18) into differential form of Eq. (11.13)

22
 d ( T )  dTh  dTc . . . (11.20) Heat Exchangers

 1 1 
to obtain d ( T )   dq    . . . (11.21)
 Ch Cc 
Substituting for dq from Eq. (11.19) and integrating across the heat exchanger,
2 2
d ( T )  1 1 
We obtain  T
 U     dA
 C h Cc  1
. . . (11.22)
1

T2  1 1 
or, ln   UA    . . . (11.23)
T1  C h Cc 
Substituting for and Ch and Cc from Eqs. (11.11) and (11.12), respectively, it follows
that

 T   Th, i  Th, o Tc , o  Tc , i 
ln  2    UA   
 T1   q q 

UA
  (Tc , o  Tc , i )  (Th, i  Th, o )  . . . (11.24)
q  

Recognizing that for the parallel-flow heat exchanger of Figure 11.18,

T1  (Th, i  Tc, i ) . . . (11.25)

and T2  (Th, o  Tc, o ) . . . (11.26)

and we then obtain

( T2  T )1
q  UA . . . (11.27)
 T 
ln  2 
 T1 
Comparing the above expression with Eq. (11.14), we conclude that the appropriate
average temperature difference is a log mean temperature difference, Tm. Accordingly,
we may write
q  UA Tm . . . (11.28)

( T2  T1 ) ( T1  T2 )

where Tm  
 T2   T 
ln   ln  1 
 T1   T2 

 T1  Th,1  Tc,1  Th, i  Tc, i 

  . . . (11.29)
 T2  Th, 2  Tc, 2  Th, o  Tc, o 
The Counter Flow Heat Exchanger
The hot and cold fluid temperature distributions associated with a counter flow
heat exchanger are shown in Figure 11.19. In contrast to the parallel-flow
exchanger, this counter flow Heat Exchanger, configuration provides for heat
transfer between the hotter portions of the two fluids at one end, as between the
colder portions at the other. For this reason, the change in the temperature
difference, T  Th  Tc , with respect to x is nowhere as large as it is for the inlet
region of the parallel flow exchanger. Note that the outlet temperature of the cold
fluid may now exceed the outlet temperature of the hot fluid. 23
Heat and Mass
Transfer Applications
Ch Th Th + dTh
dq
dA Heat Transfer

Tc + DTc Surface Area

CC TC + dC

dx

Th,i
Th, Ch
dTh

Th,o
T T dq T2
T1
Tc,i

Tc, Cc
Tc,o dTc

1 x 2
Figure 11.19 : Counter Flow Heat Exchanger

Eqs. (11.11) and (11.12) apply to any heat exchanger and hence may be used for
the counter flow arrangement. Moreover, from an analysis like that performed in
Section 11.5, it may be shown that Eqs. (11.28) and (11.29) also apply. However,
for the counter flow exchanger the endpoint temperature differences must now be
defined as

T1  Th,1  Tc,1  Th, i  Tc, o

T2  Th, 2  Tc, 2  Th, o  Tc, i . . . (11.30)

Note that, for the same inlet and outlet temperature, the log mean temperature
difference for counter flow exceeds that for parallel flow, Tlm, CF  Tlm, PF .
Hence the surface area required to effect a prescribed heat transfer rate q is smaller
for the counter flow than for the parallel-flow arrangement, assuming the same
value of U. Also note that Tc, o, can be exceed Th, o for counter flow but not for
parallel flow.
Multi Pass and Cross-flow Heat Exchangers

Although flow conditions are more complicated in multi pass cross-flow heat
exchangers, Eqs. (11.9)-(11.12) and (11.28) may still be used if the following
modification is made to the log mean temperature difference

Tlm  F Tlm ,CF . . . (11.31)

That is the appropriate form of Tlm is obtained by applying a correlation factor to

the value of Tlm that would be computed under the assumption of counter flow
conditions. Hence,

T1  Th,1  Tc, o and T2  Th, o  Tc, i . . . (11.32)

Algebraic expressions for the correlation factor F have been developed for various
shell-and tube and cross flow heat exchanger configurations and the results may be
represented graphically. Selected results are shown in Figures 11.20(a)-(d) for
common heat exchanger configurations.

24
 Ti
Ti
to ro
Heat Exchangers
t1

To
ti
1.0
Tc
0.9
1.0
Ti - To 0.8
0.9 6.0 4.0 3.0 2.0 1.0 1.0 0.8 0.6 0.2 R =
to - ti
0.8 0.7

0.7 6.0 4.0 3.0 2.0 1.5 1.0 0.8 0.6 0.4 0.2
0.6
0.6
0.5 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
ti - ti
1.0 P=
Ti - To

(a) Ti (b)
Ti

ti t0
ti t0

To
To

1.0
1.0
0.9
0.9

4.0 3.0 2.0 1.6 1.0 0.8 0.6 0.4 0.2 0.8
0.8
R
R= 0.7
0.7 4.0 3.0 2.0 1.5 1.0 0.8 0.6 0.4 0.2

b 0.6
0.6
0.5 0
0.5 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
P
P=
(c) (d)
Figure 11.20 : Correction Factors for a Shell and Tube Heat Exchanger

The notation (T, t) is used to specify the fluid temperature, with the variable t
always assigned to the tube-side fluid. With this convention it does not matter
whether the hot fluid or the cold fluid flows through the shell or the tubes. An
important implication of Figures 11.20(a)-(d) is that if temperature change of one
fluid is negligible, either P or R is zero and F is 1. Hence heat exchanger behaviour
is independent of the specific configuration. Such would be the case if one of the
fluids underwent a phase change.
SAQ 3
(a) How do you estimate the heat transfer rate for a parallel flow heat exchanger?
(b) Why counter flow heat exchanger is preferred to a parallel flow exchanger?
(c) What is LMTD? When do you use LMTD method?
11.5.2 The Effectiveness-NTU Method
It is a simple matter to use the log mean temperature difference (LMTD) method of heat
exchanger analysis when the fluid inlet temperatures are known and the outlet
temperature are specified or readily determined from the hanger may then be determined.
However, if only the inlet temperatures are known, use of the LMTD method requires to
iterative procedure. In such cases it is preferable to use an alternative approach, termed
the effectiveness-NTU method.

25
Heat and Mass Definitions
Transfer Applications
To define the effectiveness of a heat exchanger, we must first determine the
maximum possible heat transfer rate, qmax, for the exchanger. This heat transfer
rate could, in principle, in a counter flow heat exchanger of finite length. In such
an exchanger, one of the fluids would experience the maximum possible
temperature difference, Th, i  Tc , i .

To illustrate this point, consider a situation for which Cc < Ch, in which case, from
Eqs. (11.17) and (11.18), dTc  dTh . The cold fluid would then be heated to the
inlet temperature of the hot fluid (Tc , o  Th, i ) , Accordingly, from Eq. (11.12),
if Cc < Ch.
qmax  Cc (Th,1  Tc, i ) . . . (11.33)

Similarly, if Ch < Cc
qmax  C h (Th,1  Tc, i ) . . . (11.34)

From the foregoing results we are then prompted to write the general expression
qmax  Cmin (Th,1  Tc, i ) . . . (11.35)

where cmin is equal to Cc and Ch, whichever is smaller. For prescribed hot and cold fluid
inlet temperature, Eq. (11.35) provides the maximum heat transfer rate that could
possibly be delivered by the exchange. A quick mental exercise should convince the
reader that the maximum possible heat transfer rate is not equal to Cmax  (Th,1  Tc, i ) . If
the fluid having the larger heat capacity rate were to experience the maximum possible
temperature change, conservation of energy in the form Cc (Tc , o  Tc , i )  Ch (Th , i  Th , o )
would require that the other fluid experience yet a larger temperature change.
C 
For example, it follows that (Th , i  Th , o )   c  (Th , i  Tc , i ) , in which case
 Ch 
(Th , i  Th , o )  (Th , i  Tc , i ) . . . (11.36)

Such a condition is clearly impossible.

It is now logical to define the effectiveness, , as the ratio of the actual heat transfer rate
for heat exchanger to the maximum possible heat transfer rate
q
 . . . (11.37)
qmax

From Eqs. (11.11), and (11.35), it follows that

Ch (Th ,1  Th , o )
 . . . (11.38)
Cmin (Th,1  Tc, i )

Cc (Th ,1  Tc , i )
 . . . (11.39)
Cmin (Th,1  Tc, i )

By definition the effectiveness, which is dimensionless, must be in the range 0    1. It

is useful because, if , Th, i and Tc, i are known, the actual heat transfer rate may readily
be determined from the expression,  . Cmin (Th, i  Tc, i )

For any heat exchanger it can be shown that

26
 Heat Exchangers
 C 
  f  NTU , min  . . . (11.40)
 Cmax 


Cmin Cc C
where  or h , depending on the relative magnitude of the hot and cold
Cmax C h Cc
fluid heat capacity rates. The number of transfer units (NTU) is a dimensionless
parameter that is widely used for heat exchanger analysis and is defined as,

UA
NTU  . . . (11.41)
Cmin

Effectiveness-NTU Relations
To determine a specific form of the effectiveness-NTU relation 11.40 consider a
parallel-flow heat exchanger for which Cmin = Ch. From Eq. (11.38) we obtain

(Th, i  Th, o )
 . . . (11.42)
(Th, i  Tc, i )

and from Eqs. (11.11) and (11.12) it follows that

Cmin m Cp , h Tc , o  Tc , i
 h  . . . (11.43)
Cmax m c Cp , c Th, i  Th, o

Now consider Eq. (11.23), which may be expressed as

 Th , o  Tc , o  UA  C 
ln   1  min  . . . (11.44)
 Th , i  Tc , i  Cmin  C max 
 

or from Eq. (11.41)

Th , o  Tc , o   C 
 exp  NTU 1  min  . . . (11.45)
Th , i  Tc , i   Cmax 

Re-arranging the left-hand side of this expression as

T  Tc, o   Th, o  Th, i  Th, i  Tc, o 

 h, o    . . . (11.46)
 T  Tc , i   T  T 
 h, i   h, i c, i


And substituting for Tc, o from Eq. (11.43), it follows that

C 
(Th, o  Th, i )  (Th, i  Tc , i )   min  (Th, i  Th, o )
Th, o  Tc , o  Cmax 
 . . . (11.47)
Th, i  Tc , i Th, i  Tc , i

Substituting the above expression into Eq. (11.45) and solving for , we obtain for
the parallel-flow heat exchanger
  C   
1  exp  NTU 1  min 
  Cmax   
 . . . (11.48)
 Cmin 
1  
 Cmax 

27
Heat and Mass Since precisely the same result may be obtained for Cmin = Cc, Eq. (11.48) applies
Transfer Applications for any parallel-flow heat exchanger, irrespective of whether the minimum heat
capacity rate is associated with the hot or cold fluid. Similar expressions have been
developed for a variety of heat exchangers, and representative results are
summarized in Table 11.4, where Cr is the
Table 11.4 : Heat Exchanger Effectiveness Relations

Flow Arrangement Relation

Concentric Tube

Parallel flow 1  exp { NTU [(1  C r )]}


(1  Cr )

Counter flow 1  exp { NTU [(1  C r )]}

 (Cr  1)
1  Cr exp [  NTU (1  Cr )]

NTU
 (Cr  1)
1  NTU

Shell and Tube 1  1

One shell pass
 1
 1  exp { NTU (1  C r ) 2 } 
(2, 4 tube passes)   2  (1  Cr )  (1  Cr2 ) 2 
1 
 2 2 
 1  exp { NTU (1  C r ) }

n shell passes 1
(2n, 4n tube passes)
 1   C  n   1   C  n 
1 r 1 r
     1     C r
 1  1    1  1  
  

Cross Flow (single pass)  1  

0.22
(Both fluids unmixed)    exp   ( NTU ) {1  exp [ Cr ( NTU) 0.78 ]  1}
 Cr  

Cmin (mixed), Cmax   1  exp [ Cr1 (1  exp (  Cr NTU ))]

(unmixed)

All exchangers (Cr = 0)    exp ( NTU )

Cmin
heat capacity ratio Cr  . It is more convenient to work with -NTU
Cmax
relations of the form

 C 
NTU  f   min  . . . (11.49)
 Cmax 

Explicit relations for NTU as a function of and Cr are provided in Table 11.5.

28
 Table 11.5 : Heat Exchanger NTU Relations Heat Exchangers

Flow Arrangement Relation

Concentric Tube

Parallel flow ln [1   (1  Cr )]
NTU  
(1  Cr )

Counter flow 1  (   1) 
NTU  ln   (C r  1)
Cr  1  ( Cr  1) 


NTU  (C r  1)
(  1)

Shell and Tube

1
One shell pass 2   E  1
NTU   (1  C r ) 2 ln  
 E  1

2
(2, 4 tube passes)  (1  Cr )

E 1
1
2 
(1  Cr ) 2

Shell passes 1
(2n, 4n tube passes) F 1   Cr  1  n
1  and F   
F  Cr   1 
Cross flow (single pass)   1 
NTU   ln 1     ln (1   C r )
  Cr  

C (mixed), C (unmixed)  1 
NTU     ln [C r ln (1   )  1]
 Cr 
All exchangers NTU   ln (1   )

SAQ 4
(a) Define effectiveness of a heat exchanger.
(b) What is NTU?
(c) When do you use NTU method?
(d) What is the rating and sizing problem for a heat exchanger?

Example 11.1
Determine the overall heat transfer coefficient Uo based on the outer surface of a
steel pipe with an ID of Di = 2.5 cm and an OD of Do = 3.34 cm [k = 54 W/m.oC)]
for the following flow and fouling conditions :
hi = 1800 W/(m2.oC)
ho = 1250 W/(m2.oC)
Fi = Fo = 0.00018 m2.oC/W

29
Heat and Mass Solution
Transfer Applications
1
Uo 
 Do   1   Do   Do   Do  1
     Fi    ln    Fo 
 Di   hi   Di   (2k )   Di  ho

1
Uo 
(0.742  0.241  0.090  0.18  0.800)  10  3
= 487.1 W/(m2.oC)
Example 11.2
Water at a mean temperature of Tm = 80oC and a mean velocity of um = 0.15 m/s
flows inside a 2.5-cm-ID, thin-walled copper tube. Atmospheric air at T = 20oC
and a velocity of u = 10 m/s flows across the tube. Neglecting the tube wall
resistance, calculate the overall heat transfer coefficient and the rate of heat loss
per 1-m length of the tube.

Water Tm = 80oC; Um = 0.15 m/s D = 2.5 cm

Air Flow U = 10m/s1 T = 200C

Figure 11.21

Solution
The physical properties of water at Tm = 80oC are
v = 0.365  10– 6 m2/s
k = 0.668 W/(m.oC)
Pr = 2.22
The Reynolds number for water flow is
um D 0.15  0.025
Re    10,300
v 0.364  10  6

We use the Dittus-Boelter equation to determine hi for water flow:

Nu  0.023 Re 0.8 Pr 0.3

 0.023 (10,300) 0.8 (2.22) 0.3

= 47.4
k 0.668
hi  Nu  47.4  1267 W/(m 2. oC)
Di 0.025

To evaluate the physical properties of air at the film temperature, the closest
(80  20)
approximation for the film temperature is taken as T f   50o C. Then
2
v = 1.822  10– 6 m2/s
k = 0.0281 W/(m.oC)
Pr = 0.703
30
 The Reynolds number for the air flow becomes Heat Exchangers

u D (10) (0.025)
Re    13,721
v 18.22  10  6
The Nusselt number for the air flow is determine by
2
Nu  (0.4 Re 0.5  0.06 Re 3 ) Pr 0.4
2
0.5
 [0.4 (13,721)  0.06 (13,721) 3 ] (0.7030.4 )
= 70.52
k 0.0281
and ho  Nu  70.52  79.3 W/(m 2. oC)
Do 0.025
By neglecting the tube wall resistance to heat flow and the curvature effect (i.e.
assume Do  Di ), the overall heat transfer coefficient becomes
1 1
U    74.63 W/(m 2 .oC)
1 1 1 1
 
hi ho 1267 79.3

The heat loss per meter length of tube is

Q  AU T

  DU (Ti  To )
 () (0.025) (74.63) (80  20)
= 351.7 W/m
Example 11.3
Engine oil at a mean temperature Ti = 80oC and mean velocity u = 0 m/s flows
inside a thin-walled, horizontal copper tube with an ID of D = 2.5 cm. the outer
surface of the tube dissipates heat by free convection into atmospheric air at
T = 20oC. Calculate the temperature of the tube wall, the overall heat transfer
coefficient, and the heat loss per meter length of tube.
Solution
The physical properties of engine oil at 80oC are
v = 0.375  10– 4 m2/s
k = 0.138 W/(m.oC)
The Reynolds number is
uD 0.1  0.025
Re    66.7
v 0.375  10  4
The flow is laminar. By assuming fully developed flow and a constant tube wall
temperature, the Nusselt number for oil flow is given by
Nu = 3.66
k 0.138
and hi  3.66  3.66  20.2 W/(m 2. oC) . . . (a)
D 0.025
We assume that the thermal resistance for a thin-walled copper tube is negligible.
The free-convection heat transfer coefficient h at the outer surface of the tube can
be calculated from the following simplified expression : 31
Heat and Mass 0.25 0.25
 T  T   T  20 
Transfer Applications h  1.32  w   1.32  w 
 D   0.025 
 3.32 (Tw  20) 0.25 . . . (b)
The tube wall temperature Tw is not known. We consider an overall energy balance
equation with the assumption that Ai  A
hi (Ti  Tw )  ho (Tw  T ) . . . (c)
We introduce the above values of hi and ho and Ti and T into Eq. (c) :
20.2 (80  Tw )  3.32 (Tw  20) 1.25 . . . (d)
The tube wall temperature Tw can now be determined from this equation by
iteration. We find
Tw  62.28o C
Then ho is calculated from Eq. (b) :
ho  8.47 W/(m 2 .oC)
and the overall heat transfer coefficient becomes
1 1
U    5.97 W/(m 2 .oC)
1 1 1 1
 
hi ho 20.2 8.47
The heat loss per meter length of tube is
Q   DU (Ti  T )
 () (0.025) (5.97) (80  20)  28.13 W/m
Example 11.4
In a single-pass shell-and-tube heat exchanger, the inlet and outlet temperatures for
the hot fluid are, respectively, Th, i  260o C and Th, o  140o C ; for the cold fluid
they are Tc , i  70o C and Tc , 0  125o C . Calculate the logarithmic mean
temperature difference for (a) counterflow and (b) parallel-flow arrangements.
Solution
The temperature profiles for the counterflow and parallel-flow arrangements are
illustrated in the sketch.

2600C 2600C

Hot Fluid Hot Fluid

T1=135 0C
T0=190 0C
1400C
140 C
0 TL=150C
1250C 1250C
T0=70 0C
Cold Fluid
Cold Fluid 700C
700C

(a) Counter Flow (b) Parallel Flow

Figure 11.22
(a) For the counterflow
135  70
Tln   99
135
ln
70
32
 (b) For the parallel flow : Heat Exchangers

190  15
Tln   68.9
190
ln
15
Example 11.5
A counterflow shell-and-tube heat exchanger is used to heat water at a rate of
m = 0.8 kg/s from Ti = 20oC to To = 80oC, with hot oil entering at 120oC and
leaving at 85oC. The overall heat transfer coefficient is U = 125 W/(m2.oC).
Calculate the heat transfer area required.
Solution
If we take the specific heat of water as cp = 4180 J/(kg.oC), the total heat load for
the exchanger is
Q  m c p (To  Ti )  (0.8) (4180) (80  30)
= 167,200 W
The temperature profiles in the exchanger are illustrated in the sketch. The
logarithmic mean temperature difference becomes
55  40
Tln   47.1 o C
55
ln
45
1200C
Oil

T1=400C
85 C0

800C
T0=550C

Water
300C

Figure 11.23
The total heat transfer area required is
Q 167, 200
A   28.4 m 2
U Tln 125 (47.1)
Example 11.6
An oil cooler for a large diesel engine is to cool engine oil from 60 to 45oC, using
seawater at an inlet temperature of 20oC with a temperature rise of 15oC. The
design heat load is Q = 140 kW, and the mean overall heat transfer coefficient
based on the outer surface area of the tubes is 70 W/(m2.oC). Calculate the heat
transfer surface area for single-pass (a) counterflow and (b) parallel-flow
arrangements.
Solution
The temperature profiles for the counterflow and parallel-flow arrangements are
shown in the figure. The heat transfer area for each arrangements is now evaluated.
(a) Counter flow
T0  TL  25o C

 Tln  25o C C
Q 140,000
A   80 m 2
Tln U m 25 (70)

33
Heat and Mass
600C
Transfer Applications Oil 600C
Oil
TL =250C 450C
450C
35 C
0 T0=400C TL =100C
T0=250C
350C
Water Water
20 C
0
200C

(a) Counter Flow (b) Parallel Flow

Figure 11.24

(b) Parallel flow

T0  40o C

TL  10o C C
40  10
 Tln   21.64o C
40
ln
10
Q 140,000
A   92.42 m 2
Tln U m 21.64 (70)
We note that less area is required with the counterflow arrangement.
Example 11.7
Engine oil is to be cooled from 80 to 50oC by using a single-pass, counterflow,
concentric-tube heat exchanger with cooling water available at 20oC. Water flows
inside a tube with an ID of Di = 2.5 cm at a rate of mw = 0.08 kg/s, and oil flows
through the annulus at a rate of moil = 0.16 kg/s. The heat transfer coefficients for
the water side and oil side are, respectively, hw = 1000 W/(m2.oC) and
hoil = 80 W/(m2.oC); the fouling factors are Fw = 0.00018 m2.oC/W and
Foil = 0.0018 m2.oC/W; and the tube wall resistance is negligible. Calculate the
tube length required.
Solution
The specific heats of water and oil can be taken as
vw = 4180 J/(kg.oC)
coil = 2090 J/(kg.oC)
The outlet temperature of water can be calculated from an overall energy balance :
Q  moil coil (Th, i  Th, o )  m w c w (Tc, o  Tc, i )

Q  0.16 (2090) (80  50)  0.08 (4180) (T c , o  20)

So Tc , o  50o C
and Q = 10.032 W
The temperature profiles for the hot and cold fluids are shown in the Figure 11.25.
800C

Oil
TL=30 0C

500C 500C

T0=300C
Water
200C

Figure 11.25
34
 We note that the temperature difference between the hot and cold fluids is constant Heat Exchangers
(T = 30o) throughout the heat exchanger. The reason is the equal heat capacity
rates for the hot and cold fluids, that is cw mw = coil moil.
The overall heat transfer coefficient U is determined by neglecting the thermal
resistance of the tube and the curvature effects
1
U 
1 1
 Fw  Foil 
hw hoil

1
  72.2 W/(m 2 .oC)
1 1
 0.00018  0.00018 
1000 80

For T0  TL  30o C , we have Tln  30o C . The total heat transfer area
required becomes
Q 10,032
A   4.63 m 2
Tln U 30 (72.2)
Then the tube length L
A 4.63
L   60 m
 Di  (0.025)
Example 11.8
A shell-and-tube steam condenser is to be constructed of 2.5-cm-OD, 2.2-cm-ID,
single-pass horizontal tubes with steam condensing at Ts = 54oC outside the tubes.
The cooling water enters each tube at Ti = 18oC, with a flow rate of m = 0.7 kg/s
per tube and leaves at To = 36oC. The heat transfer coefficient for the condensation
of stream is hs = 8000 W/(m2.oC). Calculate the tube length L. Calculate the
condensation rate per tube.
Solution
(18  36)
The physical properties of water are taken at  27 o C as
2
cp = 4180 J/(kg.oC)

  0.86  10  3 kg/(m.s)

Pr  5.9
k = 0.61W/(m.oC)
The Reynolds number for flow inside the tube is
4m (4) (0.7)
Re    47,107
 D   (0.022) (0.86  10  3 )

The Dittus-Boelter equation can be used to determine the heat transfer coefficient
for the water side :

Nu  0.023 Re 0.8 Pr 0.4

 0.023 (47,107) 0.8 (5.9) 0.4  2556.2

k 0.61
hi  Nu  256.2  7104 W/(m 2. oC)
D 0.022
The overall heat transfer coefficient based on the outer surface of the tube can be
determined. Neglecting the tube wall resistance, we have 35
Heat and Mass 1 1
Transfer Applications Uo  
 Do   1  1  0.025   1   1 
       
 Di   hi  hc  0.022   7104   8000 

= 3509 W/(m2.oC)
The logarithmic mean temperature difference is
(54  18)  (54  36)
Tln   25.97
 (54  18) 
ln  
 (54  36) 
The tube length L is determined by writing an overall energy balance for a tube :
Qtube  ( Do L ) U o Tln  m c p (To  Ti )

Qtube   (0.025) (L ) (3509) (25.97)  0.7 (4180) (36  18)

L = 7.4 m
The heat transfer rate per tube is
Qtube  52,668 W
The condensation rate per tube is
Qtube 52,668
mtube    2.22  10  2 kg/s
h fg 2,372, 400

Exercise 11.1
(a) Illustrate with sketches the flow path arrangement for the following types of
shell-and-tube heat exchangers :
(i) Single shell pass, two tube pass, counterflow,
(ii) Two shell pass, four tube pass, counterflow, and
(iii) Three shell pass, six tube pass counterflow.
(b) Illustrate with sketches the temperature profiles for hot and cold fluids as a
function of the distance along the flow path for
(i) parallel-flow heat exchangers,
(ii) counterflow exchangers,
(iii) condenser, and
(iv) gas-heated boiler.
(c) Consider a cross-flow heat exchanger with hot and cold fluids entering at
uniform temperatures. Illustrate with sketches the exit temperature
distribution for the following cases :
(i) Both fluids are unmixed, and
(ii) Cold fluid is unmixed, hot fluid is mixed.
(d) Water at Ti = 25oC and a velocity of um = 1.5 m/s enters a brass condenser
tube L = 6 m long, 1.34-cm ID, 1.58-cm OD, and k = 110 W/(m.oC). The
heat transfer for condensation at the outer surface of the tube is
ho = 12,000 W/(m2.oC). Calculate the overall heat transfer coefficient Uo
based on the outer surface.

36
Exercise 11.2 Heat Exchangers

(a) Hot water at a mean temperature Tm = 80oC and with a mean velocity
um = 0.4 m/s flows inside a 3.8-cm-ID, 4.8-cm-OD steel tube
[k = 50 W/(m.oC)]. The flow is considered hydrodynamically and thermally
developed. The outside surface is exposed to atmospheric air at T = 20oC,
flowing with a velocity of u = 3 m/s normal to the tube. Calculate the
overall heat transfer coefficient Uo based on the outer surface of the tube.
(b) Engine oil at Tin = 50oC and a mean velocity of um = 0.25 m/s enters a brass
[k = 110 W/(m.oC)] horizontal tube with Di = 2.22 ID and t = 0.17 cm thick.
Heat is dissipated from the outer surface by free convection into an ambient
at T = 20oC. Calculate the overall heat transfer coefficient Uo based on the
tube’s outer surface.
(c) Determine the overall heat transfer coefficient Uo based on the outer surface
of a brass tube with Di = 2.5 cm and Do = 3.34 cm [k = 110 W/(m.oC)] for
the following conditions : The inside and outside heat transfer coefficients
are, respectively, hi = 1200 and ho = 2000 W/(m2.oC); the fouling factors for
the inside and outside surfaces are Fi = Fo = 0.00018 m2.oC/W.
Exercise 11.3
(a) Engine oil at Ti = 40oC at a rate of m  0.2 kg/s enters a tube with
Di = 2.5 cm ID which is maintained at a uniform temperature Tw = 100oC by
condensing steam outside. Calculate the tube length required to have an
outlet temperature To = 80oC.
(b) A counterflow heat exchanger is to be used to heat mc = 2.5 kg/s of water
from Tc, in = 20oC to Tc, out = 80oC by using hot exhaust gas
[cp = 1000 J/(kg.oC)] entering at Th, in = 220oC and leaving at Th, out = 90oC.
The overall heat transfer coefficient is Um = 250 W/(m2.oC). Calculate the
heat transfer surface required.
(c) A single-pass, counterflow, shell-and-tube heat exchanger is used to heat
water from Tc, in = 15oC to Tc, out = 80oC at a rate of mc = 1.5 kg/s by oil
entering the shell side at Th, in = 140oC and leaving at Th, out = 90oC. The
overall heat transfer coefficient is Um = 250 W/(m2.oC). Calculate the heat
transfer surface required.
Exercise 11.4
(a) A shell-and-tube heat exchanger is to cool m = 6 kg/s of oil
[cp = 2000 J/(kg.oC)] from Th, in = 65oC to Th, out = 35oC by using mc = 10
kg/s of water at an inlet temperature of Tc, in = 20oC. The average heat
transfer coefficient is Um = 600 W/(m2.oC). Calculate the heat transfer
surface required for a (i) parallel-flow heat exchanger, and (ii) a counterflow
heat exchanger. [Water : Cp = 4200 J/ kg.oC].
(b) Engine oil at a mean temperature Ti = 80oC and mean velocity um = 0.2 m/s
flows inside a thin-walled, horizontal copper tube with an ID of D = 1.9 cm.
At the outer surface, atmospheric air at T = 15oC and a velocity of
u = 5 m/s flows across the tube. Neglecting the tube wall resistance,
calculate the overall heat transfer coefficient and the rate of heat loss to the
air per meter length of tube.
(c) A counterflow heat exchanger is to cool mh = 1 kg/s of water from 65 to 5oC
by using a refrigerant [cpc = 920 J/(kg.oC)] entering at Tc, in = 20oC with a
flow rate of mc = 8 kg/s. The overall heat transfer coefficient is
Um = 1500 W/(m2.oC). Calculate the heat transfer surface required.
[cpw = 4180 J/(kg.0C)].

37
Heat and Mass
Transfer Applications
11.6 SUMMARY
Present unit is devoted to heat exchangers. Classification of heat exchangers and
applications are elaborately discussed. Different types of heat exchangers are also
described. Heat transfer mechanism in heat exchangers and method for determination
heat flow are discussed. One important aspect with old heat exchangers is the phenomena
of fouling. Importance of fouling and its effect in prediction of heat transfer coefficient is
also described. Method for finding overall heat transfer coefficient is given. Heat transfer
calculation by LMTD and effectiveness-NTU method are given in details. After studying
this unit you will be able to formulate and solve problems on heat exchangers.

11.7 KEY WORDS

Heat Exchanger : Devices that facilitate heat transfer between two or
more fluids at different temperatures.
Compact Heat Exchanger : A heat exchanger having surface to volume ratio
more than 700 m2/m3 is known as compact heat
exchanger.
Recuperator : Heat exchanger where the fluids are separated by a
heat transfer surface and ideally they do not mix.
Regenerator : Heat exchangers in which there is an intermittent
flow of heat from hot to cold fluid via heat storage
and heat rejection through the exchanger surface
or matrix.
Parallel Flow Heat Exchanger : The hot and the cold fluid enter at the same end of
the heat exchanger, flow through in the same
direction, and leave at the other end.
Counter Flow Heat Exchanger : The hot and cold fluids enter in the opposite ends
of the heat exchanger and flow in opposite
directions.
Cross Flow : The two fluids usually flow at right angles to each
other.

11.8 ANSWERS TO SAQs

Refer the preceding text for all the Answers to SAQs.

38
 Heat Exchangers
REFERENCES
Frank Kreith and Mark S. Bohn (1997), Principles of Heat Transfer, 5th Edition, PWS
Publishing Company, Boston.
E. R. G. Eckert and R. M. Drake (1959), Heat and Mass Transfer, McGraw-Hill,
Kogakusha.
J. P. Holman (2002), Heat Transfer, 9th Edition, Tata McGraw-Hill, New Delhi.
E. Achenbach (1989), Heat Transfer from a Staggered Tube Bundle in Cross-flow at
High Reynolds Numbers, International Journal of Heat and Mass Transfer, Volume 32,
pp. 271-280.
E. N. Sieder and C. E. Tate (1936), Heat Transfer and Pressure Drop of Liquids in Tubes,
Ind. Eng. Chem., Volume 28, pp. 1429.
C. A. Depew and S. E. August (1971), Heat Transfer due to Combined Free and Forced
Convection in a Horizontal and Isothermal Tube, Trans. ASME Ser., C, J. Heat Transfer,
Volume 93, pp. 380-384.
W. M. Kays and A. L. London (1984), Compact Heat Exchangers, 3rd Edition,
McGraw-Hill, New York.
B. S. Petukhov (1970), Heat Transfer and Friction in Turbulent Pipe Flow with variable
Physical Properties, Advances in Heat Transfer, Academic Press, New York,
pp. 504-564.
H. Hausen (1983), Heat Transfer in Counter Flow, Parallel Flow and Cross Flow,
McGraw-Hill, New York.
M. N. Ozisik (1985), Heat Transfer – A Basic Approach, McGraw-Hill International
Edition.
F. P. Incropera and D. P. Dewitt (2004), Fundamentals of Heat and Mass Transfer,
5th Edition, John Wiley and Sons.
P. K. Nag (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.

39
 Boilers
UNIT 12 BOILERS
Structure
12.1 Introduction
Objectives
12.2 Mechanism of Boiling Heat Transfer
12.2.1 Free Convection Boiling
12.2.2 Nucleate Boiling Regime
12.2.3 Film Boiling Regime
12.3 Pool Boiling Correlations
12.3.1 Free Convection Regime
12.3.2 Nucleate Boiling Regime
12.3.3 Peak Heat Flux
12.3.4 Film Boiling Regime
12.4 Forced Convection Boiling Inside Tubes
12.4.1 Low Uniform Heat Flux along the Tube
12.4.2 High Surface Heat Flux
12.5 Heat Transfer Relations
12.5.1 Regime A
12.5.2 Regime B
12.6 Condensation
12.6.1 Film Condensation
12.6.2 Reynolds Number for Condensate Flow
12.6.3 Dropwise Condensation
12.7 Applications of Boiling Heat Transfer
12.8 Applications of Condensation Heat Transfer : Steam Condensers
12.9 Summary
12.10 Key Words
12.11 Answers to SAQs

12.1 INTRODUCTION
In the present unit, boiling and condensation heat transfer phenomena are discussed. Both
the boiling and condensation heat transfer are complicated phenomena and considerable
work has been directed towards gaining a better understanding of the boiling and
condensation heat transfer. This is mainly because of large number of variables involved
with the phenomena and complexity in their hydrodynamics. You will notice that most of
the heat transfer correlations of boiling and condensation are still remaining in empirical
and semi empirical form. In the last part of the unit, application of the boiling and
condensation phenomena are described in brief.
Objectives
After study of this unit, you should be able to
 understand the mechanism of boiling and condensation,
 understand different regimes of boiling heat transfer,
 realise the various boiling heat transfer correlations,
 understand the application of boiling heat transfer to steam generators,
 identify various types of boilers, their uses and salient features,
 classify condensation heat transfer and different types of condensers, and 41
Heat and Mass  design a surface condenser.
Transfer Applications

12.2 MECHANISM OF BOILING HEAT TRANSFER

When a liquid is in contact with a surface maintained at a temperature above the
saturation temperature of the liquid, boiling may occur. Mechanism of boiling may be
better explained with the help of pool boiling phenomena. Nukiyama was the first
investigator who experimentally established the pool boiling phenomena. In his
experiments, he immersed an electric resistance wire into a body of saturated water and
initiated boiling on the surface of the wire by passing electric current through it
Figure (12.1). He determined both the heat flux and temperature from the measurements
of current and voltage.

T Computer

D A
CCD Camera D
A

Light

Heater Array
Window
Feed Back
Decoder
Control

Figure 12.1 : Experimental Setup for Determination of Pool Boiling Phenomena

Characteristics of pool boiling for water at atmospheric pressure is shown in the
Figure 12.2. The boiling curve describes the variation of the heat transfer coefficient or
the heat flux as a function of temperature difference between the wire and water
saturation temperatures. The boiling regime can be divided into 3 (three) broad regimes :
(a) Free-convection regime
(b) Nucleata boiling regime
(c) Film boiling regime
T = TW – Ti,OC

0.1 1.0 10 100

1000

Nucleate Film
Free Convection
II III IV V VI
Nucleate Boiling – Bubbles

I
Log h (h = Heat transfer Coefficient)

boiling and unstable

nucleate film boiling
Nucleate boiling
bubbles rise

Partial nucleate
superheated liquid

Radiation coming
to interface

Stable film boiling

condense in

into play
Spherical State
Beginning

Boiling Curve

0 1.0 10 100 1000

T = TW – TKOF

Figure 12.2 : Characteristic Pool Boiling Curve

42
12.2.1 Free Convection Boiling Boilers

In this regime, the energy transfer from the heater surface to the saturated liquid take
place by free convection and heat transfer rate can be predicted as discussed in Units 7
and 8. The surface is only a few degree above the saturation temperature of the liquid,
but the free convection currents produced in the liquid are sufficient to remove the heat
from the surface.
12.2.2 Nucleate Boiling Regime
The nucleate boiling regime in which the bubbles are formed on the surface of the heater
can be separated into two distinct regions. In the region designated II, bubbles start to
form at the favoured sites on the heater surface, but as soon as the bubbles are detached
from the surface, they are dissipitated into the liquid. In region III, the nucleation sites are
numerous and the bubble generation rate is so high that continuous columns of vapour
appear. As a result, very high heat fluxes are obtainable in this region. In practical
applications, the nucleate boiling regime is most desirable, because large heat fluxes are
obtainable with small temperature differences. In the nucleate boiling regime, the heat
flux increases rapidly with increasing temperature difference until the peak heat flux is
reached. The location of this peak heat flux is called the burnout point, or departure from
nucleate boiling (DNB), or the critical heat flux (CHF). The reason for calling the peak
heat flux the burnout point is apparent from Figure 12.2. As soon as the peak heat flux is
exceeded, an extremely large temperature difference is needed to realize the resulting
heat flux. Such high temperature difference may cause the burning up, or melting away,
of the heating element.
12.2.3 Film Boiling Regime
Once the peak heat flux is reached, any further increase in temperature difference causes
a reduction in the heat flux. The reason for this curious phenomenon is the blanketing of
the heater surface with a vapour film which restricts liquid flow to the surface and has a
low thermal conductivity. The film boiling can be separated into three distinct regions
such as unstable film boiling region, stable film boiling region and radiation film boiling
region.
In the unstable film boiling region (Region IV in the Figure 12.2), the vapour film is
unstable, collapsing and reforming under the influence of convective current and surface
tension. Heat flux decreases as the surface temperature increases because the average
wetted surface area of the heater surface decreases.
In the stable film boiling region (Region V in Figure 12.2), the heat flux drops to a
minimum because of a continuous vapour film covers the heater surface.
In the region VI in Figure 12.2, the heat flux begins to increase as the temperature
difference increases, because the temperature at the heater surface is sufficiently high for
thermal radiation effects. This enhances the heat transfer through the vapour film.
SAQ 1
(a) What do you mean by pool boiling?
(b) What are the different regimes of pool boiling?
(c) Sketch a plot for identifying different regimes of pool boiling.
(d) What is critical heat flux?

43
Heat and Mass
Transfer Applications
12.3 POOL BOILING CORRELATIONS
12.3.1 Free Convection Regime
In the free convection regime, the heat transfer take place by free convection. The heat
transfer correlations are in the form
Nu  f (Gr , Pr) . . . (12.1)
Various correlations are already provided to determine the heat transfer coefficient h in
the earlier unit. Once the heat transfer coefficient h is obtained, the heat flux for the free
convection regime is determined from
q  h (Tw  Tsat ) . . . (12.2)
where Tw and Tsat are the temperatures of heat transfer surface and saturated water
respectively.
12.3.2 Nucleate Boiling Regime
The nucleate boiling regime involves two separate processes : the formation of bubbles
on the surface referred to as nucleation, and subsequent growth and motion of these
bubbles.
The process of nucleation is very complicated. Several theories have been proposed as to
how the bubbles first form on the surface. Still there is controversy on the subject. The
distribution of active nucleation sites on the surface also must be known for the
development of theoretical models of nucleate boiling. Once the nucleation process is
completed, further heat transfer from the surface to the bubble promotes bubble growth.
Therefore, considerable effort has been expended to understand the mechanism of bubble
growth. The bubble departure size is also another important parameter that affects heat
transfer in the nucleate boiling regime.
2 4 6 8 10
2x106
5x104
q peak

105 1205 (83.5) 3x106

1603 (110.5)

1985 136.9)
P=14.7 psia (1.0 bar)
P=2465 psia (170 bar)

770 (53.1

383 (26.4)

106
Heat flux q, W/m2
q, Bru/(h-ft2

5x105

105
3x105

103

5x104

104 3x104
2 4 6 8 10 20 40 60
T = Tw – Ts oF
Figure 12.3 : Heat Flux for Boiling of Saturated Water on a Horizontal,
0.61 mm Diameter Platinum Wire
44
Clearly the heat transfer in the nucleate boiling regime is affected by the nucleation Boilers
process, the distribution of active nucleation sites on the surface, and the growth and
departure of bubbles. Numerous experimental investigations have been reported for
nucleate boiling phenomenon. Figure 12.3 shows some typical experimental data for the
heat flux q plotted as a function of T at various pressure levels for boiling of water on a
horizontal, 0.061 cm diameter electrically heated platinum wire. The dashed lines denote
the location of the peak heat flux in nucleate boiling. Many correlations are reported to
estimate the heat flux. One of the most widely used correlation was established by
Rohsenow [1] as given below :
0.33
c pl T  q g c * 
 Csf   . . . (12.3)
h fg Prln  l h fg g (l  v ) 
 
where cpl = Specific heat of saturated liquid, J/kg.oC,
Csf = Constant to be determined from experimental data depending upon heating
surface-fluid combination,
hfg = Latent heat of vaporization, J/kg,
g = Acceleration due to gravity, m/s2,
gc = gravitational acceleration conversion factor, 1 kg.m/N.s 2 (MKS), unity (SI),
c pl l
Prl   Prandtl number of saturated liquid,
kl
q = Boiling heat flux, W/m2,
T  Tw  Tsat  Temperature difference between wall and saturated water, oC,
l = Viscosity of saturated liquid, kg/m.s,
l = density of saturated liquid, kg/m3,
v = Density of saturated vapour, kg/m3, and
* = Surface tension of liquid-vapour interface, N/m.
Table 12.1 presents the experimentally determined values of Csf for a variety of
liquid-surface combinations. The value of n should be taken as 1 for water and 1.7 for
all other liquids for the liquids presented in the Table 12.1.
Table 12.1 : Experimentally Determined values of Csf for
various Liquid-surface Combinations
Sl. No. Liquid-surface Combination Csf
1. Water-copper 0.0130
2. Water-scored copper 0.0068
3. Water-emery polished copper 0.0128
4. Water-emery polished, paraffin treated copper 0.0147
5. Water-chemically etched stainless steel 0.0133
6. Water-mechanically polished stainless steel 00.0132
7. Water-ground and polished stainless steel 0.0080
8. Water-Teflon pitted stainless steel 0.0058
9. Water-platinum 0.0130
10. Water-brass 0.0060
11. Benzene-Chromium 0.0100
12. Ethyl alcohol-chromium 0.0027
13. Carbon tetrachlolride-copper 0.0130
14. n-Pentane-emergy-polished copper 0.0154
12. n-Pentane-emergy-polished nickel 0.0127
16. n-Pentane-emergy-rubber copper 0.0074
17. n-Pentane-lapped copper 0.0049
Table 12.2 gives the values of the vapour-liquid surface tension for a variety of liquids. 45
Heat and Mass Table 12.2 : Surface Tension for various Liquids
Transfer Applications
Sl. No. Liquid Saturation Surface Tension
Temperature oC *  103 N/m
1. Water 0 75.6
2. Water 12.56 73.2
3. Water 37.78 69.7
4. Water 93.34 60.1
5. Water 100 58.8
6. Water 160 46.1
7. Water 226.7 31.9
8. Water 293.3 16.2
9. Water 360 1.46
10. Water 374.11 0
11. Sodium 881.1 11.2
12. Potassium 760 62.7
13. Rubidium 687.8 43.8
14. Cesium 682.2 29.2
12. Mercury 357.2 39.4
16. Benzene 80 27.7
17. Ethyl alcohol 78.3 21.9
18. Freon 11 44.4 8.5

100

+
10
q gc α

μl h g(ρl  ρ v )
fg +

+
1.0 Atm
1
26
52.4
82
+ 109
167.7
109

0.1
0.01 0.1
Cpt T/(hfg Pr1.0)
Figure 12.4 : Rohsenow’s [1] Correlation of Addom’s [2] Data for Boiling of Water

Figure 12.4 shows the correlation of Addom’s experimental data given in Figure 12.3
with Eq. (12.3) for water with Csf = 0.013 over a wide range of conditions. It shows a
deviation in heat flux as much as 100%, but the typical error in T is about 25%. The
reason is that according to Eq. (12.3), we have q ~ T3, as a result, a large error in heat
flux corresponds to a small error in T.
46
Example 12.1 Boilers

Saturated water at Tv = 100oC is boiled with a copper heating element having a

heating surface A = 1.5  10– 2 m2 which is maintained at a uniform temperature
Tw = 115oC. Calculate the surface heat flux and the rate of evaporation.
Solution
Using Eq. (12.3) to determine heat flux q
0.33
c pl T  q g c * 
 Csf   . . . (12.4)
h fg Prln  l h fg g (l  v ) 
 
In this equation gc = 1 as we are dealing with SI system of units.
For water, exponent n = 1
For water-copper, Csf = 0.013 (refer Table 12.2)
Physical properties of saturated water and vapour are taken as
cpl = 4216 J/kg.oC
hfg = 2257  103 J/kg
l = 960.6 kg/m3
v = 0.60 kg/m3
Prl = 1.74
l = 0.282  10– 3 kg/m. s
* = 58.8  10– 3 N/m

T  Tw  Tsat  15o C
Substituting the numerical values in Eq. (12.4), we get
0.33
4216  15  q 58.8  10  3 
3
 0.013   
(2257  10 )  1.74  (0.287  10 )  (2257  103 )
3 (9.8)  (960.6  0.60) 

Then the heat flux becomes
q = 4.84  105 W/m3
The total heat transfer is

Q  q  area  (1.5  10  2 )  (4.84  10 5 )  7260 kW

Rate of evaporation becomes
Q 7260
M    3.216  10  3
h fg 2257  103

12.3.3 Peak Heat Flux

Determination of peak heat flux is of interest because of the burnout considerations. If
the applied heat flux exceeds the peak heat flux, the transition takes place form nucleate
to the stable film boiling regime in which, depending on the kind of fluid, boiling may
occur at temperature differences well above the melting point of the heating surface.
The following correlation proposed by Zuber [3], Zuber and Tribus [4] may be used to
estimate the peak heat flux
1 1
  * g g c (l  v )  4  v  2
qmax   v h fg   1   . . . (12.5)
24  v2   l  47
Heat and Mass where * = Surface tension of liquid-vapour interface, N/m,
Transfer Applications
g = Gravitational acceleration, m/s2,
gc = Gravitational acceleration conversion factor, 1 kg.m/Ns2 (MKS),
unit (SI),
l, v = Density of liquid and vapour respectively, kg/m3,
hfg = Latent heat of vaporization, J/kg, and
qmax = Peak heat flux, W/m2.
Although Eq. (12.5) correlates well selected sets of data, it is not accurate for all systems.
For example, in case of liquid oxygen, Eq. (12.5) underestimates the peak heat flux. Also
the heater geometry is another parameter that affects the peak heat flux.
Lienhard re-examined the correlation Eq. (12.5) in order to accommodate the effects of
heater geometry and size. For the peak heat flux they proposed the following modified
form of Eq. (12.5).
1 1
qmax  F (L )  0.131  v2 h fg  * g g c ( l   v )  4 . . . (12.6)
 
where F (L) is a correction factor for the effects of heater geometry and size. The factor
F (L) depends on the dimensionless characteristics length L of the heater defined as

g (l  v )
L  L . . . (12.7)
*
where L is the characteristics length of the heater.
In both the equations, the physical properties of the vapour should be evaluated at
1
Tf  (Tw  Tsat ) . . . (12.8)
2
The enthalpy of evaporation hfg and liquid properties should be evaluated at the saturated
temperature of the liquid.
In Table 12.3 recommended values of the factor F (L) for heater geometries such as flat
plate, a horizontal cylinder and a sphere are provided.
Table 12.3

Sl. No. Heater Geometry F (L) Remarks

1. Infinite plate facing up 1.14 L  2.7; L is the heater
width of diameter
2. Horizontal cylinder  L  0.15; L is the cylinder
0.89  2.27 e  3.44 L
radius
3. Large sphere 0.84 L  4.26; L is the radius of
the sphere
4. Small sphere 1.734 0.15  L  4.26; L is the
1 sphere radius
( L) 2

5. Large finite body 0.90 (Approx). volume

L  4; L 
surface area
Example 12.2
Water at atmospheric pressure and saturation temperature is boiled in a 25-cm
diameter, electrically heated, mechanically polished, stainless steel pan. The heater
surface of the pan is maintained at a uniform temperature Tw = 116oC. Calculate :
48
 (a) The surface heat flux Boilers

(b) The rate of evaporation from the pan

(c) The peak heat flux
Solution
The physical properties of saturated water and vapour are taken as
cpl = 4216 J/kg.oC
hfg = 2257  103 J/kg
l = 960.6 kg/m3
v = 0.60 kg/m3
Prl = 1.74
l = 0.282  10– 3 kg/m. s
* = 58.8  10– 3 N/m
T  Tw  Tsat  16o C
(a) Surface heat flux is calculated by applying the equation
0.33
c pl T  q g c * 
 Csf  
h fg Prln  l h fg g (l  v ) 
 
For water, n = 1.
For water-mechanically polished stainless steel, Csf = 0.0132. Now
substituting the numerical values in the above equation,
0.33
4216  15  q 58.8  10  3 
 0.013   
3
(2257  10 )  1.74  (0.287  10 )  (2257  103 )
3 (9.8)  (960.6  0.60) 


or, surface heat flux q  5.61  105 W/m 2

(b) Total heat transfer is
 
Q  area  q   0.25 2  (5.61  10 5 )  0.275  10 5 W
 4 
The rate of evaporation becomes
Q 0.275  10 5
M    3600  43.9 kg/h
h fg 22.57  10 5

(c) Calculation of peak heat flux

F ( L )  1.14
1 1
Now, qmax  F (L )  0.131  v2 h fg  * g g c ( l   v )  4
 
1 1
we get, qmax  1.14  0.131  v2 h fg  * g g c ( l   v )  4
 
which will be valid for
g (l  v )
L  L  2.7
*

49
Heat and Mass For, L = 0.25 m and other quantities as given above, we have L = 100 which
Transfer Applications
is larger than the specified lower bound 2.7. Hence, the above equation for
qmax is applicable. Introducing the numerical values yields the peak heat flux

qmax  1.14  0.131  (0.6) 0.5  (2257  10 3 )

1
3
 [58.8  10  9.8  (960.6  0.6)] 4  1.27 MW/m 2

It may be noted that the surface heat flux q = 5.61  105 W/m2 is well below
the peak heat flux qmax = 1.27 MW/m2.
12.3.4 Film Boiling Regime
Unstable film boiling region begins after the peak heat flux is reached. No analysis is
available for the prediction of heat flux in this region as a function of temperature
difference Tw – Tsat till a minimum heat flux position is reached. After the minimum heat
flux position, there will be stable film boiling. Stable film boiling has numerous
applications in the boiling of cryogenic fluids. A theory was developed by Bromley [81]
for the prediction of heat transfer coefficient for stable film boiling on the outside of a
horizontal cylinder. The resulting equation for the average heat transfer coefficient ho for
stable film boiling on the outside of a horizontal cylinder in the absence of radiation is
given by
1
 kv3 v (l  v ) g h fg  0.4c pv T  4
h0  0.62  1   . . . (12.9)
  v D0 T  h fg 
 
where ho = Average boiling heat transfer coefficient in absence of radiation, W/m2.oC,
cpv = Specific heat of saturated vapour, J/kgoC,
Do = Outside diameter of the tube, m,
g = Acceleration due to gravity, m/s2,
hfg = Latent heat of vaporization, J/kg,
kv = thermal conductivity of saturated vapour, W/m.oC, and
T  Tw  Tsat  Temperature difference between wall and saturated liquid, oC.

1
In Eq. (12.9) the physical properties of vapour should be evaluated at T f  (Tw  Tsat )
2
and enthalpy of evaporation hfg and the liquid density l should be evaluated at the
saturation temperature Tsat of the liquid.
Eq. (12.9) has been derived by assuming that heat transfer across the vapour film is by
pure conduction; therefore, it does not include radiation effects. Bromley [81] suggested
that when the surface temperature is sufficiently high for radiation effects to be important,
the average heat transfer coefficient hm can be determined from the following empirical
relation
1
 h 3
hm  h0  0   hr . . . (12.10)
 hm 
where h0 is the boiling heat transfer coefficient given by Eq. (12.9) without radiation
effects and hr is the radiation heat transfer coefficient estimated approximately from

1  (Tw4  Tsat
4
)
hr  . . . (12.11)
1 1
  1 Tw  Tsat
 
where  = Absorptivity of liquid,
50
  = Emissivity of hot tube, Boilers

Tw = Wall temperature, and

Tsat = Saturation temperature of liquid.
Eq. (10.39) is difficult to use because a trial-and-error approach is needed to determine
hm. When hr is smaller than h0, Eq. (12.10) may be replaced by
3
hm  h0  hr . . . (12.12)
4
0 500 T, oC 1000 1500
50

Experiment
250
Computed
40

200

Heat transfer coefficient

hm
30
Btu / (h.ft2,oF

w / (m2.s)
150
ho
20

100

10
hr
50

0 0
0 1000 2000
T,oF

Figure 12.5 : Heat Transfer Coefficient for Stable Film Boiling of Liquid Nitrogen
on a Electrically Heated 0.89 cm Diameter Carbon Tube

12.4 FORCED CONVECTION BOILING INSIDE TUBES

In the previous sections we considered boiling on a heated surface immersed in a
quiescent mass of liquid. If boiling takes place on the inside surface of a heated tube
through which the liquid flows with some velocity, boiling is called forced convection
boiling. Boiling of liquids in forced flow inside heated tubes has numerous applications
in the design of steam generators for fossil fuel as well as nuclear power plants and
various other power generating systems. Since the velocity inside the tube affects the
bubble growth and separation, the mechanism and hydrodynamics of boiling in forced
convection are much more complicated than in the pool boiling of a quiescent liquid. In
addition to the various boiling regimes considered in pool boiling, there are many other
boiling regimes in forced convection boiling inside the tube. The problem is complicated
further by the fact that the transition between the regimes should be understood as well.
To illustrate various flow regimes involved in forced-convection boiling inside tubes, we
consider the upward flow of water in a long vertical tube uniformly heated over its length
(Figure (12.6)). We examine separately the cases involving a low heat flux and a
progressively increasing heat flux imposed on the tube.

51
 Flow Heat Transfer
Heat and Mass patterns Region
Transfer Applications Single Phase Convective heat
Temperature H Vapor transfer treatment
profiling

Quality Liquid deficient

x=1 G Drop Flow region

Vapor Core Dry out Paint

temperature

Dry Out Point Annular flow with

F entrainment Two phase forced
convective heat
Fluid Temperature transfer through
Wall liquid film
Temperature

E Annular flow

Saturated nuclear
D boiling
Liquid Core Shag flow
temperature
C
Bubbly flow Subcooled boiling
x=o B
quality

A Single Phase Convective heat

Liquid transfer to liquid
Flow in

Figure 12.6 : Various Flow and Heat Transfer Regimes in Forced Convection Inside
a Vertical Tube Subjected to Uniform Heat Flux

12.4.1 Low Uniform Heat Flux along the Tube

Figure 12.7 illustrates qualitatively the various flow patterns and the corresponding heat
transfer regimes associated with the upward flow of subcooled liquid in a vertical tube
subjected to a uniform heat flux over its length.
Heat Flux

Subcooled Saturated Superheated

Subcooled Saturated
film boiling film boiling

Locus of
burnout
Subcooled Saturated
Convective
burnout burnout
Heat transfer to
vapor
Subcooled
nucleate Liquid
boiling region B Saturated deficient
nucleate region G
boiling regions
C and D
Convective heat Dryout
transfer to liquid Two phase forced
region A convection heat transfer
region E and F
Distance along
the tube
Quality Quality
x=0 x=1

Figure 12.7 : Effects of Increasing Heat Flux on Heat Transfer Regimes in Forced
Convection Inside a Vertical Tube Subjected to Uniform Heat Flux

The subcooled liquid entering the tube is first heated by the mechanism of forced
convection heat transfer to single phase liquid, in region A. In this region, the difference
between the wall and the bulk fluid temperatures remains relatively constant.
52
Region A is followed by the subcooled nucleate boiling region B. In the subcooled Boilers
nucleate boiling regime, the bubbles form and collapse, therefore, vapour quality remains
zero (x = 0) until the end of region B. In this region, the temperature difference between
the wall and the bulk fluid decreases linearly upto the point where
x = 0. This implies gradual increase in the heat transfer coefficient since the wall heat
flux is constant.
In the saturated nucleate boiling regime, regions C and D, the temperature difference and
hence the heat transfer coefficient, remain constant.
In the two-phase forced convection regime, regions E and F, the temperature difference
decreases and the heat transfer coefficient increases with the distance along the tube. The
reason is the reducing thickness of the liquid film as the vapour velocity increases.
At the dry out point, there is a sudden rise in the temperature difference with a
corresponding decrease in heat transfer coefficient. Beyond the dry out point, the heat
transfer coefficient is equal to that expected for heat transfer by forced convection to
saturated steam.
In the liquid-deficient regime, region G, the vapour quality continuously increases and
the temperature difference decreases with the corresponding increase in the heat transfer
coefficient.
12.4.2 High Surface Heat Flux
Suppose the flow rate of liquid in the tube remains constant, but higher heat fluxes are
applied to the wall surface. Figure 12.7 provided by Collier illustrates various other heat
transfer regions that are encountered as the surface heat flux is increasing progressively.
In Figure 12.7, the axial coordinate represents the distance along the tube length, and the
transverse coordinate, the wall heat flux. The heat transfer regime encountered along the
tube at low heat fluxes is similar to that illustrated in Figure 12.6. As the heat flux is
progressively increased, other heat transfer regimes may occur at a given location. For
example, with high heat flux, the departure from nucleate boiling may occur even in the
subcooled region.

12.5 HEAT TRANSFER RELATIONS

12.5.1 Regime A
This regime corresponds to forced-convection heat transfer to liquid only. In this regime,
flow of subcooled fluid may be in laminar or in turbulent flow. In either case, heat
transfer correlationship may be used as described in Unit 7 and/or Unit 8.
For example, for fully developed turbulent flow, the Dittus-Boelter equation may be used
hm D
Nu   0.023 Re 0.8 Pr 0.4 . . . (12.13)
k
L
where  60 and Re  10,000 . Once the heat transfer coefficient is evaluated from
D
Eq. (top), the heat flux in the tube for the heat transfer regime A is determined from
q  hm (Tw  Tb ) . . . (12.14)
where Tw is the tube surface temperature and Tb is the bulk fluid temperature.
12.5.2 Regime B
Once the boiling is initiated, the whole surface eventually becomes covered with bubbles
and the boiling becomes fully developed. In the subcooled, fully developed boiling, the
surface temperature is essentially a function of the surface heat flux and the system
pressure for a given fluid. With this consideration, equations of the same form as that
used for pool boiling can be suitable for correlating forced-convection fully developed,
subcooled boiling data.
53
Heat and Mass Therefore, heat transfer for the subcooled boiling regime B, the forced-convection, fully
Transfer Applications developed boiling can be correlated as
0.33
c pl T  q g c * 
 Csf   . . . (12.15)
h fg Prln  l h fg g (l  v ) 
 
where n = 1 for water, and
n = 1.7 for other liquids.
Recommended values of Csf in Eq. for different configurations are given in Table 12.4.
Table 12.4 : Values of Csf of Eq. (12.15) for Forced Convention Boiling
[Refer Ozisik]
Geometry Liquid-surface Combination Csf
Horizontal tube (14.9 mm I.D) Water-stainless steel 0.015
Horizontal tube (2.39 mm I.D) Water-stainless steel 0.020
Vertical tube (4.56 mm I.D) Water-nickel 0.006
Vertical tube (27.1 mm I.D) Water-copper 0.013
Vertical tube (27.1 mm I.D) Carbon tetrachloride-copper 0.013
Vertical tube (27.1 mm I.D) Isopropyl alcohol-copper 0.0022
Vertical tube (27.1 mm I.D) n-Butyl alcohol-copper 0.0030
Vertical tube (27.1 mm I.D) 50% K2CO3-copper 0.00275
Vertical tube (27.1 mm I.D) 50% K2CO3-copper 0.0054

Chen [9] proposed a correlation that covers both saturated and nucleate boiling regions C
and D and the two-phase forced convection regions E and F as shown in Figure 12.7. The
analysis is complex. It is assumed that heat transfer in these regions take place by both
nucleation and convection. Therefore, the two-phase heat transfer coefficient hT, p is
considered to be composed of the contributions made by these two different mechanisms
hTP  hNB  hc . . . (12.16)
where hNB and hc are the heat transfer coefficients due to nucleate boiling and convection
respectively. It is assumed that forced convection heat transfer coefficient hc can be
determined from the modified Dittus-Boelter equation given in the form
k 
hc  0.023  l  Rel0.8 Prl0.4 F . . . (12.17)
D
G (1  x ) D
where Rel 
l

Mass flow rate through tube

G , kg/(m2.s)
Atube
D = Inside diameter of the tube, m,
F = Convective boiling factor, and
x = Vapor mass quality.
Here the subscript l refers to the liquid phase. Figure 12.8 gives the convective boiling
factor F plotted as a function of Martinelli parameter Xn. The nucleate boiling heat
transfer coefficient hNB is taken as the modified Foster-Zuber [10] equation in the form
 kl0.79 c 0.45
pl l
0.49 
0.24 0.75
hNB  0.00122 *0.5 0.29 0.24 0.24  Tsat
 Psat S . . . (12.18)
   l h fg v 
 
54
 Boilers
where Tsat  Tw  Tsat , o C
Psat  Psat ,Tw  Psat ,Tsat , N/m2

* = Surface tension, N/m, and

S = Suppression factor.
The subscripts l and v refer respectively to the liquid and vapour phases.
Figure 12.9 gives the empirical factor S plotted against the two-phase Reynolds number
ReTP, defined as
100
8
6
5
4
Connective bio0ling factors F

3
2

10
8
6
5
4
3
]
2

1
1 2 3 4 5 6 8 10 2 3 4 5 6 8 10 2 3 4 5 6 8 100
1
Xn

Figure 12.8 : Convection Boiling Factor F

10

0.9 Extrapolation of curve

beyond the data
0.8
Suppression factors F

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0
101 2 3 4 5 6 7 8 9 105 2 3 4 5 6 7 8 9 106
ReTP = Ft.25 . Rel

Figure 12.9 : Suppression Factor S

 GD (1  x ) 
ReTP  F 1.25 Rel  F 1.25   . . . (12.19)
 l 
It may be mentioned here that the Martinelli parameter Xtt appearing in Figure 12.8 is
defined as
0.9 0.5 0.1
1  x   v   l 
X tt        . . . (12.20)
 x   l   v 
55
Heat and Mass Finally, the heat flux at the wall is related to the two-phase heat transfer coefficient hTP
Transfer Applications
by
q  hTP Tsat  hTP (Tw  Tsat ) . . . (12.21)

SAQ 2
(a) What is the difference between pool boiling and forced convection boiling?
(b) What is suppression factor?
(c) What is Martinelli parameter?

12.6 CONDENSATION
If a vapour strikes a surface that is at a temperature below the corresponding saturation
temperature, the vapour will immediately condense into liquid phase. Condensation may
be divided into two categories :
(a) Filmwise condensation, and
(b) Dropwise condensation.
12.6.1 Film Condensation
If the condensation takes place continuously over the surface which is kept cooled by
some cooling process and the condensed liquid is removed from the surface by the
motion resulting from gravity, then the condensing surface is usually covered with a thin
layer of liquid. Such condensation is known as filmwise condensation. Such phenomenon
occurs if the liquid wets the surface. The presence of liquid film over the surface
constitutes a thermal resistance to heat flow.
In most of the engineering applications vapour is condensed by bringing it into contact
with a cold surface. Typical example is the steam condenser in steam power plant.
Condensation on Vertical Surfaces
Condensation of a vapour flowing over a vertical plate is presented in Figure 12.10. In
this figure, x is the axial coordinate, measured downward along the plate and y is the
coordinate normal to the condensing surface. Let,  =  (x) be the condensate thickness.
This problem was first analyzed by Nusselt [11] with the following assumptions :
(a) The plate is maintained at a uniform temperature Tw (less than the saturation
temperature of vapour Tv).
(b) The vapour is stationary or has low velocity and hence it exerts no drag on
the motion of the condensate.
(c) The downward flow of condensate is laminar and is under the action of
gravity.
(d) The flow velocity of condensate film is low resulting in negligible
acceleration of condensate film.
(e) Fluid properties are constant.
(f) Heat transfer across the condensate layer is by pure conduction, hence liquid
temperature distribution is linear.
The velocity distribution of condensate layer at any location x is determined by force
balancing over a volume element as shown in Figure 12.10.
Force components are :
du
(a) Shear force acting upward l dx 
dy

(b) Buoyancy force acting downward (l  v ) (  y) g dx 

From force balance on the elementary volume
56
 du Boilers
l dx  (l  v ) (  y ) g dx . . . (12.22)
dy

du g (l  v )
 (  y) . . . (12.23)
dy l
At the wall surface, liquid velocity is zero.
u = 0 at y = 0
0
y

x
Tw Condensate

x
Vapor

Tp
u
  i
v

dx

 d

(l  v) (  y) g dx

Figure 12.10 : Nomenclature for Filmwise Condensation on a Vertical Plane Surface

g (l  v )  1 2
u ( y)   y  y  . . . (12.24)
l  2 

The mass flow rate of condensate m (x) through any axial position x per unit width of the
plate is given by

m ( x)   l u dy . . . (12.25)
0

Introducing u from Eq. (12.24) into Eq. (12.25) and performing the integration yield

g l (l  v )  3
m ( x)  . . . (12.26)
3l

and differentiating with respect to  gives

g l (l  v )  2
dm  d . . . (12.27)
l

Here dm represents the rate of condensation over the distance dx per unit width of the
plate, since the condensate thickness increases by d over the differential length dx.
The rate of heat released dQ associated with the rate of condensation dm is
dQ  h fg dm . . . (12.28)

57
Heat and Mass where hfg is the latent heat of condensation. The amount of heat released dQ over the area
Transfer Applications dx . 1 must be transferred across the condensate layer of thickness  by conduction,
according to the assumption 6.
Tv  Tw
Therefore, dQ  kl dx . 1 . . . (12.29)

where kl is the thermal conductivity of liquid and Tv and Tw are the vapour saturation and
wall surface temperatures, respectively.
Introducing Eqs. (12.27) and (12.29) into Eq. (12.28), we obtain the following
differential equation for the thickness of the condensate layer
d l kl (Tv  Tw )
 . . . (12.30)
dx g l (l  v ) h fg  3

The integration of Eq (12.30) with the condition  = 0 for x = 0 yields the thickness of the
condensate layer as a function of the position x along the plate
1
 4 k (T  Tw ) x  4
 ( x)   l l v  . . . (12.31)
 g l (l  v ) h fg 

Since we have established the relation for the thickness of the condensate layer, the local
heat transfer coefficient hx for condensation is determined from the definition
Tv  Tw
hx (Tv  Tw )  kl . . . (12.32)
 ( x)

kl
or, hx  . . . (12.33)
 ( x)

Introducing  (x) from Eq. (12.31) into Eq. (12.33), we obtain

1
 g l (l  v ) h fg kl3  4
hx    . . . (12.34)
 4l (Tv  Tw ) x 

The local Nusselt number Nux is expressed as

3 1/ 4
h x  g l (l  ) h fg kl 
Nu x  x    . . . (12.35)
kl  4l (Tv  Tw ) x 

The average heat transfer coefficient hm over the length 0  x  L of the plate is
L
1 4
hm 
L  hx dx 
3
hx . . . (12.36)
0 xL

or, introducing Eq. (12.34) into Eq (12.36), we obtain

1
 g l (l  v ) h fg kl3  4
hm  0.943   . . . (12.37)
 4l (Tv  Tw ) L 

The physical properties in Eqs. (12.34) and (12.37), including hfg should be evaluated at
the film temperature
1
Tf  (Tw  Tv ) . . . (12.38)
2
58
SAQ 3 Boilers

(a) What is film condensation?

(b) Find the average heat transfer coefficient for condensation over a vertical
wall.
(c) How do you estimate the fluid properties?

Condensation on Inclined Surfaces

In case of filmwise condensation over an inclined plane surface, the above analysis
can still be used with incorporation of the angle of inclination. Let,  be the angle
made by the surface with the horizontal (Figure 12.11). Then the local and average
heat transfer coefficients are given respectively as
1
 g l (l  v ) h fg kl3 4
hx   sin   . . . (12.39)
 4l (Tv  Tw ) x 

1
 g l (l  v ) h fg kl3 4
and hm  0.934  sin   . . . (12.40)
 l (Tv  Tw ) L 
y

Figure 12.11 : Nomenclature for Filmwise Condensation on an Inclined Plane Surface

Condensation on a horizontal tube

1
 g l (l  v ) h fg kl3  4
hm  0.725   . . . (12.41)
 l (Tv  Tw ) D 

where D is the outside diameter of the tube.

A comparison of Eqs. (12.40) and (12.41) for condensation for a temperature
difference Tv – Tw, on a vertical tube of length L, and horizontal tube of diameter
D yields
1
hm, vert  D 4
 1.30   . . . (12.42)
hm, horz L

The average heat transfer coefficients for a vertical tube of length L and a
horizontal tube of diameter D becomes equal when L = 2.87 D. For instance, when
59
Heat and Mass L = 100 D, theoretically, hm, horz  2.44  hm, vert . With this consideration,
Transfer Applications
horizontal tube arrangements are generally preferred to vertical arrangements in
condenser design.
Condensation on Horizontal Tube Banks
As shown in Figure 12.12, condensers are generally made of horizontal tubes
arranged in vertical tiers. Condensate from one tube drains on the tube just below.
Assuming the flow of condensate to be smooth, for a vertical tier of N tubes each
of diameter D, the average heat transfer coefficient hm for the N tubes is given
approximately
1
 g l (l  v ) h fg kl3  4 1
hm N tubes
 0.725    1 [hm ] 1tube . . . (12.43)
 l (Tv  Tw ) ND 
N4

Figure 12.12 : Film Condensation on Horizontal Tubes Arranged in a Vertical Tier

12.6.2 Reynolds Number for Condensate Flow

During the condensation over a single tube is generally laminar in nature. However, as
flow continues over the tubes downwards, flow becomes turbulent. To establish a
criterion for transition from laminar to turbulent flow, the Reynolds number for
condensate flow is defined as
Dh  m l
Re  . . . (12.44)
l
where m is the average velocity of condensate film and Dh is the hydraulic diameter for
the condensate flow, given by
4A Cross-sectional area for condensate flow
Dh   4 . . . (12.45)
P Watted perimeter
Experimental results indicate that transition from laminar to turbulent flow of condensate
occur for Re  1800. However, it depends on geometry and properties of fluid, etc.
From Eqs. (12.44) and (12.45), the Reynolds number of the condensate flow is
60
 4l um A Boilers
Re  . . . (12.46)
l P

where A = Cross-sectional area for condensate flow, and

P = Wetted perimeter.
The Reynolds number at the lowest part of the condensing surface can be expressed in a
more convenient form as
4M
Re  . . . (12.47)
l P

where M = mass flow rate of condensate at the lowest part of the condensing
surface, kg/s.
Wetted perimeter = P depends on the geometry.
P=D for vertical tubes of outside diameter D
P = 2L for horizontal tube of length L
P=w for vertical or inclined plate of width w
12.6.3 Dropwise Condensation
Such type of condensation occurs if steam contains some oil and the contact surface of
steam with condenser tube is highly polished. The condensate film breaks into droplets
(Figure 12.13). In such situation, as the entire condensing surface is not covered with a
continuous layer of liquid film, the heat transfer for ideal dropwise condensation is much
higher than a filmwise condensation.

Figure 12.13 : Dropwise Condensation of Steam under Ideal Condition

Overall heat transfer coefficient for dropwise condensation is reported to be 2 to 3 times

greated than that of filmwise condensation for a typical surface condenser. Varoius types
of promoters such as oleic, stearic, linoleic acid, benzyl mercaptan and many other
chemicals have been used to promote dropwise condensation. However, due to the
oxidation, the surface of the condenser requires special attention. Coatings of gold, silver,
palladium and platinum have been used to reduce the oxidation. Thus there is increase in
cost as well.
61
Heat and Mass SAQ 4
Transfer Applications
(a) Distinguish between the film-wise and drop-wise condensation.
(b) What are the conditions to maintain a drop-wise condensation?

12.7 APPLICATIONS OF BOILING HEAT TRANSFER

12.7.1 Steam Generators
Boiling phenomena is extensively used to produce saturated as well as superheated steam
from saturated water in steam generators. A brief discussion on steam generators are
given in the following sections.
Classification of Steam Boiler (Generator)
Utility and Industrial Boiler
Utility boilers are used for electric power generation. These are usually
sub-critical (pressure of steam below the critical point) water drum type.
Usually these boilers operate in 130 bar (approx). In case of supercritical
utility boilers, working pressure is 240 bar or even more.
Industrial boilers are used by industrial or institutional concerns. These
boilers generally produces saturated steam, process heat and even hot water.
(In case of production of hot water, it is not called a boiler).
According to the Capacity
(a) Low capacity boilers – capacity below 20 tonnes of steam per hour.
(b) Medium capacity boilers – capacity 20-75 tonnes/hr.
(c) High capacity boilers – capacity more than 100 tonnes/hr.
According to the Pressure of Steam
(a) Low pressure boiler (Pressure less than 30 atm).
(b) Medium pressure steam (Pressure 30-70 atm).
(c) High pressure steam (Pressure more than 150 atm).
(d) Super pressure boiler (Pressure 150-190 atm).
(e) Super critical boiler (Pressure more than 225 atm).
According to Design
(a) Fire tube boiler
(b) Water tube boiler
(i) Natural circulation boiler
(ii) Forced circulation boiler
Fire Tube Boiler
Fire tube boilers were used extensively for industrial power generation till late 18th
century. They are no longer used for large utility power plants. However, they are
in still use in some industries for producing saturated steam with upper limit of
pressure to 18 bar and capacity of 6.3 kg/s. Interestingly, general design of fire
tube boiler has not changed much for last three decades.
In fire tube boilers the flue gas products of combustion flow through boiler tubes
surrounded by water (Figure 12.14). The heat transferred through the walls of the
tubes to the surrounding water generates steam.

62
 Exhaust gas Boilers
To chimney

Main stop valve

Supply
Steam Safety valve

Superheater
Oil tank Manhole
Steam space
Oil pump Water feed point

P Oil recirculation
Electricity

Oil filter
Oil heaters

Damper
Steam
FD fan
Oil burner
Air Corrugated flue Blowdown
Valve

Furnace exit

Flue box

Boiler shell

Fire tubes

Insulation Water

Burner
Refractory
Flame

Fire box

Figure 12.14 : A Fire Tube Boiler

The flue gases are cooled as they flow through the tubes, transferring their heat to
water. Therefore, the cooler the flue gases, the greater the amount of heat
transferred. Cooling the flue gas is a function of heat conductivity of the tube
material, the temperature difference between the flue gases and water in the boiler,
the heat transfer area, the time of contact of the flue gases with tube wall, etc.
Fire tube boilers used today evolved from the earlier designs of a spherical or
cylindrical pressure vessel mounted over the fire with flame and hot gases around
the boiler shell. Installing longitudinal tubes in the pressure vessel and passing flue
gas through the tubes have improved this obsolete approach. This improves the 63
Heat and Mass heat transfer coefficient and increases the heat transfer area. The results are the two
Transfer Applications variations of the horizontal return tubular boiler (HRT) as shown in
Figures (12.15)-(12.16).
Smoke Box

Bridgewall

Figure 12.15 : HRT Boilers

Figure 12.16 : HRT Boilers

A parallel evaluation of fire tube boiler is the locomotive boiler with the furnace
surrounded by a heat transfer area and a heat transfer area added by using
horizontal tubes (Figure 12.17).

Figure 15 17 : Locomotive Boiler

The Scotch Marine boiler design, as shown in Figure 12.18, with the furnace a
large metal tube, combined the feature of the English Cornish type of boiler of the
1800s and the smaller horizontal tubes of the HRT boiler. This boiler originally
was developed to fit the need for compact shipboard boilers. As the furnace is
cooled completely by water, no refractory furnace is required. The radiant heat
from the combustion/waste gas is transferred directly through the metal wall of the
furnace chamber to the water. This allows the furnace walls to become a heat
transfer surface – a surface particularly effective because of high temperature
differential between the flame/high temperature exhaust and boiler water.

64
 Boilers

Combustion

Figure 12.18 : Scotch Marine Boiler

Advantages of Fire Tube Boilers

(a) Low first cost
(b) Reliability in operation
(c) No need for skilled labour
(d) Less draught is necessary
(e) Quick response to load change : fire tube boiler is having large water storage
and thus can meet sudden load demands with only small pressure change
Salient Features of Fire Tube Boilers
(a) Good for small steam requirement.
(b) Good for community hot water/domestic hot water generation.
(c) It must be used when gas pressure is high.
(d) Useful even when the gas is corrosive, the shell side is then free from
corrosion.
(e) All the tubes are immerged in water.
(f) Limited applications – In industrial plants it is used to produce saturated
steam. Scotch Marine boilers have been using for waste heat recovery.
(g) Recommended to use at a pressure below 18 bar and steam flow rate
of 6.2 kg/s.
Major Shortcomings of Fire Tube Boilers
Size and Pressure Limitations
The tensile stress of boiler drum is given by the equation
pD
 . . . (12.48)
2t
where,  is the tensile stress, N/m2,
p is the gauge pressure, N/m2,
D is the internal diameter of boiler shell, m, and
t is the thickness of shell, m.
From Eq. (12.48) it is observed the following :
(a) If the gauge pressure p is high, the tensile stress becomes high
proportionately which results in increase in thickness t of
boiler shell.
(b) If the boiler storage capacity is high, the shell diameter D
becomes more, which results in high stress demanding for
more thickness.
65
Heat and Mass It is obvious from above analysis that for both the cases mentioned above
Transfer Applications lead for high cost of boiler. Thus there is limitation on size and pressure on
boiler design.
Water Tube Boilers
Natural and Controlled Circulation Boiler
The flow of water and steam within the boiler circuit is called circulation.
Adequate circulation must be maintained to carry away the heat from the
flue gas. If circulation is caused by density difference, the boiler is said to
have natural circulation (Figure 12.19). If a pump causes it, it has forced or
controlled circulation (Figure 12.20).
Steam

Boiler Drum (Steam)

Water

Riser

Down Comer
Heat

Lower or Mud Drum

Figure 12.19 : Natural Circulation Boiler

Superheater
Outlet
header
Superheater Drum

Riser

Down-
comer

Forced
Draught fan

Forced
Draught fan

Figure 12.20 : A Forced Circulation Boiler

(a) Circulation should be adequate to maintain heat balance of waste
gas and fluid. Adequate circulation prevents any hot spot on boiler
tube.
(b) Circulation ratio (CR) is the ratio of flow rate of saturated water in
down comer to the flow rate of steam released from the drum.
(c) For maintaining the natural circulation, CR  6 for the riser. This
means that 6 kg of saturated water should be circulated through the
down comer for each kg of steam produced.
66
 (d) Too much steaming is undesirable as it creates blanket (film) Boilers
retarding heat transfer. Inserts like, steel ribbons, twisted tapes, wire
coils can reduce the film formation on riser tube. However, boiler
used for producing hot water need not contain the inserts.
(e) The downcomers are fewer in number but bigger in diameter. They
are meant to make the saturated water to fall by gravity. Bigger the
diameter, less the pressure drops due to friction, as pressure drop is
inversely proportional to the tube diameter.
f l V 2
p  . . . (12.49)
2d
where p = Pressure drop in the downcomers,
f = Friction factor,
 = Density of saturated water, kg/m3,
V = Average velocity of water in the downcomers, m/s, and
d = Internal diameter of downcomers, m.
Generally, diameter d of a down comer is chosen in the range of
150-200 mm or even higher and average velocity of flow V is in the range of
0.4-1.4 m/s2.
The mass flow rate of saturated water in the downcomers is
  
m f  n d 2  V . . . (12.50)
 4 
where mf = Mass flow rate of saturated water in the downcomes, kg/s,
n = Number of downcomers tube,
d = Internal diameter of down comers, m,
 = Density of saturated water, kg/m3, and
V = Average velocity of water in the downcomers, m/s.
The number of tubes in downcomers can be found out.
Riser Tubes
(a) Nucleate boiling should occur in the riser tube and film boiling is to be
avoided. Too much steaming and less circulation ratio in a riser may cause a
departure from nucleate boiling (DNB) and an onset of film boiling.
(b) Internal twisters and springs at the inner surface of the riser tube are
provided which break the vapor film and retard the onset of DNB. Another
method is to rib the inside of the riser tube. The ribbing creates a centrifugal
action and directs the water droplets to the vapor film clinging to the surface
and to wash it away.
(c) Adequate circulation should be maintained to prevent formation of hot spot.
(d) For the same total cross sectional area, the smaller the diameter, the larger
surface exposed to hot gas. Hence, risers are smaller in diameter and more in
number.
(e) Generally riser tube diameter is in the range of 62.5-76.5 mm
Natural circulation can be maintained upto 30 bar. Both the downcomers and riser
tubes are placed inside the furnace. However, for pressure above 30 bar,
downcomers are placed outside the furnace.
Due to the amount of gas and the space limitations, all boilers are tailor-made. The
peculiarities of the different gases have to be taken into account :
67
Heat and Mass (a) The melting point of withdrawn dust or ash has to be considered to
Transfer Applications avoid excessive fouling of the tubes. They not only reduce the
efficiency of the heating surfaces, they can also cause corrosions.
(b) Before entering the convection banks the gas has to be cooled down
to a temperature quite below the melting point. In several cases,
empty passes have to be provided.
(c) The gas velocity in convection banks depends on the gas and dust.
The velocity should not be too high to avoid erosion of the tubes. On
the other hand the velocity in horizontal passes should not fall below
a certain limit. If the velocity is too low, the dust separates into coarse
and fine particles. The coarse particle fall down in the lower part and
the fine particle gather on the upper side of the tubes leading to a
reduced efficiency.
(d) At the back end of the boiler the saturation temperature or water inlet
temperature in economizer has to be higher than the dew point of the
gases to avoid low temperature corrosion.
(e) The best cleaning method has to be chosen. Typical methods are:
steam or air soot blowers, shot cleaning, rapping or knocking,
washing and sonic cleaning. Soot blowers cannot be used in any case.
If the waste gas is used in a chemical process the air or steam changes
the composition.
(f) The mass flow in the tubes has to be checked to assure a sufficient
cooling of the tubes in all operation conditions.
Figures 12.21-12.22 present straight tube boiler. Bent tube boiler and bi-drum bent
tube boiler respectively.
Straight Tube Boiler
This is the earliest design of boiler. Straight tubes are rolled into header at each
end, since straight tubes could be made, installed and replaced easily
(Figure 12.21). The tubes were 75-100 mm O.D., inclined upward at about 15o to
the horizontal and staggered. Nearly saturated water leaving the drum flowed
through one header, called the downcomer, into the tubes. While flowing upward
in the tubes some of this water on being heated by flue gases flowing outside get
transformed to vapour and the two-phase water-steam mixture went back to the
drum through the other header, called the riser. The density of nearly saturated
water in the downcomer is larger than the density of the two-phase mixture in the
riser. This density difference causes natural circulation in the system.

Superheater

Drum

Riser
Down
Comer
Straight
Tubes

Flue gas

Stoker

Figure 12.21 : Straight Tube Boiler

68
The Bent Tube Boiler Boilers

Straight tube boilers have many disadvantages, such as :

(a) Less accessibility and poorer inspection capability.
(b) Considerable time, labour and money is required to open or close the
bolts in the headers, to replace or remove the gaskets.
(c) Leakage past hand hole caps as design is inadequate and fabrication is
imperfect.
(d) Sluggish circulation due to low head.
The disadvantages of straight tube boiler have been overcome by introducing the
bent tube boilers. Tubes enter and leave the boiler drum radially in these boilers
(Figure12.22). They have the following advantages:
(a) Accessibility for inspection, cleaning, maintenance.
(b) Ability to operate at higher steaming rates.
(c) More drier steam is delivered.
Feed water Superheated steam
Superheater
Sat steam

Gases to Steam Steam Steam

Air heater Drum Drum Drum

Baffles
Steam Hot gases
Drum

Warer tubes on Burners

Furnace walls

Figure 12.22 : Bent Tube Boilers

69
Heat and Mass Boiler Drum
Transfer Applications
The function of steam drum are :
(a) To store water and steam sufficiently to meet the varying load
requirement.
(b) To aid in circulation.
(c) To separate vapour or steam from water-vapour mixture discharged
by the risers.
(d) To provide enough surface area for liquid-vapour disengagement.
(e) To maintain a certain desired level of concentration (ppm) in the drum
by injection of phosphate or tannin and blowdown.
Blowdown
Blow down operation is essential in boiler to remove the total dissolved solids
(TDS). TDS is expressed in parts per million (ppm). Blow down operation may be
continuous or intermittent depending on the nature of operation. For utility boiler,
blow down is usually continuous.
Blow down is defined as :
Quantity of water blown down
% Blow down = 100 . . . (12.51)
Quantity of feed water

For example, 3% blowdown indicates 3% of the total amount of feed water is

removed as liquid by blowdown.
Separation Steam from Steam-water Mixture in the Boiler Drum
There are 3 (three) different methods to separate steam from the mixture of
steam-water in the boiler drum. They are :
Gravity Separation
Small boilers up to 20 bar, sufficient disengagement area is available in the
drum. Gravity is predominating and hence separation is easy.
Increase in Drum Size
This option is limited because of increase in cost and high pressure.
Mechanical Separators
Baffles, screen and cyclone separators. Cyclone separator is used when
gravity force is not sufficient to separate steam from the mixture.

12.8 APPLICATIONS OF CONDENSATION HEAT

TRANSFER : STEAM CONDENSERS
A steam condenser is a device in which steam condenses and heat released by steam is
absorbed by cooling water. The need of a condenser is to increase the specific work
output of the turbine and to recover high quality feed water in the form of a condensate
and feed it back to steam generator without further treatment.
Classification
Condensers are classified as :
(a) Direct contact type, and
(b) Surface condenser.
70
Direct Contact Type Boilers

Spray Condenser
In this type of condenser, cooling water is sprayed into the steam issued
from the turbine exhaust. Mixing of steam with cooling water condenses the
steam. As shown in Figure 12.23, exhaust steam from the turbine at state 2
mixes with cooling water at state 5 to produce saturated water at state 3,
which is pumped to state 4.
Turbine
Dry Cooling Exhaust
Tower
Non-condensables to
SJAE
2
5

Condenser

4 3
To Plant Feed
Pump
water System

Figure 12.23 : Spray Condenser

Mass balance, w2 = w4 . . . (12.52)

w3 = w2 + w5 . . . (12.53)
Energy balance, w2 h2  w5 h5  w3 h3 . . . (12.54)

w5 h2  h3
  . . . (12.55)
w2 h3  h5

Since h2  h3   h3  h5

w5   w2

Barometric Condenser
In a barometric condenser, the cooling water is made to fall in a series of
baffles exposed to large surface area for the steam fed from below to come
in direct contact (Figure 12.24). The steam condenses and the mixture falls
in a tail pipe to the hot well below. By virtue of its static head, the tail pipe,
compresses the mixture to atmospheric pressure.
Thus, patm  pcond  p f   g H . . . (12.56)

where,  = Density of mixture,

pf = Pressure drop due to friction, and
H = Height of tail pipe.
For low values of p f , H  9.5 m .

Higher is the value of H, higher is the friction.

Jet Condenser
In this type of condenser, the height of the tail pipe is reduced by replacing it
with a diffuser. The diffuser helps in raising the pressure in a short distance
than a tail pipe (Figure 12.25).
71
Heat and Mass Non-condensables to
Transfer Applications Steam-Jet Air Ejector
Turbine
Exhaust

Cooling
Cooling
Water
Water
Baffles Exhaust
Steam from
Turbine
Cascades
Mixture Mixture
Tail Pipe
H

Diffuser

Hot Well

Figure 12.24 : Barometric Condenser Figure 12.25 : Jet Condenser

Surface Condenser
These are shell and tube type condensers where steam flow from the shell
side (outside the tubes) and cooling water pass through the horizontal tube
banks. Tubes materials are cupro-nickel, aluminium-bronze, muntz metal
(60% Cu + 40% Zn) or stanless steel. Following are the features of surface
condenser :
(a) Requires large surface area.
(b) Non-mixing type (recuperative heat exchanger).
(c) Condensate can be reused as feed water as it does not mix
with the cooling water.
(d) More power required for cooling water flow.
(e) Suitable for high capacity plant.
(f) Vacuum efficiency is high.
Turbine Exhaust

Exhaust Neck Steam Dome

Support Plate Shell

Tube Sheet

Water in

Water Out

Undivided Divided water Box

Water box

Hot Well
Condensate

Figure 12.26 : Surface Condenser

12.8.1 Design of Surface Condenser

For filmwise condensation (Figure 12.27), the average heat transfer coefficient for a
horizontal tube is given by Nusselt as
72
 1 Boilers
 g l 2 h fg kl3  4
hm  0.725   . . . (12.57)
 N l  D0 

where N = number of horizontal tubes in a vertical tier

  Tv  Tw . . . (12.58)

 t

tsat

Condensate
Film

tB


Figure 12.27 : Filmwise Condensation

Heat transfer coefficient inside the tube may be obtained by applying Dittus-Boelter
equation

Nud  0.023 Red0.8 Pr 0.4 . . . (12.59)

Vdi
where Re d  , di being the inner diameter of the tube, and

Cp
Pr  = Prandtl number.
k

Now, for water hi  V 0.8 . . . (12.60)

As velocity of cooling water increases, the heat transfer coefficient also increases but
there will be simultaneous increase in pressure drop in the system. Hence with increase in
velocity requires more pumping power. Usually, velocity of cooling water is kept in the
range 2-2.5 m/s.
Overall heat transfer coefficient for a condenser tube can be estimated from
1 1 1 xw 1
    . . . (12.61)
U 0 A0 hi Ai hsi Ai k w Alm h0 A0
Since condenser tubes are thin and made of material with high thermal conductivity, the
tube wall resistance can be ignored. Then,
1 1 1 1
   . . . (12.62)
U 0 hi hsi h0

Here, h0   hi

1 1
Hence,  A  B  0.8 . . . (12.63)
U0 V
1 1
where, A 
hsi h0
73
Heat and Mass 1
Transfer Applications and B
kl Pr 0.4
0.023
di 0.2 vl 0.8

Wilson Plot
With the Wilson plot the values of A and B can be estimated. This plot is drawn
1 1
0.8
vs (Figure 12.28). The slop of the curve will give the value of B and
V U0
1
intercept of the curve with (i.e. y-axis) will give the value of A.
U0

Rate of heat transfer from the condensing vapour to the cooling water is
Q  w s (h2  h3 )  w c (tc 2  tc1 )  U 0 A0 tlm . . . (12.64)

ti  te
where tlm  . . . (12.65)
t
ln i
t e

(Ref. Figure 12.29)

I / Uo

I / V 0.8

Figure 18.28 : Wilson Plot

tsat Steam tsat

Condensing
TTD
tC2
l

C.W.

tC1

ti te
Ao

Figure 12.29 : LMTD

The terminal temperature difference (TTD) is

TTD  te

usually TTD is 11-17oC

74 Mass flow rate of cooling water is

 w s (h2  h3 ) Boilers
w c  . . . (12.66)
C pc (tc 2  tc1 )

Rise in water temperature is limited to 8-10oC. Usually, a normal industrial design

indicates that 75-100 kg of cooling water is needed for condensing 1 kg of steam.
Surface area required
w s (h2  h3 )
A0   N  d0 l . . . (12.67)
U 0 tlm
where N is the number of tubes, and
l is the length of one tube (single pass).
Typically modern condensers are having tubes as high as 50,000 numbers with
22-23 mm outer diameter.
Pumping power required
p
P  w c v p  w c . . . (12.68)

where v specific volume of the coolant m3/kg

fL V 2
and p   . . . (12.69)
d 2

12.9 SUMMARY
In the present unit heat transfer with phase change has been discussed. Both the boiling
and condensation are heat transfer phenomena associated with phase change.
Mechanisms of both the heat transfer are elaborately discussed. Applications of these
phenomena for various engineering devices are described. A design procedure for
condenser design is also presented.

12.10 KEY WORDS

Boiling : It is a phase change phenomena where liquid
phase is transformed to vapour at constant
temperature.
Condensation : It is a phase change phenomena where vapour
phase is transformed to liquid at constant
temperature.
Critical Heat Flux : Maximum heat flux at which burn out occurs.
Reynolds Number : Dimensionless number indicating ratio of inertia to
viscous force.
Nusselt Number : Dimensionless heat transfer coefficient; ratio of
convection heat transfer to conduction heat
transfer in a fluid layer.

12.11 ANSWERS TO SAQs

Refer the preceding text for all the Answers to SAQs.

75
Heat and Mass
Transfer Applications REFERENCES
W. M. Rohsenow (1952), A Method of Correlating Heat Transfer Data for Surface
Boiling Liquids, Trans. ASME, 74 : 969-975.
J. N. Addoms (1948), Heat Transfer at High Rates to Water Boiling Outside Cylinders,
D. Sc. Thesis, M. I. T., Cambirdge Mass.
N. Zuber (1958), On the Stability of Boiling Heat Transfer, J. Heat Transfer, 80C : 711.
N. Zuber and M. Tribus (1958), Further Remarks on the Stability of Boiling Heat
Transfer, Univ. Calif., Los Angeles, Dept. Eng. Rep., 58-5.
J. H. Lienhard, V. K. Dhir, and D. M. Riherd (1973), Peak Pool Boiling Heat Flux
Measurements on Finite Horizontal Flat Plates, J. Heat Transfer, 95C : 152-158.
K. H. Sun and J. H. Lienhard (1970), The Peak Boiling Heat Flux on Horizontal
Cylinders, Int. J. Heat Mass Transfer 13 : 1425-1439.
J. H. Lienhard and V. K. Dhir (1973), Hydrodynamic Prediction of Peak Pool Boiling
Heat Fluxesfrom Finite Bodies, J. Heat Transfer 95C : 152-158.
J. S. Ded and J. H. Lienhard (1972), The Peak Boiling Heat Flux from a Sphere,
AICHE J., 18 : 337-342.
J. C. Chen (1963), Correction for Boiling Heat Transfer to Saturated Fluids in
Convective Flow, ASME paper 63-HT-34, 6th ASME-AICHE Heat Transfer Conf.,
Boston.
H. K. Foster and N. Zuber (1955), Dynamics of Vapour Bubbles and Boiling Heat
Transfer, AICHE J. 1 : 531-535.
J. P. Holman (2002), Heat Transfer, 9th Edition, Tata McGraw-Hill, New Delhi.
M. N. Ozisik (1985), Heat Transfer : A Basic Approach, McGraw-Hill
International Edition.
F. P. Incropera and D. P. Dewitt (2004), Fundamentals of Heat and Mass Transfer,
5th Edition, John Wiley and Sons.
P. K. Nag (2002), Heat Transfer, Tata McGraw-Hill Publishing Company Limited,
New Delhi.
A. Bejan (1995), Convective Heat Transfer, 2nd Edition, John Wiley and Sons, INC.

76
 MRW – 002
HEAT TRANSFER

BLOCK 1 : INTRODUCTION TO HEAT TRANSFER

Unit 1 : Concepts of Heat Transfer
Unit 2 : Different Modes of Heat Transfer
Unit 3 : Mass Transfer

BLOCK 2 : CONDUCTION
Unit 4 : Governing Equations of Heat Conduction
Unit 5 : Numerical Methods to Solve Conduction Problems
Unit 6 : Heat Transfer from Extended Surfaces

BLOCK 3 : CONVECTION
Unit 7 : Convection Heat Transfer
Unit 8 : Boundary Layer Formation

BLOCK 4 : RADIATION
Unit 9 : Radiation Principles
Unit 10: Radiation Exchange

BLOCK 5 : HEAT TRANSFER APPLICATIONS

Unit 11: Heat Exchangers
Unit 12: Boilers

MPDD-IGNOU/P.O.5H/January,2022

ISBN-978-93-5568-283-3