7l)-76

Journal of the Institute of Energ)', June 1992, 65, PP

Buoyancy effects within laminar forcedconvectin flow at the thermal entrance region
of a vertical rectangular channel
R SMY

fH'

and Y K

SALMAN-'

An experimenta, srudy was conducted

,:X,ji""J"Tf"ifffiff"llilTff :: :iil3i,:!i;j[,ij'g

'

f3 ,**\*Tl,;fiff
nulfil*,H**f:+#+"il,,,!**ii,*T#":""j';r::ir;
to 4 , 10'and Reynolds,",:f
Nusselt numbers were
1ou

lisniiiunly oirtort"o m" uro"irv

o'oiiL"ti""0-3h:ll;ii:f"*

aorecrably atlected. especially in th

List of sYmbols

co

D
F,F.
Gz
c

heater plate area, mr

specific heat at constant presure, J
eauivalenr duct diameter, m
dimensionless factors in eqn (4)

kg K

uu.iug" tt.ot-t.i-ntfe.icoeflicient for duct of

spacing between healer Plates' m

heater current, A

thermalconcluctivity'Wm'

Nu.
Pe

Pr
q
q.
q,

lengthL.Wm'K'

PecleiNumber
Prandtl Number_:

= hr.D/t-

cnP/k

heatflux,Wm'

R
Rao
Reo

ha:ed on equivalenl dtrct Lliarn-

eterD=Gr,,,.Pr
nc!n"tOs Nri'iit'.t

eter D
temperature, "C
r,)
dimensionless temPerature = (1'-'l',)/(l'.,,,.,--

axialvelocitY,ms'

z"

dimensionless axial velocitY =

cross-stream co-ordinatc' m
dimcnsionless cross-slrcam dislancc = Y/H
dimensionless axial distanec = L/(D Rc,, Pr)

bulk expansion,

K'

kinematic viscosity, m- s.
dynamic viscosit-"-, kg m ' s

inlet condition
surfacc

maximum value
local axial condition

is

forced-convection heat-transfer coefh2 5 times that for

cient in the assisting situation can be up to
pure forced convectlon'

pow^er

(flat-plate bundle) in a nuclear

-itt".,roting
"ftonnels
situations could be pronc to mlxeo
.ertain
ln
..n.t.tt. *tti"tt

forcetl antl free-convection flow;

-'ina**luf tt"ut-"*changers in chemical and food-processconr"1""tlrr"i, where pioblems involving combined
ve"ction in channels may be encountered;

il:liK;j

:
:f.'J;[il
plate air heaters).

.ncrsv co'ecrors ( nar.

;:lTilJi",

situations
Thc clfect of huoylncl on lhe flo\\s in these

li :lm$:; :

ili#:[:

JJtr i'; ijr;

can

r;[ ;i#"li]!"i:: ::
[
m trow K

JllT:' 3'',1''fi r

hc mi ni m u

ril

or ct""ncccssarv to fultil,'afet1 conditions in the cvcnl

ponent failure.
c"m-

made o[
Extcnsive experimcntat sludics have been
asr\ecl
various
oI
ducls
horizontal
in
ilin.J .on""itiun
work.oi
t"cently almost all exPerimntal
i",lri,'r,t"i
"ttiiLconvectn in vertical ducts had been coni"i"t"-i-"i

Subscripts

i
s
max
z

f.,ttta-tonu"ttion flow direction

tant are:

convcctive heat flux, W m '

racliative heat flur. W m '
'
duct conduction heat loss, W m
resistancc of heater, ohm
nuyf.igft N".U"t based on equivalcnt duot diam-

Q.-u

it'.

is imporExamples of situalions where mi\ed conveclit'n

K'

: Re' Pr

vertical duct is
Mixcd-convection heat-transfer inside a
when lhe dircction ol forced-convection
"irtr"i
uo*urJ. nn61 ;55irled bv the fluid's buoyanc)' or
n* i. "i"lt,i"g

;lUllui-.ll f,:":il',i, *I T'',T:,H|

*:lx*
i".i"".
i""t,rt"

Grashol Numbcr = BgDqi(kv')

Graetz Number = ReD Pr DIL
acceleration of gravity, m s '

lensth of duct, m
uu"?us. Nutr"f t numbcr (based on D)

Introduction

;;ilil";;.;

hL

"ted-convection

'l)r

T.T l'

Sm!lh BS( PhD { tng

\llMecht

Uni'eIsrl\'^ol.Shefficld D(Parl'

::'J i'.' ","'

{ Yi..i;,}:"j Tl.: ff ii'

t Jr.g"

"i-engin""t;ng.

tlnitersil-Y of Basrah- Basrah- lrq

":-Jil;'lli "l;li:

n'

SMYTH: BUOYANCY EFFECTS WITHIN LAMINAR FORCED-CONVECTION

fined to the flowin tubes of circular cross-section. Although
this flow geometry is important, the results of these studies
are not generally applicable to ducts of rectangular crosssection with variations in the bealins of the sidewalls.

Recently a signilicant number of theoretical studies have

been made ofconvection in vertical channels of rectangular
cross-section. The most important of these are now
reviewed, although none of them has compared assisting
and opposing flows.
Early analyses of mixed convection' included an approach
appropriate to thermal boundary conditions for combined
free and forced convection, by using complex variable
theory; these were illustrated for two parallel plates and for
channels of rectangular cross-section. The method used

was

to introduce a complex function

whose real and

imaginaryparts are directly related to the velocity and temperature field respectively. In the case of two parallel
plates, the boundary layer was considered at constant temperature at the same horizontal level, with a linear axial
temoerature distribution at the wall. In the case of the rectanglar duct, the boundary layer was considered at a constant wall temperature at any peripheral section, and at a
linear temperature distribution in the axial direction.
The analysis oi fully developed combined forced and frcc

convection in vertical rectansular channels has been

reported2. The analysis shows ihat the mixed convection
can be substantially simplified in vertical ducts, with little
loss in accuracy, try the application of variational methods.
The non-homogcneous Helmholtz equation obtained' was
replaced by an appropriate variational principle. Simple
polynomials were used to reduce the heavy computational
work required in series solution, without loss of accuracy in
the results. The results of this method for rectangular ducts
have been compared with available exact solutions, and
give good agreement.

circular ducts for a uniform axial heat flux and unitbrm peripheral wall temperature. The geometries treated were
right-angle triangle, isosceles triangle and rhombic ducts.
Finite-difference and variational approximate solutions
were used. The variational method shows good agreement
with the finite-difference predictions of Nusselt number lbr
rhombic ducts with different duct angles, and for Rayleigh
numbers of 0 and 1000.

Combined free- and forcedlaminar convection in noncircular rertical ducts has been investigated:. in rectangular, elliptical and rhombic cross-sections with uniform
peripheral temperature distribution. The governing equations with fully developed flow conditions at the inlet of the

vertical duct have been solved by the finite-difference

method and by the variational metbod. The Nusselt
number for the rectangular duct has been depicted versus
the aspect ratio (width:height) of the duct, and shows that
for the pure forced convection lhe value of Nussell number
for the uniform circumferential heat flux falls slightly as thc
aspect ratio rises.

The Nusselt number for tbe uniform circumferential wall

temperature rises rapidly with aspect ralio, and has a value
of 5.6 at an aspect ratio of 5. Combined free- and forced-

convection results show that for aspect ratios from I to 4,

Rayleigh numbers varying from 0 to 3000 and uniform circumferential heat flux, the Nusselt number rises with lhe
aspect ratio for a slight amount of buoyancy. Under higher
buoyancy rates, this rise becomes appreciable.

linear interpolation polynomials for temperature and

velocity profiles was derived for ducts with arbitrary shape.

The analysis has yielded specific results for flow in square

and

angular ducts. For a square duct the computation of

the Nusselt number by the finite-element method, finitedifference method and the exact solution of Rayleigh

ilil

numbers from 0 to 314 has been presented. The cenlre-line

temperature and velocity profiles given show the effect of
buoyancy forces on the upward forced flow, which create
reversed flow in the centreline at high Re values.

a) Schematic of heated channet.

and'opposing' situations-

Analyses have been madea of fully developed combined

natural- and forced-laminar convection in vertical non-

a vertical duct. A finite-element

solution algorithm with triangular elements and piecewise

ill

Fig.1

Suction and blowing were included in an analysis of combined forced- and natural-conyection flow in horizontal
and vertical channels. and in vertical tubesr. Continuity.
momentum and energy partial differential equations were
convertd lo coupled ordinary differential equations by the
similarity transformtion method. The variations of velocity profile were shown for Rayleigh numbers (Rao) of 500
and 1600, Prandtl number of 1, assisting flow, and various
Reynolds numbers.

oped laminar flow tbrough

ltl

b) : Assisting'

71

Use has been made6 of a llnite-element technique to analyse combined natural and forced convection in fully devcl-

,'t

lr

FLOW

oppo'inc

Fr*

Numerical studies? have been made of heatlransfer by

free, forced and combined convection between parallel
plates in a vertical channel having uniform prescribed temperature. The Navier-Stokcs equations for comprcssible
flow with a discontinuity in thermal inlel profile at the
centre-line of the channel were solved by an explicit finitedifference method. An artificial compressibility term was
introduced to the continuity equation, and a central differencc scheme was used. exceDt for the diffusion term.

72

SMYTH: BUOYANCY EFFECTS WITHIN LAMINAR FORCED.CONVECTION FLOW

b
^j
LI

[\

't

tr'

z,

.E

d-.

I
*z

{
{
I

tr-'

ql

;l!-

tF
q=T-\

F-

(
*

f-

i-*

,-

jt

(t
-i\
P

F--

tf

'.3,

;L

"1

;J
r-.i

^i

ri

i-*

\L

CD

fi

tl

t*
i|

i.

sl

-,

\i

iJ

,ry

-E

's'

\ I
\

\
.9
I

\< (

=\

SMYTH: BUOYANCY EFFECTS wrrHrN LAMINAR FoRcED-coNVECTroN FLow

For natural convection, the relation between local Nusselt
number and axial distance show good agreemcnt with
experimentally available data. For combined free_ and
forc-e-d-convection results, the velocity and temperature
profiles were presented for a Reynolds numbei of 333,
Grashof number of 8 x 10a and for different axial positioni
along the channel. The pure forced-convection results fall
below the analytical solution, but for the mixed convection
the results are above the analytical prediction.

The effects of free convection ancl axial conduction on

forced-convection heat-transfer for fullv develooed flow in
a vertical channel at low Peclet numbe.r pe
1: p",rpr; 1.,ru"
been studied numerically3. The study was carrierj-oui with

73

A 30+rmpere dc supply to the electrical beaters was used,

and the power was determined by measurements of the
current cross the nxed resistance Nickrom foil, and a 60_
ampere ac supply for the guard heaters was controlletl bv

four

resistances

in

parallel. Velocity profiles weri

measured by means of a He-Ne Laser Dooler Anemometer sysrem. operating in the differential mode. Temp<:rature profiles were measured by a thermocouole
traversing between the heated surfaes on a nrecision
mechanism sliding in grooves in rhe perspex sidewall.

the assumption that the channel inlet and exit are insulated.

After being assembed. the apparatus was checked for air

leakage. Calibration lests were per[ormed to determine the
longitudinal and lateral conduition heat losses of the test
section, for both heated plates. The local convective heattransfer coefficients h, were obtained from the convective
heat flux q., which was deduced from an enersy balance
(eqn l) applied to lhe duct tesr secrion:

nodes in the domain.

1'ft

Although the above studies and otherse have contributed

significantly to lhe understanding of mixed convection,
uncertainty still exists over the situation of assisting or
opposing fl ows-particularly for rectangular ducts. Experimentai work is required to determine the effect of buovancy-induced secondary motion caused by one si<.te of the
duct heated or when both parallel sides are heated. This
paper reports on an experimental study of mixed-convec_
tion heat-transfer in a vertical rectangular duct with either

and measurements of the local wall and bulk air-temDerature difference

(equation 2):

to assess the conduction effect, and that lhe free-convection flow either assists or opposes the forced-convection
flow. The governing equalioni were cast in a finire-difference form using a central-difference scheme for the interior

opposing or assisting flows. and with ne side or two Darallel sides of the duct heared.

Experimentalapparatus

The apparatus used in the investigation is shown schemalically in Fig.l. and consisted oi a vertical channel of

rectangular cross-section in the horizontal plane; this cross\ection had an aspect ralio of 5: l. Air inducerl by a small
centrifugal fan entered a settling chambcr at the bottom of
the channel through a short rectangular perspex duct, and
exhausted through a bellmouth to the 0.O4 x 0.2 x I m test
sectron.

The settling chamber, together with the 0.04 m wide sidewalls of the test section, were constructed from perspex
sheet, and the 0.2 m wide sidewalls were constructed from
two identical heater slabs. The slabs consisted mainlv of a
rigid Teflon sheet with a composite main heater plare glued
to the duct inner surface, and the outer surface was covered
with guard heaters. The composite main heater plate was
constructed from three layers glued together with high-

temperature application glue. The three layers were

stainless-steel plate, 'Nomax' electrical insulation and
'Nickrom' (80% Ni, 20% Cr) Ioil heater strip.

a
a

Each heaterplate had 36 thermocouples along the centreof

the stainless-steel plate, eight thermocouplJs laterally 50
mm off-centre al four positions along the stainless-steel
plate, four thermocouples embedded in the main Teflon
plate directly above one side of the lateral thermocouples
In the stainless sleel, lwo thermocouples embedded in the
exil Teflon block for calculations of axial conduction heatross, and one thermo-couple in the side Teflon block
for calculations of lateral conduction heat-loss.

- Q-",r :

(c. + q.) A

(1)

^T,

^r,=(r,)._(rJ,

(2)

Hence

l, = q./AT,

(3)

To deduce the local radiation heat flux q- from the total heat
flux. the upper and lower plales were di;ided into 4q identi-

cal elements, and the temperature and shape factor were

calculated for all these elements; the local radiation heat
flux was then calculated. The overall radiation heatloss
was found to be negligiblc for the both-plates-heated condition, but was as much as 97o for the one-side-heated case
with high heat flux.
Overall uncertaintyr" in the temperatures of the wall and air
was estimated to be 0.2 K, whereas the maximum uncertainty in the mean inlet velocity was estimated to be
0.015 m s-'. The uncrtainty in the dimensionless groups
varied with mean inlet velocity and wall heat flux, and was
5.1-16.3% for Reo, 4.16-8.77u for Nu, and 12.57" for
Grrr.o'

Results

4,1

Velocity and temperature

Four sets of tests wcre carried out-two with assistins flow

and two with opposing flow. For each te\t, resulls were
obtained for the one-side-heated and both-sides-heated
conditions. The measuring position, Grashof number and
Reynolds number were approximately the same for all the
tests.

4.1.1 Assisting flows

'l'he dimensionless velocity and temperature profiles for
one-side-heated only are shown in Fig.2. Velocity profiles
exhibit the expected asymmetry in the profile. Foi lower
Reynolds numbers, raising the wall heat flux enhances the
air velocity near to the heated wall, and retards the flow

'l4SMYTH:BUOYANCYEFFECTSWITHINLAMINARFORCED-CONVECTIONFLOW
near to the other wall. Hence tbere may be a tendency to
ieversed flow near the unheated wall. Temperature proflles
show little variation in the transverse direction'

j2a
26

For vertical ducts, other studiesr'36*rr''' agree that th

velocitv is enhanced near the wall. A retarded flow near the

centre'makes a reversed-flow situation likely at low Reynolds numbers and high Grashof numbers'

4,,

profile near to the heated plates and

o"ut u"to'"iti", on
"ach
from the middle ofthe duct The
ppto*irnurety
"quidistnt
profile aPpears to be produced-as
itipe ot lt. velcity
exoected-bv the summation of a natural-convection profil and a foiced-convection profile. Since the plates wer
no1 tt."t"O aornpt.tely across to the sidewalls. the natural.onuection flow is somewhat higher at the duct centre-line
and is reduced towards the plate edges lt is for this reason
that the 6 cm off-centre profile shows almost a llat prohle
about the middle of the duct.

]**

The both-sides-heated case shown in Fig'3 displays symmetrical profiles about the middle of the duct, with two

,1,

t2

l0

3 1 5 67891

156?a9t

'ro.'

,'

nr,o

3 1.567

'.Jjo'"'

ofCnshol Numb (heatfluxt on Nusselr Number-assisring fows:

iA.a touel one side ol the duct heatel'
F.7 @eloh,l both srd?s ol thc duct heatcd

Effect

R - RI6HI PTITE
T - LEFI PLATI

4.1.2

rv

OpPosing flows

avtRAGt

Oi 80lrl

PLAIES

2A

With one side heated only. the results represented in Fig 4

tor tne
show ouite a different behaviour from that obtarned
in the 2z+
greatly
is
reduced
velocity
The
uriistitis-no* tests.
occurs
vicinitv-of the heated surface, and this reduction
I
.u.n irc relatively high Reynolds number' The duct
centre-line profile appeais to suffer greater reduction as a
result of a irigh naruial convection effect. in comParrson
flow is
*iit tn. e cnioff-centre position; thus most of th
result 3' 16
The
duct
n.i iit" .lO"t tuttler than the centre of the
oiln.r"utine the buoyancy effect by either reducing the
Revnolds number or raising the Grashof number would
.tlniuu v produce a reversd flo* near the heated plate'
esoeciallv t the duct centre, with the main forced flow
exLecteO'to exit ttrrough the duct along the unheated sides'
Ai a resutt of the natural convection effect' temperature
oiot.t ttto* a thick thermal boundary layer at the heated
wall. in comparison with the assisting-flow test shown In

220

I C 562891

a 156769t
'ro

3 1-567

a n',o 'tJio'o'"

Fie.i.
The both-sides-heated test is shown in Fig'5 Here the
velocity profile has been reduced relative to Flg J' especF
;ll;;;t'G heated walls. although the 3 cm and 6 cm off-

centre orofiles show less reduction than that at the centreiin" oi itt. duct: these profiles keep their symmetry-about
the middle of the duct. Increasing the buoyancy ellect. Dy

reducinq the Reynolds number or rarsing the -urasnol

numberiends to 6lock the duct centre position with natural-convection flow. with the forced flow being along the
unheated sides Further increasing of this effect will ulticonsematelv lead to unstedy flow in the whole duct' and

ouenriv c"utO improve

the

heat-transfer process'

i".p"'tutut. ptonles show symmetry abor'rl the centr-line

of the duct. The buoyancy flow created by natural co.nuection sets up a traniue.se temperature variation in the
temperature Profiles.

4.2

Heat-transfer results

Eiehtv-eight test runs were carried out for both assisting

un oppo-sing flows, and for the two plate-h-eating situ'Test
iuns covered Reynolds numbers from 470 to
ations.
isli-n-ctuthor numbers frm 23 x 1 ro2'2 x '101 '

4.2.1 Assisting flows

A total of 48 test runs were carried out, comprising-24 runs
ior euch of ttre two plate-heating situations' For the oneriO"-["ut"O condition, five reprsentative test tuns have
of
;;; ,!i";;; i;p;esentation.' Fig.6 shows theinvariation
heatthe
Nu. with z* and reveals the improvement
or
trarisfer process occurring below a critical value ol z"
approximately 0.0015.
Fig.7 shows three reprsentative test runs for both-sidesit"?i"i. rrt" .right uaiiations between the two heated plates
i" rrto*n, ,ogE n"r with the average result for both Plates'
i-"".""iin the Nusselt nuriber for alt heat fluxes

.to*l ..ttuin axial position' with the

rt'iirt.' for increasing z*' rhis can be
'iJin;
th; instability of natural convection at urDq
attributed to
;;l;;;;;i rot, *itl tr'i no* rnode expected at theduct

;#;;i;;;;
;f#;i

exit to debend on the forced flow. which acts either toderaJrs

;;;i*i;;i. the natural convection effect This effect
hich:l
i"i li ir'i high-heat-flux test run with a slishtlv
signrncaxr
the
Revnolds numblr. This figure shows clearly
nv"'
improvement (up ro 100%) in Nu, with increastng

flui.

SMYTH: BUOYANCY EFFECTS WITHIN LAMINAR FORCED-CONVECTION FLOW

z4
t'

3
r.@ 1600 rsoo 2ooo

l-r"d-'

.-_l-SOVOL

344

..-...-

rs

T
T
l0

(above) one side of the duct heoted.

(below) both sides of the duct heated.

^uhh..

R"o

Infuence of Reynolds Number on Nusselt Numberassisting foh,s:

Fig.8
FiB.g

26

'iudi--

21]o'

2\

.-";.;;;;;;

r.oo 6c0

t----'-------,'--,-

;'r 22,,[oL
] H*,nYt"l'*
R.ynotd.
1063

20

75

3 ( 56739r

3 ! t67A9t
xlO_!

3 1' 567

xlo'z

Z" lLl (D *Reo 'P.)

Effect of Grashol Number (heat fux) on Nusselt N umber-opposing flov/s:

FiB.10 (above) one side of the duct heated.
Fig.l (below) both sides of lhe duct heated.

o'q.

i
L AV

RIGH] PLAIE

LTFI

PLATT

AVERAGI OF BOTH PLAITS

RA1/L iAVL RAVL

Rln No. -, (6v00 50v04 53v03
ir... ll"i (u.;)
0
r37
J49
ll25
lr30
Rsv.otd6 nsmbr li36
2A

Th effect ofthe variation of Reynolds number on the Nusselt number for various Grashof numbers is shown in Fig.8

for one-side-heated, and in Fig.9 for both-sides-heated.

Both diagrams reveal the relatively high Nu, in the low

re

'6

\-*

]I

ll

range of Reynolds numbers (600-900).

t'

t2

4.2.2 Opposing flows

2 3 at 57A9i
,ro
Z'

Forty test runs were carried out, comprising two sets of 20

runs each for the two-plate-heating situations. Fig.10
shows the five reprsentative test runs chosen for the oneside-heated case. The lowest test Reynolds number has
been increased for approximately all opposing flows, to
prevent dealing with possible reversed flow. This reversed
flow, which is created by natural convection, occurs at low
Reynolds numbers (400-600) and moderate-to-high heat
flux. The onset of this effect can be checked by observing
the air temperature in the inlet settling chamber: this tem-

2 \ 1557
,ro:

{L/ lD +ReD .P.) l

.--';;;;a;;

perature rises as the reversed flow enters the chamber.

z* in Fig.10 reveals the

improvement in the heat-transfer process at larger z* as the
heat flux rises relative to the assisting-flow case.
The variation of Nu, with the

loo

For both-sides-heated, three representative tesl runs are

chosen; the results are depicted in Fig.I I , again with rcsults
for each plate and the average for both. The average heattransfer coefficient for both surfaces heated shows that the
sharp reduction in heat-transfer coefficients occurs earlier
as the heat flux increases. The variation of Nu. with z*
reveals the improvement in the heat-transfer process for
high heat flux, but with a considerably reduced effect for
the intermediate and low heat-flux resr run.
The effect of the ReD on Nu, is shown in Fig.12 tor onestde-heated and Fig.l3 for both-sides-heated. The latter
case gives a significant increase in Nu,. as Gr,,., rises as the
result of the mixed convection process-whiih improves
the heat-transfer Drocess on the whole.

1600 rSoo 2ooo

*'o

Infuence of Reynolds Number on Nusselt Number ttpposing fows:

(above) one side of the duct heated.

Fig.l3 (below) both sides olthe duct heated.
Fig.12

6r^_

.-;;;;;

FLOW
SMYTH: BUOYANCY EFFECTS WITHIN LAMINAR FORCED-CONVECTION

16

The lower Reo range with high GrD,q shows alower Nu,, as
the natural convection appears to d<iminate the heat-transf", procerr. This behaviour is completely opposite to that
obtined for assisting flow.
The results ofthis study were compared with the early work
on .i".0 .onu..tion' whi"h gave eqn (4) as valid for both
assisting and oPPosing flow:

Nul

.75

8lr/r
F' [Gz + 0.077 Fr(Gro.u Pr D/L)0

(4)

Although many investigators have sug8esled the use ofthis

rbind ro be varidI the high flux conditions of this work.

:;;;;;;. ;l';;1;;t

Conclusions

I A completely one-dimensional flow with a very.clear

surface is
il.*l* otit'. uitocity profile towards the heated
of symmerry

obtain for assisting flws. with a high degreeIor the both-sides-heated case. Raising the Plate heat nux

revcrsed
and reducing the Reynolds number may lead to a
lhe
lncreasc
factors
two
above
as
the
core,
flow at lhe duct
flow at the heated surfaces. For opposing flows' a comDlelelv contrary behaviour has been revealed l'or low-to'rnoO"iui" tt.ut'nu* the restriction on the forced flow for the
uiJ n"u. ttt" tt"ut"d surface makes th now rate at the duct
sides and corners higher than at the duct centre'

For the same Reynolds number and for low-to-moderfor

It. i.ut nu""t, ttt" hiat-transfer coefficients are higher
the
for
higher
are
no*.; U.t for high heat flux they
"iiirn*
opposirig flows. because of the increase in the natural conManinelli
i|t." Err..r. rl" single equation suggested byconvectlon'
mrxed
opposing
and
assisting
for
& Boeltero
htgh
was not found to be valid for rectangular ducts at a
heat-fl ux boundary condition.

Relerences

Trans
IAO L N. On combined frec and forccd convection in channels
ASME, J H?ttt Ttanler. lgno pp 233-238
i'"enwr-H cl a rariatinal method for combined free and forced

tnt J H?dt and M ats Ttr)ns1er ' 19b2 5 ' pP 434'j44

CARTERLFandGlLl W W As) mptol ic \olutlon tor comornco rree

.""".1." i".tt*""!,

I
ina'rliJ"o*""tion

in vertical an horizontal conduits with uniform

AlChE. teb4.loj ppJ30-339

ice,f V. nC"CeRwAL B D and |OWLER A

incrion anO blor ing

.,l

^
C

a"cts App.sci Rer. lq72 26 pp

193-208

Lamrnar comunder niduct

non-circular
vertical
in
irn"Jil". a"a'f-*o
eat and Mass Tnnsfet,1969' 12, pp^l123-t139
tr""i n"i. lr,.r H"onvection
M, KHATRY A K and AGGARWAL B D on the second
i""ar.ni"i p-ur"t of comhincd free and forced con\ection through

l-

i.t.
;- ic;t
...ti"^i

"on.i*"rrt
e LandCHENG
i--ilvr

P Finilc elcments analvsts oI laminar con/

vection heal transfer in \erlical duct wilh arbitrary closs-secllon 'nI
pp
227-2'16'
lt.
l9'15'
Tronslet,
Mas\
and
Heat
and forccd lmi" irelri p o i"d YTREHUS r' combined frec
channel 2nd Int Conf on Numerical Methods
i"
l"" "".iical
pt"ur".".
"."t:.ti
venice. llaty lg8l T't0 pp aa0- 1002
i" ir,"irn"i
.
s R and cAMPo A Eflecrs o[free con\eclron
c,
HuselN
ri liw

forced ct)nveclion heat transfer inside a vertical

;"i;;;";d;".n
;;;;;i;;i;; P;.i.' "umbcr' rrans ASME r Hcat lrunsf tsu' t0r'

to'Jl'*!*"ottt R c and BoELTER L M K. univ calir (Berkelev)

Publs EnP, 1942,5(2)' PP 23-28
i i<r-ifie sl r",i v.cl-lNrocx F A Describing unccrtdinries in

sample expcriments l Mcch En+.

iinsle
and forced
II - MAR|IER W J ANd MCMILLAN Hl?italt 'r,t;a freelemperalure
wall
w'th
conslant
vcrtlcal
lubc
In
a
laminar convcction
Trans ASME, J Hedr han{et. lel0' pP 559'562
L E rects or heat
iirin-rrir l. RdsEN E M nd KABELR
tubes /nd& En8
verlical
in
numbcrs
Reynolds
r,
i<rw
nJa
nt"
"
i".i.t

ii-

Crrd'n, 1958, 50(5),

PP 815-820

(Paper receivetl Aqust 1991 )