Chapter 1

Radiation Exchange Between Surfaces

1.1 Motivation and Objectives

Thermal radiation, as you know, constitutes one of the three basic modes (or mechanisms) of
heat transfer, i.e., conduction, convection, and radiation. Actually, on a physical basis, there are
only two mechanisms of heat transfer; diffusion (the transfer of heat via molecular interactions)
and radiation (the transfer of heat via photons/electomagnetic waves). Convection, being the
bulk transport of a fluid, is not precisely a heat transfer mechanism.
The physics of radiation transport are distinctly different than diffusion transport. The latter is a
local phenomena, meaning that the rate of diffusion heat transfer, at a point in space, precisely
depends only on the local nature about the point, i.e., the temperature gradient and thermal
conductivity at the point. Of course, the temperature field will depend on the boundary and initial
conditions imposed on the system. However, the diffusion heat flux at, say, one point in the system
does not directly effect the diffusion flux at some distant point. Radiation, on the other hand, is not
local; the flux of radiation at a point will, in general, be directly and instantaneously dependent on the
radiation flux at all points in a system. Unlike diffusion, radiation can act over a distance.
Accordingly, the mathematical description of radiation transport will employ an integral equation for
the radiation field, as opposed to the differential equation for heat diffusion.
Our objectives in studying radiation in the short amount of time left in the course will be to
1. Develop a basic physical understanding of electromagnetic radiation, with emphasis on
the properties of radiation that are relevant to heat transfer.

2. Describe radiation exchange among surfaces, in which the surfaces can be perfect
absorbers of radiation (black) or diffusely absorbing (gray).

3. Introduce the topic of radiation transfer in a participating medium that absorbs, emits, and
scatters radiation, and describe the formulation and application of the radiative transport
equation.

1
2 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

1.2 Basic Radiation Properties

For our purposes, it is useful to view radiation as the transport of energy in electromagnetic
waves. Radiation can also be viewed as the transport of energy by discrete photons, and the
basic relationship between the energy of a photon, ǫ, and the frequency ν or wavelength λ = c/ν
of the wave is given by Planck’s relation;
ǫ = h ν = hc (1.1)
λ
where h is Planck’s constant and c is the speed of light in a vacuum.
Equation (1.1) indicates that the energy of the radiation is inversely proportional to the radiation
wavelength. Thermal radiation refers to the spectrum of radiation in the visible (λ = 0.4 −0.7 µm) to
infrared (IR, λ = 0.7 ∼ 10 − 100 µm) wavelengths. Radiation at these wavelengths can excite the
rotational and vibrational energy levels of molecules and thus transfer energy to molecules in the form of
heat. That is, the absorption of thermal radiation by molecules will act to raise the temperature of the
system. Radiation having shorter wavelengths (UV, x–rays, γ–rays) can excite the electronic energy
levels of molecules and atoms and/or ionize or break molecular bonds. This spectra of radiation is often
referred to as ionizing radiation. On the other hand, longer wavelengths (microwaves, radio waves) will
not, in general, couple with the energy storage levels in molecules; although an obvious exception are the
microwave wavelengths used in the common microwave oven.
In practically all engineering–relevant applications, the source of thermal radiation is thermal
emission. All bodies at any finite temperature will emit radiation. It was theoretically established
by Boltzmann, and experimentally confirmed by Stephan, in the 19th century that the maximum
possible emission rate from a surface is given by

eb = σT 4 (1.2)
8 2 4
where σ = 5.67 × 10 W/m K . The above formula is known as the Stephan–Boltzmann law, and eb, having units of
2
W/m , is the blackbody emissive power, which depends solely on the absolute temperature T of the surface. All real
surfaces will emit radiation at a rate smaller than e b; an ‘ideal’ surface which attained an emission of e b would be
referred to as a blackbody.
It is possible to construct devices which come very close to meeting the ideal emission of a
blackbody. Typically, these devices are formed from an isothermal cavity (i.e., a hollow sphere),
with a small opening from which the radiation escapes. The radiation emitted by a blackbody at
a specified T will be distributed over wavelength λ, the spectrum of which will also depend solely
on T . Although the nature of the blackbody spectrum was experimentally known in the late 19th
century, theoretical prediction of the spectrum defied the ‘classical’ physical understanding of
the day. Planck, at the beginning of the 20th century, used the concept of discrete wavelength
energy levels to develop a theoretical prediction of the blackbody spectrum which was
consistent with experiments. His formula for the spectral blackbody emissive power is
C1
e = 5
bλ
λ (exp[C2/(λT )] − 1) (1.3)
1.2. BASIC RADIATION PROPERTIES 3

106
1000 K
1500 K
5 2000 K
10

104

103

2
ebλ, W/m2  m
10

101

100

−1
10

10−2

10−3
−1 0 1 2
10 10 10 10
λ,  m

Figure 1.1: Blackbody spectral power distribution

2 8 4 2
in which C1 = 2πhc = 3.7413 × 10 W (µm) /m and C2 = hc/kB = 14, 388 µm K. The quantity ebλdλ denotes the energy
emitted from the black surface, per unit area, within a wavelength interval dλ about wavelength λ. The energy over all
wavelengths is obtained from
Z ∞
eb = ebλ dλ = σ T 4 (1.4)
0

i.e., the total emissive power is given by the Stephan–Boltzmann law, as it must.
A plot of ebλ vs. wavelength λ is given in Fig. 1.1. As T increases two things happen: 1) the
spectral emissive power at all wavelengths increase, and 2) the wavelength corresponding to
the maximum power shifts towards the shorter wavelengths. The value of the wavelength at
which the maximum occurs is predicted by the Wien displacement law,
λ = 2898 µm K
max (1.5)
T

Equation (1.4) can be rearranged to identify a fractional distirbution function. Assume that T is
constant, then we can write
∞
1
5
σT Z0 ebλ d(λ T ) = 1 (1.6)
Now combine with Eq. (1.3), and observe that λT becomes the sole variable of the distribution. We can
therefore define
′ C1/σ

5
f (λT ) = (λT ) (exp[C2/(λT )] − 1) (1.7)
4 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

Figure 1.2: Intensity definition

so that
Z λT
f (λT ) = f ′(x) dx (1.8)
0
represents the fraction of emitted energy between 0 and λT , relative to the total emitted energy.

Intensity

The previous section identified the blackbody emissive power from a surface, which is a quantity that has units of W/m 2, that is, the same
units of a heat flux. The emissive power, however, is not a heat flux – the latter is precisely a vector quantity which can be resolved into
directional components. The emissive power, on the other hand, describes the energy leaving a surface per unit area of the surface; it does
not describe the directional characteristics of the radiation as it leaves the surface.
To describe the directional properties of radiation as it propagates through space, we need to introduce a quantity referred to as the
radiation intensity. The intensity, denoted I , is defined by use of Fig. 1.2. Radiation is emitted from a small surface element A and travels
in all directions. A portion of the radiation lands on the small area element on a hemisphere enclosing A, denoted as dA. With regard to
geometrical considerations, it should be easy to see that the net amount of energy falling on dA, which is denoted as dP , will be
proportional to

1. the projected area of the emitting surface, which would be A cos θ,

2. the area of the ‘target’, dA, and

1.2. BASIC RADIATION PROPERTIES 5
2
3. the inverse square of the distance between the source and the target, 1/R .
The last proportionality comes from the fact that the area of the hemisphere scales as R 2, and the total power falling on the hemisphere – which would equal the total power emitted by the
source

– would be constant in the absence of a participating medium. So the power per unit area on the hemisphere would go as 1/R2.
The target area dA can be related to the distance R by the polar coordinates;

dA = R2 sin θ dθ dφ (1.9)
Note that

2 Z 2π
Z dA = R Z0 π/2 sin θ dθ 0 dφ = 2πR2 (1.10)
as expected. Now dP is proportional to dA/R2, and
dA

R2 = sin θ dθ dφ ≡ dΩ (1.11)

defines a differential solid angle dΩ. The units of solid angle are the steradian (abbreviated str), and 4π steradians encompass all directions about the origin of a spherical
coordinate system.
With our solid angle definition, we now see that the power falling on the target will be pro-portional to the projected area of the source
and the solid angle subtended by the target with respect to the source. The power will also be proportional to the magnitude and
directionality of the radiation leaving the source and propagating through space. Since dP will be in watts, and dP will be proportional to A
2
cos θ dΩ, it follows that radiation propagating through space must have units of W/m str. This quantity is referred to as the intensity,
and is defined by
dP
I= (1.12)

A cos θ dΩ

taken in the limit of A−→0 and dΩ−→0.

The intensity is a difficult quantity to grasp. Although it represents the directional distribution of propagating radiation, the intensity is not a
vector in the sense that it cannot be resolved into vector components. Rather, the intensity is a scalar that is dependent on the directional coordinates

θ and φ as well as the usual spacial coordinates and time. In a sense, I is a scalar which is relevant to the 5–D ‘space’ defined by x, y, z and θ, φ.

It was implicitly assumed that the intensity, defined in the Eq. (1.12), represents a wavelength– integrated (or total ) quantity. The
spectral intensity Iλ is defined by a relation similar to Eq. (1.12), except now on a per–unit–wavelength basis;

dPλ
Iλ = (1.13)
A cos θ dΩ dλ
6 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

where the limit is again taken, and dPλ is the power, in W, falling on the surface dA within the wavelength interval dλ about λ. It follows that
Z∞
I= Iλ dλ (1.14)
0

Blackbody Intensity

If the surface A is a black surface – meaning that it emits as a perfect blackbody – the intensity distribution leaving the surface will be

independent of direction, i.e., the blackbody intensity Ib landing on the hemisphere in Fig. 1.2 would not be a function of θ. It is important to note
that the power per unit area landing on the hemisphere, dP /dA, would be a function of θ; this quantity would be proportional to the projected
area of the source, A cos θ. The intensity, however, represents the power per unit projected area, and so this quantity would remain constant.
The total power leaving the black source would be P = eb A, and in the absence of an intervening medium the total power leaving the source
would be the total power landing on the hemisphere. If we use Eq. (1.12), and note that the intensity I = Ib is constant, then

P = eb A

Z dP = I A Z cos θ dΩ
= b Ω=4π

2π Z
= Ib AZ0 dφ 0 π/2 cos θ sin θ dθ
= π Ib A (1.15)
or
eb
Ib = (1.16)
π

Radiative Heat Flux

As mentioned previously, the intensity, which is a scalar, describes the directional distribution of radiation energy transfer. The
radiative heat flux, qR, is a vector which describes the net direction and magnitude of radiant energy transfer in space.
Say we take a small surface element in space, dA, oriented so that its normal points in the positive z direction. Radiation, arriving from
all directions with an intensity distribution I (θ, φ), passes through the element. The net rate at which energy is transported across the
surface, denoted as dP , will be
Z 2π Z π
dP = dA I (θ, φ) cos θ sin θ dθ dφ
φ=0 θ=0

Note that the cos θ term appears because dA cos θ is the projected area of the element with respect to the θ direction. The quantity dP will be either positive, negative, or zero; if it is
positive, it
1.2. BASIC RADIATION PROPERTIES 7

means that the net transfer of energy across dA is in the positive z direction, negative means
the net energy transfer is in the negative z direction, and zero means that there is no net
transfer. The last condition does not necessarily imply that I = 0; rather, it implies that the
positive and negative contributions cancel out. Such would be the case if the intensity
distribution were isotropic, i.e., independent of direction.
The radiative heat flux, in the z direction, would be dP /dA. The three components of
radiative flux with respect to a cartesian coordinate system are given by
2π π

Z Z 2
qR,x = φ=0 θ=0 I (θ, φ) sin θ cos φ dθ dφ (1.17)
2π π

Z Z 2
qR,y = φ=0 θ=0 I (θ, φ) sin θ sin φ dθ dφ (1.18)
2π π

Z Z
qR,z = φ=0 θ=0 I (θ, φ) cos θ sin θ dθ dφ (1.19)
1.2.1 Surface Properties
Surfaces can emit, absorb, and reflect radiation. You are certainly familiar with the basic
concepts of surface absorptivity and reflectivity, i.e., the absorptivity α represents the fraction of
incident radiation on a surface that is absorbed by the surface. We will need to develop more
precise definitions than this to account for the directional and spectral dependencies of the
radiation falling on or emitted by a surface.

Emissivity
As mentioned in the previous section, the intensity emitted by a blackbody is independent of direction and
given by Ib = eb/π. Likewise, the spectral (or wavelength–dependent) blackbody intensity is independent of

direction and given by Ibλ = ebλ/π. The emitted spectral intensity from a real surface, into some direction θ, φ
and within a small wavelength interval about λ, will always be less than the spectral blackbody intensity for a
surface at the same temperature. We can therefore define the spectral directional emissivity ǫ′λ as

I (θ, φ)
ǫλ′ = e,λ (1.20)
I
bλ

in which Ie,λ is the emitted intensity from the surface; it does not include components due to other
sources such as reflected intensity. The spectral directional emissivity represents the most
fundamental, or irreducible, information on the emission properties of a surface. In general, it
would be a function of direction, wavelength, temperature, and the physical and chemical
properties of the surface.
8 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

We can obtain averaged emissivity properties by integration of ǫ′λ. The spectral hemispherical emissivity represents the
directional average of ǫ′λ, and is defined by
2π π/2

Z
ǫλ =Z φ=0 θ=0 Ie,λ(θ, φ) cos θ sin θ dθ dφ
2π π/2

Z Z
φ=0 θ=0 Ib,λ cos θ sin θ dθ dφ
2π π/2

Z
Z φ=0 θ=0 ǫλ′ (θ, φ) cos θ sin θ dθ dφ
=
π
= eλ (1.21)
eb,λ

That is, ǫλ is the ratio of the emitted spectral power from the surface, eλ to the blackbody power
eb,λ.

′ ′
Likewise, the total directional emissivity ǫ is obtained from an appropriate wavelength average of ǫ λ;
Z∞
Ie,λ dλ
′ 0
ǫ =Z∞
Ibλ dλ
0
Z
∞

ǫλ′ Ib,λ dλ
0
= (1.22)
Ib

Finally, the total hemispherical emissivity ǫ is obtained from either a directional average of ǫ′ or a wavelength average of ǫλ;
either would yield the same result, which is
Z ∞ Z 2π Z π/2
Ie,λ(θ, φ) cos θ sin θ dθ dφ dλ

ǫ= λ=0φ=0 θ=0

Z∞Z 2π Z π/2
Ibλ cos θ sin θ dθ dφ dλ
λ=0 φ=0 θ=0
Z 2π Z π/2
ǫ′(θ, φ) cos θ sin θ dθ dφ
= φ=0 θ=0

Z π
∞
ǫλ Ibλ dλ
λ=0
=
I
bλ
e
= (1.23)
eb
1.2. BASIC RADIATION PROPERTIES 9

Absorptivity

−
Consider now a surface that is exposed to an incident source of spectral intensity, denoted as Iλ (θ,
φ). The – superscript indicates that the radiation is moving downwards onto the surface. When the
radiation strikes the surface a fraction of it will be absorbed by the surface, the remainder will be
1 ′
reflected . Denote the absorbed intensity as Ia,λ. The spectral directional absorptivity αλ is defined by
α′
I
= a,λ
(1.24)
λ Iλ−

That is, it is the fraction of incident spectral intensity that was absorbed by the surface.
′
Similar to the spectral directional emissivity ǫ λ, the spectral directional absorptivity describes
the fundamental absorption properties of the surface. And as was done with the emissivity, we
can define averages, with respect to direction, wavelength, or both, by appropriate integrations.
The spectral hemispherical absorptivity αλ is

2π π/2

Z
φ=0 Z θ=0 Ia,λ(θ, φ) cos θ sin θ dθ dφ
αλ = 2π π/2

Z Z
φ=0 θ=0 Iλ−(θ, φ) cos θ sin θ dθ dφ
2π π/2

Z
φ=0 Z θ=0 αλ′(θ, φ) Iλ−(θ, φ) cos θ sin θ dθ dφ

=q qλ−
a,λ

= qλ− (1.25)

where qλ− and qa,λ are the downward spectral flux on the surface and the spectral absorbed flux. The total
directional absorptivity α′ is
Z ∞
Ia,λ dλ
0
α′ = ∞
Z
Iλ− dλ
0
Z ∞

αλ′ Iλ− dλ
0
−
= I (1.26)
1 Some of the radiation might also be transmitted through the surface, but at this point we will not make this
distinction; if the radiation is not reflected, then it went into the surface material and was absorbed by it
10 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

and the total hemispherical absorptivity α is

∞ 2π π/2

Z Z Z
λ=0 φ=0 θ=0 Ia,λ(θ, φ) cos θ sin θ dθ dφ dλ
α=
Z ∞ Z 2π Z π/2
Iλ− cos θ sin θ dθ dφ dλ
λ=0 φ=0 θ=0
Z 2π Z π/2
α′(θ, φ)I − cos θ sin θ dθ dφ
= φ=0 θ=0 −
q
Z
∞
−
αλ qλ dλ
λ=0
= q−
q
= a (1.27)
q−

All these definitions look analogous to those for the emissivity. An important distinction, though, is in regard to the ‘weighting
function’ used to obtain the averages. The blackbody inten-sity Ib is not a function of direction, so it could be removed from the integrals
over direction in Eq. (1.21) and (1.23) for the hemispherical emissivities. On the other hand, the incident intensity I − (either spectral or total)
is, in general, a function of direction, and it cannot be removed from the corresponding formulas for hemispherical absorptivity in Eqs.
(1.25) and (1.27). This points out an important fact: the hemispherical absorptivity will be a function of the directional intensity distribution
falling on a surface. For example, a surface illuminated by the sun from the normal direction would, in general, have a different
hemispherical absorptivity than the same surface il-luminated by the sun at an oblique angle, or by the sun on a cloudy day (diffuse
illumination). Likewise, the total absorptivity (either directional or spectral) will be a function of the spectral distribution of the incident
radiation. A surface illuminated by visible light would likely have a different total absorptivity than the same surface illuminated by IR
radiation.

Reflectivity

Reflectivity is one step more complicated than absorptivity. On the most basic level, the reflectivity will depend on the angle of the incident radiation as
− + ′ ′
well as the angle of the reflected radiation. As before, denote as I λ (θ, phi) the incident intensity, and let Iλ (θ , φ ) denote the reflected intensity from the
′ ′ ′′
surface in the direction θ , φ . The spectral bidirectional reflectivity ρ λ is defined by

′′ ′ Iλ+(Ω′)
ρλ(Ω, Ω )= − (1.28)

πIλ (Ω)
1.2. BASIC RADIATION PROPERTIES 11

The reason for the π in the denominator will become obvious shortly. The spectral directional– hemispherical
reflectivity ρ′λ is obtained by integration of the reflected intensity over the hemisphere;
2π π/2
Z + ′ ′ ′ ′ ′ ′
′ φ′ =0 Z θ′ =0 Iλ (θ , φ ) cos θ sin θ dθ dφ
ρλ = πI −(θ, φ)
λ
=
π
Z
φ ′ =0
2π
Z
θ′ =0
π/2
ρλ(θ, φ, θ , φ ) cos θ sin θ dθ dφ (1.29)
1 ′′ ′ ′ ′ ′ ′ ′

All of the radiation incident on the surface is either absorbed by the surface (again we assume
transmission through the material counts as absorption) or reflected. Consequently,
α′ + ρ′ =1 (1.30)
λ λ

Formulas for the hemispherical–hemispherical reflectance are analogous to used for the
hemispher-ical absorptivity.

Kirchoff ’s Law
Only when one goes to the spectral directional level does the absorptivity become independent of
′
the properties (spectral, directional) of the intensity falling on the surface. For a given surface, αλ , at
a particular direction and wavelength, would not depend upon whether the incident intensity at this
direction and wavelength was produced from, say, a laser or an incandescent source.
The spectral directional absorptivity αλ′ is therefore a function solely of the surface material properties – as is the case
with ǫ′λ. Indeed, it can be shown from thermodynamic principles that the emissivity and absorptivity are equal on the spectral
directional level,

α′ = ǫ′ (1.31)
λ λ

This equality is commonly referred to as Kirchoff ’s law.

It is easy to show, using Kirchoff ’s law and Eqs. (1.25–1.27), that if the incident intensity arriving
at a surface originates from a blackbody that a) completely surrounds the surface, so that the
incident intensity is independent of direction, and b) is at the same temperature of the surface, so
that the incoming intensity has the same spectra as a blackbody emission from the surface, then
emissivity and absorptivity will be equal on the hemispherical and total levels. This condition, i.e.,
equal temperatures of source and target, would correspond to thermal equilibrium for which there
could be no net heat transfer between the source and target. In most engineering applications of
relevance, the incident intensity on a surface will be directionally depend, and will originate from a
source that is not at the surface temperature. And for such cases Kirchoff ’s law, in general, will not
hold at either the directional or the hemispherical levels.
12 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

We can, however, apply approximations to extend the application of Kirchoff ’s law. Firstly, a
surface with emission and absorption properties that are independent of wavelength will have
equal emissivity and absorptivity on the total level. From Eqs. (1.22) and (1.26), it can be seen
that if ǫ′λ = ǫ′ =6 func(λ), then ǫ′ = α′. Surfaces with wavelength–independent properties are
referred to as gray surfaces.
Likewise, if the spectral directional emissivity is constant for all directions, i.e., ǫ ′λ = ǫλ =6 func(theta, φ), then ǫλ = αλ.
Surfaces with directionally–independent properties are referred to as
diffuse.
Only for surfaces that are both gray and diffuse can Kirchoff ’s law be applied at the total
hemispherical level, i.e.,
ǫ = α, gray, diffuse surfaces (1.32)

The diffuse approximation is relatively accurate for surfaces that have a roughness on the
scale of the radiation wavelength or greater, such as oxidized metals, wood, paper etc. Diffuse
absorbers/emitters will also be diffuse reflectors, meaning that the reflection of intensity from a
surface is isotropic (independent of direction), regardless of incident direction of the intensity.
The gray approximation is more of a stretch of reality. It is fairly accurate when the source of
incident radiation is a blackbody at a temperature close to the temperature of the surface onto
which it is falling; for such cases the spectra of the incident and emitted intensities will be about
equal.
The gray assumption can fail miserably, however, when the incident radiation has a
significantly different spectra than the emitted radiation. A common example is sunlight falling
on a solar collector. The radiation spectrum of sunlight is similar to that of a blackbody at T s ≈
5800 K, and is concentrated mainly in the visible wavelengths. Solar collectors, on the other
hand, will typically operate at a temperature of around Tc ≈ 350 K, and emission at this
temperature will be concentrated in the mid IR wavelengths. For such conditions, a collector
with a high spectral emissivity in the visible yet a small spectral emissivity in the IR (which is a
desirable quality for collectors) would have α ≫ ǫ.

1.3 Radiosity and irradiance

The basic idea of this section is as follows: given N surfaces, which can exchange radiation heat
transfer among each other, calculate the net rate of heat transfer to each surface.
We will assume that the surface exchanging radiation have diffuse surface properties, in that
emissivity and absorptivity are not a function of direction. However, the properties are initially
taken to be wavelength–dependent, which implies that, in general, α =6 ǫ. Once the
formulations are complete, we will examine the simplified situation of gray surface properties.
We begin by defining some basic quantities for use in radiation exchange. Consider a surface,
denoted ‘surface 1’ and having area A1. The properties of the surface, including the temperature
1.3. RADIOSITY AND IRRADIANCE 13

T1, total (hemispherical) emissivity ǫ1 and absorptivity α1, are assumed to be uniformly constant over the surface.

The irradiance, H , is the flux of radiant energy falling on a surface, averaged out over the surface area of the surface. That is, the total radiant energy falling on the
surface is
Z Z T
H1A1 = I − cos θ1 dΩ dA1 (1.33)
1

A1
− −
in which I1 denotes the intensity falling on the surface, which is (implicitly) a function of incident direction θ, φ. Brewster uses the symbol q
for the irradiance. The radiosity, J , is the flux of radiant energy leaving the surface, averaged over the surface area. The total rate of radiant
energy leaving the surface is
Z T
Z

J 1A1 = I + cos θ1 dΩ dA1 (1.34)

A1
The radiosity can be related to the irradiance by
J1 = ǫ1eb1 + (1 − α1)H1 (1.35)
4
i.e., radiosity will consist of emission from the surface (in which eb1 = σT1 ) plus the reflected part of the irradiance (with ρ1 = 1 −
α1). On the other hand, irradiance H1 will depend explicitly on the incoming radiation field at the surface, which, in turn, will depend on
the outgoing radiation fields from all the other surface which can ‘view’ surface 1. We’ll encounter the explicit formulas shortly.
The net average radiative flux from the surface, denoted q, will simply be the difference between the flux leaving the surface and the flux arriving at the surface, i.e.,

q1 = J1 − H1 (1.36)

Note that this formula is not, explicitly, a function of the surface properties ǫ or α. However, an alternative formula for q can be stated, in
which q is the difference between the emissive flux from the surface and the absorbed incident flux,
q =ǫ e −α H
1 1 b,1 1 1 (1.37)
If the surfaces are in steady state (which we assume to be the case), then the net radiative heat transfer rate from the surface, q1A1, will equal the net rate of heat
transfer to the surface by other means such as conduction or convection. That is, q1A1 is the rate of ‘external’ heat transfer required to keep the surface at a constant
temperature. If the surface is adiabatic, then q1A1 = 0.
By eliminating H1 among the previous two equation, we get

1
q1 = 1 − α1 (ǫ1eb1 − α1J1) (1.38)
This equation is not too useful for a black surface (i.e., α = ǫ = 1). For this special case Eq. (1.35) shows that J = eb, but we will need to use either Eqs. (1.36) or (1.37) to get the heat transfer q1.
14 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

1.4 The Configuration Factor

Say our ‘system’ contains N surfaces, on each of which the temperature Ti is specified. We want to calculate the net heat transfer rate to
each surface per Eq. (1.38). To do so, we need to determine the radiosity Ji at each surface (assume, for the moment, that the surface are not
black). To get the radiosity, however, we’ll need to know the irradiance at the surface, per Eq. (1.35). And to get the irradiance, we need to
know how radiation is exchange among the various surfaces. The relevant formula to evaluate is Eq. (1.33), repeated here as
Z Z T
H1A1 = I − cos θ1 dΩ dA1 (1.39)
1

A1

To simplify the evaluation of this (without a tremendous loss in generality), take the system to consist of a pair of surfaces, 1 and 2. The
‘background’ (i.e., what surrounds 1 and 2) is taken to be black at zero K, so that no radiation originates from the background. In this case
all of the radiation arriving at 1 originates (either through emission or reflection) at 2, and the integral over solid angle in Eq. (1.39) will
include only those directions which point towards surface 2. Say a point on 2 is located a distance R from a point on 1. The differential solid
angle dΩ in Eq. (1.39) will be, by definition,
cos θ2 dA2
dΩ = (1.40)
R2
in which θ2 is the angle between the normal on 2 and the direction vector from 1 to 2. Alternatively, cos θ2 dA2 is the projected area of dA2 as
seen from the point on 1. The medium between 1 and 2 is non–participating; it does not absorb or emit radiation along the path. Consequently,
− +
for a specified path between 1 and 2, I1 = I2 . That is, the intensity arriving at 1 along the path is the same as the intensity leaving 2 along the
same path. Finally, the surfaces are assumed to be diffuse, so
+ J2
I2 = (1.41)
π
Now replace the two previous equations into Eq. (1.39). The radiosity J2 is not a function of position on surface 2 (recall that it is averaged
over the surface area), so it can be taken out of the integrals. We get
cos θ1 cos θ2

Z Z
H1A1 = J2 A1 A2πR2 dA2 dA1 (1.42)

The cluster of integrals depends only on the geometrical configuration of surfaces 1 and 2, and defines a configuration factor F2−1
cos θ1 cos θ
2
F A ≡Z Z
2−1 2 A1 A2 πR2 dA2 dA1 (1.43)
so that
H A =J F A
1 1 2 2−1 2 (1.44)
1.5. EXCHANGE EQUATIONS 15

for our 2–surface system.

The configuration factor Fi−j represents the fraction of radiant energy leaving i that arrives at
j Basic properties of the configuration factor are reciprocity,

F A =F A
i−j i j−i j (1.45)
which follows directly from Eq. (1.43) by exchanging the subscripts 1 and 2, and summation,
N
X

Fij = 1 (1.46)
j=1

in which N is the total number of surface that can view surface i. This latter property simply
states that all the radiation leaving i must end up somewhere. Note also that j = i must also be
included in the summation, as Fi−i is not necessarily zero.

1.5 Exchange equations

Equation (1.44) can be generalized to an N –surface system,

N
X
H A =J F A
i i j j−i j (1.47)
j=1

We can now use reciprocity, i.e., Fj−iAj = Fi−j Ai, in the above and cancel out the area Ai;

N
X

Hi =
JF
j i−j (1.48)
j=1

Replacing this into Eq. (1.35) gives a system of equations for the radiosities,

N
J F
Ji = ǫiebi + (1 − αi) j=1
j i−j (1.49)

X
4
with i = 1, 2, . . . N . If we know the temperature of each surface (from which e bi = σTi ) and we also
know the configuration factors, then the system of equations can be solved for the radiosity at each
surface. And once we know the radiosity, we can get the heat transfer per Eqs. (1.36–1.38).
Frequently, the heat transfer to a surface is known, and the temperature of the surface becomes
an unknown. A typical example is the insulated surface, for which qi = 0. For such cases Eq. (1.49)
16 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

will not be useful, since ebi will be unknown. To remedy the problem we use Eq. (1.36) to formula the
radiosity equations, to get
N
X

q = J +J F
i i j i−j (1.50)
j=1

Once we solve for the radiosities, the temperature of the surface can be obtained from Eq. (1.37).
X

This equation has a network interpretation: if we multiply the Ji term by Fi−j = 1 (recall the
j

summation property), then

N
X

qiAi = (Ji − Jj )Fi−j Ai (1.51)

j=1

The heat transfer from i can therefor be interpreted as a sum of currents flowing from i to all other
surfaces, with Ji − Jj being the potential (or voltage) difference and 1/Fi−j Ai the resistance between i and j.
The general procedure is as follows: say our system has M < N surfaces on which the tem-perature is prescribed, and N −
M surfaces on which q is prescribed. We apply Eq. (1.49) to the
M surfaces with specified temperature, and Eq. (1.50) to the N − M surfaces with specified q. Altogether we obtain N
linear equations for the radiosities J1, J2, . . . JN . And once we have solved for these quantities, we can calculate either
the heat flux or the temperature of the surface.

1.5.1 Spectral considerations

The previous formulas can be applied on a spectral level by simply appending the λ subscript to all relevant
quantities. In doing so, however, the spectral heat flux qλ,i to surface i can no longer be viewed as the heat transfer
rate to the surface by external means. Rather, the total flux, i.e.,
Z ∞
q = q dλ
λ,i λ,i (1.52)
0

is the heat transfer by external means. This means that, in general, it is difficult to explicitly state
specified heat flux boundary conditions in the spectral exchange equations, because we don’t
(beforehand) know the heat flux on a spectral level. In particular, an adiabatic surface has a total
heat transfer rate of zero, yet the spectral flux to the surface could be positive or negative at various
wavelengths in such a way that the total, when integrated out per the previous formula, is zero.
It is also difficult to accurately apply the exchange equations on a total (wavelength integrated)
basis to surfaces that strong wavelength variations in emissivity/absorptivity (i.e., non–gray sur-
faces), especially in conditions in which significant temperature differences exist among the
surfaces. The problem here is obtaining an estimate of the total absorptivity on each surface prior to
solv-ing the equations. The total absorptivity of a surface depends on the spectrum of the incident
radiation, yet the incident radiation (the irradiance) will depend on the emission and reflection of
1.6. GRAY APPROXIMATION 17

radiation throughout the entire system. Indeed, the exchange formulation presented above, with
the corresponding configuration factors, does not offer a clean way of predicting the radiation
spec-trum that falls on a given surface. To see this, note that the absorbed flux (on a total basis)
at surface i will be
N N
=αH =
X X
q
abs,i i i Jj Fi−j =(ǫj eb,j + (1 − αj )Hj ) Fi−j
j=1 j=1

Now αi depends on the spectrum of Hi, yet Hi is seen to depend on emission and reflection from
all other surfaces. We can predict the emission spectrum, but we cannot predict the reflection
spectrum without considering the incident radiation on surface j. And so on....
In view of these problems, the common practice is to either apply exchange equations on a
gray basis (α = ǫ), or to use a spectral formulation and solve on a wavelength–by–wavelength
basis. Alternatively, the Monte Carlo procedure (to be discussed in the near future) offers a way
of modeling a non–gray yet wavelength–averaged system.

1.6 Gray Approximation

1.6.1 Two surface systems
Often we deal with simple problems in which radiation is exchanged between a pair of surface,
such as two parallel plates or a pair of coaxial cylinders. In this case the exchange equations
reduce to a pair of equations for J1 and J2. Because the overall system is in steady state, the
heat transfer rates must be balanced by
q1A1 + q2A2 = 0
It is easy enough to solve two linear equations for two unknowns, yet for the general non–gray
case (α =6 ǫ) the resulting equation for the heat transfer rate q 1A1 is algebraically complex and
need not be presented here.
A considerable simplification will occur if we examine the gray simplification, for which αi =
ǫi. For this case, the heat flux, from Eq. (1.38), becomes
ǫA
1 1

q1A1 = 1 − ǫ1 (eb1 − J1) , gray approx. (1.53)

which has the same network form (current=voltage drop/resistance) as Eq. (1.51). The
resistance in the above, i.e., (1 − ǫ)/ǫA, can be viewed as a surface resistance, whereas the
resistance in Eq. (1.51), 1/F1−2A1, is a geometrical (or space) resistance. In the two–surface
problem the network is equivalent to a series circuit with three resistances; two surface
resistances and a single space resistance. And by adding up the resistances, we get
e −e
b1 b2
q1A1 = 1 − ǫ1 1 1 − ǫ2 (1.54)
+ +
F A
ǫ1A 1 1−2 1 ǫ2A 2
18 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

The multiple reflection model

An alternative way of modelling radiation exchange in a simple, two–surface system is to view

the exchange process as a series of reflections. For simplicity, consider a system of parallel flat
plates, for which F1−2 = F2−1 = 1. Assume also that surface 1 is at a finite temperature yet 2 is
at zero K, so that emission occurs only from surface 1 This does not limit the generality of the
approach, for we can model heat transfer between two surfaces at finite temperature as a
superposition of two heat transfers, with each of the two heat transfers corresponding to one
surface at zero K and the other at the finite temperature.
Surface 1 emits heat at a rate ǫ1eb1. This emission travels to surface two, and a fraction ρ2
is reflected. The reflected fraction travels back down to 1, and a fraction ρ1 of this is reflected
back towards 2. And so on. With this picture, the net rate at which radiation leaves 1, i.e., the
radiosity at 1, is
1
2 =ǫ e
J1 = ǫ1eb1 1 + ρ2ρ1 + (ρ2ρ1) + . . . 1 b1 1 − ρ1ρ2
The second line comes from the power series expansion of 1/(1 − x) for x < 1. Likewise, the
irradiance on surface 1 will consist of the first reflection of the emission, ρ2ǫ1eb1, plus all
multiple reflections,
1

H1 = ρ2ǫ1eb1 1 + ρ2ρ1 + (ρ2ρ1)2 + . . . = ρ2ǫ1eb1 1 − ρ1ρ2

The heat transfer is q1 = J1 − H1, and using ρ = 1 − α we get
ǫα
1 − ρ2 1 2
q =ǫ e
1 1 b1 1
− ρ1ρ2 = eb1 α1 + α2 − α1α2
If you now set α = ǫ (the gray approximation) and perform a little extra algebra, the above result
will be equivalent to Eq. (1.54) with eb2 = 0.

1.6.2 More than two surfaces

For gray, nonblack surfaces the exchange equations become
N
X

ǫiebi = Ji − (1 − ǫi) Jj Fi−j , specified Ti (1.55)

j=1

N
X

q i = Ji + Jj Fi−j , specified qi (1.56)

j=1

These can be solved for Ji, i = 1, 2, . . . N by standard methods for linear equations (matrix inver-
sion, iteration). Once the radiosities are obtained, the heat transfer fluxes at surfaces with specified
temperature are obtained from
ǫi
qi = 1 − ǫi (ebi − Ji) (1.57)
1.6. GRAY APPROXIMATION 19

and on surfaces with specified qi, the emissive power (and, from which, the temperature) would
be obtained from
ebi = 1 − ǫi qi + Ji (1.58)
ǫi
1.6.3 Gray/diffuse exchange example
Consider the system illustrated below. The system consists of a cylindrical cavity, of outer
radius R = 3 m and height L = 6 m. A circular surface on the lower end (surface 1) has a radius
of R1 = 1 m and properties of ǫ1 = 0.8, T1 = 1000 K. Surfaces 2 and 3 are adiabatic and L3 = L4
= 3 m. Surfaces 4 and 5 are at 400 K and have ǫ = 0.2.

2 1

We want to compute the net heat transfer rates from surfaces 1,4, and 5 and the
temperatures of surfaces 2 and 3.

Configuration factor calculation

The formula for parallel concentric circles on P. 500 of the text can be used to obtain all of the
needed configuration factors. For a pair of circles (1 and 2, say) separated by a distance L, the
formula is

R1 R2 1+Y2
X= L , Y = L , Z=1+
F1−2 = 2Z − Z2 − 4 X
X2 !

2 1/2
1 Y

Applying this formula gives

F1−5 = 0.1965, F(1+2)−5 = 0.1716

R2
F = 1 F = 0.0218, F = F = 0.1716
5−1 R2 1−5 5−(1+2) (1+2)−5
20 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

F =F −F = 0.1497, F = R2
5−2 5−(1+2) 5−1 2−5 R − R2
2 F5−2 = 0.1685
1
Let surface 6 be the imaginary circular surface formed by the dotted line.

F5−6 = 0.3820, F1−6 = 0.4861

Then
F5−4 = 1 − F5−6 = 0.6180, F1−3 = 1 − F1−6 = 0.5139
R2 R2
F4−5 = = 0.3090, F3−1 = 1
F1−3 = 0.2855
2RL4 2RL3
By symmetry and summation,

F =F =F +F :
3−(1+2) 4−5 3−1 3−2 F3−2 = F4−5 − F3−1 = 0.2805
2RL3
F2−3 = = 0.6311
R2 − R2
1
Now use summation:

F1−4 = 1 − F1−5 − F1−3 = 0.2896, F5−3 = 1 − F5−4 − F5−2 − F5−1 = 0.2104

F2−4 = 1 − F2−5 − F2−3 = 0.2005

2
R R2
F4−1 = 1 F1−4 = 0.0161, F3−5 = F5−3 = 0.1052
2RL4 2RL3
R2 − R2
F 1

= 2RL F2−4 = 0.0891

4−2
4
By the symmetry of the problem and summation,

F =F =F +F : F = 0.2805
3−(1+2) 4−5 3−1 3−2 3−2
A3

F2 3 = F3 2 = 0.6311

What leaves 3 and lands on 6 (the imaginary surface) must land on either 4 or 5, so

F =F +F
3−6 3−4 3−5
But F3−6 = F4−5 by symmetry, so

F3−4 = F4−5 − F3−5 = 0.2038 = F4−3

Only surfaces 3 and 4 can see themselves. Use summation:

F3−3 = F4−4 = 1 − F3−1 − F3−2 − F3−4 − F3−5 = 0.3820

1.6. GRAY APPROXIMATION 21

Exchange equations
For surfaces 1, 4, and 5, upon which the temperature is specified, the exchange equations are
N
X

Ji − (1 − ǫi) Jj Fi−j = ǫiebi, i = 1, 4, 5 (1.59)

j=1

and on the adiabatic surfaces, 2 and 3, the exchange equations are

N
Ji − j=1
Jj Fi−j = 0, i = 2, 3 (1.60)

X
The following is a listing of the Mathematica code used to solve the equations. I had
previously calculated the configuration factors and stored them in the arrays f[i,j]. I use the
symbol jf[i] for Ji in the code.

In[29]:=sigma=5.67*^-8;eps[1]=0.8;eps[4]=0.2;eps[5]=0.2;
t[1]=1000;t[4]=400;t[5]=400;
eb[1]=sigma t[1]^4;eb[4]=sigma t[4]^4;eb[5]=sigma t[5]^4;

In[37]:=vars=Table[jf[i],{i,1,5}]
Out[37]={jf[1],jf[2],jf[3],jf[4],jf[5]}

In[40]:=eqns=Table[
If[i>1&&i<4,
jf[i]-Sum[jf[j] f[i,j],{j,1,5}]==0, jf[i]-(1-
eps[i])Sum[jf[j] f[i,j],{j,1,5}]==eps[i] eb[i]
]
,{i,1,5}]

Out[40]=
{jf[1]-0.2 (0.513878 jf[3]+0.28963 jf[4]+0.196491 jf[5])==45360.,
jf[2]-0.631053 jf[3]-0.200488 jf[4]-0.168458 jf[5]==0,
-0.0285488 jf[1]-0.280468 jf[2]+0.618034 jf[3]-0.20382 jf[4]
-0.105197 jf[5]==0,
jf[4]-0.8 (0.0160906 jf[1]+0.089106 jf[2]+0.20382
jf[3] +0.381966 jf[4]+0.309017 jf[5])==290.304,
-0.8 (0.0218324 jf[1]+0.14974 jf[2]+0.210393
jf[3] +0.618034 jf[4])+jf[5]==290.304}
22 CHAPTER 1. RADIATION EXCHANGE BETWEEN SURFACES

In[42]:=soln=Solve[eqns,vars][[1]]

Out[42]={jf[1]->47060.5,jf[2]->8808.58,jf[3]-
>9753.96, jf[4]->7088.69,jf[5]->7314.03}

In[44]:=t[2]=(jf[2]/sigma)^(1/4)/.soln
t[3]=(jf[3]/sigma)^(1/4)/.soln

Out[44]=627.814
Out[45]=644.02

In[48]:=
q[1]=eps[1]a[1]/(1-eps[1])(eb[1]-jf[1])/.soln
q[4]=eps[4]a[4]/(1-eps[4])(eb[4]-jf[4])/.soln
q[5]=eps[5]a[5]/(1-eps[5])(eb[5]-jf[5])/.soln

Out[48]=121133.
Out[49]=-79693.6
Out[50]=-41439.7

In[51]:=q[4]+q[5]
Out[51]=-121133.