HEAT TRANSFER

Lecture 26: Natural Convection

April 7 2020
 Natural Convection

Natural convection refers to the movement of fluids induced by buoyancy forces.

09.qxd 2/24/11 3:15 PM Page 595
Buoyancy forces can arise in a fluid in which there are density gradients and a body
force that is proportional to the density.

In heat transfer,9.1
density gradients are due to temperature gradients, and
Physical Considerations
! 595 the force of
the body is given by gravity.
T1 ρ1 ρ1 T1

ρ ( x) Unstable x T ( x)
fluid Stable
circulation ρ ( x)
T ( x)
g
ρ2 T2 T2 ρ2

dT dρ
__ > 0, __ dT dρ
__ < 0, __
<0 >0
dx dx dx dx
(a ) (b)

FIGURE 9.1 Conditions in a fluid between large horizontal plates at

different temperatures: (a) Unstable temperature gradient. (b) Stable
temperature gradient.
 2 #T #T #2T #x #y #y2
T* #T* 1 # T* u #v !$ 2 (9.8)
% v* ! Note
#x
that
#y
viscous #y
dissipation has been neglected (9.11)
in the energy equation, (9.8
x* Rethat
L Pr
2
Natural Convection: Dimensionless
#y* Note #y*dissipation
viscous tion that
hasisbeen
certainly reasonable
neglected in thefor the small
energy velocities
equation, (9.8),associated
an assump- with free c
the mathematical
tion that is certainly reasonable sensevelocities
for the small the appearance of the
associated buoyancy
with term in Equation
free convection. In 9.7
ameter in the first term on sense
the mathematical thematters.
right-hand
the No longer
appearance maybuoyancy
ofside
the the hydrodynamic
of Equation problem,9.7given
9.10
term in Equation is by Equations 9.
complicates

Parameters
matters. No longer mayuncoupled from and solved
the hydrodynamic problem,to the exclusion
given of the thermal
by Equations 9.6 andproblem,
9.7, begiven by
he buoyancy force. The
uncoupled fromreference The to
and solved velocity
solution
the exclusion u0thecan
to the momentum
of thermalbeproblem,
specified
equation depends
given byontoknowledge
Equation of T, and
9.8.
solution to2equation
the energy equation. Equations 9.6 through 9.8 are therefore stro
quation. It is convenient
The solution to
to choose
the momentum
and must u 0
be ! g!(T
solved
depends " onT
simultaneously.
s )L,
knowledge
! so that
of T,
solution to the energy equation. Equations 9.6 through 9.8 are therefore strongly coupled
the
and hence on the
3 2 obviously
1/2
mes unity. Then, and
Remust
L becomes
be solved [g!(T T
Free convection
simultaneously.
in which s " ! )L
! is obtained
/"
effects] . It is customary
depend on the expansion coefficient !
age 605
Grashof's Number:
Free convection effects obviously dependdepends on the fluid.
on the expansion For an ideal
coefficient gas, manner
!. The " ! p/RT and
ber GrL as the square
in whichof! isthis Reynolds
obtained depends on the number:
fluid. For an ideal gas, " !#"p/RT and p
! !" 1 " # ! 1
" #T p " RT 2 T ! 1
3
g!(Ts " T!)L where T is the"absolute
! ! " 1 #"
" #! 1
"
p
2
#T p temperature.
RT
! 1 (9.9)
T For liquids and nonideal gases, ! mus
GrL " 2 absolutefrom appropriate
(9.12)
where T is the " temperature. Forproperty
liquids tables (Appendix
and nonideal A). ! must be obtained
gases,
from appropriate property tables (Appendix A).
1/2
tions 9.10 and 9.11 9.3 is replaced
Similarity by Gr
Considerations
.6 ! Empirical Correlations: External Free Convection L , and we see that the
Flows 605
e precisely, GrL1/2 )Considerations
9.3 Similarity plays the same role in free convection that
Let us now consider the dimensionless parameters that govern free convec
in forced convection.
Rayleigh Number: Based on the
heat resulting
transfer for the form
vertical of Equations
plate. As for forced9.10
convection (Chapter 6), th
where the Rayleigh number,
Let us now consider the dimensionless parameters that govern free convective flow and
ransfer correlations of the
heat transfer form
for the
may Nu
be
vertical
obtained
L !
plate. Asf (Gr
by
for ,
forced
L Pr) in
convectionfree
nondimensionalizing theconvec-
governing
(Chapter 6), the
equations.
parameters
Introducin
may be obtained by nondimensionalizing the governing equations. x Introducingy
olds number provides a measure of the ratio g!(Ts % of Tthe
!y)L inertial
3 x* !
L
to vis-
y* !
L
uid element. In contrast, the RaL Grashof
# GrL Pr # number
x* ! x
"# is
y* !a measure of the (9.25)
L L
es to the viscous forces acting on the fluid.
s based on the characteristic length L of the geometry. Typically, n # 14 and 13 for laminar
convection effects are comparable, the situation is more –
complex.
nd turbulent flows, respectively. For turbulent flow it then follows that hL is independent
f boundary layer
L. Note that of Figureare
all properties 9.3, but with
evaluated a non-zero
at the free stream
film temperature, T ! (T ! T )/2.
 t flows, respectively. For turbulent flow it then follows
9.5 that
ThehL is independent
Effects of Turbulence
hat all properties are evaluated at the film temperature, Tf ! (Ts ! T!)/2.

It is important to note that free co

he Vertical Plate
he Natural Convection: Vertical Plates
Correlations:
of Effects
the form
External Free Convection Flows
of Turbulence
given by Equation 9.24 have been developed for the vertical
flow. As with forced convection,
605 603 in the flow may be amplifi
bances plate
minar flow (104 " RaL " 109), C # 0.59 and n # 1/4, and for turbulent This process
flow is shown schematicall
he10
"gh 13
),
Rayleigh C # 0.10
number, and
whichn #
is 1/3.
simply A correlation
the product of that
the may
Grashof be
and applied
Prandtl num-
Transition
over the in a free convectio
number,
of the buoyancy and viscous forces i
Ra
vertical
TheL has been
plates the
critical recommended
numberby
critical Rayleigh
Rayleigh Churchill
number
for and Chu
is plates
Vertical is: [8] and is of the form
g!(Ts % T!)L3 3
RaL # GrL Pr # g"(T s % T!)x1/6 (9.25)
$ %
"#
!# Ra L ! 10
9
Rax,c # Grx,c Pr # 0.387 2 (9.23)
Ts > T∞
NuL # 0.825 ! 9/16 8/27 1 1
(9.26)
haracteristic
sive For
length
vertical
discussion ofplates
of the
L the
stability [1 ! are
geometry.
following
and transition(0.492/Pr)
Typically,
correlations
effects ]nto#determine
is provided 4 and 3 for laminar Quiescent
–by Gebhart et al. [2]. fluid, T∞
ws,
n forcedrespectively. For turbulent
convection,coefficient:
the convection flow it then follows that h is independent
transition to turbulence has a strong effect on heat transfer.
L
all
uationproperties
9.26 is are evaluated
suitable for at
most the film
engineeringtemperature, 9 Tf ! slightly
calculations, (T ! T )/2.
better accu-Turbulent
e results of the foregoing section apply only if RaL ! 10 . To obtain appropriate s ! g
obtained for laminar
ns for turbulent flow
flow, emphasis
Laminar Flow by(Ra
isusing
< 10[8]
placed 9) on experimental results.

Vertical Plate xc
Transition
0.670 Ra1/4
L 9
Rax,c ≈ 109
formNu
he9.1
E given 0.68
L # by !
Equation 9.24 have been Ra " 10
developed Lfor the vertical
9/16 4/9 plate (9.27)
4 [1 9! (0.492/Pr) ] Laminar
ar flow (10 " RaL " 10 ), C # 0.59 and n # 1/4, and for turbulent flow
a ),0.25-m-long
13
C # 0.10 vertical
and n #plate
1/3. that
A is at 70"C. The
correlation that plate
may isbesuspended
applied in quiescent
over the x F
ayleigh
at number
25"C. is the
Estimate moderately
velocity large, thelayer
boundary second term and
thickness on the right-hand
maximum upwardside of ve
a26L has
andbeen
9.27recommended
For any range
dominates, byofChurchill
and Rayleighand Chu [8] and is of the form
the correlations are the same form as Equation
at the trailing edge of the plate. How does the boundary layer thickness compare
that the constant,
thickness is replaced
that wouldC,exist if the air by a 1/6
were function of Pr.
flowing over the Equation 9.27stream
plate at a free is then in
$ %
antitative
of 5 m/s?Nuagreement with the 0.387
analytical
2
Ra L solution given by Equations 9.21 and
# 0.825 !
L 9/16 8/27
(9.26)
rast, when the Rayleigh[1number
! (0.492/Pr)
is small,] the first term on the right-hand side
N9.26 and 9.27 dominates, and the equations yield the same behavior since
on 9.26 presence
8. The is suitableoffor most engineering
leading constants incalculations, slightly
Equations 9.26 andbetter
9.27 accu-
accounts for
Vertical plate is in quiescent air at a lower temperature.
 9.17 Consider a vertical plate of dimension 0.25 m % 0.50 m
perature Tf . that is at Ts " 100!C in a quiescent environment at
(b) Consider a highly polished, 150-mm square alu- T! " 20!C. In the interest of minimizing heat transfer
minum alloy (2024) plate of 5-mm thickness, ini- from the plate, which orientation, (A) or (B), is preferred?
tially at 225!C, and suspended in ambient air at What is the convection heat transfer from the front sur-
25!C. Using the appropriate approximate correla- face of the plate when it is in the preferred orientation?

Problem I
tion from Problem 9.11, determine the time
required for the plate to reach 80!C. Ts
(c) Plot the temperature–time history obtained from Ts
part (b) and compare with the results from a g
lumped capacitance analysis using a constant free Orientation A
convection coefficient, ho. Evaluate ho from an Orientation B Air, T∞
appropriate correlation based on an average surface
temperature of T " (Ti $ Tf)/2.
9.18 During a winter day, the window of a patio door with a
3 A square aluminum plate 5 mm thick and 200 mm on a
height of 1.8 m and width of 1.0 m shows a frost
side is heated while vertically suspended in quiescent
line near its base. The room wall and air temperatures
air at 40!C. Determine the average heat transfer coeffi-
are 15!C.
cient for the plate when its temperature is 15!C by two
methods: using results from the similarity solution to
the boundary layer equations, and using results from an
empirical correlation.
4 An aluminum alloy (2024) plate, heated to a uniform
temperature of 227!C, is allowed to cool while verti-
cally suspended in a room where the ambient air and
surroundings are at 27!C. The plate is 0.3 m square
with a thickness of 15 mm and an emissivity of 0.25.
(a) Develop an expression for the time rate of change
of the plate temperature, assuming the temperature
to be uniform at any time.
(b) Determine the initial rate of cooling (K/s) when the
plate temperature is 227!C.
(c) Justify the uniform plate temperature assumption. Window,
1.8 m × 1.0 m
(d) Compute and plot the temperature history of the
plate from t " 0 to the time required to reach a
temperature of 30!C. Compute and plot the corre- (a) Explain why the window would show a frost layer
sponding variations in the convection and radiation at the base rather than at the top.
heat transfer rates. (b) Estimate the heat loss through the window due to
5 The plate described in Problem 9.14 has been used in free convection and radiation. Assume the window
an experiment to determine the free convection heat has a uniform temperature of 0!C and the emissiv-
transfer coefficient. At an instant of time when the ity of the glass surface is 0.94. If the room has elec-
plate temperature was 127!C, the time rate of change tric baseboard heating, estimate the corresponding
of this temperature was observed to be #0.0465 K/s. daily cost of the window heat loss for a utility rate
What is the corresponding free convection heat trans- of 0.18 $/kW ! h.
fer coefficient? Compare this result with an estimate 9.19 A vertical, thin pane of window glass that is 1 m on a
Problem I: Solution
Problem I: Solution
Problem I
 Fluid, T∞

(a) (b)

Natural Convection: Inclined and

CH009.qxd 2/24/11 3:15 PM
Fluid, T∞
Page 610

Plate, Ts

Horizontal Plates
Plate, Ts FIGURE 9.7 Buoyancy-driven flows on
horizontal cold (Ts % T!) and hot (Ts & T!)
plates: (a) Top surface of cold plate.
610 Chapter 9
∞
! Free Convection
Fluid, T (b) Bottom surface of cold plate. (c) Top
The buoyancy force is normal, rather than parallel, to the plate. surface of hot plate. (d) Bottom surface of
(c) (d) hot plate.
Heat flux and transfer depends on whether the plate is hot or cold and whether if it is positioned facing up or
Fluid, T∞
down.
where As and P are the plate surface area (one side) and perimeter, respectively. Using this
The following arecharacteristic
the different correlations
length, based oncorrelations
the recommended the position ofPlate,
for the the Tplate:
average
s Nusselt number are

Upper Surface of Hot Plate or Lower Surface of Cold Plate [19]:

Plate, T
NuL " 0.54 Ras 1/4
L (104 # RaL # 107, Pr $ 0.7) (9.30)
Fluid, T∞
NuL " 0.15 Ra1/3
L (107# RaL # 1011, all Pr) (9.31)
(a) (b)

Lower Surface of Hot Plate

Fluid, Tor
∞ Upper Surface of Cold Plate [20]:

NuL " 0.52 Ra1/5

L (104 # RaL # 10 9
, Pr
Plate, T $ 0.7) (9.32)
s

Additional correlations can be found in [21].

Plate, Ts FIGURE 9.7 Buoyancy-driven flow
EXAMPLE 9.3 horizontal cold (Ts % T!) and hot (T
plates: (a) Top surface of cold plate.
s T >T ∞
0.75 Tm∞ wide and 0.3 m(b)high
Bottom surface of cold plate. (c)
Airflow through a long rectangular heating duct that is Fluid,
∞ s
surface
tains the outer duct surface at 45!C. If the duct is uninsulated and exposed to air
T >T
main-
of hot
at 15!C in plate. (d) Bottom surf
(c)
the crawlspace beneath a home, (d) duct per meter ofhot
what is the heat loss from the plate.
length?
 characteristic length, the recommended correlations for the average Nusselt number are

Natural Convection: Inclined and

Upper Surface of Hot Plate or Lower Surface of Cold Plate [19]:
1/4 4 7
09.qxd 2/24/11
Nu " 0.54 Ra
3:15 PM
(10 # Ra # 10 , Pr $ 0.7)
L
Page 610 L (9.30) L

Horizontal
Nu " 0.15 Ra Plates
(10 # Ra
L
1/3
# 10 , all Pr)
L
7
(9.31) L
11

Lower Surface of Hot Plate or Upper Surface of Cold Plate [20]:

1/5 4 9
610 Nu " 0.52 Ra (10
ChapterL 9 ! Free Convection
L # Ra L # 10 , Pr $ 0.7) (9.32)
Additional correlations can be found in [21].

Fluid, T∞
EXAMPLE 9.3
Plate, Ts
Airflow through a long rectangular heating duct that is 0.75 m wide and 0.3 m high main-
tains the outer duct surface at 45!C. If the duct is uninsulated and exposed to air at 15!C in
the crawlspace beneath a home,Plate,what
Ts is the heat loss from the duct per meter of length?
Fluid, T∞
SOLUTION
(a) (b)
Known: Surface temperature of a long rectangular duct.
T∞ > Ts
Fluid, T∞
Ts >T∞
Find: Heat loss from duct per meter of length.
Plate, Ts
 CH009.qxd 2/24/11 3:16 PM Page 613
The insulation therefore provides a 76% reduction in heat loss to the ambient air by
naturalExternal
rrelations: convection. Free Convection Flows 613

Natural Convection: Horizontal

2. Although they have
d 2/24/11 3:16 PM Page 614 been neglected, radiation losses may still be significant. From
Equation 1.7 with ! assumed to be unity and Tsur ! 288 K, q"rad ! 398 W/m for the
uninsulated duct. Inclusion of radiation effects in the energy balance for the insulated
duct would reduce the outer surface temperatures, thereby reducing the convection
heat rates. With radiation, however, the total heat rate (q"conv # q"9.6
rad) would increase.
! Empirical Correlations: External Free Convection Flows 613

614
Cylinder
q" ! (17.5 # 22.8 # 15.5) W/m ! 55.8 W/m
9.6.3 The Long Horizontal Cylinder
herefore provides a 76%
Chapter 9 ! reduction
which yields
in heat loss to the ambient air by
Free Convection
on.This important geometry has been studied extensively, and many existing correlations have q" ! (17.5 # 22.8 # 15.5) W/m ! 55.8 W/m
been reviewed by Morgan [22]. For an isothermal cylinder, Morgan suggests an expression
have
of thebeen
form neglected, radiation losses may still beThe significant. From provides a 76% reduction in heat loss to the ambient air by
insulation therefore
ith ! assumed to be unity andhDTsur !n288 K, q"rad !natural convection.
398 W/m for the
NuD ! ! C Ra D 2. Although (9.33)
they have been neglected, radiation losses may still be significant. From
t. Inclusion of radiation effects kin the energy balance for the 1.7insulated

Nuθ
Plume Equation with ! assumed to be unity and Tsur ! 288 K, q"rad ! 398 W/m for the
ucewhere
theC outer
and n aresurface temperatures,
given in Table thereby
9.1 and RaD and Nu D are basedreducing the
on the cylinder convection
diameter.
uninsulated duct. Inclusion of radiation effects in the energy balance for the insulated
In contrast, Churchill and Chu [23] have recommended a single correlation for a wide
duct would reduce the outer surface temperatures, thereby reducing the convection
radiation, however,
Rayleigh number range: the total heat rate (q"conv # q"rad) would increase.
0 πheat
/2 rates.
π With radiation, however, the total heat rate (q"conv # q"rad) would increase.

! "
0.387 Ra1/6
D
2
NuD ! 0.60
Ts # RaD & 1012 (9.34)
[1 # (0.559/Pr)9/16]8/27
Ambient fluid, T∞
g Horizontal Cylinder
The foregoing correlations
Boundary provide the average Nusselt number over the entireThe
9.6.3 Long Horizontal Cylinder
circumfer-
layer
ence of an isothermal cylinder. As shown in Figure 9.8 for a heated cylinder, local Nusselt FIGURE 9.8 Boundary layer development
numbers are influenced by boundary layer development, which begins ThisatCH009.qxd
important
" ! 0 and geometry
and
con-
2/24/11 PM has
Page been
3:16Nusselt number
613 studied extensively,
distribution and many existing correlations have
on a heated
metry hasat "been
cludes studied
$ # with formation extensively, θ and from
of a plume ascending many existing
been
the cylinder. correlations
reviewed
If the have
by horizontal
flow remains Morgan [22]. For an isothermal cylinder, Morgan suggests an expression
cylinder.
laminar
organ over the
[22]. Forentire
an surface, the distribution
isothermal of the local
cylinder, of the form
Nusselt suggests
Morgan number with an" is expression
char-
acterized by a maximum at " ! 0 and a monotonic decay with increasing ". This decay
Morgan correlation:
would be disrupted at Rayleigh numbers sufficiently large (RaD % 109) to permit transition Churchill and Chu correlation: hD ! C Ra n
9.6 NuCorrelations:
Empirical D! External FreeDConvection Flows 613 (9.33)
to turbulence within the boundary layer. If the cylinder is cooled relative to the ambient k
!

Nuand XAMPLE hD
fluid, boundary layerEdevelopment begins at "n ! #, the local Nusselt number is a maxi-
9.4 C Rafrom where C and n are given (9.33)
in Table 9.1 and RaD and NuD are based on the cylinder diameter.
D! the plume!
which yields
mum at this location, descends D the cylinder.
k In contrast, Churchill and Chu [23] have q" !recommended a single
(17.5 # 22.8 # 15.5) W/m correlation for a wide
! 55.8 W/m

The fluid of Example 2.2 is characterized by number

Rayleigh a thermal conductivity,
range: density,
The insulation specific
therefore provides heat, in heat loss to the ambient air by
a 76% reduction
iven in Table
TABLE 9.1and
9.1 Constants andofdynamic
Ra D and
Equation 9.33 D are
Nufor based
free of
viscosity onW/m
convection
0.705 the !cylinder diameter.
K, 1146 kg/m 3
, 3587
natural convection.
J/kgthey! K,
2. Although have and 1/6962 !
been neglected, 10"6
radiation losses may still be significant. From

! "
on a horizontal
ll and Chu [23]Nhave circular cylinder [22] 0.387 Ra 2
! s/m ,recommended
respectively. Ana experiment
single correlation forNuaDin!wide
2 Equation 1.7 with ! assumed to be unity and T ! 288 K, q" ! 12398 W/m for the
is conducted which #a long
0.60uninsulated aluminum
duct. Inclusion
D
of 9/16
radiation
8/27
rod
effects of
in Ra
the
sur
&
energy
D 10
balance
rad
(9.34)
for the insulated
nge:Ra Cdiameter D n # 20 mm and initial temperature T # 32$C is suddenly [1 # (0.559/Pr)
immersed ]
however, horizon-
duct would reduce the outer surface temperatures, thereby reducing the convection
D
i heat rates. With radiation, the total heat rate (q" # q" ) would increase.
conv rad
'10
10 –10 '2
tally into0.058
0.675 a large bath of the fluid at a The
temperature of T! # 22$C.
foregoing correlations provideAt # 65 s,Nusselt
thet average the mea-
number over the entire circumfer-

! "
0.1481/6
10'2–102
0.387 Ra sured temperature
1.02
D
2
of the rod is Tf #ence
23$C.
of anDetermine
12
the9.6.3
thermal
isothermal cylinder. As expansion
shown
The Long Horizontal coefficient
in Figure 9.8 for a heated cylinder, local Nusselt
Cylinder
D ! 0.60 # 10 are influenced
RaD &numbers (9.34)
102–104 0.850 0.188 by boundary layer development, which begins at " ! 0 and con-
of the fluid !.
0.250 9/16 8/27
[1 # (0.559/Pr)
104–107 0.480 ] This important geometry has been studied extensively, and many existing correlations have
cludes at " $ # with formation of a by
been reviewed plume
Morgan ascending from cylinder,
[22]. For an isothermal the cylinder. If the
Morgan suggests flow remains
an expression
107–1012 0.125 0.333
laminar over the entire surface, the distribution of the local Nusselt number with " is char-
of the form
cal Correlations: External Free Convection Flows 617
CH009.qxd 2/24/11 3:16 PM Page 617

eres
Natural Convection: Sphere
correlation due to Churchill [10] is recommended for spheres in fluids of
11
or Ra D " 10
Churchill .
correlation:
9.6 ! Empirical Correlations: External Free Convection Flows

0.589 Ra1/4
D
NuD # 2 $ (9.35)
9.6.4 Spheres
[1 $ (0.469/Pr) 9/16 4/9
]
s RaD l 0, Equation 9.35 reduces
The following correlation due to Churchill [10] is recommended
to NuD #
Pr ! 0.7 and
2, which corresponds to heat
for RaD " 1011.
nduction between a spherical surface and a stationary infinite medium, in a
CH009.qxd 2/24/11 3:16 PM Page 613

tent with Equations 7.56 and 7.57. 0.589 Ra1/4

D
NuD # 2 $
nded correlations from this section are summarized in Table 9.2. Results for 9/16]4/9
[1 $ (0.469/Pr) 613
d geometries and special conditions are presented in comprehensive reviews
9.6 ! Empirical Correlations: External Free Convection Flows

In the limit as RaD l 0, Equation 9.35 reduces to NuD # 2, wh

10] and Raithby and Hollands [21].
which yields
q" ! (17.5 # 22.8 # 15.5) W/m ! 55.8 W/m
transfer by conduction between a spherical
The insulation therefore provides a 76% reduction surface
in heat loss to and a airstationa
the ambient by

manner consistent with2. Equations

natural convection.
Although they have been 7.56 and
neglected, 7.57.
radiation losses may still be significant. From
Equation 1.7 with ! assumed to be unity and T ! 288 K, q" ! 398 W/m for the
Recommended correlations from thiseffects
section
in the energyare
balance summarized
sur rad
uninsulated duct. Inclusion of radiation for the insulated
duct would reduce the outer surface temperatures, thereby reducing the convection

Summary of free convection empirical correlations

other immersed geometries and special conditions are presented in
heat rates. With radiation, however, the total heat rate (q" # q" ) would increase. conv rad

rsed geometries by Churchill [10] and 9.6.3

Raithby and Hollands [21].
The Long Horizontal Cylinder
This important geometry has been studied extensively, and many existing correlations have
been reviewed by Morgan [22]. For an isothermal cylinder, Morgan suggests an expression
of the form
Pr ! 0.7 and for RaD " 1011.

0.589 Ra1/4
D

Natural Convection: Summary of

NuD # 2 $ (9.35)
[1 $ (0.469/Pr)9/16]4/9
In the limit as RaD l 0, Equation 9.35 reduces to NuD # 2, which corresponds to heat
transfer by conduction between a spherical surface and a stationary infinite medium, in a
manner consistent with Equations 7.56 and 7.57.

Correlations
Recommended correlations from this section are summarized in Table 9.2. Results for
other immersed geometries and special conditions are presented in comprehensive reviews
by Churchill [10] and Raithby and Hollands [21].

TABLE 9.2 Summary of free convection empirical correlations

for immersed geometries
Recommended
Geometry Correlation Restrictions
CH009.qxd 2/24/11 3:16 PM Page 618

1. Vertical platesa

Equation 9.26 None 618 Chapter 9 ! Free Convection

2. Inclined plates
Cold surface up or hot
surface down TABLE 9.2 Continued
Recommended
CH009.qxd 2/24/11 Geometry
3:16 PM Page 613 Correlation Restrictions
Equation 9.26 0 % ! " 60&
g l g cos ! 4. Horizontal cylinder

3. Horizontal plates
(a) Hot surface up or
cold surface down
Equation 9.34 RaD $ 1012
5. Sphere
9.6 ! Empirical Correlations: External Free Convection Flows 613

which yields Equation 9.35 RaD $ 1011

Equation 9.30 104 " RaL " 107, Pr ! 0.7 q" ! (17.5 # 22.8 # 15.5) W/m ! 55.8 W/m Pr % 0.7
Equation 9.31 107 " RaL " 1011 a
The correlationThe
mayinsulation
be applied therefore provides
to a vertical cylinderaif76%
(D/L)reduction
% (35/Grin
1/4 heat loss to the ambient air by
L ).
natural convection.
(b) Cold surface up or 2. Although they have been neglected, radiation losses may still be significant. From
hot surface down Equation 1.7 with ! assumed to be unity and Tsur ! 288 K, q"rad ! 398 W/m for the
uninsulated duct. Inclusion of radiation effects in the energy balance for the insulated
9.7 Free ConvectionductWithin Parallel
would reduce the outerPlate Channelsthereby reducing the convection
surface temperatures,
heat rates. With radiation, however, the total heat rate (q"conv # q"rad) would increase.

A common free convection geometry involves vertical (or inclined) parallel plate channels
that are open
9.6.3to the
Theambient at oppositeCylinder
Long Horizontal ends (Figure 9.9). The plates could constitute a fin
array used to enhance free convection heat transfer from a base surface to which the fins are
Equation 9.32 104 " RaL " 109, Pr ! 0.7 attached,This important
or they geometry
could has been
constitute an studied extensively,
array of and many
circuit boards existing
with correlations have
heat-dissipating electronic
been reviewed by Morgan [22]. For an isothermal cylinder, Morgan suggests an expression
components. Surface thermal conditions may be idealized as being isothermal or isoflux
of the form