CONDUCTION AND RADIATION

K. Muralidhar
Department of Mechanical Engineering
Indian Institute of Technology Kanpur
Kanpur India

J. Banerjee
Department of Mechanical Engineering
Sardar Vallabhbhai National Institute of Technology
Surat India

January 2010
Overview

Thermodynamics adopts the following viewpoint: Heat is energy in transit and is trans-
ported by virtue of a temperature difference. The subject of heat transfer is an extension
and is concerned with the calculation of the rate of heat transfer for a given tempera-
ture difference. This step brings into picture the physical mechanisms in the intervening
material (or, space) when it is subjected to a temperature difference. If the medium is
a solid body, the dominant transport mechanism is conduction; conduction is also the
prevelant mechanism in a stagnant fluid. In an engineering context, fluids are often in
a state of motion, driven by an external mechanism or by internal density differences.
An additional (bulk) transport mechanism (called advection) is thus introduced over
and above conduction. Jointly, the heat transfer mechanism in fluids comprising con-
duction and advection, is referred to as convection. If temperature difference is applied
across surfaces that enclose evacuated space, heat transfer occurs radiation, namely by
exchange of energy between surfaces in the form of electromagnetic radiation.
In the mathematical description of heat transfer, it is necessary to assume that the
transported quantities such as temperature and heat flux, as well as material properties
vary continuously from one point to another. In view of the particulate nature of
matter, it is appropriate to question the validity of this assumption. Equations derived
on the basis of this assumption have, however, withstood the test of time (mainly via
experiments) and the treatment of the physical medium as a continuum has firmly
become established. The continuum approximation breaks down if the density of the
medium is so small that molecular motion occurs on the same scale as the total distance
between the energy exchanging surfaces. If λg is the mean free path of the molecules and
L the characteristic dimension of the macroscopic system, the continuum approximation
is valid if λg /L  1 and not valid if λg /L ∼ 1, order of unity (or greater). In much of the
discussion to follow, physical phenomena are assumed to take place at the continuum
scale.
The ratio λg /L is called Knudsen number. It is of order unity in rarefied gases, for
example in the upper layers of the earth’s atmosphere. The formulation of the heat
transfer problem at this limit of the Knudsen number is not covered here. A short
discussion on microscale heat conduction is however, presented.
Typical atomic length scales are of the order of 10−9 m, time scales are smaller
than 10−9 seconds, and frequencies are in excess of 109 Hz. In view of the extreme
values of these scales, the continuum approximation has wide applicability. Consider
the following example. Under atmospheric conditions near the surface of the earth,
viii Overview

1 mm3 volume of air contains around 3 × 1016 molecules. This is a very large number.
Over the length scales involved in engineering problems, 1 mm3 is a small volume and
yet it is sufficiently populated for the continuum approximation to be valid.
The continuum scale is also called the macro-scale. The commonly observed laws of
physics such as conservation of mass, Newton’s second law of motion, and conservation
of energy are applicable here1 . Length and time scales that are smaller are called
the micro-scales. While the principle of conservation of energy is applicable at the
micro-scale, one can expect a significant departure in terms of the material behaviour.
When the atomistic scale is reached, departures are to be expected in terms of material
behaviour as well as the physical laws.
In a problem involving continuum scale heat transfer, mass and energy are unrelated
quantities. Hence, the conservation of mass principle is not required in the development
of the mathematical model. In conduction and radiation heat transfer, the velocity
vector is mostly zero. Hence, the momentum equation is also not required during the
theoretical development.
The three modes of heat transfer, conduction, convection and radiation can be jointly
categorized in terms of a mean free path. It is a representative distance over which
elementary particles travel before transporting thermal energy to their neighbourhood.
For conduction, it is a small value; for radiation it is large. In solids, the mean free path
can be described by theories analogous to kinetic theory of gases. In contrast, thermal
radiation transports as electromagnetic waves below the infrared range of wavelengths.
The mean free path takes on intermediate values for convection and is determined by
the nature of fluid flow. It is also to be emphasized that all three mechanisms can be
jointly present in certain applications.
Principles of thermodynamics necessarily must appear in heat transfer analysis. The
first law is enforced in the form of a conservation of energy principle. In applications
presently considered, the work contribution is generally negligible, or identically zero.
The second law prevails in a more subtle manner. It dictates the direction of energy
transfer in relationship to the temperature gradient. In other contexts, it can require
material properties to satisfy certain constraints.

1
These laws also hold at large scales though there is ambiguity at truly astronomical length and time
scales.
Contents

Preface v
Overview vii

Part I CONDUCTION
1. Energy Equation 3

1.1 Reynolds transport theorem 3

1.2 Derivation of energy equation 5
1.2.1 Constitutive equations 8
1.3 Thermal energy equation in a solid medium 10
1.4 Remarks on energy equation 10
1.5 Fourier’s law revisited 13
1.6 Thermal conductivity as a tensor 13
1.7 Expanded forms of the energy equation 14
1.7.1 Governing equation in Cartesian coordinates 14
1.7.2 Cylindrical and spherical coordinates 15
1.8 What is thermal conductivity 16
1.9 What is specific heat 18
1.10 Initial and boundary conditions 20
1.10.1 Composite media 23
1.11 Non-dimensionalization 25
1.12 Fin equation 27
1.13 Phase transformation 29
1.14 Closure 33
Bibliography 33
Problems 35

2. Steady Heat Conduction in One Dimension 37

2.1 Slab and circular geometries 37

2.2 Heat conduction in a fin 43
Problems 48
x Contents

3. Steady Heat Conduction in Two Dimensions 51

3.1 Heat conduction in a square domain 51

3.1.1 An example with a convective boundary condition 56
3.1.2 An example with contact resistance 59
3.1.3 An insulated bar with heat generation 62
3.2 Heat conduction in cylindrical coordinates 64
3.2.1 Example where boundary conditions are homogeneous in the φ-direction 65
3.2.2 Example where boundary conditions are homogeneous in the r-direction 68
3.2.3 An example in r-z coordinates 69
3.2.4 Note on Bessel functions 72
Problems 73

4. Unsteady Heat Conduction in 1 and 2D Regions 77

4.1 Bounded one dimensional domains 77

4.1.1 Important remarks 80
4.1.2 Slab with heat generation 80
4.1.3 Principle of superposition 81
4.1.4 Temperature dependent thermal conductivity 82
4.1.5 Thermal resistance 84
4.1.6 Cylindrical coordinates 85
4.2 Transient response of linear systems 88
4.2.1 Time-dependent boundary conditions 91
4.2.2 Time-dependent source term 92
4.3 Semi-infinite solid 93
4.3.1 Time-periodic boundary conditions 96
4.4 Two dimensional regions 102
4.4.1 Polar coordinates 104
4.4.2 Time-dependent boundary conditions 105
Problems 108

5. Phase Transformation 115

5.1 One dimensional phase change 115

5.1.1 Stefan condition 116
5.1.2 Special cases of Stefan condition 117
5.2 Movement of a phase front in a supercooled liquid 119
5.3 Heat transfer in solid and liquid phases 121
Problems 123

6. Diffusive Mass Transfer 125

6.1 Remarks on mass transfer 126

6.2 Higher order effects 127
Contents xi

6.3 Conservation of mass equation 128

Examples 129
Problems 131

7. Inverse Heat Transfer 133

7.1 Parameter estimation 133

7.2 Applications to heat transfer 135
7.2.1 Method of sensitivity coefficients 136
7.3 Least squares approach to Example 7.1 139
7.3.1 Analytical solution of the least squares problem 140
7.3.2 Example with heat generation 143
7.4 Linear and nonlinear inverse problems 144
7.5 Nonlinear diffusion equation 146
7.5.1 One-equation model 149
7.5.2 Two-equation model 149
7.6 Inversion of the 1-equation model 150
7.6.1 Construction of the objective function 151
7.6.2 Sensitivity functions 151
7.6.3 Conjugate gradient method 152
7.6.4 Determination of the step sizes 154
7.6.5 Adjoint problem 155
7.6.6 Convergence check 156
7.6.7 Summary of the numerical algorithm 157
7.7 Results and discussion 157
7.7.1 1-equation model 158
7.7.2 2-equation model 166
7.7.3 Closure 167
References 167
Problems 168

8. Microscale Heat Conduction 171

8.1 Hyperbolic heat conduction 173

8.2 Micro-structure of metals 177
8.2.1 Pulsing heat sources 179
8.2.2 Sample solutions 179
8.2.3 Semi-infinite solid 183
8.3 Two-equation model 184
8.3.1 Heat capacity 185
8.3.2 Thermal conductivity 186
8.3.3 Simplified two-equation model 188
8.4 Pulse heating of thin films 190
References 194
Problems 195
xii Contents

Part II RADIATION
9. Introduction to Radiative Transfer 199

9.1 Nomenclature 201

9.2 Three dimensionality 206
Bibliography 207

10. Black Body Radiation 209

10.1 Planck’s law 210

10.2 Special forms of Planck’s law 212
10.3 Radiation intensity 214
10.4 Total hemispherical emissive power 216
Examples 217
Problems 219

11. Properties of Non-black Surfaces 221

11.1 Emissivity 222

11.2 Absorptivity 224
11.2.1 Integrated absorptivities 225
11.2.2 Diffuse-gray surfaces 228
Examples 228
Problems 234

12. Energy Exchange between Black Surfaces 235

12.1 Enclosure theory 240

12.1.1 Heat transfer in a three surface enclosure 242
12.1.2 Network analysis 244
12.1.3 Spectral calculations for enclosure theory 246
12.2 Calculation of shape factor 246
12.3 First principles calculation of shape factor 255
12.4 Variable wall temperature 262
Problems 263

13. Energy Exchange between Diffuse-Gray Surfaces 265

13.1 Energy exchange calculation 266

13.1.1 Energy exchange in an enclosure with N surfaces 268
13.1.2 Energy exchange in a three surface enclosure 271
13.2 Enclosure with a specular surface 277
Problems 279
Contents xiii

14. Treatment of Variable Wall Temperature 281

14.1 Diffuse-gray enclosure 283

14.1.1 Energy exchange in a two dimensional enclosure 285
14.1.2 Numerical integration 287
14.1.3 Numerical solution of integral equations 289
Examples 291
Problems 294

15. Treatment of Non-Gray Surfaces 297

15.1 Spectral radiosity formulation 297

15.2 Band approximation 299
15.2.1 Energy exchange between two spectral surfaces 299
15.2.2 Radiosity formulation 302
15.3 Specified heat flux boundaries 303
15.3.1 Heat flux - temperature formulation 304
Examples 304
Problems 305

16. Gas Radiation 307

16.1 Equivalent beam length 310

16.2 Enclosure theory in the presence of a radiating gas 311
16.3 Spectral calculation of energy exchanges 316
Examples 318
Problems 320

17. Combined Conduction and Radiation 321

Examples 321
Problems 327

18. Radiative Transfer Equation 329

18.1 Introduction 329

18.2 Radiative exchange in a participating medium 330
18.2.1 Attenuation by absorption 330
18.2.2 Attenuation by out-scattering 331
18.2.3 Augmentation by emission 332
18.2.4 Augmentation due to in-scattering 334
18.3 The equation of transfer 335
18.4 General solution of RTE 336
18.5 Conservation of thermal energy 339
xiv Contents

18.5.1 Radiative heat flux 340

18.5.2 Divergence of radiative heat flux 340
18.5.3 Radiative equilibrium 341
18.6 Boundary conditions 344
18.6.1 Opaque surfaces 344
18.6.2 Semitransparent and transparent boundaries 345
18.7 Closure 346
Bibliography 346
Problems 347

19. Exact Solutions of the Radiative Transport Equation 349

19.1 Introduction 349

19.2 Gray medium approximation 350
19.3 Plane parallel medium 354
19.4 Non-scattering plane parallel medium 356
19.4.1 Non-scattering plane parallel media with a specified temperature distribu-
tion 356
19.4.2 Non-scattering plane parallel medium under radiative equilibrium 361
19.5 Plane parallel isotropic scattering medium 367
19.5.1 Isotropic scattering medium with specified temperature field 368
19.5.2 Radiative equilibrium in an isotropic scattering medium 368
19.6 Plane parallel anisotropic medium 369
Examples 370
Bibliography 376
Problems 377
MATLAB code 378

20. Numerical Solution of the Radiative Transport Equation 381

20.1 Introduction 381

20.2 Discrete ordinates method 382
20.2.1 Formulation using discrete ordinates 382
20.2.2 Representation of boundary conditions 384
20.3 Selection of directions for discrete ordinates 384
20.4 Discrete ordinates method for Cartesian coordinates 388
20.4.1 One-dimensional plane parallel medium 389
20.4.2 Two-dimensional configuration 400
20.4.3 Three-dimensional configuration 407
20.5 Closure 407
Bibliography 407
Problems 408
MATLAB codes 409
Contents xv

21. Combined Conduction, Convection and Radiation 419

21.1 Introduction 419

21.2 Combined conduction-radiation heat transfer 420
21.2.1 Conduction-radiation heat transfer in a one dimensional geometry 424
21.2.2 Two-dimensional conduction-radiation heat transfer 431
21.3 Treatment of combined conduction, convection and radiation 435
21.4 Closure 436
Bibliography 436
Problems 437
MATLAB codes 438

Index 449
Part I: CONDUCTION
Chapter 1

Energy Equation

The first law of thermodynamics, equivalently the principle of conservation of energy

can be expressed in analytical form by first utilizing the mathematical identity called
the Reynolds Transport Theorem.

1.1 Reynolds transport theorem

This theorem relates integrals of transport properties evaluated over a definite mass of
material occupying an arbitrary volume with integrals over fixed well-defined volumes.
It is a useful mathematical identity because physical laws are stated for fixed mass
systems, while it is easy to identify fixed control volumes in engineering applications.
Let Φ be a transported quantity associated with the physical material occupying
a space SP and φ be its intensive value, that is, Φ per unit mass. Since Φ is defined
with respect to a fixed mass of the material, conservation principles such as the first
law of thermodynamics are applicable to it. Let CV be a control volume fixed in space
and CS be the surface (control surface) enclosing it. The control volume is chosen to
overlap the material space SP at a time t. The quantities Φ and φ are related as
 
Φ= ρ φ dV = ρ φ dV
SP (t) CV

where ρ is the material density. Let u be the velocity vector and n, a unit outward
drawn normal on CS (Figure 1.1). The Reynolds transport theorem (RTT) can be
stated in this context as
 
dΦ ∂
= (ρ φ) dV + ρ φ u · n dS (1.1)
dt CV ∂t CS

i.e., the rate of change of the total property Φ (= dΦ/dt) of a particular mass of material
enclosed by the space SP is the sum of contributions from changes in φ throughout the
control volume and the net outflow of φ at the control surface.
Chapter 2

Steady Heat Conduction in One

Dimension

In the present chapter, we review analytical solutions of the steady heat conduction
equation in various coordinate systems.

2.1 Slab and circular geometries

In many instances, the geometry of the physical domain is such that temperature
changes rapidly with respect to one coordinate but is relatively constant in the other
directions. Since the corresponding fluxes (for nearly uniform temperature) would be
small, energy transfer is predominantly in the direction along which large changes in
temperature take place. It is then sufficient to work with the heat conduction equation
that neglects temperature gradients in some directions but retains them in others.
Consider first a slab geometry (Figure 2.1) where temperature changes take place pri-
marily in the x-direction. Assuming the slab material to be isotropic and homogeneous,
the steady three dimensional heat conduction equation

∂2T ∂2T ∂2T

+ + =0 (2.1)
∂x2 ∂y 2 ∂z 2

reduces to the one dimensional form

d2 T
=0 (2.2)
dx2
For the boundary conditions shown, we have

T (x = 0) = T1 T (x = L) = T2
Chapter 3

Steady Heat Conduction in Two

Dimensions

Temperature distribution in a two dimensional homogeneous, isotropic region can be

obtained by solving the heat conduction equation in an appropriate coordinate system.
Cartesian and cylindrical coordinate systems are of interest to the present chapter.
The analytical technique is broadly based on the method of separation of variables.
If the physical region is inhomogeneous, the thermal conductivity (and specific heat)
would be functions of position. The methods discussed below will then, not be directly
applicable. However, for certain configurations where the partitioning is conveniently
along constant coordinate lines, an analytical solution should be possible. If thermal
conductivity is a function of temperature (but not position), the governing equation can
be linearized using Kirchoff’s transformation (Section 2.1). The resulting equation is
once again analytically solvable. If the region has a thermal conductivity that depends
on temperature as well as position, the equation can only be numerically solved.

3.1 Heat conduction in a square domain

Consider determination of temperature distribution in a square region shown in Figure
3.1(a). The mathematical problem can be stated in Cartesian coordinates as follows:
Txx + Tyy = 0 (3.1)
where suffix indicates partial differentiation. In Figure 3.1(a), the boundary conditions
for temperature are
T (0, y) = T (x, 0) = T (x, 1) = 0; T (1, y) = 1
Here, temperature is represented in dimensionless form (with maximum corresponding
to unity and minimum being zero). The edge of the square region is used as the length
scale. The above mathematical problem can be solved by separation of variables, a
Chapter 4

Unsteady Heat Conduction in 1

and 2D Regions

In the previous chapter, steady heat conduction in one and two dimensional regions
was analyzed. The analytical tool used was essentially separation of variables, lead-
ing to an exact solution in the form of an infinite series. This approach is applicable
to materials that have a constant thermal conductivity. Of course, it is restricted to
simple geometries such as rectangles, circles and spheres. Extension to composite re-
gions in which thermal conductivity is piecewise constant is possible by applying flux
and temperature continuity conditions at the interfaces. For materials that show a
temperature-dependent thermal condictivity, the steady heat conduction equation can
be linearized using Kirchoff transformation.
The separation of variables approach continues to hold for unsteady heat conduction
in orthogonal coordinate systems when the material has constant thermal conductivity.
Nonlinear problems require special treatment and the solutions are at best approximate.
For linear heat conduction, the eigenvalue problem will emerge purely in the spatial
direction since the time axis stretches to infinity. These issues are discussed through
specific examples in the following sections.

4.1 Bounded one dimensional domains

Consider heat conduction in a slab that is initially at a temperature T2 ; the left face
temperature is suddenly raised to a temperature T1 and a thermal front diffuses through
the slab as a function of time. The steady state temperature distribution is linear,
varying from T1 on the left face to T2 on the right face. The distribution is not linear
during the transient evolution process.
In a one dimensional region such as a constant conductivity slab (Figure 4.1), the
unsteady heat conduction equation along with initial and boundary conditions is stated
Chapter 5

Phase Transformation

For pure metals as well as eutectic alloy compositions, the phase change boundary is
a sharp surface1 . On one side of the surface, the material is in liquid form, while it is
a solid on the other. For alloys, phase change occurs over a region of finite thickness2 .
The analysis of phase change in alloys is not the subject of discussion.

5.1 One dimensional phase change

The present chapter addresses heat transfer with phase change in one dimensional ge-
ometries. Here, the solid-liquid interface is planar and moves strictly in one direction.
Thus, the phase boundary overlaps a constant coordinate line. This requirement is es-
sential for the derivation of analytical solutions. In two and three dimensions, the phase
boundary is curvilinear, in general and applying boundary conditions can be a source of
difficulty. Thus, analytical methods discussed in the context of one dimensional prob-
lems do not carry over readily to two and three dimensions. Real-life phase change
problems are invariably handled via numerical simulation by a suitable discretization
procedure.
At a first glance, change of phase of the material can be thought of as a situa-
tion analogous to contact resistance between dissimilar materials. There are important
differences, however. These are:

1. Temperature at the phase change front is the fusion point temperature of the
material and is pre-determined. Additionally, the temperature field is continuous
as one proceeds from the liquid phase to the solid. This trend is in contrast to a
contact resistance wherein a jump in temperature is possible.

2. Unlike temperature, heat flux is not continuous since latent heat is released or
absorbed at the solid-liquid interface. The change in heat flux across the interface
1
of zero thickness
2
called the mushy zone
Chapter 6

Diffusive Mass Transfer

Consider fresh water floating over a concentrated salt solution at an initial instant
of time as in Figure 6.1 With the passage of time, salt diffuses from the base of the
beaker into the rest of the liquid. In a modeling framework, the physical property that
is transported is salt concentration, with water serving as a solvent. The transport
process is in the direction of diminishing concentration. In this respect, concentration
is a material property analogous to temperature. In fact, once can conceptualize mass
diffusion (of salt) as a gradient diffusion process quantified by the mass transfer rate

ṁ = −ρD∇C (6.1)

where C is salt concentration (kg of salt/kg of solvent), ρ is the density of the solvent1
(kg/m3 ), D is the diffusivity of salt in water (m2 /s) and ṁ is the mass flux in units
of kg/m2 -s. The above expression, called Fick’s law is analogous to Fourier’s law of
heat conduction. Mass diffusivity is a property of the solute as well as the solvent. The
product ρ × D plays a role akin to thermal conductivity in a heat transfer process.
An analogy refers to the a situation wherein two distinct physical phenomena give
rise to identical mathematical models. One often encounters analogies in fluid and
thermal sciences. For example, transport of mass, momentum and thermal energy
are quite similar under certain conditions. Of interest to the present discussion is
the mathematical analogy between diffusion of thermal energy and diffusion of mass
in a physical domain. When an analogy holds, analytical techniques applicable to
one physical problem can be fruitfully used to solve the other. Such analogies across
dissimilar disciplines is also common-place. Examples are electromagnetic field theory,
semiconductor physics, and transport phenomena in porous media. Once the governing
equations are setup, methods of solution of one discipline carry over to the other.

1
water, in the present example
Chapter 7

Inverse Heat Transfer

Engineers are often asked to estimate the performance of a given device, either via
simulation or through experiments. The implication here is that one would like to gage
those quantities that constitute the device output for a given set of specified (input)
parameters. This information would be called the performance characteristics of the
device. An example is efficiency as a function of head, for a pump or a turbine, at
specified flow rates and RPM. This is referred to as the direct or the forward problem. In
an independent scenario, the engineer is asked to build a system wherein the performance
characteristics are pre-determined. This is the inverse problem; clearly it is the central
theme of the design process.
The direct and the inverse problems are inter-related since they refer to the same
application, but differ vastly in the manner in which they can be addressed. For example,
it should be clear that the inverse problem need not have a unique solution. Several
configurations can be arranged to generate a given set of output characteristics. To
generate specific results, it is necessary to impose additional constraints such as size,
weight, and cost to force a unique solution. In an extraordinary situation (for example,
when the number of constraints is large) no configuration may be acceptable at all;
here the solution simply does not exist. These factors make the treatment of the inverse
problem quite complex. In the present chapter, some examples of the inverse formulation
in an analytical framework are presented. The method of solution has been discussed.
The material is presented from a thermal science perspective, though tools referred to
have a broader applicability1 .

7.1 Parameter estimation

In words, the direct problem can be posed in the following manner: Given complete
information regarding the input data, in general a function of time, the process param-
eters and the material properties, find the output response of the system (pointwise
1
The extension of the analytical techniques of the present chapter to the design of full engineering
system is a research topic under consideration.
Chapter 8

Microscale Heat Conduction

The term micro in the chapter title refers essentially to the smallness of the length scale,
though it is connected to small time scales as well. In metals and alloys, the distinc-
tion between small and large length scales is with reference to lattice dimensions. For
example, thin films used in electronic devices or as optical coatings may be a few mi-
crons thick (or smaller). Small linear dimensions are encountered when the overall heat
transfer in the physical domain occurs for a very short duration or occurs repeatedly at
a high frequency1 . Small time scales encountered in applications go from microseconds
(10−6 s), nanoseconds (10−9 s), picoseconds (10−12 s) to femtoseconds (10−15 s). High
frequencies would be megahertz (106 ), gigahertz (109 ) and terahertz (1012 ). These fre-
quencies should be compared with those encountered in traditional applications, say IC
engines that run at 2400 rpm and gas turbines that run at (say) 24,000 rpm. In units
of Hz, these values are 40 and 400 respectively; classical heat conduction is entirely
satisfactory in this context.
The relationship between distance and time (and hence, frequency) can become clear
from estimates of the thermal boundary-layer thickness. For a semi-infinite solid, it was
shown that √
δ ≈ t (step response)
and
1
δ≈√ (oscillatory response)
ω
The proportionality constants in the above expressions involve the transport properties
of the material. These expressions show that the size of the thermally affected region
increases as the square root of time in a transient process and as the reciprocal of the
square root of the forcing frequency in a pulsatile heating problem. In addition, small
time scale, small length scale and high frequency correlate with one another.
At the length and time scales referred above, it can be shown that the principle of
conservation of energy continues to be applicable as for the continuum limit. However,
it is quite important that the structure of the material is accounted for.
1
a process referred to as ultrafast
Chapter 9

Introduction to Radiative
Transfer

One way of distinguishing radiative energy transfer from heat conduction is as follows.
We associate a long mean free path to the radiant energy carriers, as opposed to a
short mean free path for heat conduction. The energy carriers in radiation are photons;
in heat conduction, they are normally electrons moving in a phonon environment. In
semiconductors, phonons (lattice vibrations) may serve as energy carriers as well. A
large mean free path has the following implication. The energy carriers travel long
distances before experiencing a change, either due to interaction with other photons,
with the intervening medium, or a surface. The change referred here could be that in
wavelength, energy density or direction. A simple example where the mean free path is
clearly large is energy transfer between surfaces held in vacuum. Here, a photon leaving
a surface travels unhindered till it strikes the second. The heat exchange calculations
in this context must refer to surface-to-surface exchanges, temperature profiles across
surfaces being less meaningful in this context. A similar situation is encountered during
energy transfer in media transparent to the wavelength under question. Of course, there
are examples such as dense particle-laden fluids, cloudy atmosphere and ionized gases
where radiation interacts with the surrounding fluid during its passage from one surface
to the other. The medium is then said to participate in the energy transfer process. It is
also conceivable that, under extreme conditions, all radiation emerges from the body of
the fluid and none from the bounding surfaces. A temperature profile will then appear
in the physical domain, as in the context of heat conduction.
The initial discussion on radiative transfer (Chapters 9-15) assumes that the inter-
vening space is vacuum or is purely transparent to energy exchange. Thus, the analysis
is restricted to surface-to-surface heat transfer. Chapter 16 onwards, radiative transport
in a participating medium is described.
Consider heat transfer from and to an opaque solid object first. The physical mech-
anism responsible for energy release from a surface is electronic transitions within the
material. In the interior, the emitted radiation is internally absorbed or scattered and
is accounted for in the conduction mechanisms. If the surface is exposed to vacuum
Chapter 10

Black Body Radiation

A true black body is an idealization but a close representation is possible with carefully
constructed laboratory models. The advantage of using the idea of a black body is that
it provides a standard against which real surfaces can be compared. This comparison is
possible because emission from the black body also permits a first principles derivation
using quantum mechanics1 . In the present chapter, we utilize results of such an analysis,
but the derivation is beyond the scope of the book To this extent, the emission law from
a black body is stated here as a constitutive relationship for radiative heat transfer.
Engineering problems in radiation can then be solved by combining the constitutive
relation with energy balance laws.
Prevost’s law states that energy is emitted by all surfaces whose temperatures exceed
zero kelvin. The emissive power of a surface is calculated on the basis of the energy
emitted per unit area of the surface and unit time. Emissive power is expressed in units
of W/m2 . As in the case of the surface properties, it is possible to define a spectral
directional emissive power to be the radiation emitted per unit time and unit area in
the wavelength interval λ to λ + dλ and angle dω centered around the direction ω. The
spectral hemispherical emissive power is the spectral directional emissive power summed
over all angles, equivalently integrated over all directions2 . The total hemispherical
emissive power is the spectral directional emissive power integrated over all ω and all
wavelengths 0 < λ < ∞. The present chapter deals with the emissive power of a black
body. Properties of real surfaces are discussed in Chapter 11. The suffix b is used to
denote properties of a black surface. Hence, for a black body, the following notation is
made use of:
1. eλb is the spectral directional emissive power (W/m2 -μm-sr).
2. eλb is the spectral hemispherical emissive power (W/m2 -μm).
3. eb is the total directional emissive power (W/m2 -sr).
1
The analysis is based on the work of Max Planck, a German physicist.
2
The direction ω referred here is in spherical coordinates defined in terms of θ and φ. This point is
discussed in Chapter 9.
Chapter 11

Properties of Non-black Surfaces

A black surface has spectral emission characteristics that follow Planck’s law. A non-
black surface, by definition has emitted radiation whose spectrum does not follow the
black body spectrum. It can then be shown that the emissions of non-black surfaces are
always less than that of a black body (for each wavelength and direction)1 .

Remarks:
1. All real life surfaces are non-black. Some of them approach black body behaviour
under certain limiting conditions2 .
2. Since emissions from real surfaces are wavelength and direction dependent, the
surface properties are also functions of wavelength and direction. Under excep-
tional circumstances, this functionality can be ignored and one can work with
average values.
3. For a black surface, α = = 1 and ρ = τ = 0. This statement holds for every
wavelength and every direction. Hence, for a black surface
λb = λb = b = 1
αλb = αλb = αb = 1
ρλb = ρλb = ρb = 0

τλb = τλb = τb = 0
For a real surface (each wavelength, each direction), α < 1, < 1, ρ > 0 and τ > 0.
In addition, real surface properties can also depend on temperature. In practice,
surfaces degrade with age and hence the radiation properties are time-dependent
as well.
1
This remark is related to the statement of Chapter 9 that a black body is the greatest absorber of
radiation in every direction and at every wavelength.
2
Examples are surfaces with coatings and roughened surfaces.
Chapter 12

Energy Exchange between Black

Surfaces

Radiative exchange of thermal energy between surfaces that are at distinct tempera-
tures will be of interest to the present as well as succeeding chapters. For the present
discussion, the surfaces are taken to be black. The surface properties do not appear in
the formulation since α = = 1 while ρ = τ = 0. The intervening medium is taken to be
vacuum (or gas) for which the refractive index is (practically) unity. Thus, energy ex-
changes accounting for all wavelengths and all directions (hemispherical total exchanges)
are considered1 . For a black surface, knowledge of the total emissive power eb is equiv-
alent to knowing temperature via the Stefan-Boltzmann relation. With T given, the
spectral emissive power ebλ (λ, T ) can be recovered from Planck’s black body function.
Hence, for black surfaces, spectral and total exchange calculations are equivalent.
Figure 12.1 shows two surfaces in the vicinity of each other that exchange thermal
energy by radiation. These surfaces are labeled 1 and 2 in the figure. Consider point P1
on surface 1 where the temperature is T1 . An elemental area dA1 is identified on surface
1 with point P1 at its center. The corresponding location on surface 2 is P2 with an
elemental area dA2 and a temperature T2 . The line joining points P1 and P2 is P1 P2 , a
distance indicated by the symbol r. The angle made by P1 P2 with the unit normal on
surface 1 is θ1 ; the corresponding angle on surface 2 is θ2 . These angles are measured
in the clockwise direction.
Proceeding with the energy exchange calculations, the expression for energy emitted
can be derived as follows (also see Problem 11).: For the element dA1 in the direction
P1 P2 per unit solid angle per unit time, the emission is eb1 dA1 =ib1 dA1 cos θ1 . This
expression is written in terms of radiation intensity because of the property of isotropy
of black body radiation. Dependence of emissive power on wavelength is presently
ignored (see Chapter 15 for a spectral formulation). The quantity dA1 cos θ1 is the
elemental area projected normal to the direction of P1 P2 . In addition, we can use the
result ib = eb /π = σT 4 /π of Chapter 10.
1
Consequently, the emissive power of a surface at a temperature T is σT 4 .
Chapter 13

Energy Exchange between

Diffuse-Gray Surfaces

Diffuse-gray surfaces have absorptivity and emissivity that are constants with respect
to wavelength and direction. The constant values are, of course, less than unity, the
maximum permissible value. In addition, Kirchoff’s law α = continues to be applicable
(see discussion in Section 11.2.1). In the present chapter, we continue to look at examples
where the transmittivity of the surface is zero. Thus, the incident radiation is partly
absorbed while the rest is reflected. The reflectivity of the surface is simply ρ = 1 − α =
1 − . The reflected radiation is taken to be diffuse1 , namely distributed uniformly in all
directions. In this respect, the emitted radiation is indistinguishable from the reflected.
For a diffuse surface, a given surface will view all other surfaces without preference, to
the extent permitted by the shape factor.
Real surfaces can only be approximated as diffuse-gray since some dependence on
wavelength and specularity (non-diffusive behaviour) is invariably obtained. In addition,
real-surface properties can be functions of temperature itself.
Apart from a non-zero reflectivity, a second point of departure of a gray surface
from black is the amount of radiation leaving the surface. The emitted radiation is
the product of emissivity and the emission of a black body at the surface temperature,
namely × eb . The reflected part is the product of reflectivity and the radiation incident
on the surface, namely ρ×G. Here, the symbol G refers to the incident radiation (Figure
13.1) and is called irradiation. The total radiation leaving the surface, called radiosity
is given a symbol J and expressed as
J = eb + ρG (13.1)
Both G and J are heat fluxes with units of W/m2 . An energy balance applied to the
surface shown in Figure 13.1 yields
Q
=J −G
A
1
as opposed to specular
Chapter 14

Treatment of Variable Wall

Temperature

When some or all surfaces of an enclosure are prescribed with a temperature distribution
(instead of a constant temperature), it is convenient to work with elemental areas and
the energy exchanges among them. Each elemental area at a location x can be taken
to be at a uniform temperature T (x). Over this elemental area, the local heat flux q(x)
is also uniform. The energy exchange calculation for the enclosure is then obtained by
integrating local heat fluxes over the area of the entire edge. As an example, let the
k th surface of the enclosure be given a temperature distribution Tk (xk ), where xk is a
local coordinate over the kth surface. The local heat flux is denoted as qk (xk ). Let us
also assume that the enclosure surfaces are all flat and black. An equation of the type
(Equation 12.7)

j=N
Qk = σTk4 Ak − σTj4 Ak Fkj
j=1;j =k

is now re-written as

qk (xk )dxk = σTk4 (xk )dxk − σT 4 (xs )dxk dFdk−ds
s

Here, dxk is an element on surface k where the energy balance calculations are being
conducted. The coordinate s is tangential to the enclosure surface everywhere. The
interpretation of this equation continues to be: external heat addition = energy leaving
the surface − energy incident on it. The above representation is exact since the energy
balance expression is now applied to a differential element rather than a full edge of an
enclosure.
It is convenient to divide the enclosure into N surfaces where temperatures could be
spatially constant, as with the ambient, or have surfaces of known radiation properties.
Chapter 15

Treatment of Non-Gray Surfaces

Non-gray surfaces have properties such as absorptivity and emissivity that are dependent
on the wavelength of radiation. The radiosity and heat flux formulations of Chapters
13 and 14 carry over with certain modifications1 . These aspects are discussed in the
present chapter.

15.1 Spectral radiosity formulation

When surface properties exhibit strong dependence on wavelength, the following changes
need to be incorporated into the governing equations.
1. The surface properties are wavelength dependent, i.e. we use functions such as
λ (λ) in the radiative energy balance equation.
2. The wavelength-dependent radiosity Jλ is defined as follows: Jλ dλ stands for the
net radiation leaving a surface in the wavelength interval λ and λ + dλ. Jλ has
units of W/m2 -μm. The external heat transfer is analogously represented in a
spectral calculation as Qλ dλ where Qλ has units of W/μm and the spectral heat
flux qλ has the units of spectral radiosity.
3. The black body emissive power eb = σT 4 (in units of W/m2 ) is now replaced by
the spectral hemispherical emissive power eλb (in units of W/m2 -μm). It is given
by the Planck’s formula of Chapter 10, namely
2πC1
eλb (λ, T ) =
λ5 (exp(C2 /λT ) − 1)

4. Calculations for the kth surface of an enclosure repeated over each wavelength
can then be combined to determine the total external heat exchange (or total
1
The problem statement of heat exchange in terms of wavelength is often called spectral.
Chapter 16

Gas Radiation

Several gases and liquids are practically transparent to visible radiation. The transmit-
tivity is close to unity, the associated absorptivity being close to zero. In the presence
of suspended particles, light is scattered and transmittivity is no longer unity. Scatter-
ing is generally uniform in all directions1 ; it redirects energy from a given direction to
others. Consequently, energy transmitted in a particular direction is reduced. One can
then think of scattering as equivalent to absorption of energy.
Even in the absence of particles, molecules of a gas can scatter radiation at specific
wavelengths. Thus, when thermal radiation is exchanged, it is possible that the gaseous
medium shows a non-zero absorptivity for certain wavelengths, being purely transparent
to others.
The process of absorption of energy in a gaseous medium is well-modeled by the
expression (Figure 16.1)
Iλ
= exp (−κλ x) (16.1)
Iλ,0
Equation 16.1 is presented in the context of one dimensional radiative heat transfer.
Here, I is the intensity of radiation in the direction along the x-coordinate and suffix
λ indicates a dependence on wavelength. Suffix 0 refers to the value of intensity at
x = 0, namely the initial intensity before radiation enters the region filled with gas. The
parameter κλ is called absorption coefficient; it is dependent on the chemical composition
of the gas being studied as well as wavelength. Equation 16.1 is called Beer’s law.
Equation 16.1 shows that the gas has a transmittivity equal to

τλ = exp (−κλ x) (16.2)

Hence, it has an absorptivity

αλ = 1 − exp (−κλ x) (16.3)

1
unless the particle is transparent or unsymmetric in shape. In the former, considerable anisotropy
is generated by refraction, internal and external, in radiation leaving the particle surface.
Chapter 17

Combined Conduction and

Radiation

In several applications, heat transfer occurs by the mechanism of conduction as well as

radiation. Heat transfer by joint mechanisms of conduction, convection, and radiation
is also possible, but is not within the scope of the present book. The features seen in
mathematical models of combined heat transfer are revealed in the following examples.

Example 17.1 Heat transfer from a radiating fin.

Consider steady heat transfer from a rectangular fin, whose base temperature is Tb ,
Figure 17.1. The cross-sectional area is A, perimeter P , and the length is denoted
as L. A segment of the fin of thickness Δx at a distance x from the base is at a
temperature T (x). The ambient temperature is specified as Ta . The fin approximation,
namely uniformity of temperature across the fin cross-section is taken to be applicable.
The relevant physical mechanisms of interest are heat conduction within the fin and
convection+radiation to the environment1 . The goal of the analysis is to predict the
temperature distribution in the fin along its length.
For an infinitesimal segment of length Δx located at point 1 (Figure 17.1), the net
diffusive heat flux, namely the difference of heat fluxes crossing its left and the right
faces is
d2 T
net diffusive flux = q(x + Δx) − q(x) = −k 2 Δx
dx
Diffusive heat transfer through the fin of cross-section A is −kA(d2 T /dx2 )Δx. The
difference is accounted for as heat losses to the ambient by convection and radiation.
The former is hc P Δx(T (x) − Ta ), with hc representing the convective heat transfer
1
In the present chapter, convection is given a simplified treatment via a heat transfer coefficient,
which is in turn a prescribed quantity. Explicit determination of the heat transfer coefficient is not
under discussion.
Chapter 18

Radiative Transfer Equation

Radiative heat transfer between surfaces separated by vacuum or by a transparent

medium has been discussed in the previous chapters. In many engineering applications,
the enclosure is filled with an absorbing gas, a semi-transperant solid or liquid, or a
medium with absorbing and scattering particles including gas bubbles in a liquid phase.
Nuclear reactors, plasma and fusion reactors, crystal growth applications involve such
participating media where radiation intensity and possibly its spectral distribution are
altered during the passage of energy from one surface to the other. Radiative energy
entering such a domain may be augmented or attenuated due to absorbtion, emission
and scattering within the material contained in the enclosure. The present chapter
formulates the equation that describes changes in radiation intensity during its passage
in a participating medium 1 .

18.1 Introduction
The radiation intensity in a given direction is constant along its path while traveling
through a transparent medium. This property of intensity makes it a suitable quantity
for description of absorption, emission, and scattering of energy within a radiatively par-
ticipating medium. Any change in intensity along a given path must now be attributed
to one or more of these phenomena. The radiation intensity field within a participating
medium is denoted as Iλ (r, Ω̂). It is a function of location (position vector r), direction
of travel of the beam of radiation (direction vector Ω̂) and wavelength (λ).
At the particulate level, absorption and emission are surface phenomena, analo-
gous to those discussed in the preceding chapters. Scattering, on the other hand, ex-
hibits greater complications both in terms of scale and physical origin. The scattering
mechanism depends on the ratio of the particulate diameter (dp ) and wavelength. For
dp /λ >> 1, the particle casts a shadow in the beam direction. For dp /λ ≈ 1, radiation
1
Material which contributes to augmentation and attenuation of intensity is denoted as participating
medium
Chapter 19

Exact Solutions of the Radiative

Transport Equation

19.1 Introduction
The conservation of energy principle, in general, leads to a non-linear partial differen-
tial equation1 . The radiative flux in the energy equation requires the calculation of
irradiation, an integral that is evaluated in terms of the radiation intensity over the
applicable range of solid angles. The solution of the overall energy equation yields the
temperature field but requires the estimation of intensity field in the domain of inter-
est. The equation of radiative transport (RTE) derived in the previous chapter is an
integro-differential equation for radiative intensity. Thus, for domains involving partic-
ipating media, the contribution of thermal radiation to the energy balance results in a
non-linear integro-differential equation.
For low temperature applications, the contribution of thermal radiation is negligible
with respect to the overall heat transfer. More generally, however, radiative intensity
is a function of temperature. Hence, the equation of radiative transport (RTE) and
the overall energy equation are coupled via the temperature dependent intensity field.
Decoupling of these two equations are possible for two simplified cases: (a) known
temperature field in the domain, and (b) for radiative equilibrium conditions when
thermal radiation is the only mode of heat transfer. In the absence of other modes of
heat transfer except radiation, the divergence of radiative heat flux is zero. This results
in an integral equation relating temperature with the intensity field via irradiation.
The steady state equation of radiative transfer (RTE) involves six independent vari-
ables: the wavelength of radiation, three space coordinates and two coordinates describ-
ing the direction of travel of photons. In addition, absorption, scattering and emissive
properties of the medium are required to be known for a complete solution. Radiative
properties are strongly dependent on wavelength and are usually difficult to measure.
1
Nonlinearities may be related to dependence of material properties on temperature, nonlinear
boundary conditions, and dependence of the velocity field on temperature.
Chapter 20

Numerical Solution of the

Radiative Transport Equation

20.1 Introduction
The mathematical treatment of heat transfer through radiatively participating media
is complex. This is because a separate equation of transport (RTE) is required to be
solved to determine the contribution of the radiative heat flux to the combined energy
equation1 . The radiative transport equation is of the integro-differential type involving
two independent direction coordinates and three spatial coordinates. For a non-gray
medium, wavelength will appear as an additional independent variable. In addition,
the equation of radiative transport is coupled to the combined energy balance equation.
Decoupling is possible for two simplified cases. One, the temperature field is known
and so the energy equation need not be solved. The solution of radiative transport
equation provides the intensity field. The radiative heat flux can then be obtained from
the intensity field. Secondly, if radiative equilibrium prevails, the temperature field can
be obtained from the intensity field.
Strictly, neither of these approximations is realized in engineering problems. Hence,
an iterative procedure needs to be adopted. The intensity field is first obtained by
solving RTE using an initially guessed temperature field. The divergence of radiative
heat flux is then estimated from the intensity field. The overall energy balance equation
involving the divergence of radiative heat flux is then solved to improve the estimate of
the temperature field. Iterations continue till the temperature field attains convergence.
Along with coupling between RTE and the combined energy equation, it is seen
that RTE is an integro-differential equation that can seldom be analytically solved.
The present chapter deals with the approximate solution of RTE by the process of
discretization. The temperature field is taken to be known, at least as an approximation,
that can subsequently be improved by iteration.
1
The combined equation includes all modes of heat transfer - conduction, convection, and radiation.
Chapter 21

Combined Conduction,
Convection and Radiation

21.1 Introduction
To estimate the temperature field in a radiatively non-participating medium, the energy
equation that includes the diffusive and the convective terms is solved. The contribution
of thermal radiation enters the calculation in the form of a nonlinear boundary condition
at the enclosure surfaces. When the surface temperatures are known, the conductive,
convective, and radiative, heat fluxes at the enclosure walls can be obtained individually
and simply added to obtain the total heat flux1 . This approach was discussed in Chapter
17. If the medium is a stationary solid exposed to the ambient, heat transfer will occur by
conduction with a combined convective-radiative boundary condition for heat transfer
to the environment. If the medium is a non-participating flowing fluid, heat transfer
will be by conduction and convection with a convection-radiation boundary condition
for representing heat transfer to the ambient.
In many instances, the convective field may depend on temperature distribution
itself, as can happen when fluid motion is by the action of buoyancy. Here, radiation can
alter temperature gradients and hence the strength of convection itself. The additive
approach to estimating heat fluxes referred above will not be appropriate. Here, it
would be necessary to solve the system of equations of flow and heat transfer with
radiative boundary conditions all together, and estimate wall fluxes from the resulting
temperature distribution.
For a radiatively participating medium, the radiative heat flux originates from the
surface and the medium contributes to its attenuation by absorption and scattering. The
flux is augmented by local emissions from the medium itself. These mechanisms lead to
the incorporation of the divergence of radiatve heat flux into the overall energy balance
for an elementary volume in the domain. The divergence of radiative heat flux involves
1
If the material velocity is zero at a wall, the convective flux is also zero. On the other hand,
convection steepens temperature gradients at the wall and enhances heat transfer.