1

LAMINAR-FLOW HEAT-TRANSFER IN NON-CIRCULAR DUCTS.

by

P. Wibulswas, B.Se.(Eng.), A.C.G.I., D.I.C.

A thesis submitted for the degree of Doctor of

PhilosOphy in the University of London.

Department of Mechanical Engineering,

University College London,

Gower Street, W.C.1. May, 1966.

 2

ABSTRACT

A. numerical method is employed to obtain solutions

for laminar flow heat transferwith fully developed velocity
profiles and invariant fluid physical properties for rectan-
gular ducts of various aspect ratios with the thermal bounda-
ry conditions of constant wall temperature and constant heat
input per unit length of the duct. Since an analytical solu-
tion for the fully developed velocity profile in a reotangu-i
lar duct is available, the varying temperature profile
remains to be solved numerically from the energy equation
which is transformed into a finite difference form by means
of two finite difference operators in two dimensions. Numeri-
cal values of the initial and boundary temperatures are fixed
by choosing a suitable dimensionless temperature depending
upon the, thermal boundary condition. As computation involved
is very lengthy, a fast digital computer is required. Numeri-
cal results obtained from an I.C.T. Atlas computer are
presented as the variation of the Nusselt number with the
Graetz number.
The numerical method is extended to analyse heat
transfer with simultaneously aeveloping velocity and tempera-
ture profiles. To determine the development of the velocity
profile, some simplifications of the Navier-Stokes equation
 3

are made. Results are presented for various aspect ratios

with the Prandtl number of 0.72. The effect of Prandtl number
on heat transfer is also illustrated by numerical results.
The numerical method is also used to solve for heat
transfer in right-angled isosceles and equilateral triangular
ducts with the same hydraulic and thermal boundary conditions
as in the previous cases.
The predicted results are compared with experimental
data. For constant wall temperature, they agree well for
Graetz numbers under 70; for constant heat input per unit
length, closer agreement is shown over a much wider range of
the Graetz numbers. Accuracy of the numerical method is
confirmed by the facts that variations of the predicted
Nusselt numbers obtained here follow the same trends as
those for cii'cular ducts and parallel plates and at the
Graetz number of zero, they approach values of the limiting
Nusselt numbers obtained by other mpthcds.
 14.

ACKNOWLMGEMLNTS

The work described in this thesis was carried out

in the Department of Mechanical Engineering, University
College, University of London, under the supervision of
Dr. S.R. Montgomery, M.A., S.M., Sc.D., A.M.I.Mech.E, I
would like to express my sincere gratitude for his guidance,
help, advice and criticism during the course of this work.
I would also like to thank members of the workshop especially
Mr. E. Farmer, for their helps in building up the test
apparatus.
Finally, I am grateful to the Ministry of Educa-
tion of Thailand for awarding the scholarship for research
in Mechanical Engineering in the United Kingdom, which has
made this work possible.
 5
CONTENTS

Page
Abstract 2
Acknowledgements 4
Nomenclature 8
CHAPTAAI INTRODUCTION 11

1.1 Circular Ducts 13

1.2 Infinite Parallel Plates 15
1.3 Rectangular Ducts 17
1.4 Other Cross Sections 18
1.5 Scope of the Present Work 19
CHAPTLR 2 APPARATUS 21
2.1 General Description 21
2.2 Production of a Fully-Developed
Velocity Profile. 24
2.3 Rectangular Test Section For Constant
Heat Input per Unit Length of Duct 26
2.4 Rectangular Test Section for Constant
Wall Temperature 27
2.5 Measurement of Temperatures 25
2.5 Equilateral Triangular Test Section
for Constant Heat Input per Unit Length
of Duct 32
 6
Page

CHAPTER 3 HEAT TRANSFER IN THE THERMAL ENTRANCE

REGION OF RECTANGULAR DUCTS 33
3.1 Basic Differential Equations 33
3.2 Analysis of Parabolic Equation 54
3.3 Exact Solution of the Navier-Stokes'
Equation 58
3.4 Numerical Solution of the Energy
Equation 39
3.5 Solution for Constant Wall Temperature 41
3.6 Solution for Constant Heat Input per
Unit Length of Duct 51
3.7 Experimental Results 64
CHAPTER 4 'PRAT TRANSFER FOR SIMULTANEOUSLY
DEVELOPING VELOCITY AND TEMPERATURE
PROFILES IN RECTANGULAR DUCTS 72
4.1 Theoretical Solutions 72

4.2 Experimental Results 77

CHAPTER 5 RECTANGULAR DUCTS 86
5.1 Right Angled Isosceles Triangular Duct 86
5.2 Equilateral Triangular Duct 95
CHAPTER 6 DISCUSSION AND CONCLUSION 108
6.1 Predicted Numerical Results 108
6.2 Comparisonwith Experimental Results 111
6.3 Conclusion 114
 Page

APPENDIX 7.1 Limiting Nusselt Numbers 116

7.2 Variations of Predicted Nusselt Numbers
with Graetz Numbers for Circular Ducts
and Parallel Plates 117
7.3-4 Predicted Nusselt Numbers for Fully
Developed Velocity Profiles in Rectangular
Ducts with Constant Wall Temperature and
Constant Heat Input per Unit Length 118
7.5-6 Experimental Results for A Rectangular
Duct of Aspect Ratio of 2.0 with Constant
Wall Temperature and Constant Heat Input 122
7.7 Computer Program
7.8 Predicted and Experimental Results for
Simultaneously Developing Velocity and
Temperat.re Profiles in Rectangular Ducts 133
7.9-10 Predicted Results for Heat Transfer in
Right Isosceles Triangular Ducts 139
7.11-12 Predicted Results for Heat Transfer in
Equilateral Triangular Ducts 141
7.13-14 Experimental Results for Heat Transfer
with Fully Developed Profile and Simulta-
neousl Developing Profiles in Equilateral
Triangular Ducts with Constant Heat Input 143
REFERENCES 146
 8

NOMNCLATURE

A Area
A c Cross sectional area
As . Surface area
a Short side of rectangular cross section
a Short side of right-angled isosceles triangular

6 cross section
Loiig stde of aectatirlod C4010 Section.
Constant
Specific heat at constant pressure
dh Hydraulic diameter (= 4Ac/P)
Ratio of finite step to hydraulic diameter (=h/dh)
Unspecified function
Ii Finite step in x and y directions
Coefficient of heat transfer
H Partial Differential operator for square network
K Partial differential operator (= H+I )

3 Partial differential operator for triangular network

C/JfficiJnt of thormal conductivity
Length of duct
1 Length
Number of step
Integer
P Perimeter of duct
Pressure
4 Rate of heat transfer
 9

Temperature
Velocity in x direction
3 Volumetric flow rate
Velocity in y direction
Velocity in z direction
X Partial differential operator for square network
x,y,z Cartesian coordinates
Distance along duct

or. Thermal diffusivity (. k/Cpe)

' Aspect ratio of a rectangular duct (= b/a)
i3 Dimensionless temperature for constant heat input per
unit length of duct
Incremental sign
Laplacian operator
O Dimensionless temperature for constant wall temperature
joi, Dynamic viscosity
y Kinematic viscosity
Summation sign
e Density

pjaerkpiojes_g....groups

Gz Graetz number (= Re.Pr/(z/dh))

Nu Nusselt number (= hdh/kb)
Re Reynolds number (= ewbdh440
Pr Prandtl number (=AiCp/kb)
 10

w+ Dimensionless velocity in z direction

(. w/Onit. d,2)
dz'

gulagarints.

Bulk, average value

o Centre, correction
Final
Value distant h from wall
1 Logarithmic
Mean
o Initial
Peripheral
Wall
Local
or LimitiOng value
 11
CHArTLR 1
IMODUCTION

Heat transfer by forced convection occurs when a

fluid is induced by mechanical means such as a pump, fan
etc. through a duct with wall temperature differing from
fluid temperature. Since heat transfer occurs as a result
of interaction between the velocity and temperature
profiles of the fluid, the rate of heat transfer depends
upon the shapes of the two profiles.
For a duct with a hydraulically smooth entrance
such as a bell-shaped inlet for the circular duct, velocity
at the entry plane is uniform. As the fluid flows along the
duct, local velocities near the wall are retarded by fluid
viscosity and a boundary layer grows. When the boundary
layer reaches the central axis of the duct, the flow is

said to be fully developed. In a circular duct, the fully

developed velocity profile for a laminar flow is parabolic,
but in a non-circular duct, the shape of the profile
depends upon the shape of the duct.
The temperature profile also exhibits a similar
boundary layer growth depending upon the thermal boundary
condition. Most common ones encountered in practice are
constant wall temperature, constant heat flux and constant
wall telaperature gradient.
 12

Investigations of heat transfer by forced convec-

tion in circular ducts have been extensively conducted by
many research workers and their analytical and experimental
results may be found in various text books and journals on
heat and mass transfer. In most theoretical analyses,
physical properties of the fluid were assumed invariable
with respect to temperature.
Recently, the need for compact heat exchangers
in many fields such as air conditioning units, rocket power
plants etc. has accelerated the work dealing with foxced
convection heat transfer in non-circular ducts, but the
knowledge in this field is still incomplete. Most analyti-
cal solutions were obtained for fully developed flow and
very few experimental data are available. Recommendations
for further investigation have been made in an extensive
literature survey up to 1961 by Montgomery and Weiss (1)
for both laminar and turbulent flows.
In order to provide up-to-date informationl, a
brief survey of publications on laminar flow heat transfer
from 1961 onwards is given below.
 13
1.1 CIRCULAR DUCT.

Numerical solutions for laminar flow beat

transfer with simultaneously developing velocity-and
temperature profiles were obtained by Uirichson and
Schmitz (2) for the boundary conditions of constant wall
temperature and constant heat flux. The work is a refine-
ment of an earlier work of Kays (3) to include the radial
component of the velocity in the entrance region and it
shows a significant decrease in the local Nusselt number
from that obtained by Kays.
Hudson and Bankoff (4) solved a transient
problem wherein a new wall temperature is Suddenly
impressed upon an initially isothermal laminar flow by
double Laplace transformation. For small time or large
axial distance, the solution is found to be independent
of axial distance and for small distance, it.becomes
independent of time.
The effect of the variation of fluid physical
properties with temperature was investigated theoretically
by Bradley and Entwistle (5) for fully developed air flow
in a circular duct. Numerical solutions are presented for
a range of air temperature from 350 deg. K. to 2500 deg.
K. for the conditions of constant wall temperature and
constant axial temperature gradient. The solutions also
show that the effect of axial conduction on the tempera-
 14-

ture profile becomes important at low Reynolds numbers and

the effect of axial momentum on the velocity profile owing
to different fluid densities is considerable for large
temperature differences between the wall and air.
For laminar flow heat transfer with variable
properties in the thermal entrance region, an analytical
method was used by Xoppel and Smith (6) for the thermal
boundary condition of constant heat flux, the radial velocity
being neglected. The method was applied to supercritical
carbon dioxide whose properties vary rapidly with temperature
The results show fluctuations of the heat transfer coefficien
for particular values of heat flux, fluid temperatures and
flow rates.
Experimental results were obtained by Kays and
Nicoll (7) for large temperature differences between the wall
and air for the cases of constant heat flux and constant wall
temperature over a narrow range of Reynolds numbers. Data was
found to agree well with analytical solutions baed on
constant fluid properties.
 15

1.2 INFINITE PARALLEL PLATES

A few theoretical papers were published recently on

simultaneously developing velocity and temperature profiles

in the entrance region of two infinite parallel plates.
Stefan (8) used an approximate series solution for the

thermal condition of constant wall temperature. The case of

constant heat input per unit length of the duct was solved

by Han (9). He assumed Langhaar's solution (10) for the

varying velocity profiles and derived an integral method to

solve for the variations of the wall and fluid temperatures

along the duct. Numerical solutions are given for Prandl

numbers of 3.2, 1.6, 0.8 and 0.4.

A more rigorous analysis of the simultaneously

developing flow was done by Hwang and Fan (11). They applied

the finite difference analysis to the Navier-Stokes equation

and the continuity equation to solve the developing velocity

profiles which are then substituted in the energy equation

to dertermine tea,.erature profiles. Nusselt numbers are given

for krandt1 numbers in the range of 0.01 to 50. For the

constant wall temperature case, their solutions agree very

well with those of Stefan (8); as for the constant heat flux,

their results differ appreciably from Hart's results (9), but

agree closely with an earlier work by Siegel and Sparrow (12)

who used Langhaar's velocity profiles.

 16
Solutions of heat transfer in the thermal entrance
region with an interesting thermal boundary condition of
unequal wall temperatures were obtained by Hatton and Turton
(13). They show that the Nusselt numbers for the two walls do
not reach the limiting value until the fluid temperature
gradient becomes linear after a very long entry length. The
entry length also depends on the magnitude of the fluid entry
temperature in comparison to the wall temperature.
A transient problem on heat transfer for an
incompressible laminar flow was theoretically analysed by
Perlmutter and Siegel (14). The transient is caused by sudden
changes of the fluid driving pressure and the wall temperature.

Solutions are given for both zero and finite wall thermal
resistances.
 17

1.3 R3CTANGULAR DUCTS.

Since laminar flow heat transfer in a rectangular

duct is strongly affected by the aspect ratio of the duct,
theoretical analysis is more complex than those for the
circular duet and parallel plates. Solutions for flows with
fully developed velocity and temperature profiles were
obtained by various authors and their worksttoru fully compiled
by Montgomery and Weiss (1).
Variations of wall temperature for fully developed
laminar flow in channels with the aspect ratios from 1 to oo
were investigated by Savino and Siegel (15). Solutions are
available for various ratios of heat fluxes between the short
sides and the broad sides from 0 to 1, while the total heat
flux per unit length of the channels is kept constant.
Results show that the peak temperatures occur at the corners
owing to low fluid velocities. When only the broad sides are
heated, the peak temperatures are lowest and decrease rapidly
as the aspect ratio increases.
 18

1.4 OTHER CROSS SBCTIONS.

Fully developed Nusselt numbers for isosceles

triangular, right triangular and circular sector ducts were
computed by Sparrow and HajiSheik (16) and results are
presented over a wide range of opening angle of the cross
sections for the boundary condition of constant heat flux.
For small opening angles, results for the sector and the
isosceles triangular ducts are very close to each other.
Among the triangular cross sections, the equilateral
triangular duct gives the highest Nusselt number. Nusselt
number for the circular sectors reaches a maximum value
when the opening angle approaches 180 deg., i.e. it
represents an infinite plate.
Analytical solution of heat transfer in a cone with
a small opening; ;Ingle was obtained by Cobble (17).
Variations of the Nusselt numbers with the Graetz numbers
for a finite cone angle of 0.1 radian are plotted with
various cone lengths as a parameter.
 19

1.5 SCOkAl OF MA kfUSAINT WORK.

As already mentioned in (1.3) and (1.4), analyti-

cal solutions are available only for fully developed flows
in a few types of non-circular ducts, but in practice, heat
transfer usually begins simultaneously with the development
of velocity profile, and in some ayplications, though the
velocity profile is fully developed when the heat transfer
process begins, the whole region where heat transfer occurs
must be taken into consideration. Since no theoretical and
very few experimental data exist for the stated conditions,
the present work attempts to provide them.
Theoretical analyses here are concerned with
laminar .flow heat transfer in rectangular ducts with
different aspect ratios and in triangular ducts of various
cross sections under the following boundary conditions :
(1) Fully developed velocity profile and develop-
ing temperature profile. This is also known as. the
' thermal entrance resion.' At the entrance of the duct,
the fluid temperature is uniform. The exact solutions for
the fully developed velocity profiles in rectangular and
triangular ducts can be found in a text book by Rouse (18).
(2) Simultaneouly developing velocity and
temperature profiles. Both of them are assumed uniform at
the entrance.
 20

Thermal boundary conditions of constant heat

input per unit length of the duct and of const4mt wall
temperature are considered. In all cases, fluid properties
are assumed invariable with respect to temperature.
Experimental results were obtained for a
rectangular duct of aspect ratio of 2.0 with constant heat
input per unit length and with constnnt wall temperature,
and for an equilateral triangular duct with constant heat
input per unit length.
 21

CHAPTER 2

APPARATUS

2.1 GANERAL IJE1CR11TION,

A general flow diagram of the apparatus is given in

fig.1, p.22. Before entering the test section, atmospheric
air was sucked through an unheated duet to produce a fully
developed velocity profile. Various test sections were
employed for the thermal boundary conditions of constant
heat input per unit length and constant wall temperature,
and their details will be described later on. On leaving the
test section, the heated air entered an insulated mixing
chamber, fig.2, p.23, which was a box containing a wire mesh
to even out the final temperature of the air. A rotameter
was provided for measuring the flow rate of the air which
could be controlled by varying the opening of a by-pass
chamber, fig.3, p.23, Circulation of the air was maintained
by a high speed centrifugal fan (21), the fluctuation of
which was damped by inserting a piece of wire wool in the
duct between the rotameter and the by-pass chamber.
 22
Air in

Fig. 1 General Flow Diagram of

Apparatus.

--Divergent nozzle

Straight duct

Rotameter
AMA

Wire wool
Test section
Air
in Air pump
Mixing AM.
1 \
1=r By-pass iMotor
chamber
chambe
(J I
r.-n

Air C
. .
t Air in 23
Main supply

Test section Thermocouples

..--'

Wire
mesh

Fig. 2 Mixing Chamber.

Divergent
nozzle
Flexible tube

Straight Shutter
v-duct

Manometer To pump

I'

Fig.3 By-pass Chamber.

Pitot tube

Fig. 4 Production of Fully-developed

Velocity Profile.
 24

2.2 PRODUCTION OF A FULLY DEVALOtED VELOCITY PROFILE.

In order to produce a fully developed velocity

profile for laminar flow in a non-circular duct, a length
of about 100 hydraulic diameters uas required. However, if
air is allowed to flow through a divergent nozzle first and
then through a straight duct, the production of the fully
developed velocity profile can be greatly accelerated (22).
In the case of the rectangular duct used in the present
experiments, the cross sectional area of the nozzle varied
from 0.5 x 1.0 in to 1.0 x 2.0 in with a length of 30 in.
The air then entered a straight rectangular duct of 1.0 x 2.0
in2 in cross sectional area and 3.0 ft. in length, fig.4
p.23. The velocity profile of the air at the end of the
straight duct was determined by means of a pitot tube and a
micromanometer which gave an accuracy of about 14 % for the
worst measurement.
Fig.5 shows a comparison of the central line
velocities between the flow in the straight duct and the
flow in the same duct with an additional divergent nozzle
over a range of Reynolds numbers from 840 to 24-80 The
central line velocity was appreciably increased by the
divergent nozzle.
 25

o
o
o
X' \ 0
4
4-3

0
0

rd
-I-)
0
0 PI
a) CiS
o b.0 H
0 P
N a) tO
0
-1 cd
r d 4-)
0
H 0
4 Cti p
rd o
o ctS
4 .r.1
3 -I-) 0
ri .1-1
Q. rd
rd a)
cd H
ii -1-1
0 fi-i

0 4 0
0 a) 4-3 PI
0 rZ p4
r. cl-i
o F.,
4-)
4-) ri
0 0
CD 0
H
H
c a)
Pi

II \

to
1-1

CO kD 4 N 0
09SP4,7 S4TOOTGA OUT' re,141.1@0 = 3/41
 26

2.3 RJCTANGULAR TOST SECTION FOR CONSTANT BEAT INPUT.

To obtain a constant heat input per unit length of

the duct, alternating current was fed through a metal duct
made of an electrical resistance material called Ferry Metal
(23) of 0.006 in. thick. A vertical mounting of the duct was
chosen because, in a . horizontal position, the temperature
profile of the heated air is distorted owing to natural
convection.
The Ferry Metal was supplied in the form of a strip
of 1.0 in. wide. To construct a rectangular duct of I x 2 in2
ei
in cross sectional area, a wooden formkof the required cross
section was employed as a guide. Six metal strips were placed
around the wooden form and joined together with thermosetting
plastic tapes. To make sure that the cross section of the
duct would remain rectangular after the withdrawal of the
form, the outer surface of the duct was covered with a double
bond resin called Hermetal (24) which set hard afterwards.
A brass flange was soldered to each end of the duct to form
an electrical terminal, fig.6, p.28. The whole duct was
heavily insulated with glass wool to reduce heat lost to the
atmosphere to a minimum.
The heating circuit is shown in fig.9, p.29. The
main voltage was stepped down by a transformer from 240 V.
to a voltage under 2.5 V. The input power was controlled by
 2?

a variable transformer and measured by an ammeter, (0-'0A),

and a voltmeter, (0-2.5V). By comparing the input power to
the total change of enthalpy of the heated air, heat lost to
surroundings could be estimated.

2.4 RBOTANGULAR UST SZCTION FOR CONSTANT WAL.LJ TI3MERATURE

To obtain a condition as close to a constant wall

temperature as possible, the test section was made of copper
which is a good heat conductor. Its dimensions were original-
ly 2 ft. in length and 2 x 1 in in cross section. For
rigidity, a wall of 1/8 in. in thickness was chosen. The
copper , duct was placed inside a jacket which was filled with
heating water at constant temperature, fig.?, p.28. A pump
heater was employed to circulate heating water the tempera-
ture of which was kept constant by a thermostat.
To cover a wide range of Graetz numbers, the test
section was later on shortened to 73 and 13 in. and the
same test procedure was repeated. The reason is explained in
(3.7.2).
 28

Unheated section

Seal

Ferry metal duct

Main Supply

Hermetal cover

Glass wool insulator

FIG. TEST SECTION FOR CONST.

HEAT FLUX.
Wire mesh Mixing chamber

- Hot water Pump heater and.

jacket Unheated section temperature control

4]
Heating element
Pump

Copper duct

Hot water

FIG. TEST SCTION FOR COLOi.

WALL TEAriPbRATURE.

Eixing chamber
 29
Hot surface

Lead Standard Hot junctions

accumulator cadmium cell
N / Cold
00 junction
0
o 0
"A o
o
o
0
o
0

o O 0 0
0 O 0
o 0 000

\\
Selector switch

00 00 Series resistance box

v
Tinsley two-dial
precision potentiometer

00
0

FIG. 8 TH3RMOCOUPLE CIRCUIT Galvanometer

Variable Step-down
transformer Transformer Voltmeter
IltU,,
/
Main ...... Resistance
-,-
supply
metal duct) -

:.
Ammeter
FIG. 9 HEATING CIRCUIT.
 30

2.5 MEASUREMENT OF TEMPERATURE

Chromel-constantan thermocouples of 0.01 in. Cia.

were used for measuring initial and final temperatures of
the air and wall temperatures. As a large number of thermo-
couples were used, they were connected to a common reference
junction via a selector switch the terminals of which were
gold-plated to ensure good electrical contact. The ice point
was used as the reference temperature and its construction
is described in (25).
Fig.8, p.29 1 shows a null-typed thermocouple
circuit which consisted of a Tinsley two-dial precision
potentiometer (26), a moving mirror galvanometer, a Weston
standard cadmium cell and a series resistance box for
controlling the sensitivity of the galvanometer. Accuracy
of the measuring circuit was 1/AV and consequently, with
chromel-constantan thermocouples, could measure temperature
within about 0.05 deg. P.
For the thermal boundary of constant heat input
per unit length, measurement of wall temperature presented
some difficulties, Taermocouplescould not be attached
directly to the outside surface of the duct because it
carried an electric current. To form an electrical insulator,
a thin film of Hermetal was painted on the surface of the
metal where the measurement of temperature was required
 31

and then the thermocouple junction was attached on the

Hermetal film. As the film was very thin and. Hermetal is a
moderate heat conductor, the error in measurement of the wall
temperature owing to the presence of another medium was very
small and the measured temperature was assumed equal to that
of the wall.
For the constant heat input per unit length, the
wall temperature varies around as well as along the duct.
Though the variation around the duct is rather small when
the duct is made of a good heat conductor, at each distance
along the duct, five thermocouples were placed around the
perimeter in order to obtain a good average of the wall
temperature at that distance.
The final temperature of the heated air was measured
by four thermocouples at different points in the mixing
chamber and an average value was determined.
 32

2.6 EQUILANRAL TRIANGULAR TEST SECTION FOR CONSTANT HEAT

INPUT PER UNIT LENGTH OF Tffig DUCT

The unheated section for producing the fully

developed velocity profile consisted of a divergent nozzle
and a straight duct similar to the combination employed in
(2.2). The heated section was made of three strips of Ferry
1 in. long. Along the
metal, each being 1 in. wide and 154
section, chromel-constantan thermocouples were placed at
3, 4471 , 6 and 12 in. from the entry plane of the duct. At
each distance along the duct, four thermocouples were placed
around the perimeter in order that a small variation of the
wall temperature could be measured. An aluminium equilateral
triangular form was provided as a guide In construction of
the duct. Constructional details were similar to those of
the rectangular test section, (2.3).


33

CHAPTER 3

BEAT TRANSFER IN THE TIORMAL ENTRANCE

REGIONS OF RECTANGULAR DUCTS

3.1 BASIC DIFFERENTIAL EQUATIONS

The Navier-StokesP equation for a laminar flow of

a fluid in the z direction is

z (f04 + d_ + 3 2 w +u+v.w+w
ay 2 az2) az 3x 3y 8z
where u, v, w = velocities in the x, y, z directions,
P = pressure.

If the flow in a duct is hydraulically fully

developed, the velocity profile does not vary with the
axial distance, ze

aw/a z 0, u, v 0,
a2w/az2 0, 3p/3z constant,

Equation (1) is then reduced to

a2w a2w
(1c.)
()z

The energy equation for an incompressible fluid

with invariable physical properties is
A24. A2t A2t 3t at at
oc (t33
`-j- + S32. + w---- (2)
dx dy az
 34-

where local temperature,
thermal diffusivity, kiepe

Since the conduction term in the axial direction,

d 2 tp 2
oz is negligible in comparison to those in the x and y
directions, and for a fully developed flow, the velocity
components u and v are zero, the energy equation is there-
fore reduced to
azt 82t
(2a)
crj

3.2 ANALYSIS OF PARABOLIC EQUATION

Equations (la) and (2a) are parabolic partial

differential equations. Their solution can be conveniently
obtained by numerical methods. The left hand sides of both
equations contain second derivatives in two dimensions and
these can be replaced by finite difference approximations.
For a duct of rectangular cross section, a square network is
most convenient for computations.
First, the following partial differential
operators are defined t-

Hf(x l y) ; f(x+h,y) + f(x-h l y) + f(x,y+h) + f(x,y-h)

- 4f(xly)

Xf(x l y) = 1 ( f(x+h,y+h) + f(x-h,y+h) + f(x-h,y-h)

+ f(x+h,y-h)*- 4f(x,y) )
 35

The two operators can be represented diagramma-

tically by

MB
X

624. f
A relationship between + and the two
6x' ay'
operators, H and X, can be found (19), as follows :-

Let OC E h. aox
-- and f3 h

The operators, H and. X, can be written in terms of

cc and. 13 as
H . 2 cosh cx + 2 coshp . 4
X . 2 cosh cx cosh /3 - 2

Solving for cosh cc and coshfl and then substitu-

ting for these in terms of sinh 2oli./2 and sinh 213/2 gives

2 sinh c /2 . (i(H +Ale+ 8H - 8X) ) -1 =A

2 Binh p /2 . (i(H +1/4H2 + 8H - 81) )i

But cc . 2 sinh-1 A/2

2 2A4 2 2 2 A 6 n 2 A8
Hence oc 2 A- 4! ' 6. 8.
and similarly,

B2 2B4 + 2.2 2 .B 2.22

- 4. 61
 36
2 a2 cx 2 4. ,32
But h Tct r5,2)

B2 ) 442, (A4 + - 4 ) I
+ go (A6+ B6
6) +

Replacing A and B in the above equation by their

values in terms of H and X

h2( a2
+ a 2 I 2
) H - 17 1 3 + 6H 2
(H + 4H - 4X ) + go(H
ax2 ay 2 a
6HX)

. -3-
2HX
+ 3. + ay + HX +....

In the present work, the first order terms will

give sufficient accuracy so that

/4 2 f oN2f
o
--IT (411 + 2X) f(xa)
a7 6h'

It is convenient to define a new operator, K,

where K E (4H 2X)

The operator K can be represented diagrammatically

as

' --
62f a2i.
K f(x,y) (3)
g 6h'


37

Finite dilIpmence eapliplico Simp son's Rule in two

dimensions.

The average value of a quantity over an area such

as velocity or temperature, can be computed numerically by
means of Simpson's Rule in two dimonsions. For a function
f(x,y), the average value can be written in finite difference
form (20) as :
A
rf(x l y)dA = h2 2:( 4f(x+h t y) + 4f(x-h l y) + 4f(x, y*h)
j
+ 4f(x,y-h) + f(x+h,y+h) + f(x+hly-h)
+ f(x-h,y+h) + f(x-h l y-h) + 16ffx,y) )

where N is the number of squares of side, h, into which the

area A is divided.

Diagrammatically,
(71) (T)

rf(x,y)dA = h
7 1

. 2
f(xly)
J
Di
77-
II
0,.
3.3 EXACT SOLUTION TO THE FAVIER-STORE'S EQUATION

For fully deveLoped laminar flow in a rectahgular

duct, equation (la) can be solved exactly (18).
In a duct having dimensions a and b in the x- and
y-directions respectively, the velocity, w, at any point
(x,y) on the cross section is given by :
00
= C(x2 -bx) + [ sin mi-
wc- ( Amcosh EF + B sinh rilb
m a )
m.1

ywhere C I dP

Am 2b2C ( cos mn - I )/ m3n3

Bm . -Am ( cosh mica - I )/ sinh Ff.

It is convenient to define a number of dimension-

less parameters as follows :

aspect ratio, cc = b/a,

velocity, w la2 2
_wa z
co-ordinates x + x/a, and y+ y/a.

Hence, co
i(mtxt., x +i) + L + +2
W + mi (cos mn - 1) sin --- . 2cr
MAX
33.
cr ;m n
MR
+ cosh - 1
X (cosh EMI sinh ale))
+ ) (5)
cr
sinh mn cc
 39

To determine the average dimensionless velocity,

wb' the cross sectional area of the duct is divided into a
square network and velocities at nodal points are calculated
from the above equation.

The average velocity is defined as

j(fw
+ d dy
wb x

By applying the Simpson's rule in two dimensions,

equation (4), to the velocity distribution, the average
velocity can be computed.

3.4 NUMERICAL SOLUTION TO THE ENERGY EQUATION

By combining equations (2a) and (3), the energy

equation can be put into finite difference form as :

g_t....(1.3.
c Mai 'm w t(x.7.z+ Az) - t(x.1.,.z)
6112

Rearranging this, the energy equation becomes

t(x,y,z+4z) t(x,y,z) w
-b 0c4z -.5 K t(x,y,z)
w w b h 6
(6)

where e ratio of the side of the square grid, h, to the

hydraulic diameter of the duct, dh.
Using this equation, the temperature distribution
at any cross section of the duct can be computed from the
 40

temperature distribution at the prectding finite step in

the z-direction, i.e. along the axis of the duct. The
computing procedure is as follows :

First, the cross sectional area is divided into

a square grid of side h.
From equation (6), ortiz
77 can be written as z/d h
wbd"h Re.Pr
ewbdh
where Re is the Reynolds number Pr is the Prandt1
number Cp/A/k and Re.Pr/(z/dh ) is the Graetz number, Gz.

Hence, 2 is a small increment of the reciprocal of

ochi ziwbdh

the Graetz number, and it must be chosen so that

tf
W-b + ccAz I-2 is small' than i in order that the solution to
w wbdh e
the equation (6) is in a steady state.
The ratio wb/w in the equation (6) at any nodal
point in the cross section of the rectangular duct can be
calculated from equation (5) since wb/vt =

To solve equation (6) numerically, initial and wall

values of temperature are also required, and they can be
obtained by using a suitable dimensionless temperature, the
form of which depends upon the thermal boundary condition.
 1

3.5 SOLUTION FOR CONSTANT WALL TEMPLRATURE

For a duct with constant wall temperature, a

dimensionless temperature, 0, is defined as

G = t
w - t

f. - t
w o

where t = local fluid temperature,

t w = wall temperature . constant,
t o . fluid initial temperature = constant.
. n
im 1 and Q w,o . O
by() w .

The energy equation (2a) can be written as

0) 20 4) 2 8 w

a7
- .S;
and consequently, equation (6) becomes

ocAz KG(x,y,z)
G(x,y,z+ilz) G(x,y,z) .1!
(7)
w w
b dh 6

Temperature distributions along the duct can be

computed step by step by means of the above equation and
hence, the Nusselt numbers along the duct are deduced.
However, because the computation involved is very lengthy,
a fast digital computer is required.
 42

ITAturatZuralus Nix.

By definition, the coefficient of convective heat

transfer, h 4/ As (tw - tb)
where 4 . heat transfer rate,
A s . surface areafromwhich heat is transferred.

Consider an elementary length, dz, of the duct.

Heat transfer by convection, d4 h * dAs (tw - tb)

= h P (tw - tb ) dz

where P is the perimeter of the duct.

Rate of change of enthalpy, dR = tzwbAcCp dt

ab dz

where A c is the cross sectional area of the duct.

By the first law of thermodynamics, d4 dH, since
no shaft work is done.
Hence h P (tw tb ) ewbAcCp dt

and h k wb 4
d11
dt b / ( tw tb )

or in dimensionless form,

h kw d
or b-h4 --b
Az / 0b

By definition, the peripheral Nusselt number,

Nu . h.dh / kb

Hence, Nup w bd2h Ifig

b (8)
 . A2
In the above equation, "b"h has already been
cca z
chosen in order to solve the energy equation as described
in (3.4). The bulk dimensionless temperature, O b , at any
position along the duct can be determine by applying the
Simpson's rule in two dimension, equation (4), to the tem-
perature distribution at that position which has already
been computed numerically from equation (7).
For design purposes,a mean value of the Nusselt
number is more suitable than the local one. The mean Nusselt
number, Num , is defined as :

Num O J NuPdz

The mean Nusselt number can be computed numerical-

ly from the peripheral Nusselt numbers, Nup , determined at
small equal intervals along the duct.

Num = Nup/N (9;
where N . number of intervals considered = z/tiz.

Experimentally, it is inconvenient to determine

the mean Nusselt number defined above, and the logarithmic
mean Nusselt number, which is based upon the logarithmic
mean temperature difference, is employed instead. This can
be calculated as follows:
 44-

Consider a length z of the duct.

h P zat 1 . wb e Cp ( tb - t o ) Ac

where at 1 = logarithmic temperature difference,

(t t ) (t w t )

in (tw - to)
(tw tb)

= (t b - t o )/ ln tw to
w b

Nu l . h.dh c wbA c du in t - t o
. w
,____,
-k
b b 'TT " tw - tb
"12
w
b"h 1 in 1/Gb (10)
= 4aAz N

It has been found from the results that the mean

Nusselt numbers predicted by equations (9) and (10) agree
within 0.5 %.

Fas_aLn.p,le o Splamtatign

For the aspect ratio, e, of 2.0.

Let the dimensions of the duct be a x 2a.
4 x 2a x a
Hydraulic diameter, dh 6a

Choose the finite step, h a/10 . 3dh/40.

Equation (7) becomes



45

Wb 5Az 1600
. KO(x,y,z)
"b h

ChoosealiSz
9
2 I'
w d
b h 32000

wb
- 1246.11.E1
w

8000 Ib
From equation (8), Nup -3;
'b
8000 in 1/01)
and from equation (10), Nu =

With all the necessary variables obtained numeri-

cally above, a computer program can be written to solve for

the temperature distributions and the local and mean Nusselt

numbers along the rectangular duct. The flow diagram of a

computer program can be seen on p. 46 and 47. Computer

programs were written for aspect ratios from I to 6 and some

of the results obtained by using an I.C.T. Atlas computer

are shown in figs. 11 et 12. Because of similarity, only a

quarter of the rectangular cross sectional area had to be

considered during the computation.

 46

FIG. 10 FLOW DIAGRAM OF COMPUTER PROGRAM

FOR HEAT TRANSFER IN THE ICAERMAL ENTRANCE
REGIONS OF RECTANGULAR DUCTS.

read in the chosen sizes of the

I network and steps , alongtheduct

set initial values of dimensionless

temperatures of the wall and fluid

calculate distribution of fully developed

velocity ratios from the exact solution, equation(5)

proceed one step along the duct and compute

a new distribution of dimensionless
temperatures by means of equation (7)

apply the extended Simpson's rule in two

dimensions, equation (0, to determine
the dimensionless bulk temperature.

iconstant heat inputl

constant wal calculate the exact bulk tempera-

temperature ture from equation (12), p.53.

1.
 compare the predicted and exact

I dimensionless bulk temperatures

if they are not equal, correct the

di p tilbution of the predicted
temperature, see p.54 .

if they are equal, compute the wall

-1 temperature from equation (13)

calculate the Graetz number, see p.40, local

mean and logarithmic Nusselt numbers from

equations (8), (9) and (10) eqautions (14), (15) &

for const. wall temperature (9) for const. heat
input per unit length
1

print out number of the present step, Graetzl

number, bulk temperature and Nusselt numbers
1 I
if Graetz number z 10, print out the distribu
tion of dimensionless temperatures of last step
i
 _

MOMAIIMM NSIMEIMMIMOMUOMOMMO Emu mmmmil a'
INIMEMEMBOMMINIUNOMMIEWHIMMOMMEEMMUOMMEMMOmm====mmil
OROMMOOMMEMMEMMOMMINIMEMEM gmmommommummirmommum g mmgmegm
ffi
ing smummumgmommggimmmum logglImmummomm egggmaimummommammil
1 1 '
guessugggsmommil
m oug miumumumgasmo
m mg g
............. ..........e.... IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIPAIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIIL .9p. .............71.17.
EMMEMMOMMEMOMMEN
OMEMEMOMMENE
fROMMEMMINNUMEMEMONOMMEN
m imeommummumummummum
gm
r mum N 1111
NAM
MI
la
1
MM
Jolgolimm
MM
O OM
mu ER
mum
momummummilmmummommumme MENEM A=MENEM
MOMMUMMEMMEMIMMOMMMEMM17:1171MMOIMMINMENUMM
4-3 ....mos....."
MM. ..........m. II; um. .. IIMUOOMM
gum
A I, M IENI NE r m..
,
*1-1
,
ISM /MEM
, MMMI
EMO UMEM
mum millogvg, mmu
mummum Momm4m110 iiihmmrnm .mommoll
rd
.,--1 co o millimpl 11 muoillumplium
.16...c..E.N. Arm us....
/.n 16dulim.
,---1 o 1111 MU Ewa num mIEB IN
MIIIIMMUMMOMM MM UAW IIIEm11111
0 . 1-)
d
FA
MMOMMEMEMOMMW
EMOM MEIMONOMIW
M
IONIUMEMEMIC:7
riNIMOmmill IMEMENEE
Annino mi m illu mmom
f NOMMOMOMOMMI
T
OMMOOMME

HOMO MMOMM gm i N AMMEMME M MMEMEM

-1-)
4-D IIMMENEMEME , N IMIUMMINOM .ERIMEM
o ffimmommummi WM Irmummilimrimmu
e4-1
0 0 Egmmommomm igibm g T uun
P4 mouggAgmm Elf mammlagE um
um m
(1)
FA
-1-)
cd
He-N
Clf

r4
\
0
0
M EMEMEMMEMM
=
NIMEM
IF
/
IERIFIF


..........
IIMMOMEMEM,
R.............
l
ournun
H
0
IN

p.
A

un
Fi.m...
....e.
mgmgmommog

M
mum
WhEOMINIMMErnim MUM=
i
MEMMMIIMMOMMI .11111111MMM
a)
sp,
0
0
rci
\
ro
limmmir wili,
mum sl
MENEM
mogggpmmummggrummomm
A
AMEmmEmmEMOMEmmuimilINEME
mil, ummognimmommummommommgm
N EN Num Ail 111111.
C \J s
"1 IIIIIIIIIIIIIIIII1

4.3 ....,
=MUNN =EMMEN
N ^ \--- MIIMMIMMITAIr
71 i gm
H P gum" AVM MIII M
0
ro
HINER
NOM= , Ma mor
A
11nM
MUNE
Sim 111111111MITIIIIIII

=1 , EMI MU

g
, CD

_
0 111111Prirli
MUM =MU
iF gninenlinall 1211ffar
ArjaMMEMOMMO
_
0
CO

H N
11t_
MI

1111MA
, mommommompommg 1111111114411111M
WM

a
mg 2 m
1M Mill
,II!
MOM

_ g 4-D 111111111111.111111....1110.11, SW AMMEIMMEMEMEMENIMMO 01.111611.111.

____
_ _ ri o 0 0 1111Kr
111111E1111111111 glif An 111111w.4
nelERE11
_
_
_
rn
o
0
0
.1-1
CA
0
a)
4-)
1101111111ow
111111.111
asperanurrir
MNIMMEMMUMME
MOMOMMEMEMEMEM
IMMOSEMEMEMME /4/
Fir
Alinsuprpurnoriprqn:
nn
-
A 11110111

AMME MEM i MUMMEMOM M . ME

EICHEMERMEMEME MEM
US
Mik11

___ rd
ARMEMUMEMMEMMENUMMIIME
Nrimummuommussomglimwmuss
_ MENNEN W , /111 10 In NNE
immumman sow I' AIM MBE lIl ME
0
_
___
_
4-)
H n NW
wor Ago mug 11111M Num pet111111
A m1111Rum Le 'IMMUNE
_
_
q-4
0
0
0 EMERMOMMOME
SUMMOOMME
i/
/ mummmigniggnimummugsgmormomm
MOMMUMMEMEMEMMIIMOOMMEMME %MIMEO
_ UMW= MOOMMEOMMUMMEM gglgommil mum
o

_

0
0
P
Cal
r-4
Emma
MUM 11M11
E M 9mu
"UM
l megg 111u"11'mum
m molimmum
(..) MEMO MEM m
MENUMMOMMEMMEEMERO
EMMMMOVIE M
_ -r-I
P 0 MUM INIMMEMMICUMEEM INNEMOMMONMOREIMMEMOMME
MMOOKOMMOMMUMMEMEMOMMEEMOMMOMMMEHE
EE
d
-,1

CA
CA

0
NIB
UM
tmogrommugnmigggimaggmgo
riumenunrrearen9"111U
MOM
mgrimpg MEM MI MU -ME

_...
Emma
gm
mu
. ..
ME
A MEM MIEMUSEN MINOMUMME MEM II
mu milmilmomm mmommummumm mu
111116111:111111LIIIIIiiiilii
111===111
mm mgmmmm mgmaggsgssommumgommgmmu ngli
mmlimm
mom
1111111
O U SEMEM MEEMBEEMOMMOMMEOMMMEMI w En MO
SMOMOMMEMER immOMOImmunMEMOMMUMMIIII OM 011
MEMMOMOMMO MUMMONOMMEMEMOREMIE MON
MO Moms MEMORmsaMEMMUMEMEINUM NOMEINIE
1 110111111
+4'7
MTh MUM_ MEM MEM IMORE gum
so, - -MMus '1 110== MUM
' L am
0 MIMI ,, ill Immgmlim . gg moggig gm
ilium 11 ma 411111 11 IIIII I llm

1 , 1.. - ...


a-
11)

a
a ;19
0
Cs1 I>
I-I
r-4
0
4-I
t

1 0 F4
CO 0
e+-i
CO
4r4
O r-I
.0 0
I
0
to S
I O 4)

I O g
-P -P
I O 0
i 0 0 0
.. .7 r.4 0
t. ( 0
I CO ,0
e
I
O 4.)
lf) - 0 ni
CV I F-I
..... 0 a)
1
CV 0
C 4,
VI -P 13 0
0 0 . PI 0
I
r1 a) rcl
0 .
_
N .0 ;.
-P . 'CI - I .....
cd . El H
F-I PI 0 0
A . 8.-- I:1 I-I
4) r-I 0
0 0 a, 4) ta0
0 0
. I-I g
ID a) 0
Pi F-I
. co 4)
CD r-I CL
0 CO 0
aS . 0 0 CO
co I I Z F-1

N H 14
. CIS eri
i 0 k
'. O CO
I .0 0
I_ . 0
%.o er-I r1
K1
% PI cH
... O 0
N
I
PI PI 0
t 1Pi C-1
TS 0
O PI 4)
co 4) 4) 0
1
.1 0 n-1 PI
ri 0 0
rd 0 04
O r-i El
r-1 0 0
1
P4 P -P
1 0
C.4
1

1 .
1
I_

a CO 40 .7 tsi 0

cl LI d

)1 / P dLi = nN
 so

CO -P
0
4-1 S
rd
4-'
CI
c-I r-I
05
4-1 r-i
0
a) c.)
gli ci
0 r-I
1 0 0
i


I
1
_
_ \ .
.-
i
I
1 ti1 I
. I
.. I
I
i
I
I

- 1
%
a.
ID c
.

rd
a) all
4, Pi
% 0 0
1-41-1
%
rd a)

% a)
Pi rid
%

.
d CV
%
'C
.
%

n r1
%
_

CO CO n7 CSI

qmit4pLULI
nN
 51

3.5 SOLUTION FOR CONSTANT HEAT INPUT PER UNIT LENGTH

For this thermal boundary condition, it can be

shown that the bulk temperature of the fluid rises linearly
along the duct as follows :

From the first law of thermodynamics, heat input

per unit length of the duct . rate of change of enthalpy.

4 w c d:b . constant.
b R CP A

For a given flow rate of the fluid assumed incom-

pressible, wb , R , Cp and tic are constant.

dt . constant.
aib

When the heat input per unit length of the duct

is constant, the wall temperature varies along and around

the duct. However, if the duct is made of a good heat

conductor, the wall temperature can be assumed constant
around the pOmeter at any cross section and to vary only
with distance along the duct. At each step of computation,
a new value of wall temperature has to be computed from the
boundary condition of constant heat input.


52

346.1 Determination of Fluid Temperatures

For constant heat input per unit length of the

duct, it is convenient to define a dimensionless temperature

t to
N
dtb_
d11h b
<is ct

where t o initial temperature of the fluid assumed

uniform.

At any cross section, the distribution of fluid

temperatures can be predicted numerically from the fluid
temperatures at the previous section by combining equations
(2a) and (3) and substituting for t in terms orig.

p (x,y,z+41z) 13 (x,y,z) = ri b
74- ,
orAz 1013(x,:v.z) (ii)

" wb`411 6

. An approximate bulk temperaturei5 b at the new

position can be evaluated from the distribution of tempera-
tures by means of equation (4).
However, by definition, the true bulk tempera-
ture is
t - t
)5 _
b dtb wbd 4h
dz
tb - to
dtb T.,
wbdh2
dz cc


53

dtb
As already shown, = constant for cons-
tant heat input per unit length, hence
dt b t b - t o
a-z-

p OC Z a ZN (12)
b
wbd'
h wb d2
h

where N = the number of step being considered.

2 )/a4lz has been chosen to have a
Since ( wb dh
definite value when equation (11) is solved numerically, the
bulk temperature at any position can readily be calculated
from equation (12).
If the prediction of fluid temperatures by means
of equation (11) were accurate, the predicted bulk tempera-
ture,N, would be the same as/3b calculated from equation
(12). In computation, it will be found that p L differs from
AI) and a correction is needed for the predicted fluid
temperaturesp, such that IOL .)01).
Because of non-linea4ity of the temperature ana
velocity profiles, the correction will have to be varied
from point to point on the mesh with the maximum correction
at the centre. The following expression would be most suit-
able.
(/5 7)9)
Corrected temperature =13 + :/3)
7/3b)
 54-

But the factor,

(Pw - , cannot yet be deter-
(fiw -Pb)
mined since the wall temperature,p w , is still not known.

A good approximation can, however l be obtained by using the

velocity ratio, w/w b , for fully developed velocity profile

instead of 0?w-/3)/(13 w -A b ). This is justifiable

. because the two terms differ only slightly, and moreover,
(.1.71) -[3) is quite small in comparison to/3 during any
step.
.*. corrected temperature 213
wb (pb -pig))

A new value of the approximate bulk temperature,

b , is then computed from the new distribution of fluid
p
temperatures, and if it is still not equal to the
same procedure is repeated.

3.6.2 Determination of Wall Temperature

To predict the temperature at the wall,from the

values at other points inside the duct, it is necessary to
satisfy the energy equation. It is convenient to consider

a surface B at a constant distance h/2 inside the wall of

the duct and of axial length dzs fig. 13, p.55,

By the first law of thermodynamics, the heat tranb-

ferred across B must equal the change of enthalpy of the

fluid flowing inside B. The latter can be evaluated as the

 Pig.13

difference between the change of enthalpy for the entire

duct less that for the gap between the surface B and the
wall of the duct.
These quantities can be expressed in the finite
difference form. Across any section of B of area h.dz I the
temperature gradient is (t w th )/h, where th refers to the
fluid temperature at a distance h from the wall of the duct
and the corresponding heat transfer rate is
k.h.dz(tw th)/h

Similarly the rate of change of enthalpy throug:1

an element of area h. 2. between the wall and surface B can
be expressed as
h2 pw C p uz
dt dz

Summing up round the perimeter, one can equate

the total heat transfer through B to the net change of
enthalpy inside B as
 56

dth 2 e w Cp d
dt
Yk(tw -th)dz =Wb Cp A aib dz - Z 2 z

Certain assumptions must be made to solve this

equation. For a metal duct, the temperature variation around
the periphery is small and it is assumed that t is constant.
From experimental results, this is justifiable.
If the gap between the surface B and the duct wall
h
. is small compared to the duct dimension d , the mean veloci-
h
ty in any element can be taken as w /4 and the axial
temperature gradient as dth/dz.

dt h20 whC p dth

w
. knt h
it wbk 1.--b -
A uz
C pc 8 a--z-

where n is the grid points on the duct wall.

This can be written in terms of the dimensionless
temperature /3 and solved for the wall temperature.

.
e 2 A2 wh
wh,h
(13)
/3w - -1- To: I-13 h tri rvi; 4A

In the first summation for, h the values at th.3

four corners have to be included twice so that there are n
terms in the summation. In the second summation, each value
is included only once.
In this equation, the values of e l n and wb dh
azirz---

can be calculated as soon as h is fixed; the velocity

h )
w in has been obtained above, whilebp h and lph can be
 5.?

determined from the figures found in (3.6.1) . Hence p(4,

can be determined at any point in the duct.

3.6.3 Nuspelt numbers

By the same approach as that for the condition of

constant wall temperature, the peripheral heat transfer
coefficient, hp , for constant heat input per unit length
is
kw b d
412 dt
aib /(tw - tb)

or in terms of dimensionless temperature,i6

h 4-w

. . peripheral Nusselt number, Nu p a 122kdh

(14)
4(13
w -f
i b).

The logarithmic mean temperature difference,AN t1

(tw- tb ) - ( tw 0- t0)
a
tw - t b
in

2 dt
7/4 1 fiwo wbdh b
b cr dz
wo
 5s

in /3w - /3
h
1 = Aiwo
j3w -Ab -.043w,o

idw - b
ln---
and Nu . 1 _,0
17-
(15)
4. w 73bw,o

The mean Nusselt number, Nu m , can be computed

by means of equation (9) like that for constant wall
temperature.

Example of Computation.

For the aspect ratio, or +, of 2.0.

Let the dimensions of the cross section be a x 2a
Hydraulic diameter, dh . 4a/3
Choose the finite step, h = a/16
Ratio, e = h/dh =3/64
Energy equation (11) becomes
1k a6z '641N2
)3(x l y,z+Az) 13(xly,z)
wv w d 1/4-3-
b h

Choose the step a4z

A 2 1 9
(.64) - 8192u
wbdh

/(.3 (i,y,z+Az) = w
-b K 1 xasz)
 5.9

Equation(13) for the wall temperature becomes

_
ie w 23.6 pi

At the first step13 0 = 0, and/I wo = 4.6

As the initial temperatures and finite steps now

have numerical values, a computer program can be written
for solving the temperature distributions and the Nusselt
numbers along the duct. Computer programs were written for
aspect ratios from 1 to 4 and results obtained from the
I.C.T. Atlas computer are plotted on p. 60 to 62, and
tabulated in Appendix(7.4). The flow diagram of such a
program can be seen on p.46 & 47. A comparison of the
Nusselt numbers between the thermal conditions of cons-
tant heat input per unit length and of constant wall

temperature is shown in fig.17, p.63.

11 MIREMOM IMMMERIMMEMMOMMEMEMEMOOMMOOMME a
IMMEEMME MMEMMEMMOMEMEOUREMMEMEMMUMEMMUI 111 1/ 1
.........rum....r....... MEOMMMONMEMMME
WIEMMEMOMEMEREMOMMEMMOOMMEM EMEMMEENEUMIO

111111111 111111111 IIIIIIII IIREM -.'-t;.=1 PIM

MINHIMMEMEMMUMEMEMEMINEmMINIME 11111111111111 ., IMMO

0 1 1111111111111111111 1111
q 111 1111111111111
LNIIIIIIIIIIIIILIIIIIIIIIIILIIIII 1111
A ci IIIIIiiiii

Tommumuipommo quiripi
. .
11111111
mom.
........
-P,-) cL4
iVMEMEEMEMEMMUMMI MISMEMMEROM
MB MI ONENEM UBE IMMEMEM
MENNE
ME IMME EN FA ,---,
I.
I
HUM
Mir
I

1 n 11111111 1111 1111111M 1 1
IL 0 OM IMEMEMEMERIMMEMMUMBIEUMMEM
MAL EL
11
.116.1 0
Pi
0
rid
'...
__C

PIA
MEINIMOMOM
IMUMMEMEMMEMEMOM MORIMUMMEOMMEE 0 0 MRMEMIMII
OEM. i
maum
I thirallellitan
1
4.) ..../

LNK
NE
iiliklairri
MEW SUM
MEE= RE MU
,H ,
CA
1711000
1 Urn,
MEM
Elm MiAMWOMIMENEMEMEMMIMMEmismorn
MON c-1.
.., ,oc)
Ill U imm
NEMEM
irmIMEMEMMEMMEM MOINEMONMEMMMEEME
SHOMMEM II MIKOMMUOMMEMON MUMMEMMOMMEIMMINI
Inal........... .......m.
.......
-
co
co
9 101111111
....1...
M ILIIMOMMEMMUMMEMMUMEMMIRNMEMEM
.........17.1z ...............m......m.
MEMMI :EMILM 11111 MOMUMMOMMEM
EMOMOMMEMMEMOMM H
0
0
mu ism

MOM
ORM
MEM retlikleMUMMMEMMEMMUMMIM
EMEMENEOMMUMEMEMEMOMMEM
0
.1-1
4-1
0
0
1

I '
MAI
11111111

EWE OM
IIMMIUMMEMEMEMEMEMEMEMEME
FE1=11 Imilliki
..........0 mmo timm. li nurpourannorm
milm.MEENIIME.MMI
2
a.) 0
11"114111
Li! .qi ,-1 lim 11111
111111119111:1 1P11111
I I
MMUMMIMMMEME 1
MIIIMUMEM
NOMMEM
M
IMNUEM
9
NUMMEMM IIIMMEMMUT
111111111111:
OEMMOM ,e.)
4-)
cH
o
INI
IIMOMMEMOMENMEM
SOMMEMMEME
,MOMMEMEMENEMEMMIE NE
MEILMEMMEMOMEMIUMMUM ME 4-3 H 1111111111
ITN MEM MOMMOM OW a.likik
IMAM IIMIHMIRE LIM 4 i - d
M 1111
IMODONIM11111=11 IVE01116111111111MEMOMI
MEMMONIMEMEIMEMMIIV61n1 VINEMEMMUMMEMEMEM
0 0
0
I MP
MEM
ff
E UMENINIMMEMOMMEME MOWN MEMBEEMEMEMEMME M P MIWINIMMEM
MMONOMMEMMOMMOMME11:111 VIIIIIIMMIIMEM 0 Cal MUM=
M
MENNMEMEMBOMMEMEM IIMMUNUMNOMMOME 0 .ri
MEMOMMMEM
EMLNUMMEMMUMMMENEME cormams.m. .r1 0 IMMIIMM
U
MMEMMMISOMENMEM
IMOMMEROMMIMEEREM
EMOMERMINOMMEMEM Urn
ME 1
Eall
MEMMENMEMONO
MENSOMMUEEMME
Tell
4-)
d
H
0
CA 11EMEM
WOMEN
NEWOMMEMOMMUMMONOMIUM IMIUMINION
IIMMEEMMEMMOMENEHMEN
momMEMMEMEMIMMMEMMOMEM
somi
WMammum
III U ``'.
0

-c'4-3
EMEmmffi
MEM m
MEMMOMIIM

NMEMUMMEMEMMEMENEM En
\
EMEMMEEMONERMEMEMEMONO Mil
1
NEMOMEIMM

MEM
ME M
NEMICOMMEM
IMIIMMUOROMMOMMINUMM
c---
IMIEME
OM ROI
Milliallinill1111111111111aNk MIN NM
IIIIIMEMNOPEOMOMMMOMMUMU
IMMEMENEONOMMEMEOMMOREENE ME
............m....1.....1.........
Mir=
MR MU
UR
b.O
.
.r1
cr.4
'MEE=
MEME

BRIMIMILINUMUINB
MOMODMOMMIUMME MUM
IMEMITIMEMMOMMIONESTEMEn
imall

inillh
MN,
MO=
MO
mmilEammilm
MEMBETEM IMEN
EMONEMMEMNI meg MOM
whith

i - - 1 11111111111111111111111111 gil 11 11111111111IM

I! -- , 21mi nst ra
E rn .
mum

Ea 4111E1111111

F
inm
SI uIME
U MM
mom
U MNIM
,.
+.
mum
milm
' ' ME
mum
IMMEMM Rp m. Ips
E MREM
EME
o.
EU
KM
OM
ONOOMMEMEMO
111111 MEMOMMEir
OUNIOMMOMMWO ISME MORIMMMEMEME MO
MEMMINI
0 ME =Om 4.
I I MAO
_WM a NM I OMEMOMM 11 ME
ap. r - MN
111111
M IME
IIMMEEMEM
MEMONMERM
E
imp .....
REIM
ME

_ III 1
1 -711 III I 11M
,_ --
 C. sI
a;
1:..3
H
H
0
4-1 4-)
C
<
F-I g
0 4-3"
4-1 0
0
CO 0 0
Ck.I P 0
".../ G)
, I
CO 4.)
0 0
-ri 0
-1--1 rd 0
Cd N
;4 4.) 4.)
Pi i O 0
as
QS 0
0 0
P rd
0 0 0
fa t P
F-I

CO n-1
Cd C.)

, 1

.-
141

0.
N (I]
a CD ,0
H 4)
ri t.0
4-I 0
ON 0 0
t kl F-4
,.../ . pi
H

4-)
% 1)..2 vi
4-)
9-I

0
Csi

0

co
(.0
0

q 4pd d
li 4 = nN
 a

a 4'
03

0
0
a
0
to 0
Ci-i

CD
o 43

0
ta
4-,
rI
crj
ts3 4'
.c 0
0

F-1
"P
0
11-1

4t
4-)
ELO

0 a)
r-4
P14

1.) rI

1-1
a)

1C:5
04
0
3I
-P
GS
co 0

C4.1 CD 03 tD
nnnn

/ 14 P wLI wnN


n 0 63
CV

43
0 .
04 0.
in g
c-. H s-I CV
. ad
O 43 . 4-I
F-1 cd 0
a) 0
.0 .0 0
4-3 ri
43 4.3
a) g cd
o .0 cd F-1
Is) 4-3 -P
. co -1-3
O 0 0
a) o a)
a) 0 P.
3 a)
4-.) rd cd
a)
.0 g 4-1
II) 0
C's1 ..-.. CII 0
_C c-1 rc
13
a P 43 C.)
....., ,r) -I-1 0
aS 'd
IQ 0 F-1
....... g a) Ps
....... Pi ad
42 0 i-I

>
o
a a.
L'
H
0) -sp
0 0
ED
al 03 0
Ct a) H 0
O H 43
Is ad 0
3 a)
N nct A
O 43
0 4.)
In o g - 0
r--. -14 -P F-1
4:1 a) o
a) 0 44
- F-i o
P. 0 .0
-1-1
44 44 ED
O 0 g
0 a)
it) 0 0) ei
O g
c) 04-'
ri r1 .1-4
4,
Cd
Pi rcl
O 0 r-4
O o o
1/7 0 0 P.
et1

N.
r'
ED
-g-i
Pi
cg OP .7 CV 0
tU rIN drIN
3.7 EXPERIMENTAL RESULTS

3.7.1 C9Agimt_yea.11_p_
n ut Perjit LenKth

Tests were carried out over a range of Reynolds

numbers from 735 to 1960, which is well in the transition
regime for non-circular ducts. The wall temperatures were
measured at 7.5, 12.0 and 23.5 inches from the entry
plane of the duct. The initial and final temperatures of
the air were also recorded.
To determine the NUsselt number at a position
along the duct, the bulk temperature of the fluid at that
position must be known. In experiments, though intermediate
values of the bulk temperature could not he measured
conveniently and accurately, they could still be deduced
from the measured values of the initial and. final tempera-
tures of the air.
For constant heat
Input per unit length of
ti the duct, it has been shows.
in (3.6) that the bulk
temperature of the fluid
rises linearly along the
duct.
dt b
dz = constant
Fig. 18

tf - to
 65

Hence the bulk temperature of the fluid at any

position along the duct can be obtained directly from a
linear relationship between the initial and final bulk
temperatures of the fluid and the total length of the
duct, see fig. 18, p.64.

Cal,ulatk

Data

Aspect ratio, a + = 2.0

Cross sectional area A d = 1.0 x 2.0 in 2
Total length, L = 29.5 in.

Volumetric flow rate of air, V . 1.7 c.f.m.

Initial bulk temperature, t o = 71.6 de6.x.

Final bulk temperature, tr . 94.0 deg.F.

Initial wall temperature, t wo = 84.0 deg.F.

Position along Local wall Average wall Fluid bulk

the duct temperature temperature temperature
inch. deg.F. deg.F. deg.F.
115;0
113.8 114.5 76.6
7.5 114.0
115.2
124.3
12.0 125.1 125.6 80.0
125.6
127.3
136.4 319('
23.5 136.3 137.2 89.3
138.5
139.8


66

Hydraulic diam6ter, dh 4Ao/P = 4x2x1 = 1.333 in.

Mean velocity of the air, wb 60V/A.. E.OX7

4

= 7.33x103 ft/h
Rate of heat transfer by convection per unit length of
the duct hp P (tw - t b ) B.Th.U/h.ft.

Rate of change of enthalpy per unit length of the duct

Cp t wb dt b .Th.U/h.ft.

.*. by the first law of thermodynamics,

C e w, A (t. - to)
h
a -P -P L (tw tb) B.Th.U/h.ft

Nu = laRkb
d h = 0.24 x 0.075 x 60 x 1.7 (94.0 - 71.6)4
126. x (tw tb ) kb x 12

= 3.73/k ( tw tb)

Logarithmic mean temperature difference, (tw- tb)1

A t max tmin
in (iltmax./atmin)

tw tb - 84.0 + 71.6
ln (tw - tb ) / 12.43

h
d1 h 3.73 ln (tw tb)/12.4
Nil kb
kb (tw tb - 12.4)
 67

Reynolds number, Re = R wb ah

1385

Graetz number, Gz Re.Pr = 1385:4030.231/
= 1330/z

Dimensionless wall temperature /6 = w - t o T ----2

ct
t r - t o wb db.

tw - 71.6 ,.h 6 0.82 x 12 2

g4-70.:7775 7.33 x 103x 1.3332

= 0.997 x 1O - 71.6)

z Gz tw tb 4t1 Nup Nui lew

in. deg.F deg.F deg.F x lo3

7.5 177.0 114.5 76.6 22.7 6.7 11.2 3.78

12.0 111.0 125.5 80.0 25.5 5.5 9.8 4.54

23.5 56.5 137.2 89.3 26.4 5.2 9.3 4.78

Fig.20 p.70 shows a rise of the measured

dimensionless wall temperature,/3, with the reciprocal
of the Graetz number, (z/dh)/Re.Pr. Variation of the
measured Nusselt numbers with the Graetz numbers
 68

is plotted in fig. 19, p.69 and tabulated in Appendix

(7.5).

3.7.2 Cmistant Wall Tern erature

Details of the test section were described in

Section(2.4). Tests were conducted over the same range
of Reynolds numbers as for constant heat input. Since
the rise of the bulk temperature along the duct was not
linear as in the previous case and it could not be
measured conveniently and accurately, the local heat
input could not be estimated. The initial and final bull:
temperatures of the air were, however, measured and
recorded, hence the total heat input and the logarith-
mic mean Nusselt number were deduced.
Because of the same reasons as above, ducts of
different lengths were employed in order to cover a
wide range of the Graetz numbers.
Results were computed by a similar procedure as
that in the preceding section(3.7.1). Variation of the
logarithmic mean Nusselt number with the Graetz number
is shown in fig.21, p.71 and tabulated results are
available in Appendix(7.6).


'C
F-I
6

+ 0

-P 0 (1.1
rn cIS
+ -P II
9-1 0
U a) 0
FA H
FA cii -P
+ cii 0
0

H
+ 6 -P
-P
()
() O 0 0
..--7 H H P-
-I-) -r-I co
C) cH
g-i
.0 ce) a o
V to c.n a) P-4
N U) cp -P
-I-) LO
5. 4)
+ 0 1.0 En r-1
,
e4 0 H
I 0 0

9
V H -P
+ . 0 9-I

0
g
rd
0 + -I-) a) F-4
cD L H PI a)
cp CL o
0 O H
Cl) -P
0 C
Cl) RI
II
rr:i 0

O P.-a
O H
u) 1E1 r-1
Cl)
! O 4-1 ,521

0
cf--1
0
F-1 rn
cc/
6.0 o
O ,o

rd 0 -P

+3

Cl) 4-3
cii ai 0
F-i

r-1
 _ _
NO III III MO ME

1 111111 11 MN MIL
r ApprINN
mmilm
1111
MTN
1011 E
rill
IIIIIIIIIIIIIIIIIIIIIIIIIEII
Mr
NM MIN
MEE OM ME
t OMM1111 11 EMI MIMI NENE ME
UNE ll
,11Praril
TA11111
....... -
1111111111111
11 111110p1111
-P
M
li
P-I ,d
- 1111111
EMEMMIN
ERMWOMMEMMEMMEME MOM ON MEW In W 3
4- 1111111111E
--0- n 11111111111111E1111MMI 111111MI 11E111 111 1111 *H d LO UMW
.L IIIIIMIIMMI111111111111 111111=111 M memo11111111 d I......I
ME1111171111
momENINN MUM 1111111111EMM MMEE111115 MO
11111111111111M m mmmmm Immo
9 c-1 0 ENNEENP
Q 0 H- . WM
IIIIIIII 1111111 1111:111111
1111111"9191rimpli
......11.1 , .1 ... .. mill
.. Il li pm:
40 . !
11.11111121111 491111.:
IIIII 11111 rd
4--
4,

ro
-d.FA MIA
-P

-,,
Au 11
mm I

ibh
1111111.01.0111k1
1.1111111 . 181
=UM= UM WINBOOMMUILU
111:11111/4110Milll
.......e...fromm..1126.11.1111
1 1191111111100111M mIlrialliman
Nimmam.
r9
"'
0
c i
td3
1

v
FA
Cli
-,
2,
_
- . UM
mu

HMO
LER

ull
ME UM WM MIR=191111.11.11
all 111.11111.1 MAI I ME IMO fi; (6) .r1

j , c(e) 111:11
IIIIIII
MEM p_ i _p
milui
... Illimm
.. ..., mm illollimmill
illi mmm ..... ......
cD4

E
MIRE

En=

mum= mmm m mammmommmmommom
mum= 1111 mmm mm mmmm 1 NEMEEMEMEMEIIMIIIIMMINIM
4-3
b.0 .0mmu
IMEM. n 1 smormun urimmum :=11 .g rum
WI
trimu
ENNIN
FENN Elli
... . . 1 mom
IEBEEEMPNI
LEM
1101rn MMEMEMME
NEREEEEIEEEEE
ImpOMBI morn
1.1 ME EMEMEM
E3
..
m
H-
c6 -I'
W
...
UMW
MU%
1111pripi
IF .
Milli
..
111111_Mll
rlirEUIrrlii
RENNEENNIIIN Nor
... ...
IIMMEE111111.11 MUM Ell
1111111111111111
m
0 00 0 lissA
ummo.
MONIE
Aga
H

MMUNIMERIMIUMMEI HOLE
ImlimmmImmimmIlmommillimmi mmomplimmummo
m ma 0 um , mor mum
.2 1
111114011111111=11111M IIMIIIMII
EIIIIIMMEM1111111111111 111.91111111%
01111111111.1111
1
114111111111111E
co
rd . , .-- _

..
0
moms mmIlmmomm um mumm um= mo 0 M MIS ME
MI111,111M4 Mil Emmmalmmmmmmm W 0
7 El l
MILNE
EIMIEURN
ENE Em
Ernrur
mimmlow
E m MOW
_EN E N NE =NEN il il MEM:
nwr morw mp. i nsinp.
111111111111=
mminummOwas empoulimm
r
rd
1

rd
0

0
_.:.
"
4-1
_ mmr
.
DiN
i
uprit elm mom immilm m . m millimm
mg
MI

moms mom mmummommum Emm
ME MEM REMMUMMUME MO
MIM MIMI
NUMIRLIrromummum
i Um
ME W
1j1111
IWO
OMMIll
MUM
o
:E:
"
-1-1
rd p
El

V
d

0
Penn

11111
pm= Im

OMMEEM
MOM
1
WON
UnnindiqUILIMLI NUI.
ThArmmouloor
mu "ANN
IVA .mm44
mm 0
c\J
mom
.smog
mom
1m
iq mommum
mommem ol or m or
EMI '
1 4 mum
im ;Ism.
immum
Numen munimm0 0 E
_-00 g miam
inammu 111111111MIWOMEEMMEll TO 11
to
.H
mmusim
MUM.
_
NIMMINM
M IIIMMONM ESIMMIMMONNE
mummy p OM MOMMEMINIMMOMME EMIR
MO M Ift
Mil
imi IMMO=
MSNIMMIOMMO MOM IIMMEMMO
"MMIIIMM III Millir":1110."111
-TIII MMEMEMINIMM MU Mum MEW= milimmill1111 =OM

II
I1
PIM N
ONOOMMEMIMMOOMMOW
NEffiENTENIENNEE,
mmultzull
4111:04111 411 III 1.
_, . m ommilm Jummmopp0
um Inumummullums
MMOMOMMIUMUMMEMEM0
NumENTIINEEPENE
MEM EMI 111111EMENE11110O
Nor
ol

.... ..... ..
Mr,
. mommispEpum EmmummimmummumMUE

i
IMUOMMONME
.....4. .....1.81.
III:
some= mmiummomplimmum
simummummum
imgmmimmumm
m'
4 9
II
9 r
_ 411
Al
Irmo
....
immommo mmum 6. 1. IMAIMIIIIIIiIIIII
02simpummulinummumo mmummummmom
mu
mil mosimmumm
mOMMEMEginm
impimill mm U MW MEME
11 ER

1101 EIMER UM 111111=1111
IIMIIMOM MIIIIIIIIINOMMO011
IIIMIE111111111 EMEMINOMMMEMEMEMOMERREMEEEME WM
MOMENIIIMME OM um mmumplummummumpinumplimmimmor
MAE
Rum ENNEEN !ER

mil limmo

01
RIM
mmmmmmmmmmmgmsmgmmsmmmgmmm
us. niIl lIfilnMLIII on
1
 MESEEEEEEEEEEEMEEMEEEMEMEEEEEMEMEMEMEEEMMERMEMEEE MEIMMEEMEEEEE
NEENMEMMENEMEmmEEEEEEMEEEEEEEEEEMEEEEMOEMMEMMEMENEINNEMBEENUMEMEEMENE
EMEMAIEUMEEEEEEEEMMENEEMEEWEENEENEEEMEEEMEEEEEEMINEEMmEEEEUMMEEEENEE
EMEEMEEMEEEEEEEEEEEEMUEEEMEEEEMEEMEEMMEEEEEEEMMEMEME EMEEUMEIMEEME
EMEEMEEEEREEEMEEEEMEEEEEEEEEEEMEEEEEEMMEEEEMEEEEEEE mEEmPl
EEMEEMEEEEEEMEREEEEMNEMEEEMUMEUMMEEEEREEE n
NEEMEEMEEEEMEEEEEREEMEEEEMEMEMMEIMMEMEMEMEEMEME UM
11111rnmmummummlnimmsmuumunummumnium -:I , EME
EEEEMMENOWENEMEEEENEMEEMENOMEMEMEEMONEE 0 d e
EMENEEENEEMEENEEEEEEEEuEEEMMEMENEEEMMEEEEEEE P H

REEEEEMEEEEEEEPIEEEMEMEEEMEEMEEMEEEEMEEEESEMMEE
im n
:::
IMEEMOMEUMEMEEMINE N ENMEMENEMEMITOM MEMEMMEN1
: 0 B
WARM

ma
umossmummiummummum uumumus
InummunnommunnnownnImmoumm onnumuninsumosso
4-' 0 la
m .0
mnimummummimmommo mmommgp
mommomommompaimmu 4-, MM.
MUN EEEEEENEEMENEMEOUNEEEEUE
EM EMM EM MU
MEMMEUMMUMEE *r-i
0 0
0 0
OMMEMMINIMMINIMMEMEMMEMMIMMMEMIMM sEEEEMMEEMEERION
ummommmommommummmumminnummumppapummimpsomm t3 P k
ENLEE11111111 n
1 1111 cd a
minimmummorimmulmmu mommullso mmillmmilium
EMEEEREEMEN NEON impopmnum mop
omm m

rd
0 0
d
OMMINIENEMPO EMEMEMO EMEEEEEEEMEMEM ER= ME. N NIP
pplimummommommommumenummommin ummonsmommum 4-) ' F-1 0 Milli M
EmmEEMBEEEEEEMEEMMEENREEEMEEREEMM MEEREEEESEEEEMO 4-3 Pt mom
mummummimmsmimmummrimummummum EMEEEMSEEEMUNE o 0 5
mmummummommimmummimi gammummumm 411 momplimmium ,--1
Pi
H 5? . imam
mnin
rannignia min=1 0116 11 1111111 ingaur--
MEMEMMUMEMMUMMUMMEMEMEMMIWUNIAEMM
v__ gra
MUER =mom. - o
H +?

H El
mosummessimmlimmummunsti
niummummummnolummummumm MEER
..7; r--,1.4
c,- 0
H lailO
nm m
immWESENEMMEMEMEMEINftMEMEEEREE INIERNUEM
ME . '...- t.-
OMMEMEREEMEEMEMEMMEMEMMEMEME EMEMEMNEMEINI mmmm
m mmmmmEmmmmommmmnummmmmommommill mmmmmmonno mmn
p - 4_, 4_,
NEMMEEMENMEMEEEEEEEEEELIMEEMEEEMEEE EMMEEEME .H NUM
MEEMEEMMEMEMEREEEMME ITME MENU MEEMOM _ammo
miummommippmonimppoal
me 4 mumemmiummummm
lim m
0
r0
0
0
o
4-
co
INWEEI9r=ilintain InininUilli= a
-1
,--1
0 .
limmommommummopppplomme _pplimpp p mplimmomemmp o 4-- -
Immogymmummommommmilmmilo mommollagimmommismir 0
immusommommummemummum iimmip ii III: ME MO _p rd
mumu
ME
mmummummumom mism
MEMEMEMEMEMEMEMEMEME lim
MEM
A mo
MEM
gpm
MM.
m am
MEE
m H
0
0
P.-4
,d 1111 .4
- MEEMEMEMEEMEEEMOWEEMEVEIRAMM
E MAEEENEMEEEMEMENS:11=
IMEMMERNER
To......1
EMEEEM 0 cr) 0 .H:i-: -1-)
_
IMMEMEEEEEMMEMMEMEMERM OM 1 EEEMEN r-I
1
IMMEEMEEMEMEEMEMMENNEUMMEEE Lifil 416:111mmin
Z
0
> 0
mmunmommummilmmummommommmumm mum o .
EMEEEMEEMMEUEEEEMENEEMEUMMEME:13 rd (\.1 All
VEMMOOmmEMEUEEMOUEM MEMMEMEEEME ME It:III:III cd
migmEmOMOREEEEMEM OlimmEMEEEm WESid IRM o l,- cH
MENEM =EMMEN NOMmEMEEEEEEMESEI EMU E H 0 -
NEEESIMME EEMIONMEEMMEEMMEREEEEMEE RUMS 5
H
EMMENEMEMONINNIMMNIMU MEMMMEETNEEMEREMENIE
OO o 0
MEEEEMEMMEMEMEUMMEUENRS . MEM CH
a4
EIEREMEIFEEMEEEEEMEEEERE MMONEMOI .H .H
EMI
MEM
MIUMMEENMEINIMMOI
EMEEEMMEREEEEERE
a. 11111
EME
ars .-5 C4
4-,
0 - .
um ullmlai
EMEEEEMM
EM
mm....m
IORMEMME
Ir
mmommomumppmmom
EMEEMEEKEE m ESEEEEE
mu
Ern= ME
r Igiiiiiii
mom=
MIMEO
4-,
.,--1
P
0
to
o
c=-1
-
4-,
0
0

U.
,ppgmmumm _el um _j__ imommulummumm
II - MAW
0
H
N
:_.
'at
u)
MM EMEMMERM
IMMEMIIMI"04_t r ImIlmmumm liumm 0:5
II 011111
-
minommmmum r
_
m mamma mom= rd ..

EMEEREEEENE .11 .) n mu m u 0
P P
ci-i
O Irill
mmummmoimummputtaz -111111111
Irla= 4 IIII C)
ME= MUMUM MEMUMEMEMEMUOMMEEMMIUMEMEME
ENMEESIM MMEMEEEMO mommimmimmmommommomm
ME ,0
5,
co
d
_ MO
EEMEEESE ENVIREEEEE EMMEMMEEMEMEmmommE i .0
Immummappmiummum imponlimmuummummummum e , ...L. 0 '
EMEEMINEEMENE mm EREM E UMW pploomplimps
po M
E
MUUMMEEEMMEUEMENRIMEMIMEOME NENE EWEN ME
MEEEEEEMusEEEEEM MEE MEM momps c- WWI
EMEEEEREEMNEEMERE E REMEEMEEME MEE= CV. RTEMEE
MEEMEEEEEM MEEEMEMEM IEMEREEEEEREEEMMEERNMENVIM 4.4
, L-1-
- 771- EMEEEMEMEEEEMER ESMEEMMEMEEMEHEMN EEmom to
r___

r_ ,--- milimummum mommummiumplumposammirm .1-1

II ..............................mm
EMEMEEMEEEEMENEEEENEEMSEMEEmEm MBA
MMEEE
EE
ma..
RU.
,.!1.. tIEUITA
li

-Almalimilifillial
i n ia l
jaggENNEN
l ain pill
EN Eam
UNIMBERIMATIIR
ENIINEENEIN
momm0
oft
Emu
LE
immismaimilu
4--
_
_ 1
,- 4-
milim
EMS
VI Eli= .
UM=
ME MUOMO
1
1
,
cp I MI L RE' MI
_ mu ,
---t-
1._.

.
---r-
-4 -
- WE= 4- 0. * Fl Ea =El
. 1-4- ' -t- -
MEW
' EMIll NE
EEMBEIBEE

1111 1111111111EIN MUM ME 1 i
 72

CHAPTER L.

HEAT TRANSFER FOR SIMULTANEOUSLY DEVELOPING VELOCITY

AND TEMPERATURE PROFTL-FIS IN RECTANGULAR DUCTS4

4.1 THEORETICAL ANALYSIS.

In the previous chapter, fully developed velocity

profiles have been assumed and this condition exists in
practice only when a long duct is allowed for the flow to
become fully established before the heat transfer process
begins. In many cases, heat transfer takes place as soon
as the fluid enters the duct, so that both temperature and
velocity profiles are developing simultaneously.
The Navier-Stokes' equation (1) for the velocity
w, in the z-direction is

ew ew
I Tx7 v.& + wSw
ax ay 8z

Langhaar (10) obtained an approximate solution

for a flow between two parallel plates by using the
. following assumptions :
(i) The term 32w/h 2 , which represents the rat A of

change of the viscous force in the z-direction with respect

to z may be neglected in comparison to those with respect
to x and y.


73

(ii) The velocity components u and v are

negligible in comparison to the main stream velocity, w.
(iii) The pressure gradient, dp/dz is a
function of z alone.

. . a2w a 2w (IR +w aw (16)

e dz az

The above assumptions will also be used for

solving the developments of velocity profiles in rectangu-
lar ducts.
The pressure term in equation (16) may be
eliminated by considering the flow at the central axis of
the duct.

k ) e1 dz + w c az
where the suffix 'c' refers to the central axis. Hence

d'Alk ( rw )2w ) ( 3 2w ew aw
S"? ) Trc + 5-; w

The above equation can be put in a finite

difference form by means of the operator 'K' . defined on
p.36 .

K w(x,y,z) - K w(x l y,z) c + %lc

dw = w
Aw
6ehr- 'Llz 6z
 w w(x,y,z+4z) w(x,y,z)

ift, du 2 A4
6 1;4 K w(x l y,z) - K w(xly,z)
ewbdh

+ we w(x,y,z+ b z) e w(x,y,z)e (17)

Equation (17) will be solved simultaneously with

the energy equation (2a) which can be written in the
finite difference form as

1 Ch
t(x,y,z+Az)-t(x,y,z) . 6 w % (c46 2) K t(x,y,z)
lh)211
"bA'h

As already mentioned in the previous analysis, the

2 ), which represents the dimensionless
term (abz/wbdh
finite step in the z-direction will be given a suitable
value when the energy equation is solved numerically,
( see p.40 ). The same value can be applied to the term
(/iAziew bdh2 ) in equation (17) as follows :

"Az Zi z k C7)AK
( .52.1 ) pr
7 -----
Qwbd h wbdb. CI) k wb dh

Define the velocity ratio, w* w/wb
Equation (17) can be written in terms of w* as :


75

w*(x,y,z+/Az) i Pr f9A!.) K w*(x,y,z) KW*(xly,z)03

r17*-7
e w d
b h
we
w w*(x l y,z+ilz) e w*(x,y,z)e

w*(x,y,z) (18)

Boundary conditions
(i) With a bell-shaped inlet, the velocity of the
fluid at the entry plane is constant everywhere and equal to
the bulk velocity, wb , hene at z 0, wit; . wt . 1.
(ii) The fluid at the wall is stationary,
w* = 0.

In the above equation, the value of 'e' is fixed

2 ) has been
from the size of the network chosen; (o(Aziwbdb
given a value when the energy equation is solved in the
previous section; the Prandtl number, Pr, has to be given a
numerical value depending upon the type of fluid under consi-
deration, for example 0.72 for atmospheric air; lastly the
term w*(x,y,z+4Nz) can be approximated by the following
steps :
(i) Solve equation (18) numerically for all nodal
points in the network assuming that

w*(x l y,z+6z) e w*(x l y,z) 0 . 0.

 76

Hence, the first approximation of the velocity

distribution at (z+ANz).
(ii) Compute the dimensionless bulk velocity, wt,
from the first approximation. As the correct wt . 1, the
error of the velocity at each nodal point is (1 -wt).
(iii)Add (1 -w*)
b to the term w*(x,y,z+Az) c -
w*(x l y,z) c in (i) and resolve equation (18).
(iv) Repeat the same procedure until wt N
When the velocity distribution at (z+Z1z) is known
the energy equation can now be solved by the same procedure
as in the previous analysis (3.4).
Computer programs were written for rectangular
ducts with aspect ratios from 1 to 4, for the thermal bounda-
ry conditions of constant wall temperature and constant heat
input per unit length of the duct. An example of a computer
program is shown in Appendix (7.7). Results obtained from th-
computer are plotted in fig. 22-25, p.78-81. A comparison of
the predicted Nusselt numbers between the hydraulic boundary
conditions of fully developed velocity profiles and simulta-
neously developing profiles for rectangular ducts of aspect
ratio of 2.0 with the two usual thermal boundary conditions
is shown in fig. 26.
Prom equation (18), it can be seen that the effect
of the Prandtl number on the development of the velocity
 77

profiles is quite significant. As a result, solutions of the

heat transfer for simultaneously developing profiles contain
the Prandtl number as a parameter. Fig.27 & 28, p.83 & 84,
show variations of the predicted peripheral and mean Nusselt
numbers with the Graetz numbers and the Prandt1 numbers of
0.72, 0.10, and 10 for rectangular ducts of aspect ratio
of 2.0 with constant heat input per unit length of the duct.

4.2 EXPERIMENTAL RESULTS.

Test equipments were the same as those in section

(3.7) but without the unheated inlet section so that the
velocity and temperature profiles could develop simulta-
neously. Tests were performed for the thermal boundary
condition of constant heat input per unit length over the
same range of Reynolds numbers as in section (3.7). Results
are shown in fig.29, p.85.


78
0

4-3 S-A
u) rct 0 N
P C-1 cf-i r--
ri ci 6
0)
c4
h3 ..,...) ,Q)
II
cd -P "ri '-r
H -P a_
rd G F-(
a) o (1) .,
-P H PA 0
4-3 G) c ;.-t
0
O
H
PA ::1
PA -P
0
S.

Q., t)".) d
g ,--1
c;I.)i .a d
o c_2
(1)

a;
0 r-1
c..
..
0H cr; (1)
E 0 -P T-4

.-.,
0 PA I-1
--o H
+3 4-) 0

O H d
Cl) CI . r1 4-3

C d cd4
4 a) -P
W-1 Ci%

g cd . r-1 g
al -P o
(1.) ,---i -2 0

M
0H
.ri
-P

c0;)

4-3 0
.1-1 C4H -P
cs
O 0

w
04d
O 0

a g-1 oci
--(2)
0
4-3 N.0

O N ;-s

-P d
Ts a) Pi +3
(/ :A 0
LI Li a) a)
4-, f-f

!AO

(st

i/
9 , 1"N

- 'IF 111,
 79

co
0 0
o - /.1 4)/( 0, 4 M4) =q0
MEENIMAIMMEIMUMEMEMMEMSEMMINEMMEEROMMEEMMUMinOUOMMEMMEM177
MEMEmmOIMMUMUNIMEMSEMIUMEMEMMOUOMMEEMEMarnmummomelmmamilmmEMEM mall
EMENNIIIINUMMEMEmmoMMINUMMEMEMEMEMMEM MEM MIMMEMUMIRMUMERNEN .,IMATIIM
104.1
MEMOIMEMOMMEMMENRIONEMEMMEMOMMEMEMEMMEMMEMEMMEMMEMEMOMMEMMEMgg
EMEMMINNI all NMEMEMENSEMERESEMMIMEMEREMM IIIIMOMEMminapleANNEM
011 INIUMMEM I MMEINIMMOMMUMOMEMOOMEMEMEME EMENEEME mum=
UM RIHIMMEMEM OMMINIUMMEMEIMUMMENEMMEMOME EMMEMONE 111111.
UM 101REENMEEN IONMEMILIMMENEMI MOIERME ME
ME
.... UA
VI W "
ME
1..... 111.
1111111111111111
wommummummummplymmumm P
-
mu-1- mums
ROI
mpg mmummilisms I imEMOMMUMMEMENOMmomMEROMM o ME
EMEEMIREMEM 1 I mammummummiummums, ,0
MIUMMIIIN
MN , MN
mumMEEMMEMMEMMEMEM
IN M 5 4-)
111=11aMOMME
IIMMEI IIMEMINUMMI a 17ROMMOMMINOMERINSINIEN rri c\ I 11ft-AMINI
MU IMEMEMMEMME 1111mmEMEMEMMINIEmmEMENE 0 7-= EN- MEM EEEMMEM
B M mimmumm i mmmummummummumm munirum
mmum smssiumigi i mmiggiumgmmummummum
lk a EMOmmUMEEMMOMMONIMMEmoll pu m 0 mon mum=
Urn
m um ummammummomm
nip maspraharramormapon
1
ME Mill IMMO HEM
m
mummummem_mbionim
t

d
4-0)

rd
CH
0
11 nrir
ill
=WM MMEMEMENENO
ummusiummu k
sumo= 111111101MMUUTUMEMEMOMMiirMil
WEEMNIMMEONINIMEM
mminummummukommtwom 0P P
P
0

mum
MO
MUM 2,-) 0 r0 ONIN
MINIMMOMEM._ IMMUISITIMINIMI "EmMiall 0 H ME
MEM WI RUM MINIEMMIMMER Mid= mu
limummumm_Amplip
mmumsmaggssm mum W milMMEmaillUmoraamMuUmm -1--1 b..0 0 Um
MMUMMEMMEEREMENUrna Mu
I 111 1 ElarlAntirra.igraSIMMO 0 0 Mill.
=L W
0
d
4-)
H
4-) mom
....
EMMA-0
. a Il
w, 1ft IIMMOMEM
. .....
luummumma
. MINE..-1
....... 0 . .
. mummumm far
.
SadMil
ORMEN
OMM U
rd
0
0
0
P
,
0
0
sm
Ell 1111 ,11 MINEMMEMINW
EMIEUIrammnad 111 4'

m1
TEMOIM
mil. . ........
i Inalkulinn AI tis-
.......... ,.
- -P
H
0 --I

rd
, . q
m
"Epsom. ik "lam=
..
.. ......... n...... H
0 d 11141.

yommumummimmumm will EMMEN KU :WM l ulls limmEnO EMmum

g4
0
9-1 ' ,-1
cf--1 40 1-- .
EMMIIMMOMEmEMIMME ro o LO
IONMENTrOMME WILIMmo rMI
MEM 1111
MEMNON II
ITN m ammumm WI m- -i
PA 0
0
4._. ' .
limmummumm 1 limin e
II
sa EN
LO
H
T ins=
IIIIIMININBRIMOUN s INIIMI
MUM 2,-, 4..) mum mum
.....IIIIIIIE il ma ., ,-1 .1-; -r-1 II MU
mum= mu um 0 p4 q IMMI
1 .11111 rill 0 o
IMMOMMUMMOMMOk.
ummummmummum

mum
mmumummummumummommum fl usurromnim
man.
mmummumumm
. r-i
0 k
mi .
am-
mmurnommummumm r---. 0
nial
ummimiemu
ll111111111111K lik 11111 MINIMNIEN
MMIRLINI III.211 H
0
rd
P4 M

on
EIMI-tirlii=1111111111111 II IMILHIMI
lem...ua C6
P
-P
ROM
ME

. 1riti ..........
NM OM Igen ME Il
l k 0 H 04 t--- MN
INOMEMERM MN MINIMMON t,o,
mpg= mum mows P4 -, .H -1.--
SlImmi.011
INIMMOMME
MEMENRE
E MOSIMEEMOM
MERMITIMOM 11111 WM
WM=
1111= EU i
l
mom= um ,
-ri
p 0
0
0

0
4-,
0
-
t
ERMEMEEMEMINUNME mogiumft 11 UWE
Q.4 C6 0
EU
IMMUMMEMESUMNOM MEMMENIUMENEL
INIMMUMMINIUMMEM
11WM11MINIMMI
....m......
.. .. .......r....MA
INIMEMMEMERENNEENMEMER mourn
mummmummummummmumm
OUSERNIMI

IMMEMBE MN SI
immu NIIMMON_
MN=
mums
ME ME

INIMOREM
ll
rd H
0
_p
4-)
.,. .
0 yl-I
,s74
-P
0
0
-
WM
MM.
immg.
MIME O
mummi
ME
M
lansuonula=111 111111flows=1111 I.
mom
.H CO 4-'

I0=111
g rd m -
MU MIUUMMEMEMXIIMMUMMEr IHMill 0 P 0
ME
ME
M N11111
IHMIEMMEMMEMON WIMINE
IIMMUMERIMMEN WM
::::::::::1111111
MIIMMIMMUNIIIMMMIIIIIM
IIMMIMMOM IMO
118111:1111
....
1111111:101rn
Mg
1111
k ilk P
p4
4-
N
0
cH
0
0

_
MINUIMMU
NEMSMOMOMMOME
MONMEMEMMO
M EOM
MMOI
MMM1:
MEMIN11:31 liiikirn
mmilli
mum
il
mAgull W
-I-4 11111
MMEMEMINEUMUMENEME
inumunrmumiumminunnunma imimmg mo weik
mumummumimmumimmumummum milmmum u la
Imommuummummummilmumummg um
1 .11,1
m Piti
- UM
Ism
mem
011111.11111111:111111
noa mm
". muummmummii l
mum mmemmiummel
mummilmmallitlinilligm
um
am
mum
II
Ilm
_
1-a m I llinniraml -k muumuu t.

ill ilk a vavou . -
mul
mom
Ea amp....
RIUME MOM=
2-
-- ..-_-, --,--
4_4 -1- .--,
-
_

. .
, T-1
,a
.: 1

A
' n .
 81

cs4
(r)
F-1 ,g
-P

0 0 (\J
rd
Ui 0
g-1 F-1 0
Ui e-1
H 0
-P
g
f-1 Ui ,0
cij
-P
ED 0
g
0
'CC
0 g
rd
g 4-3
-p .1-1 rd
4-3
0 Ea (1
H F-4
Cal H P-I

FA 0 g
C) PI C'd
ra

9 O -P
g
-1-D Q a)
H 0 H
0 H
U) -P
(1)
rd
g
UI H
ti)
0 4-)
rd a)
C) ci p1
Cd
0 -P
H
rd -P
C)
F.4 0
p.i co

U1
CNI

Pt,

1,4
0

1Z 4 7/V in N


....1.1nn Ai.

82

0
4-) 4-> Co
Co
O 0
0

O 4-)
o 0
a) 0
P
ci 4-)
C-c
4.
rcp
04
0
Co
C.L./
ZLn "1")
o
H
0
-0 4-) 0
iv C. -/-1
0.- cs1 rti
i i- o
C? a 4--)
-0 _.., a o e3
S.
. .....-1
t) a tIS
o
0- . 0- S)
0 \
2
0 0
." \ --, C) P -P
0 \ -0 -0
v
Qi (Ti 0
Gs s D 3d cti
IP :r
....-j ...P0
._,...
'0 irl P
co
.. r .7i
7.-) O 0
I.'..... 0 --k-, ,0 0
kt 0
, r-I
-ile-, !Z
0
a cc5
tO

a
---
1, r xi
4 -P
4-I 4-)

0
o
Co /E1
o
P
H
O c6

LO
H
ci

z:t c\.10 c0 q31

q31/11 P =


83
0
0
0 -P
4-4 CILS
CO
F-1 4-)
(1)

5 CO 0
0
K
4-3 4-i
o
a)
co 4, rd
(I)
a)

4-3
c-1
d e4-1
O H o

o r0
r d c -1-.)
G) 0 LO
4-) 4-) 0
C.) 0 (1) -0
ri (1) LN
rd F-1
a)

a) ri
4-) U) CI)
a) Pi 0
H
O -P
4-1
U) 0 Pi
F-I
0 Pi H
,C)
o 4-)
0
ri (1)
r0
H 0
-P H
rd a) 0

d 0 4-)
F4 rd CO
P-1 0
o
C) H
,d CO
-P
o
(1) H
0
4-) 4-) 0
C.) H
N
(4-1
CH H
t+-1
P:1 CO 0

1-1
PT-4

q w
P = -1111
 84
0

rd
o CA rd
M 0 P
U) H
0
ED 0
H
H cd
Cd -P

O
.
r-1
-P
0
0
*1-1

PI4
r1 UI c-
P 0 0
O A 0
A
rd C))
O 0 A
P H
H
cH
.1-1

rid O
-r--1
0 P
PA d 0
u\
N A 0
A
w
O 0 P rd
P A 0
o P
H
O 0 0 c-1
o pt2
U)
rd
O -51 12
P
H 9-1
W 3
O 0
H 0
P (\J
r
d 0
ci P
O H 0
E 0
H
O (1)
0 0
N
0
q0-1 P
m
N A
C) 0 m
O e0 0
ef-t

171 0 0

00

ELO
.ri
PT-4

0 UN 0 tIN 0
Nc- c--
q T1 d
V P 11
 co
g-4
40 4-3
.0 0
El 0
0
P4 C\J
O (0 LN--
cii

4-3 CH 0
o o
g-4 4-3
0
ch 0 la4
-1-3 rd
g4 d
H
cd H .0
cri Elo

rd czS
ci 4-3 H
-P
4-) a) 43
o g-4 r-1
04 0 0
ci) g-4
C-3
a)

9
ra Cl)

H P4
cH
-1-3 0
H $-1
0.3 01 4-)
cn
LO CD
g
04 4-)
g 0
C'd H
(1) 0 4)
g
rd 0

t>3
rd H
4-3 -P

F-1 0
O 0
0
H 4-3 (N-1
"
o 0
*1-1
U) 0
ci) r1
4,d g-4
CD 0 d

t^4
 86

CHAPTER 5

HEAT TRANSFER IN TRIANGULAR DUCTS.

5.1 RIGHT ANGLED ISOSCELES TRIANGULAR DUCT.

The exact solution of the Navier-Stokes equation

(2a) for a fully developed laminar flow in a right angled
isosceles triangular duct of sides x+y = 0, x = a Ind
y . a, is given in ref.(30) as :
co
K [a(x + y) 4-(x +7) 2 - ( a ) (--)n N-3cosech(NA
n=0

sinh(Nx)cos(Ny) + sinh(Ny)cos(Nx)

where 2aN = (2a + 1).a

Put w+ a 2w/Ka2 ,0 x m xia and yt y/a.

co
w
+ (x++ y+ ) - i(x + + y+ ) 2- 16 (-1)11 cosech(2n+1)n/2
n=0 (2n-0050

sinh (2n+l)n (2n+l)n + + sinh(2n2 1 )7t y+.

2 x.+cos
2 ,y

cos (2n+l)n x + (19

2

The network used In computations is shown in

fig.30, p.87. Velocity ratios at nodal points in the
 87

a
A 1n

x = a

/1//f
x + 0
X
0. a

\ I

V2h

Fig.30 Network for the Right-Angled Isosceles

Triangular Duct.
 88

network can be determined by calculating the dimensionless

velocittes from equation(19) and then computing the average
dimensionless velocity, 4,0 by means of the extended
Simpson's rule in two dimensions, equation(4).
Numerical solutions of heat transfer for the
hydraulic boundary conditions of uniform velocity profile,
fully developed velocity profile and simultaneously
developing profiles in the right angled isosceles triangular
ducts were obtained by the same procedure as those for the
rectangular ducts in sections (3.4), (3.5) and (3,6), with
the thermal boundary condition of constant wall temperature.
Results obtained from the Atlas computer are shown in fig.31&
32 and Appendix(7.9).
For constant heat input per unit length of the
duct, determination of the wall temperature presents some
difficulties. From fig.30, p.87, it can be seen that nodal
points adjacent the sides AB and BC are at a finite step, h
from the duct wall, hence the analysis in section (3.6) is
applicable, but the nodal points adjacent to AC are at a
distance h/V2 h' from the wall, thus a modified analysis
is required.
Consider a surface S at a constant distance h/2
from the sides AB and BC. By a similar reasoning as in
section (3.6), heat transfer by conduction through S is
 89

k ( t w -th) dz

where n = number of nodal points on AB and BC.

Rate of change of enthalpy through the area

between S and the duct wall is

h2 e wh c, tit h dz
8 EF

Now consider a surface S' at a distance h/2V2 from

the duct wall AC.
Temperature gradient at any nodal point on S'
(tw th,)/(h/V2)

Rate of heat transfer by conduction through S'

1
k V2h dz (tw - th t )/ (12/t/2)

yn' 2k (tw -th ,) dz

where n' = number of nodal points on AC.

Rate of change of enthalpy through n' elementary

areas, each being (1/2h.h/2i/2), between S' and the duct wall
n' 2
,2 h e whi cp dth , dz

From the first law of thermodynamics, rate of heat

transfer through S and S' = rate of change of enthalpy
inside S and S' in the absence of shaft work.
 90

n'
yk (tw -th ) + 2k (tw -thi ) e wb Ac Cp dtb

a-E
n'
h2.e wh Cp dth -e wh , Cpdth,

As in
in section (3.6.2), the wall temperature, tw
is assumed constant around any perimeter. If the cross
sectional area, A ct is substituted by + lan')h.dh/4 1 and
the temperatures,t aro replaced by the dimensionless tempera-
tures, , defined on p.52, the above equation can be
written as n'

n +
$17 4
4- 21!
2n')
11,
Pw

w A 2 n n'
2'.1.11. -13
e2 "b =41 h + 2. w" h'LA h*
Cu + 2n') 8criNz "b b
( 2 0)

Solutions for constant heat input can now be found

for the hydraulic bounclary conditions of uniform velocity
profile, fully developed velocity profile and simultaneously
developing profiles by the same procedures as those in
sections (3.6) and (4.1) and by using equation (20) instead
of equation (13) to solve for the wall temperature. Numerical
results obtained from the Atlas computer are plotted in fig.
33 & 34, p.93 & 94.


91

, c)
CJ

0
0
7-
-P

O rd
.0 0 0
1 0.

H
7-1 t-43
O -P
C15 CO
-P I C-4
-P
0 .0 PA
E
C1) H
A-)

H r1 H

Cl)
o
0
0.4
4-1 Ct
-P
0 PI Cl)
0.4 Ci
0 0
'CI ri
O El,

0 H -P

rd
O 0
PI 4-1 CO

0
4-4 PI
O 0 rd

9
0 g
cii

-P N
Cti 4-) 0
.ri 0 CO

Cd P P
-P
T-
N" \

CV 0 CD l0 - CV
c- 0
C13 CI ci
r/qP IT = tilI


92

0
co
v-

0
4-3 rd
H 0 0 v-
c-1
1.3
-P g
0 0
0 I C-1
ra -P

g H 0
F-I
-P
H
(1) H
U)

-P
H
ef--I
-P
o g-i
FEI cr3
PI 0
rd r-i 0

-P c ,g
0 H -P
rd
(1.) 0 C
o
F-1 cH
CD1 -P

4-1
0 rd

LU
0

-P NEt0

.H
c- CS H 0

-P

rc\

ID
r-I 0
rr-4

cj 0 co 0

w
(431/14 PWIT =
 0
0
Cu

En
cj 0
rf CO
ct
JD 3
cts g

I'd H
0
-P 4-D

0 Cn
H (I) .p
H
o
0 itll o_
0
124 cn
o

H H

rd
0
0 H 0
er0 b..0
-0 Cu
(-1

I 4.)

-P rd
ta0 g
H
C.-1
.H a
-0 o
Cr)
N
ci
0 0:1)

cH ,c1
rot
4)
N 0 -/
co
CO 'N

p-i
ri
0
0

rj 4-3
0
r-1
-P
Cl)
(-7
rd 0 -P
0 ci 0
g-1
P-1 rd

ri Cu
C-1-1

0 CO Cu

q, d
4/ P =
co
a qd
P = 7ritq
 95

5.2. EQUILATERAL TRILULAR DUCT.

5.2.1 Exact Solution of the Fully Developed Velocity Profile.

The fully developed velocity profile for a laminar

flow in an equilateral triangular duct has been determined
exactly in ref.18. In fig.35, p.96, if the length of each
side of the equilateral triangular duct is 'b' and the
origin is at the centre of one of the sides which is on the
y axis, the velocity, w, at any point on the cross section
is given by
b) V3y b)
W
."GV4
X (x + V3Y-

Define the following parameters :

Pimensionle p s velocity, w + z

Co-ordinates x+ 2 x/b and

The above equation becomes

w + x+(x4.+ 1/39+- )(x+- in7+- (21)

Fig.35 Network for the Equilateral Triangular
Duct.
 97

.2.2 Finite Difference Operator for Equilateral

Triangular Network.

To solve the energy equation numerically, for this

particular cross section, it is convenient to use a network
of equilateral trianglos. A new finite difference operator
K' is defined as :

K'f(x l y) = fi + f 2 + 3 + f4 + f 5 + f 6 - 6f0

where f o , f 11 f 2 f 6 are functions of x and y.

f 2 Jo
f

Diagrammatically, K i f(x,y) = I
]
i
f4...O f6

A relationship between the operator K' and

2 which will be represented by V 2 , has

been obtained on p.23 of Southwell's book (31) as :

f
1 Kirx,Y, h2 ( 372 fO l 7 2fi
+s ( + 9 2f2 + ...cy2f6)

(22)
prom the above equation, it can be seen that a
direct evaluation of the term 17 2f(x,y)' at each nodal
point requires six neighbouring C7 2 f' terms which are also
 98

not known. However, this difficulty can be overcome by the

following steps of computations :

(1) For the first approximation, put

2 2 2
9'2fl 7 f2 + f6 6 fo

Equation (22) becomes

h2 172f ( x 1Y) Vf(x95)

The above equation is used to compute approximate
values of '1192 1" terms over the whole cross sectional area.
(2) Use the values of V 2 f obtained in (1) in
equation (22) and recompute a new distribution of %72f.
(3) Repeat the computation of the distribution of
c7 2 f by using equation (22) and the values of 2 f obtaineE
previously until steady values are obtained.

5.2.3 Numerical Solution of the Energy Equation.

The energy equation (2a) in the finite difference

form is V2t(x,y,z)

Replace V 2 t in the above equation by equation (22).

ZS t = t(x,y,z+Az) t(x,y,z)

8 w ccASz d h 2 V t(x,y,z)% h 2
Tic (i7 2,v i + C7 2 t2
g
w wbdh
+ V 2 t3 + + 47 2 t6 ) (23)
 99

Values of 72 t 1 c7 2 t2 , ... and 7 2 t 6 are computed

by the method described in the preceding section, values of
2 are deter-
2 ) and (dh/h),
other terms, i.e. wb/w, (o;Az/w b dh
mined by the same procedures as those in section (3.4).

Example,

For the triangular network shown in rig.35, p.96,

the finite step, h b/30
Hydraulic diameter, d h 4Ac/F b/11/3
b .161
e g h/dh I/I0A/3

It (Q41z)/(wbd121 ) is chosen to be 15000, oTaation

(23) is reduced to
h2
t(x,y,z+Ada) IC't(lcoriz) - 1-6 ( 0 2t1 +

2 t 2 + v 2 t 6 )3 + t(x,y9z)
(24)
Initial and wall values of temperatures are required
to solve the above equation numerically and they can be deter-
mined by using the same dimensionless temperatures as those
in sections (3.5) 8G (3 .6)
Hence for constant wall temperature, the dimension-
less temperature 0 r ( tw t)/(tw . t o ), and the boundary
values are
0 0w 0 and 0b,o . 1.
 100

Numerical solutions for the hydraulic conditions of

uniform veloeity profile, fully developed velocity profile
and simultaneously developing profiles were obtained by
means of the Atlas computer and results are plotted on p.IO2
and 103, and tabulated in Appendix (7.11).
The thermal condition of constant heat input per'
unit length of the duct requires a slight modification for
the determination of the wall temperaturo, but the basic
approach remains the same as that of the rectangular ducts.
Since the nodal points adjacent to the duct wall
are at a distance 1/3h/2 from the latter, see fig.35, p.96,
a surface S at a distance V34014 from the duct wall has to
be considered instead of h/2 for the rectangular duets.
By using the first law of thermodynamics and tho
same approach as that in section (3.6.2), it can be shown
that
dtb dt h
2 kntw - 72.5 k th
%/5 a nC wb A C h2 Iwh .n-
1 P

The above equation can be written in terms of

dimensionless temperature,, defined on p.52 and rearranged
into a2 n w
. 2
/3w a 12..11A_ "b'h
.Az v7- 413h
h
(25)

Numerical solutions were obtained for uniform velo-

city profile, fully developed velocity profile and simulta-
 101

neously developing profiles by the same computing methods as

those in sections (3.6) and (4.1) and by using equation (25)
to determine the wall temperature. Variations of the Nusselt
number with the Graetz number obtained from the computer
are plotted in fig.38 & 39, p.104 & 105 and tabulated in
- Appendix (7.12).

5.2.4 Enftrimental Results.

Details of the equilateral triangular test section

for constant heat input per unit length of the duct have
already been described in section (2.6), p.32. Tests were
carried out for the boundary conditions of fully developed
velocity profile, i.e. with the unheated inlet section, and
of simultaneously developing profiles. Results were computed
by the same method as those of the rectangular duct in
section (3.7.1) and they are presented in fig. 40 and 41, p.
106 and 107.


0
4-3 H
o H
cd

4-3
0
ozt
4-3 0
co cd
O 4-3
-ri (J)
QS 0
b,.0 0
cd 0

rd .0
o 4-3
-P
4-3

Pi
cf)
4-3 0
0

Q rd

cd

.0 0
9 cd
rl 8
4-3
-P
o (d
CO
Cl)
4-3
cii

d 0
01
o 0
104
-ri 0
FA CH
0

rd

-P
0
(ll
rd 0
a)
9
c0

PTA

Ci 0


rtr
Jo 3

cn1

4-, En
CO -P
g

rd

O P

o
4-)
O 0
H
Pt4 0
-P
EH
P
0
cj C) C
P H P
o H
-P
5 10
0 P rd

P 0 N
I-) 0 E
H 0
CH
-P
0
S.
co
N
tic
H
P-1

Co
P -P
O 0
rO
E 4-D
CO
O 0 g
-P
N
0
r-I 4--,
rd 0
0 0 -1-7
P
P-1 0

(cs

h0

C\1


a .

0
1,4

_
C4 2
cio
a
co
IN // t4ir,1
. 4
n cs4 0

I, 1 '4
 tst

4-,

ci
Co orJ
0 0

4-) o

g
.H 4-)
c6

Co
rd 4-)
o
4-3

0
H

ri 0
E NN
O rd
cti a) 0
C .; 4-)
C.24N
4-)

o H 0
0.1
4 CC
04-'
-P [20 0
ci st
-P H 0
H H H

CO C1 -P
Co .r1
0
a)
0
cH
24
N
rj
0
4-3
o c4 F-1
r-1 0
rd -P
ci
r-1 a)
g 4 'Tr
cr
re/

PT-4
c-4

kr 0 0
01 / _
1,
2 r -
n .) ?i .1,1.
 106

C.)
'71

4-3
0

rd
0
/1"
-P a)
0H
0 4.3
p
c!3 0 cH
-1-3 0
4-) ej
0 H 4.)
0
hO U) 0

rd UI .4) 0

4-) T-1
0
p
H 0
,-1 0
tr)
04-'
0 C-1
Pi

0 4-)
4-
4D 0 as
0 0

ta
4-)

rd as
0 4.)
P4 CO
cii 0
(1) H 0
0 0

rd 4_,

-P .p
-r-I H
0
UI cH
hO rd
0 FA
o

rd
0 0 d
C-1 FA bp
C])
cd
a) 9 4_)

hO
C-4

3i/gp.Tg
 f07

0
0
0s1

0
co

0
kta

op o

U-

6 un
cs4

1117901 = 4.1N
 108

CHAPTER 6

DISCUSSION AND CONCLUSION

6.1 PREDICTED NUMERICAL SOLUTIONS.

In the same manner as the exact solutions for the

circular ducts (26,28) and parallel plates (27,29), the
numerical solutions obtained in sections (3.5,3.6) for fully
developed velocity profiles in rectangular ducts with
different aspect ratios show that for a given aspect ratio,
the Nusselt number has a maximum value at the entry plane of
the duct and decreases as the Graetz number decreases. As
the Graetz number approaches zero, i.e. at large distances
from the entry plane or very small flow rates, the Nusselt
number approaches a limiting value which represents fully
developed velocity and temperature profiles. The NUsselt
number increases with increasing aspect ratio and has its
greatest and least values for parallel plates and square
ducts respectively.
The results also show a strong effect of the
thermal boundary conditions. The comparison of different
conditions in fig.17, p.63, shows that for the same aspect
ratio, constant heat input per unit length gives higher
Nusselt numbers than those of constant wall temperature, a
 109

result similar to that obtained for circular ducts.

Accuracy of the immerical method depends upon the
sizes ot the squares in the computing network and the finite
step along the duct. In computation, it has been found that
when comparing the numerical solution of the velocity profile
to its exact solution found from equation (9, an accuracy
within one per cent can be obtained if the size of squares
in the network is less than one tenth of the hydraulic dia-
meter of the duct. The same size of squares has been used
for predicting fluid temperatures, hence the same order of
accuracy may be expected. For the steps along the axis of
the duct, the dimensionless group, (wb/w)(q4z/dfiwb)(dh/h) 2
in equation (6) must be less than 1/2, otherwise the
solution becomes unsteady. On the other hand, if the sizes
of the network and the finite steps are too small, too much
time will be required for computation.
Predicted Nusselt numbers for simultaneously
developing velocity and temperature profiles in rectangular
ducts in fig . 22-25, p.78-81, exhibit variations with the
Graetz numbers similar to the solutions for fully developed
velocity profiles in fig. 11-17. Theoretical curves for
circular ducts ( 2 1 3) and parallel plates (8,11) are also
shown as a comparison and they show the same trends as the
predicted results obtained in this work. At a high Graetz
 110

number, the Nusselt number for the former is greater than

that of the latter and their difference decreases as the
Graetz number decreases until at the Graetz number of zero,
both solutions are as otic to the limiting Nusselt number,
fig.26, p.82.
As already mentioned on p.76-77, the Prandtl number
has a strong effect on the Ntsselt number for laminar flow
with simultaneously demeloping profiles. Fig.27 & 28, p.83 &
84, show that at the same Grsetz number, a small Prandtl
number gives a higher Nusselt number than a larger one. This
can be deduced from equation (18) that the velocity profile
of a fluid with a small Prandtl number develops more slowly
than that of a fluid with a larger Prandtl number. Hence,
the solution for the uniform velocity profile t or Pr = 0,
represents the upper limit and the solution for the fully
developed velocity profile, or Pr = co, corresponds to the
lower limit.
Predicted Nusselt numbers for the right-angled
isosceles triangular and equilateral triangular ducts in
fig.51-54 & 36-39 vary with Graetz numbers in the same
manner as those for rectangular ducts.
It may be noticed that all numerical solutions in
the present work are presented within a useful range of the
Graetz numbers from 0 to 200. For a Graetz number over 200,
 111

the duct is too short for any practical use, or else the
flow is too large for the laminar regime, for example, with
a flow of atmospheric air at Reynolds number of 1600 and
Graetz number of 200, the dimensionless length of the duct
(z/dh ), is only 5.76.

6.2 COMPARISON WITH EXPERIMENTAL RESULTS.

Experimental data for fully developed velocity

profiles in circular ducts with constant wall temperature
(26) shows that at high Graett numbers, the measured Nusselt
numbers are greater than those predicted by Graetz. Experi-
mental results have been obtained for a rectangular duct
with the same boundary condition and they showed a similar
deviation, fig.21, p.71. For a Graetz number under 73,
experimental and predicted results agreed quite well, at
Graetz numbers of 100 and 150, the deviations were about 28
putty
and 51 % respectively. The discrepancy was due to an effect
of the variation of fluid thermal conductivity with tempera-
ture. At a high Graetz number, i.e. a short distance from
the entry plane, the wall temperature and fluid temperatures
close to the wall were much higher than the bulk temperature,
and as a result, thermal conductivities of the formers were
much larger than that of the latter. From equation (6), p.39,
it can be deduced that the fluid temperatures near the wall
 112

increased more rapidly than the predicted values obtained by

assuming constant properties at the bulk temperature. Though
the opposite effect occurred at the central part of the cross
section where fluid temperatures were lower than the bulk
temperature, their deviations were less than those of the
temperatures near the wall, and hence the overall effect
resulted in ahighermeasured value of the bulk temperature
than predicted value At a low Graetz number, i.e. a large
distance from the entry plane, the temperature difference
between the wall and the fluid bulk was much less than that
at a high Graetz number, the effect of varying thermal
conductivity was therefore very small and a good agreement
between the measured and predicted results was obtained.
For ths thermal condition of constant heat input per
unit length of the duct, experimental data was obtained for
the rectangular and equilateral triangular ducts with fully
developed velocity profiles and simultaneously developing
profiles. Results in fig.19, 29, 40 and 41, p.69, 85, 106 and
107, show closer agreements over wider ranges of the Graetz
number than those for constant wall temperature. At a,low
Reynolds number, the measured NUsselt number is slightly
lower than the theoretical value. This appears to be due to
the opposing free convection effects which were present in
all the shown experimental results and became more apparent
at a low flow rate. By inverting the test section used in
 113

section (3.7.1), a few tests with additive free convection

were performed and it was found that at Reynolds numbers of
1060, 1400 and 1800, the mean Nusselt numbers increased by
about 17, 10 and 3% respectively.
The effect of varying thermal conductivity, which
caused a deviation of the measured Nusselt number from its
predicted value, was apparent at a low Graetz number, i.e. a
long distance from the entry plane, where the temperature
difference between the wall and fluid bulk was much larger
than that at a short distance, ( c.f. constant wall '6mpera-
ture). The same result was also obtained in ref.(33) for
Graetz numbers up to 14.
In a test with constant heat input per unit
length, it was assumed that the presence of a thin film of
HerMetal between the duct wall and thermocouple junction
introduced a negligible error in the measurement of wall
temperature, section (2.5), p.30. This was confirmed by an
experiment in which a thermocouple was directly attached on
to the outside surface of a metal can containing boiling
water while a thin film of Hermetal was present between
another thermocouple and the surface of the can. Readings of
both thermocouples were taken with a circuit the accuracy of
which was + 0.2 deg.F. and no difference was observed betweer
the two readings which were subjected to a higher temperature
gradient than those of the ducts in the experiments.
 114

6.3 CONCLUSION.

In the present work, numerical solutions have been

obtained for laminar flow heat transfer with fully developed
velocity profiles and simultaneously developing velocity and
temperature profiles in rectangular ducts of various aspect
ratios and in right-atgled isosceles and equilateral triangu-
lar ducts for the thermal boundary conditions of constant
wall temperature and constant heat input per unit length of
the duct. The solutions give the limiting Nusselt numbers
which approach those calculated by other methods,(1) and
section (7.1), and moreover, their trends are compatible
with those of the corresponding theoretical solutions
obtained by various authors for circular ducts (26, 28, 2
and 3) and parallel plates (27, 29, 8 and 11). This
indicates that the numerical method employed here is reasona-
bly accurate for both square and triangular networks. Since
the method is straight forward and effective, it may be
extended to heat transfer in other cross sections such as
circular sectors, various triangular and elliptical cross
sections. Though suitable computing networks such as rectan-
gular ones for rectangular ducts with very large aspect
ratios have to be used and some modifications of the finite
difference operators and of the expression for the wall
temperature will be required, solutions can still be obtained
 115

without too much difficultire with a help of a fast digital

computer.
It can be finally concluded that
(1) for the same Graetz number and thermal boundary
condition, a rectangular duct with aspect ratio greater than -
2 gives a higher Nusselt number than a circular one, but all
triangular ducts give lower Nusselt numbers than circular and
rectangular ones;
(2) for the same Graetz number and thermal boundary
condition, a flow with simultaneously developing profiles
gives higher Nusselt number than that with fully developed
velocity profile;
(3) since most experimental values of the Nusselt
number are higher than the predicted ones, the latter can
ba used as a lower limit in design.
(4) a more compact heat exchanger can be designed 1.7
using a large number of short tubes rather than a few long
tubes.
 - - 4.N.WarINIOrai.,....

1
mllotatall mk
I { II it
_ -
1 II SIMPHI
0111. M MEME III

i
1 ".

noun
Fill
...
EMMIIIII
pmfelfill - griureari
.

_ aubrainum i__
Am
ME_ FillEMMME _

- . 11-i- muipsi nil IIIIIIIIT

u:
- 1 - hi 11-
11 1 - -' 1 1
a ill MPH
punurailungmn
I + -

1 . - 11 1 Iiiiiiiiiiiiilliiiimiiiim
Illikk, 10ii- t'll' II .. : .espr

:i-

-4
.

ma

,
:1!

0.9116. -"IR
nal 1M. i lA1111..11211111
A um . ...
Prirer
111:1Lun ISI
.. ME M

pes powtiPh . a ...........2111111

...tan.....,Fm .1. .
I :UM rilffni. LlillIblitimm iligu
imm. ___ Irafffe EMM

mum Frupsurplapirankrappurarpoit
raumgrimpungiLkdkimparmil
mai
I 71tiir- euppripippmmon
rim impurp.... IIIIIIIIIIVE I

- N.L0111. il 119111. , i 111 1 11 E111

WeENDIX 7.1 Limiting Nusselt numbers, Nu o) , for fully developed

laminar flow in ducts of various non-circular cross
sections.
MEMMEMEMEME M MU= ME ME ME

11111111111111111511111111111151111111111111111111 allIM
m 1111:21,111111:119 grappplin9111.1.11111111.
11111111
li 1j1 Ai
I' '11.11 MM b
n &p% 1.4111.1151114111
ILulpillipliummii.
a M IIMMEHIPITIrjurriellillnellorl
BM 2520 milm =mom mmui mosi
qpro pronaimpipmehrip.
m

WM
ilag%
pill
EMI 11111 ..
MIME
MEEMEMEmonmEME
1111111111/1MISIN
ME MIME ME
qtrellelERIBLIIPPIIIIIMI
114111114111upilddrinlippliliblid1111
ME MOM NEM UM ME

.11112rellig 110111 OM OM nn Whia...44111MMOWIMMIMMIMMEMOMI

MUM
marrir-F.211ENIIIIM1111111111 MEI EMEM Om
lam
1 1111.19:11M0111,11.9p111111.19111
MEE MILIMM.2..EUEUEMALEMEM
mm limm

Weis -Endt111.1 1
mommummommilmmommum ilmnam, mi. Ilm lim mimmomm
--,-- m. , l'411111,111Filjarglinplirieprir 0111E ME 111.
OM MI. mmillrnms
PAMME Of
MEMMUUMMEIMMEMEEMEWBUISM MMINAMMAMMOMMOOMMINO m
limaammoMME moths
amnia' m
upwasswing 41.110.1row.0 OM IrallIMMM 41.1111111
1'111 111. 111111111
- IllifiiIIIIIigilltki . . III'
1 11 NM MIMME ME =EOM
ME MEM
__
M
1_4_ 4
ME M
_
 MIN
som s
.
,- Mgr fill ' 1111111111111
-m 11111 :ism Illiiiiimillill
_._4. fn
IIIIIIIIIIIIIIIIIIIIII:11
m Crirragir - I-
1 '111.1
_
I:11.11111i
mmillmml_
m.n. mum
ma
II
IIIIIII
"8 I
. imiillz 11110:41.1.-li
IIIIIII I 41 girl'
1111114E M -4111111MONIFIN
I
IIMM.01111_1
t i 5lila
II_
KIP
Rigirill
MR

-
s
II
111111Ww=11
to
l
nisi
lin JILI
ow-. ,E"'"'
MUM
Il
1-111110iiiiilliiilliliii
11.0_MOMOMMOOM 3 11 OMMIMMENM
so 3 Ming m mommommgmmmilessammimmesaamOdsolommom.
4 / __
IIal .-'--T--
a-
m
U..
.1.1 Mr'
gm ----
"
I mommommossmomomm om
Filklmimmommommosolmollos
a.m..
. mill .....m.. mom so
os mem mum
ORMIIIMM
polunidinutmarasi
IMMEMEmmi
MENEENEN --lailimENE RNA IlinirrniliniarVilan
ERNE IIINNIIINMENI II NUNN NOM gm mom

O
SMMERNON N
tommoommog
moossommos.
=gam
NENUTENR
mo mom. momoggsmgg mg m. . oggloommommgsgsmom
mmommums..ms.. smm. mos omm...mommommEmm
mamommogsms
mosgsmom m mso
msmosomgmgmmommoggs rmimmosogsmommommgmmimmmgmoms
so..mogomm...
.m
isimomomgrarmgm mo
Immo. lgomommog
m m mmm.mgim
magmogommom
limmENWI
mAEMENNEms
ARNI m
I1N MORMON
MEM
g .mmommgmommos. NOMMEN
11111 mgg_gommomplggmmommommomm
ImmEMMINImmomEMENNERNEENME

on so iris, ollilir_NITHINNUIFTSTIL
MANNER
mama
MONRRIIITI IME1111111
mg'
ril
m mommo - IMMOMuzNAMMIlm
IIII.. mom am . mom . m. giIL j ounr AL imikimg mmgm ommomommom
MOMMUMMEMO WM O Imml
awsmENNEm NENE ENNENMEENEME
ENE NE ER NE NON ENNNENNE ANNA MOE MEN EN
IIIIIIIIIIINION NE

'Variations of predicted Nusselt numbers , with Graetz number Nu

for circular ducts and parallel plates; .-------- fully develope(
velocity profile, ----- simultaneously developing profiles.

, ru
m mop Isom Ilmml
m um glom momg.ss goomm gm mom mo mo rums
mmo oil 0 mill imm
smommg gm
am moms ..mmogommim.msmsommomm
1
1111 911 1111119111
mm.110 m mi....Immli m.s.
111111111191111IIIIII
1
msmgggs mom mommomposomogg.m.m
11 IIIIIIIIIIIIIIIIIII1
mmmommommommo.somommospo.wommomm
I
4 mmilimissimmosamr
ersigmegintompommogp,
mum ma m.m.mommlammanu mm mm.lasuis-vsomme....
Immo+ m. eariggsommig
WHIM 11111111111111mOmmoslislimmilimr.41
mulu Illo IIIIIIIIIII
mmomel:111111
illommmommg
Illegriliipilas o plomallIliiii
m...
im
I NM
111:11
in 3
i IIIIMMAIlligairmerillImIlliallImAIIIIII9111011
PuMil.p.e.1Z mom momommosongOomm..mommgomm.mgg
Mr'mm m 11 mom..1 mmommomo
"""111-c"imarviainualumumranno
.....ralmom
mommommgmgr oimmgmlimm g
ggssomm. m ammo
mmomommm
modUMONNEO NE NIMA:1 11A INNEN
MINagail
M atr
cles
..mmommommsommomommommommommoms
ME
m oommomm
MOMMINMNIIONENIREMEN
mom mm
o
NIMENNRI NONN2 ME U
NIMANNIMERINNENEM
NUMI
ENONNR
ONNMEMMEORmE
RENE miNENNIN
EMER MNN
ENENNErn =Mamma NENNMEMOMMINEMOOMME ONOMENNEMEENMOMME U=
MEMIANN
MONNE
mom
Eti
IIII1NN
mo
PrallIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
Alm
mommillussiimmoil
...gm momm me mo !mom g mmgom IIIIIIIIII
mom
g WINN ERNMONR OINM
OMM NENE
NUMMI IMINEM AI MIBMIL MENNE MENIANNIMINEMNIENMOO NENONNOMAM IINE NINE
NENNONORENNEMINNEIONENEEENNE=NINENE MENNIORMIMENNENERMENEMOREENMEENRME
ERE ENTENIMMONE MR NM MONRMIIRRIMIERREMOINEMUMEMR
ENDMIN MUM NENNOM mum mu NNENINNEON INIONNMENE
NEMERI M
MOM
MEM mum
11111
ENE IIIERNENNNIMEN OM NNONINROM ORMENNORN
mmENEMEMEME MENEMONmEmmENEEMENNMEN
MEN ONENENN , '1410 i-06 I II, . III LTI :WI
. II. _mu ..
, AS
^mMERMOIMMOMMIWP"""--"--
, --,--
 118

7.3 Predicted Nusselt numbers for fully developed velocity

profiles with constant wall temperature.

Peripheral Nusselt numbers, Nu
P

Aspect Ratio
Gz
. 1.0 2.0 . 3.0 . 4.0 5.0 . 6.0

0 2.65 3.39 3.96 4.51 4.92 5.22

10 2.,E-6 3.43 4.02 4.53 4.94 5.24
20 3.08 3.54 4.17 4.65 5.04 5.34
30 3.24 3.70 4.29 4.76 5.31 5.41
40 3.43 3.85 4.42 4.87 5.22 5,43
60 3.78 4.16 4.67 5.08 5.40 5.64
80 4.10 4.46 4.94 5.32 5.62 5.86
100 4.35 4.72 5.17 5.55 5.83 6.07
120 4.62 4.93 5.42 5.77 6.06 6.27
140 4.85 5.15 5.62 5.98 6.26 6.47
160 5.03 5.34 5.80 6.18 6.45 6.66
180 5.24 5.54 5.99 6.37 6,63 6.86
200 5.41 5.72 6.18 6.57 6.80 7.02
 119

Lean Nusselt numbers, Num

Aspect Ratio
Gz
. 1.0 2.0 . 3.0 4.0 5.0 . 6.0

0 2.65 3.39 3.96 4.51 4.92 5.22

10 3.50 3.95 4.54 54)0 5.36 5.6E
20 4.03 4.46 5.00 5.44 5.77 6.04
30 4.47 4.86 5.39 5.81 6.13 6.37
40 4.85 5.24 5.74 6.16 6.45 6.70
60 5.50 5.85 6.35 6.73 7.03 7.26
80 6.03 6.37 6.89 7.24 7.53 7.77
100 6.46 6.84 7.33 7.71 7.99 8.17
120 6.86 7.24 7.74 8.13 8.39 8.63
140 7.22 7.62 8.11 8.50 8.77 9.00
160 7.56 7.97 8.45 8.86 9.14 9:35
180 7.87 8.29 8.77 9.17 9.46 9.67
200 8.15 8.58 9.07 9.47 9.79 10.01
 120

7.4 Predicted Nusselt numbers for fully developed velocity

profiles with constant heat Input per unit length.

Periperal Nusselt numbers, Nup.

Aspect Ratio
Gz
. 1.0 2.0 3.0 4.0

0 3.60 4.11 4.77 5.35

10 3.71 4.22 4.85 5.45
20 3.91 4.38 5.00 5.62
30 4.18 4.61 5.17 5.77
40 4.45 4.84 5.39 5.87
60 4.91 5.28 5.82 6.26
80 5.33 5.70 6.21 6.63
100 5.69 6.05 6.57 7.00
120 6.02 6.37 6.92 7.32
140 6.32 6.68 7.22 7.63
160 6.60 6.96 7.50 7.92
180 6.86 7.23 7.76 8.18
200 7.10 7.46 8.02 8.44
 121

Mean Nusselt numbers, Num

Aspect Ratio.
Gz
1.0 2.0 . 3.0 . 4.0

0 3.60 4.11 4.77 5.35

10 4.48 4.94 5.45 6.03
20 5.19 5.60 6.06 6.57
30 5.76 6.16 6.60 7.07
40 6.24 6.64 7.09 7.51
60 7.02 7.45 7.85 8.25
80 7.66 8.10 8.48 8.87
100 8.22 8.66 9.02 9.39
120 8.69 9.13 9.52 9.83
140 9.09 9.57 9.93 10.24
160 9.50 9.96 10.31 10.61
180 9.85 10.31 10.67 10.92
200 10.18 10.64 10.97 11.23
 122

7.5 Experimental results for heat transfer with fully

developed velocity profile in a rectangular duct of
aspect ratio of 2.0 with constant heat input per unit
length of the duct.

Re Gz t wo . t o . t tw tb . u1

deg.F, deg.F. deg.F. deg.F. deg.F.

735 94 81.0 71.0 97.0 110.5 77.7 8.1

59 121.0 81.5 7.0
30 135.5 91.6 6.6

813 104 81.0 70.0 96.0 111.0 76.7 8.4

65 121.6 80.5 7.4
33 134.0 90.6 7.0
897 115 80.0 70.0 94.0 109.8 76.3 8.9
72 119.4 79.7 7.9
37 131.7 89.1 7.4
980 125 80.5 69.8 93,5 110.0 76.7 9.3
78.5 120.0 79.3 8.2
4n 131.5 88.3 7.8

1060 136 81.6 70.5 94.0 110.5 76.5 9.6

85 121.5 80.1 8.5
43 132.5 89.2 8.1
 123

Re Gz two to . tt. . tw tb Nui

deg.F. deg.F. deg.F. deg.F. deg.F.

1140 146 78.5 68.5 90.5 109.0 73.7 10.1

91 119.0 77.0 8.8
46.5 130.0 85.2 8.4

1220 156 82.5 70.7 93.5 111.5 76.7 10.5

97.5 122.5 80.0 9.3
50 133.4 83.8 8.8

1300 166 77.5 67.5 87.7 107.1 72.5 10.8

104 117.0 75.6 9.5
53 126.5 83.5 9."

1390 177 84.0 71.6 94.0 114.5 76.6 11.2

111 125.5 80.0 9.8
56.5 137.2 89.3 9.3

1470 188 84.0 71.6 94.0 115.3 77.6 11.6

117 126.0 80.6 10.1:
60 137.5 89.5 9.8
1560 199 83.5 71.0 93.0 114.0 76.6 12.1
125 125.0 80.0 10.3
63.5 136.5 88.5 10.2

1650 211 83.5 '70.5 93.0 115.0 76.2 12.8

132 125.5 79.5 11.5
67 137.5 88.5 10.8

1720 220 83.7 70.5 93.5 113.5 76.5 13.7

138 124.0 80.0 12.3
70.5 138.0 89.0 11.3
 124

7. 6 Experimental data for fully developed velocity profile

in a .Nactangular duct of aspect ratio of 2.0 with
constant wall temperature.

Z/dh . Re Gz t o . tf tw ...ui
deg.F. deg.F. deg.F.
18.02 655 26 70.0 107.5 143.3 4.5
735 29 70.5 107.5 145.5 4.7
820 32,5 69.8 106.0 145.8 4.9
900 36 69.0 104,5 145.8 5.1
980 39 69.0 103.5 148.0 5.25
1060 42 69.0 102.5 149.2 5.4
1140 45 68.0 101.5 150.5 5.55
1220 48 69.0 100.5 149.8 5.7
1300 51.5 69.5 100.0 150.8 5.9
1390 55 69.0 100.0 151.3 6.2
1470 58 69.0 99.5 151.0 6.5
1555 62 69.5 100.0 148.0 6.8

9.76 655 48 70.5 102.5 155.0 5.3

735 54 70.5 100.1 151.5 5.6
820 59 71.0 99.1 152.5 5.9
900 65 71.0 98.5 153.0 6.3
980 71 71.0 97.8 153.0 6.17
1060 77 72.0 101.0 160.0 7.2
 125

zidh Re Gz to tw Nu1
tr
deg.F. deg.F. dega.

9.76 1140 82..5 72.0 101.0 160.0 7.7

88 71.0 99.0 159.0 8.1

94 70.0 98.5 159.0 8.6

100 69.5 97.5 158.5 9.0

1560 106 69.0 97.5 158.5 9.7

5.63 655 821. 66.0 93.5 150.5 7.9

66.0 93.0 150.7 8.5
95
105 66.5 92.8 150.5 9.3

900 116 67.0 93.0 151.5 10.1

128 66.5 92.5 151.5 10.8

1070 137 66.0 91.0 151.5 11,4

. 14-8 66.0 91.0 151.0 12.0

159 66.0 90.5 153.0 12.5

1320 169 65.5 89.5 153.0 13.0

 126

7.7 COMPUTER PROGRAM

JOB
LSM44EC1, WIBULSWAS RUN 40 APRIL 66
COMPUTING 20000 INSTRUCTIONS
OUTPUT
0 LINE PRINTER 500 LINES
STORE 32 BLOCKS
COMPILER EXCEL?

TITLE
SIMULTANEOUSLY DEVELOPING FLOW IN RECTANGUIAR DUCT
TITLE
CONSTANT HEAT INPUT ASPECT RATIO = 2.0

ROUTINE 1

BI=UCF'/AI+D'A152/AI+AI >>the Raider -Stokes equation in

RETURN >>finite difference form
CLOSE R

ROUTINE 2

*>method to compute velocities

C'=-20A152+8.4135+8A151+4A134 >>K W(x.yor) at the central axis

D'=B152 -A152 >> W at the central axis

J=18(17) 120
K=J+14
I=J(1)X
C= -20AI+O(I -1)+4A(I+1)+4A(I -17)+A(I -16)+A(I -18) >> K W(x.y.E)
C=4A(I+17)+A(I+16)+A(I+18)+C -CI
JUMPDOWN(R1)
REPEAT
REPEAT

,)velocities at central lines

1=33(17)135
C= -20AI+4A(I -17)+0(1+17)+8MT -1)+2A(I -18)+2A(I+16)-C2
JUMPDOWN (R1)
REPEAT

1=137(1)151
C=-20AI+4A(I-1)+4A(I+1)+8A(I-17)+2A(I-16)+2A(I-18)-C'
JUMPDOWN (R1)
 127

REPEAT
>>to evaluate bulk velocity
V=0
J=18(34) 120 >>apply the Simpson's rule in two dimensions
E=J+14
I=J(2)K
W=16BI+4B(I-1)+4B(I+1)+48(I-17)+4B(I+17)+W
W=13(I-16)+B(I-16)+B(I+16)+B(I+16)+W
REPEAT
REPEAT
W=W/1152 >> bulk velocity
RETURN
CLOSE R

ROUTINE 3
DI=F t C/BI+CI >> Fourier-Poisson energy equation in
RETURN >> finite difference form
CLOSE R

ROUTINE 4
NEWLINE 3
CAPTION
STEP N GZ BULK TEMP, WALL TEMP, NUP NUM
NEWLINE 2
RETURN
CLOSE R

ROUTINE 5
to evaluate bulk temperature, V
V=0
J = 18(34)120 >>the Simpson's rule in two dimensions
K=J+14
I=J(2)K
V=16BIDI+4B(I-1)D(1-1)+4B(I+1)D(I+1)+4B(I-17)D(I-17)+48(I+17)D(I+17)+V
V=13( 1-16) D( I-16)+B( I-16) D( 1 .46) +B ( I+16)D( I+16)+B( I+18) D( I+18)+V
REPEAT
REPEAT

V=V/1152 predicted bulk temperature

RETURN
CIDSER
 128

ROUTINE 6
>>printout instructions

NEWLINE 3
CAPTION
DIMENSIONLESS VELOCITIES
NEWLIN3 2
J=0(1)16
K=J+136
MEWLINE
I=J(17)K
PRINT(AI)0,2 >> velocity distribution
REPEAT
REPEAT

NEWLINE
CAPTION
DIMENSIONLESS TEMPERATURES
NEWLINE 2
J=0(1)16
K=J+136
=MINE
I=J(17)K
PRINT(CI)0,2 >> temperature distribution
REPEAT
REPEAT

JUMPDOWN(R4), NP

RETURN
CLOSER

CHAPTER 0
Arb300
B'300
0300
D>100

READ(E) >> 3dh/h

READ(F) (wbdhdh)/(Q4z)
READ(P) >> number of steps required
READ(A) accuracy
READ(u) >> Prandtl number

E=3/E >> e b/dh

Et =9DIVIDE(Y EE)
 129

F' =6p
10 =1/10 " (01CAlz)/(6e e wbdhdh)
>>wall initial temperatures and velocities
1=0(1)16
Ai=0
BI=0
CI=E/4
REPEAT

1=0(17)136
Al=0
BI=0
CI=E/4
REPEAT

initial velocities and temperatures

J=18(17) 137
K=J+15
I=J(1)K
AI=1
CI=0
REPEAT
REPEAT
R=1
H=4/E Nu

JUMPDOWN(R4)
0152=1
M=0
N=1

>>to compute velocities

1)JUMPDOWN(R2)
2)M=M+1
C=1-W >>error in bulk velocity

D=l+A
JUMP 3, W>D

D.pmoD(c)
JUMP 4, A>D
JUMP 3, ip25
8152=8152+1.00= correction of the central axis velocity
JUMPDOWN(R2) >> recompute velocity distribution

JUMP 2
 130

3)J=18(17)137
K=J+15
1=J(1)K
BI=B !+C
REPEAT
REPEAT

M=0

to compute temperatures

4)N=N+1
J=18(17)120
K=J+14
I=J(1)K
C=-20CI+4C(I-1)+4C(I+1)+4C(I-17)+4C(I+17) >> K t(x,y,z)
C=C(1-18)+C(I-16)+C(I+16)+C(I+18)+C
JUMPDOWN(R3)
REPEAT
REPEAT

temperarures at central line

1=33(17)135
C=-20CI+4C(I-17)+4C(I+17)+8C(I-1)+2C(I-18)+2C(I+16)
JUMPDOWN(R3)
REPEAT

1=137(1)151
C= -20CI+4C(I -1)+4C(I+1)+8C(I -17)+2C(I -16)+2C(I -18)
JUMPDO7N(R3)
REPEAT

C=-20C152+8C135+8C151+4C134
D152=11"C/B152+C152

10 =NEE/F >> actual bulk temperature, equation(12)

A/=0,001Aft

5)JUMPDOWN(B5)

c=40 v
Eftwors(c)
JUMP 6, P>1)

3=18(17)137
K=J+15
1=J( 1)K
DI=1.001CBI+DI
 131

REPEAT
REPEAT

JUMP 5
>> computation of wall temperarure

6)C=0
1=18(17)120
C=DI+C
REPEAT

1=18(1)32
C=DI+C
REPEAT

c=0.9)33+0.9)137+c
C=C/23 i3 h/n
D=0
1=18(1)32
D=BIDI-BICI+D
REPEAT

1=35(17)120
D=BIDI-BICI+D
REMAT

D.0.9333D33-0,5B33c33+0.513137D137-0.93137c13744) Li
rw h
r=u2/8 b

C= -D/22+C
C=E/4+C >> wall temperature, equation(13'

1=0(1)16
DI=C
REPEAT

1=17(17)136
DI=C
REPEAT

>>Nusselt numbers, X,Y

X=C-Y,
X=0,25/X >> local NUp

H=H+X
Y=H/N mean NUin
 13 2

>>Graetz number
G=E' /N

output

JUMP8,N=P
JUMP71p200>11
JUMP10,R#30
JUMP9
7)JUMP8,100>11
JUMP10,R#10
JUMP9
8)JUMP9,6>N
JUMP10,R#5
9)NEWLINE
PRINT(N)0,2
PRINT(G)0,3
PRINT(V)0,3
PRINT(C)0,3
PRINT(X)013
PRINT(Y)0,3

R=0
10)R=R+1

mincmN(R6), N=100
JWAPDOWN(R6), N=P

JUMP11, N=P

1=0(17)136 >> transfer predicted velocities and temperatures

K=I+16 >> to prepare for computations at the next step.,
J=I(1)K
AJ=BJ
CJ=DJ
REPEAT
REPEAT

JUMP4, B152>105
JUMP 1

11)END
CLOSE

64 20 995 0.0005 0.72



133

7:8 Heat Transfer 'or Simultaneously Developing Velocity and

Temperature Profiles in Rectangular Ducts.

7.8.1 Predicted 2ean NUsselt numbers for constant

wall temperature .and Prandtl number of 0.72.

Aspect Ratio
Gz
1.0 2.0 3.0 4.0 6.0

3.75 4.20
10 4.67 5.11 5.72
5 20 4.39
.56 4.79 5.17 6.13
30 4.88 5.23 5.60 5.93 6.47
40 5.27
6.27 5.61 5.96 6.78
50 5.63 5.95 6.28 6.61 7.07
60 5.95 6427 6460 6.90 7.35
80 6.57 6.88 7.17 7.47 7.90
100 7.10 7.42 7.70 7.98 8.38
120 7.61 7.91 8.18 8.48 8.85
140 8.06 8.37 8.66 8.93 9.28
160 8.50 8.80 9.10 .9.36 9.72
180 8.91 9.20 9.50 9.77 10.12
200 9.30 9.60 9.91 10.18 10.51
220 9.70 10.00 10.30 10.58 10.90
 134
7.8.2 Predicted Nusselt numbers for the thermal condition or
constant heat input per unit length of the duct and
Prandtl number of 0.72.

Peripheral Nusselt number, Nup.

Aspect Ratio
Gz
1,0 2.0 3.0 4.0

10 4.18 4.60 5.18 5.66

20 4.66 5.01 5.50 5.92
30 5.07 5.40 5.82 6.17
40 5.47 5.75 6.13 6.43
50 5.83 6.09 6.44 6.70
60 6.14 6.42 6.74 7.00
80 6.80 7.02 7.32 7.55
100 7.38 7.59 7.86 8.08
120 7.90 8.11 8.37 8.58
140 8.38 8.61 8.84 9.05
160 8.84 9.05 9,38 9.59
- 180 9.28 9.47 9.70 9.87
200 9.69 9.88 10.06 10.24
 13 5

Predicted mean Nusselt number 1 Num.

Aspect Ratio
Gz
1.0 2.0 3.0 4.0

5 4.60 5.00 5.57 6.06

10 5.43 5.77 6.27 6.65
20 6,60 6.94 7.31 7.58
30 7.52 7,83 8.13 8.37
40 8.25 8.54 8.85 9.07
50 8.90 9.17 9.48 9.70
60 9.49 9.77 10.07 10.32
80 10.53 10.83 ' 11.13 11.35
100 11.43 11.70 12.0 12.23
120 12.19 12,48 12.78 13.03
140 12.87 13.15 13.47 13.73
160 13.50 13.79 14.10 - 14.48
180 14.05 14.35 14.70 14.95
200 14.55 14.88 15.21 15.49
220 lt.03 15.36 15,83 16.02
 136

7 .8.3 Effect of Prandtl numbers on predicted mean Nusselt

numbers, Num , for simultaneously developing profiles
in a rectangular duct of aspect ratio of 2.0 with
constant heat input per unit length of the duct.

Prandt1 Number, Pr.

Gz
co 10 0.72 0.1 0

25 5.60 6.15 6.94 7090 8.65

40 6.64 7.50 8.54 9.75 10.40
60 7,45 8.40 9.77 11.10 11.65
80 8,10 9.20 10,83 12.15 12.65
100 8.66 9.90 11.70 13.05 13.50
140 9.57 11.05 13015 14.50 14.95
180 10.31 11.95 14.35 15.65 16.15
220 10.95 12,75 15.35 16.70 17.20
260 11.50 13,45 16.25 17.60 18.10
300 12.00 14.05 17,00 18.30 18.90
350 12.55 14.75 17.75 19.10 19.80
400 13.00 15.40 18.50 19.90 20.65
 13 7

7.8.4 Experimental data for simultaneously developing velocity

and temperature profiles in a rectangular duet of aspect
ratio of 2.0 with donstant heat input per unit length of
the duct.

Pa Gz t w0 to tf tw tb. Nu 1
1
deg.F. deg.F. dega.
755 94 76.5 70.3 100.3 110.6 78.2 11.2
59 122.0 82.6 9.8
30 136.6 94.5 9.2

820 104 77.2 71.0 97.8 108.7 711.1 11.6

65 118.8 80 10.4
33 132.8 92 \ 9.5
900 115 72.7 65.7 94.6 107.8 73.0
72 119.5 77. 5
37 1335
.. 88 8
980 125 73.3 66.3 94.0 106.5 73.3 13.1
78.5 117.7 77.6 11.4
40 132.0 88.3 10.7
1060 136 74.5 67.5 95.0 108.6 74.5 13.7
85 120.1 78.7 11.9
43 135.1 89.4 11.0
1140 146 77.5 71.4 95.6 109.0 77.8 14.4
91 120.0 81.3 12.5
46.5 134.5 90.6 11.3
 13 8

Re w,o too tf tw tb Nu
deg.F. deg.F. deg.F. deg.F. deg.F.

1220 156 77.2 71.0 94.5 108.5 77.1 14.9

97.5 119.5 80.5 12.8
50 134.4 89.9 11.7
1300 166 77.5 70.6 94.9 109.0 77.2 15.5
104 119.8 80.5 13.5
53 134.0 90.0 12.3
1390 178 77.5 70.7 94.0 107.5 76.7 16.1
111 118.5 80.2 14.1
56.5 132.5 89.3 12.6
1470 188 76.8 70,2 92.7 108.0 76.0 16.8
117 118.6 79.3 14.4
60 133.0 88.2 13.0
1560 199 76.3 69.5 92.2 107.5 75.3 17.4
125 118.6 78.8 15.0
132.5 87.8 13.7
1650 211 76.1 68.0 91.8 108.0 74.2 17.9
132 118,6 77.8 15.5
132.5 87.1 14.2
 139

.7.9 Predicted Nusselt numbers for right-angled isosceles

triangular ducts with constant wall temperature.
FzieSI
+At} cLmdLroulk Nagreiks
lr E.

Peripheral Nusselt Number Mean Nusselt number

Fr co 0.72 0 co 0.72 0
Gz

10 2.40 2.52 3.75 2.87 3.12 4.81

20 2.53 2.76 4.41 3.33 3.73 5.85
30 2.70 2.98 4.82 3.70 4.20 6.48
40 2.90 3.18 5.17 4.01 4,58 6.97
50 3.05 3.37 5.48 4.28 4.90 7.38
60 3.20 3.54 5.77 4.52 5.17 7.73
80 3.50 3.85 6.30 4.91 5.69 8.31
100 3.77 4.15 6.75 5.23 6.10 8.80
120 4.01 4.43 7.13 5.52 6.50 9.18
140 4.21 4.70 7.51 5.78 6.82 9.47
160 4.40 4.96 7.84 6.00 7.10 9.70
180 4.57 5.22 8.10 6.17 7.33 9.94
200 4.74 5.49 8.38 6.33 7.57 10.13
 140

7.10 Predicted Nusselt numbers for right-angled isosceles

triangular ducts with constant heat input per unit
length of the duct.

4
14. 0Ltt JJ2.0.12)wort.NA Arao-cAl rka ftLe. Firt S

Peripheral Nusselt Number Mean Nusselt Number

Pr co 0.72 0 co 0.72 -0

Gz
10 3.29 4.00 501 4.22 5.36 6.86
20 3.58 4.73 6.27 3.98 6.51 7.97
30 3.84 5.23 6.85 .5.50 7.32 8.68
40 4.07 5.63 7.23 5.91 7.95 9.20
50 - 4.28 5.97 7.55 6.25 8.50 9.67
60 4.47 6.30 7.85 6.57 8.99 10.07
80 4.84 6.92 8.37 7.14 9.80 10.75
100 5.17 7.45 8.85 7.60 10.42 11.32
120 5.46 7.95 9.22 8.03 10.90 11.77
140 5.71 8.39 9.58 8.40 11.31 12.14
160 5.95 8.80 9.90 8.73 11.67 12.47
180 6.16 9.14 10.17 9.04 12.00 12.75
200 6.36 9.50 10.43 9.33 12.29 13.04
 141

7.11 Predicted Nusselt numbers for equilateral triangular

ducts with constant wall temperature.
et41T-LA 4)-*-)41reA
Peripheral Nusselt Number Mean Nusselt Number

0. Pr co 0.72 0 co 0.72
Gz
10 2.57 2.80 3.27 3.10 3.52 4.65
20 2.73 3.11 3.93 3.66 4.27 5.79
30 2.90 3.40 4.46 4.07 4.88 6.64
40 3.08 3.67 4.89 4.43 5.35 7.32
50 3.26 3.93 5.25 4.75 5.73 7.89
60 344 4.15 5.56 5.02 6.08 8.36
80 3.73 4.50 6.10 5.49 6.68 9.23
100 4.00 4.76 6.60 5.93 7.21 9.98
120 4.24 4.98 7.03 6.29 7.68 10.59
140 4.47 5.20 7.47 6.61 8.09 11.14
160 4.67 5.40 7.88 6.92 8.50 11.66
180 4.85 5.60 8.20 7.18 8.88 12.10
200 5.03 5.80 8.54 7.42 9.21 12.50
 142

7.12 Predicted Nusselt numbers for equilateral triangular

ducts with constant heat input per unit length of
the duct.
4,911 A4whainaratsk ,wagre.JCI -prk-fAA Rts S a
Peripheral Nusselt Number Mean Nusselt Number

Pr co 0.72 0 CD 0.72 0
Gz
10 3.27 3.58 4.34 4.02 4.76 6.67
20 3.48 4.01 5.35 4.76 5.67 8.04
30 3.74 4.41 6.14 5.32 6.80 9.08
40 4.00 4.80 6.77 5.82 7.57 9.06
50 4.26 5.13 7.27 6.25 8.20 10.65
60 4.49 5.43 7.66 6.63 8.75 11.27
80 4,85 6.03 8.26 7.27 9.73 12.35
100 5.20 6.56 8.81 7,87 10.60 13.15
120 5.50 7.04 9.30 8.38 11.38 13.82
140 5.77 7.50 9.74 8.84 12.05 14.46
160 6.01 7.93 10.17 9.25 12.68 15.02
180 6.22 8.33 10.53 9.63 13.27 15.90
200 6.45 8.71 10.87 10.02 13.80 16.00
 14 3

7.13 Experimental data for heat transfer with fully developed

velocity profile in an equilateral triangular duct with
constant heat input per unit length of the duct. e
ta- c:
r 1 `T`',
Re Gz to tf tw tb Nu 1
t wto
deg,F. deg.F. deg.F. deg.F. deg.F.

650 90 79.6 66.3 106.5 138.0 74.0 5.5

64 149.6 77.3 5.0
45 156.7 82.0 4.8
22 183.4 97.5 4.3

810 112 75.2 64.2 101.5 133.1 70.3 6.9
80 144.3 73.0 6.3
56 151.7 76.7 6.0
28 177.5 89.1 5.3

970 134 74.5 62.5 98.8 129.7 69.4 7.8

96 141.0 72.3 7.3
67 149.5 76.7 6.9
33.5 175.2 90.5 6.1

1135 157 78.0 63.7 101.7 135.5 70.8 8.6

112 147.0 74.0 8,0
79 155.8 78.4 7.6
39 182.7 93.1 6.8

1300 180 80.0 66.5 103.0 141.8 73.6 9.4

128 153.5 76.5 8.7
90 163.2 80.8 8.2
45 190.3 95.0 7.3
 144
Re Gz to
wt to tf tw tb Nu1
deg.F. deg.F. deg.F. deg.F. deg.F.

1460 202 79.0 65.2 101.4 143.0 72.1 10.0

144 155.0 75.0 9.3
101 165.2 79.2 8.8
50.5 192.2 93.2 7.9
1620 224 79.4 64.8 100.2 142.5 71.4 10.9
160 154.0 74.4 10.0
112 165.0 78.5 9.4
56 191.0 92.0 8.5

7.14 Experimental data for heat transfer with simultaneoua_Ly

developing velocity and temperature profiles in an
equilateral triangular duct with constant heat input
per unit length of the duct. AAA_ F*1 4-1

Re Gz to tf tw tb Nu
wto

deg.F. deg.F. deg.F. deg.F. deg.F.

650 90 69.6 63.3 97.0 112.0 69.7 7.7

64 120.5 72.3 7.0
45 127.7 76.3 6.7
22 151.0 89.4 5.9
810 112 69.7 62.8 97.5 115.0 69.5 9.2
80 123.8 72.2 8.4
56 131.9 76.2 7.9
28 155.8 89.6 6.9
 145

Re Gz t,,,0 tf t w t b Nu1
deg.F. deg.?. deg.F. deg.F. deg.F.

970 134 69.8 62.1 97.2 115.2 68.2 10.5

96 123.7 71.0 9.7
67 132.2 75.2 9.0
33.5 156.2 89,1 8.0
1135 157 75.0 68.0 100.9 120.5 74.3 11.9
112 129.0 77.0 10.9
79 137.8 80.8 10.3
39 162.0 93.5 8.8
1300 180 72.7 65.7 97,5 117.0 71.7 13.2
128 125.6 74.3 12.2
90 135.7 78.0 11.2
45 158.7 90.2 9.9
1460 202 73.0 65.7 97.5 118.5 71.7 14.4
144 127.0 74.3 13.4
101 136.7 78.0 12.3
50.5 160.3 90.2 10.7

1620 224 73.0 64.8 96.6 117.4 70.8 15.6

160 125.5 73.5 14.5
112 134.7 77.2 13.4
56 156.7 89.3 11.8
 14-6

REFERENCES

1. Montgomery, S.R. and Weiss, H.K. "Forced convection

heat transfer in ducts of non-circular cross-sections",
N.E.L. Report No.1 0 June 1961.

2. Ulrichson, D.L. and Schmitz, R.A. "Laminar flow heat

transfer in the entrance region of circular ducts",
Int. J. Heat & Mass Transfer, 1965, 8, 253-258.
3. Kays, W.M. "Numerical solutions for laminar flow heat
transfer in circular tubes", Trans. ASME, 1955, 77,
1265-1274.

4. Hudson J.L. and Bankoff, S.C. "Asymtotic solutions for

the unsteady Graetz problem", Int. J. Heat & Mass
Transfer, 1964, 7, 1303-
5. Bradley, D. and Entwistle, A.C. "Developed laminar
flow heat transfer from air for variable physical
properties", Int. J. Heat & Mass Transfer, 1955, 8,
621-638.

6. Koppel, L.B. and Smith, J.M. "Laminar flow heat trans-

fer for variable physical proporties", Trans. ASME,
J. Heat Transfer, 1952, 84, 157-163.
7. Kays, W.M. and Nicoll, W.B. "Laminar flow heat trans-
fer to a gas with large temperature differences",
Trans. ABM, J. Heat Transfer, 1963, 85, 329-338.
 1 47

8. Stefan, K. "Warmeuborgeng und Druckabfall bei nicht

aurgebildeter Laminarstromung in Rohren un in ebnen
Spalten" Chem. lug. Tech., 1959, 31, 773-778.
9. Han, L.S. "International Developments in Heat Transfer",
International Heat Transfer Conference, Colorado, U.S.A.,
1961, 591-597.
10. Langhaar, E.L. "Steady flow in the transition length of
a straight tube", J. Appl. Mech., 1942, 9, A55-A58.
11. Hwang, C.L. and Fan, L.T4 "Finite difference analysis of
forced convection heat transfer in the entrance region of
a flat rectangular duct", Appl. sci. Res., 1964, A, 13,
401-

12. Siegel, R. and Sparrow, E.M. "Simultaneous development

of velocity and temperature distributions in s flat duct
with uniform heating", A.I.Ch.E., 3.5, 1959, 73-75.
13. Hatton, A.P. and Turton 3.8 "Heat transfer in the
thermal entry length with laminar flow between parallel
walls at unequal temperatures", Int. J. Heat & Mass
Transfer, 1962, 5, 673-679.
14. Perlmutter, M. and. Siegel, R. "Two-dimenzional
incompressible laminar duct flow with a step change in

wall temperature", Int. J. Heat &I Mass Tranafer, 1961,34

94-107.
 148

15. Savino, J.M. and Siegel, A. "Laminar forced convection

in rectangular channels with unequal heat addition on
adjacent sides", Int. J. Heat & Mass Transfer, 1964, 7,
733-741.
16. Sparrow, E.M. and Haji-Shiekh, A. "Laminar heat trans-
fer and pressure drop in isosceles triangular, right
triangular and circular sector ducts", Trans. ASMS,
J. Heat Transfer, 1965, 87, 426-428.

, 17. Cobble, M.H. "Nusselt number for flow in a cone",

Trans. ASMW, J. Heat Transfer, 1962, 84, 264-265.

18. Rouse H. "Advanced Mechanics of Fluids", J. Wiley,

New York, 1959.

19. Milne, W.E. "Numerical Solutions of Differential Equa-

tions", J. Wiley, New York, 1953, 131-133.

20. Crandall, S.H. "Engineering Analyses", J. Wiley, Nev

York, 1956, 246.

21. Secomak Hi-Velocity Fan No.350. Service Electric Co.

Ltd., Secomak Works, Honeypot Lane, -Stanmore, Middlesex.

22. Schlichting, H. "Boundary Layer Theory", McGraw-Hill,

New York, 1960, 89-91.

23. Ferry Metal - Nickel-copper alloy, 45% Ni, 54% Cu, and
1% other elements. Henry Wiggins & Co. Ltd., Wiggins
Street, Birmingham 16.
 149

24. Hermetal - Double bond resin. Keniworth Manufacturing

Co. Ltd., West Drayton, Middlesex.

25. Coxon, W.F. "Temperature Measurement and Control",

Heywood & Co. Ltd., London, 1960.

26. McAdams, W.H. "Heat Transmission", McGraw-Hill, New York,

1954, Kogakusha Co. Ltd., Tokyo, 229-233.
27. Norris, R.H. and Streid, D.D. "Laminar flow heat transfer
coefficients for ducts", Trans.ASME, 1940, 36, 525-533.
28. Sellars, JR., Tribus, M., & Klien, J.S. " Heat transfer
to laminar flow in a round tube or flat conduit-the Graetz
problem extended". Trans.ASME, 78, 1956, 441-.
29. Cess, R.D. & Shaffer, E.C. "Heat transfer to laminar flow
between parallel plates with a prescribed wall heat flux",
Appl. sci. Res., Hague, A, 1959, 339-344.
30. Dryden, H.L., Muraghan, F.D .., Bateman, H. "Hydrodynamics",
National Research Council, Washington, D.C., 1932, 197-.
31. Southwell, R.V. "Relaxation Methods in Theoretical
Physics", The Clarendon Press, Oxford, 1946, p.22-23.

32. Scarborough, J.B. "Numerical Mathematical Analysis",

johns Hopkins Press, Bultimore, 1955.
 15 0

33. Clark, S.H. and Kay, W.M. "Laminar-flow forced convec-

tion in rectangular tubes", Trans.ASME, 75, 1953, 859-.