Physica I X , no 8 S e p t e m b e r 1942

T H E DISSIPATION OF HEAT BY F R E E CONVECTION

T H E I N N E R SURFACE OF VERTICAL TUBES
OF D I F F E R E N T SHAPES OF CROSS-SECTION
by W. ELENBAAS
Natuurkundig Laboratorium der N.V. Philips' Gloeilampenfabrieken Eindhoven-Holland

Zusarnmenfassung
Die WArmeabgabe bei ffeier K o n v e k t i o n der i n n e r e n OberflAche yon
v e r t i k a l e n ZyIindern m i t kreisfSrmigem Q u e r s c h n i t t wird b e r e c h n e t i m
Falle die HShe h gross ist im Vergleich z u m Durchmesser. I n Analogie m i t
der frfiher fiir paralelle P l a t t e n abge!eiteten Formel, wird die ftir kleine
Durchmesser geltende Formel so ausgebreitet, dass sie fiir grosse Durch-
messeriibergehtindieFormelvon Schmidt, Pohlhausen und
B e c k m a n fiir die vertikale Platte. Es wird d a n n eine F o r m e l aufge-
stellt fiir Z y l i n d e r von nicht-kreisfSrmigem Querschnitt, wobei als P a r a -
m e t e r die Gr6sse + Re auftritt, welche ftir kreisf6rmigen Q u e r s c h n i t t den
W e r t 16 h a t u n d fiir rechteckigen Q u e r s c h n i t t von B o u s s i n e s q
angegeben wurde. Ffir gleichseitig dreieckigen Q u e r s c h n i t t wird ~bRe be-
r e c h n e t u n d der W e r t 40/3 gefunden. Wird ccr/Xw (r = hydraulischer Ra-
dius) doppelt logarithmisch als F u n k t i o n von (r/h) Grr, w ._Pr aufgetragen
so e n t s t e h t eine Schar von Linien mit + Re als P a r a m e t e r , welche ffir klei-
ne Werte von (r/h) Grr, w . P r eine Neigung 1 h a b e n u n d flit grosse W e r t e
yon (r/h) Grr, tv. P r z u s a m m e n l a u f e n m i t der Neigung 1/4 (Gebiet der freien
v e r t i k a l e n Platte). Die Messungen 'im Zwischengebiet zeigen einen etwas
a n d e r e n Verlauf wie aus den friiheren Messungen der paralellen P l a t t e n
(~ R e = 24) e r w a r t e t wurde.
Fiir jede Q u e r s c h n i t t s f o r m 1/~sst sich der W e r t von (r/h) Grr, w.. P r ange-
ben, bei welcher die je cm 2 Querschnittsfl/~che iibertragene Wi~rme maxi-
mal ist. Vergleicht m a n diese giinstigsten Werte der W i t r m e a b g a b e ftir die
verschiedenen Querschnittsformen, so liegt dieser W e r t hSher je n a c h d e m
Re kleiner ist. (Abgabe je cm 2 proportional (+ Re)-'h). Die Werte yon
( W g r m e a b g a b e / c m2, ° C) sind fiir die gtinstigsten Werte yon r u n t e r sonst
gleichen Umst/~nden u n a b h a n g i g v o n d e r Querschnittsform.

§ 1. C i r c u l a r cross-section. I n a p r e v i o u s p a p e r 1), i n t h e f o l l o w i n g ,
r e f e r r e d to as I we d e s c r i b e d m e a s u r e m e n t s o n t h e h e a t d i s s i p a t i o n

-- 865 --

Physica IX 55
-866 w. ELENBAAS

b y free convection of parallel plates, and derived the formula:

I b
Nub, w -= ~ ~ Grb,w P r E1 - - e--C~hlbC"b,~P')C']c" (1)

with the condition

C2 x Ca = i (2)

Of the two indices of N u and Gr, the first refers to the linear dimen-
sion which is used (b = distance between the piffles) arid the second
to the value of T used in determining the values of the quantities
depending on temperature, •, ), and ~ (the index w indicates t h a t the
wall temperature is used).
We will now calculate the heat dissipation of the inner surface of
tubes, and as a first example we consider a vertical tube with circu-
lar cross-section. In order to derive a formulafor the heat dissipation
of the inner surface of the tube, we first consider a tube of which the
length is large compared with the diameter. In this case the heat
dissipation m a y be calculated in the same way as t h a t in which the
heat dissipation of parallel plates was derived in I for the case where
the distance b between the plates was small compared with the
height h. The inner diameter of the tube 2R being small compared
with the height h, the flowing medium will over the whole cross-
section attain the same temperature as the tube wall. The heat Q
dissipated per sec. equals therefore
Q = M C , 0~, (3)

where M is the mass of the medium flowing through per sec. and 0w
the temperature difference between the tube wall and the medium at
the moment of flowing in. As in this case the lifting force is constant
over the whole cross-section, the flow in the tube as long as it is
laminar is the well known P o i s s e u i 11 e flow. Hence :

M = ~ ~R 4 dp
dz • (4)

With dp/dl=--g£~, ~O~ (~=expansion coefficient of the medium)

and (4) we find for (3):
Ra - 2 2
s- w (5)
 T H E D I S S I P A T I O N OF H E A T BY F R E E CONVECTION 867

According to the defnition of c¢:

Q - 2re Rh ~0~. (6)
F r o m (5) a n d (6) it follows w i t h Pr ~ Cp ~ , / ~ t h a t
~R R 4 g ~2 ~0~ Pr
~ 16h~
or

1 Rh GrR,,o Pr.
NuR,w -- 16 (7)

Eq. (7) is analogous to eq. (31) of I. Since for R -+ oo eq. (7) m u s t

become identical w i t h the equation of S c h m i d t, P o h 1 h a u-
s e n and B e c k m a n n 2) for the free vertical plate :
Nuh,w = 0.6 (Grh,w Pr) ~' (8)
we write in analogy to (1) :

NUR'w -- 116 Rh GrR't° Pr [I -- e-(20MRGrR,t°Pr)c*]c', (9)

C2 a n d C3 satisfying (2~.

§ 2. Other shapes o~ cross-section. In the case where the tubes do

not have a circular cross-section it is usual s) to characterise the
r e s i s t a n c e of the l a m i n a r flow b y the value of + Re, where ~b =
(2r/p~2)dp/dl and R e = ~ r/~. Here ~ is the m e a n velocity over the
whole cross-section and r the hydraulic radius, defined b y :
r = 2F/U (10)
where F is the area of the cross-section and U the circumference *).
F o r the circular cross-section + Re = 16. F o r rectangular cross-
sections w i t h cathetes a a n d b (a > b) B o u s s i n e s q 3) 4) derived
for the mass M flowing t h r o u g h per sec. :

1 ab3 dp ~ 192 b (tanh =a 1 3rca }

M= 12 "O dl _1 ~ a ~ +-~tanh-~-+...)_(ll)

from which the value of + Re m a y be calculated as a function of' b/a

*) By English physicists tile hydraulic mean depth F/U is often used. We prefer to u~e
2F/U, since for the circular cross-section this equals the radius R and for the infinitely
long parallel plates it equals the distance h between the plates.
868 W. ELENBAAS

(fig. 1) 5). For b/a=O (infinitely long parallel plates) d/Re -~ 24 and
for b/a -~ 1 (square cross-section) + Re : 14.9.25. Apart from the
circular and four rectangular cross-sections (b/a = 0, x/5, 1/2 and 1)
we also carried out measurements on tubes with equilateral triangu-
lar cross-sections. The value of ~bRe for this cross-section is derived
in the next section.

YRel~ all I

i i f
0 O.2 t14 O6 0a tO

Fig. 1. T h e v a l u e of d/Re for r e c t a n g u l a r c r o s s - s e c t i o n s .

§ 3. The lami.nar/low in a tube o/triangular cross-section. When

the axis of the tube coincides with the Z direction (fig. 2) and the
velocity in the Z direction is denoted by w, the following equation
must be satisfied:
d2w d2w 1 dp (12)
dX 2+-d-fly2 : ~ dz

The velocity components in the X and Y direction are zero, and w

is zero at the wall. The boundary conditions for (12) are therefore :
y~0
w -----0 for x + y/a/3 - - k = 0
x--y/~/3 + k : O

The function:

~/3 Y k) (x Y k) 1 dp (13)
w : ~-~-y (x-I- v'3 ,v'3 + ,~ dz
 THE DISSIPATION OF HEAT BY FREE CONVECTION 86~

satisfies (12) and the boundary conditions, as m a y be seen easily.

The fluid flowing through per sec. is 6 times the a m o u n t flowing
through the hatched part of fig. 2. Hence:
i
h (k--x)/~/3

20 ~ dz
~q o o

Introducing the hydraulic radius r ~ 2 F l U ~ k/M'3 we find:

M= 9~v/3 p dPr4" (14).

20 ~ dz

With M----Sa/3 pr 2 ~ and + Re-- - - (2r2/~) dp/dz we have with (14)

+ Re = 40/3.

J
0~ I

Fig. 2.

§ 4. General/ormula/or the heat dissipation o~ the inner Surlace ol

tubes o/di//erent shapes o/cross-section. With dp/dl - = - gp= ~0,~ we
find for ~bRe:

+ Re = 2r2 gP~ [~0= (15)'

For a long narrow tube the heat dissipation per see Q is:
Q = ~F O= Pw.Cp.
870 w. ELENBAAS

W i t h ~ from (15) we obtain :

2r 2 gp~ ~0~ FCp ( I 6)

As Q =- ~0. Uh, a n d w i t h Cp ~w/X~ =- Pr, (l 6) gives :

1 r
Nu,,~, -- ~bRe h Gr,,w Pr. (17)

F o r the case where r is not small compared with h, we write in

a n a l o g y w i t h (9) :

1 r Gr, w P r [1 - - e-(c'hl'C",,we')c'] c'. (18)

Nu,,~ = d~Re -h '
Requiring t h a t for r -+ oo (18) shall change into (8), we have :

t~ Re
-- 0.6 or Ci = 0.5 (d~ Re)'/'.
We find, therefore, for the h e a t dissipation of a vertical tube of
a n y given shape of cross-section :
1 r
Nu,,~, = + Re h Gr,,w P r [ 1 - -
e__lO.Sh@Re)°h/rC;rr,wpr}C, ] C, (19)

with C2 a n d Ca satisfying (2).

§ 5. The measurements. Cylinders were made with borings of

circular cross-section, of rectangular cross-section w i t h ratios of the
cathetes of 1 : 1, 1 : 2 and 1 : 5 a n d of equilateral t r i a n g u l a r cross-
section. F o r each shape of cross-section the value of the hydraulic
radius and the height was varied (r from about ½ cm to 2½ cm and h
from 15 cm to 50 cm). The outer surface of the cylinders was w o u n d
r o u n d w i t h a h e a t i n g wire which was covered with a layer of asbestos
on which a thin A1 plate was fixed in order to diminish the radiation
loss. The cylinders were placed in the vertical position in air of
atmospheric pressure. The m e a s u r e m e n t s were carried out b y de-
termining the t e m p e r a t u r e of the cylinder for a given input while the
boring of the cylin'der at the upper and the lower end was closed with
A1 covers. The A1 covers were t h e n removed and the w a t t a g e in-
creased until at t e m p e r a t u r e equilibrium the same t e m p e r a t u r e rise
was obtained. The difference A W b e t w e e n these two inputs gives the
h e a t dissipation of the inner surface of the cylinder. Two corrections
 THE DISSIPATION OF HEAT, BY FREE CONVECTION 87!

however must be applied. AW must be incraesed b y the convection

loss of the two covers and decreased b y the difference between the
radiation losses of the two covers and the two holes. As these two
corrections are of the same order of magnitude the total correction is
only small. Since, however, AW is small compared with the input,
especially for small values of r, individual errors of 50 % m a y Occur
in this case. An advantage, compared with the measurements of I,
is that no corrections for edge effects have to be applied. 0w was
varied from about 25°C to 300°C.
In fig. 3 the measured values of Nu,,~, are plotted as a function of
(r/h) Gr,,~, Pr on a double logarithmic scale. The measurements of i
are also plotted. These measurements give the closest agreement with
(19) for C2 ---- 1 and C3 = -~-.For the cylinders, however, C2 = } and
C3 = 1 give a closer agreement. As the difference between these tw6
cases is always smaller than 10 % and in the case of the parallel plates
the edge effects cause an addit.ional uncertainty,, we shall assume
that C2 = -~, C3 = 1 isalso correct for the parallel plates. The 6 lines
drawn in fig. 3 therefore obey the equation:

1 r Gr, ~ Pr [1 - - e-~RcC°'shl'c",,,oP°'l'] (20)

Nu,,~ -- qJRe h '
Eq. (20) holds within about 10% for all shapes of cross-sections.

§ 6. "Discussion. In the same manner as in I we m a y derive from

(20) the value of r which gives the highest energy dissipation per unit
of cross-section. To this end we must make ~ hU/F a maximum.
Giving the quantities referring to this oi~timum arrangement the
index m, we find from (20) :
t

( h Gr,,~Pr),n=0.72(, Re)'/'*). (21")

The value of N u belonging to this value of [(r/h) Gr,,w Prim is:

Nu,n = 0.385 (+ Re)'/'. (22)
From (2 l) and (22) it follows that :

Num = 0.42 r Gr,,w Pr .(23)

~ m

*) W i t h ~Ie Rc = 24 we f i n d [(r/h)Grr, w Pr] m = 50 w h e r e a s in I we f o u n d 46 (eq. (49)).

T h i s d i f f e r e n c e is c a u s e d b y t h e f a c t t h a t we n o w u s e d C3 ~ 1 a n d C3 = ~. T h e difIere.nce
c a u s e d in t h e o p t i m u m d i s t a n c e s b e t w e e n t h e p l a t e s is o n l y 2°/o.
 872 W. ELENBAAS

~ -..-..
I I I I I I I I I I I I I I I I I I I ll l I

i
I 0

~ '~ o >- o + .

il

d U
,r.,.l.c::

f
J~
o

.D. ~:~,

0 ~"

,...,

u~

~-ill

0
.~ j
,.-, ~

~ 4.o
0
• 0
,,, , ,,, , I I I I I I I f 1 1 I I I 1 1 1

.~
 THE DISSIPATION OF HEAT BY FREE CONVECTION 873

Thus, if we draw in fig. 3 the line N u = 0.42 [(r/h) Gr,,wpr]'h, this

line intersects the curves 1 to 6 at the points where the cooling per
cm 2 of cross-section is a m a x i m u m .
The energy dissipated per cm 2 of cross-section at the o p t i m u m
value of r is :

(Q)m= (~h U Ow),, __ O:775 h,~ O~(~b

Substituting rm from (21) we have:

(Q) = 0.9 x~ h'' O~"g't' Pw~'' Pr'' (24)

,,, (~b Re)'l, .~,~
The cooling at the o p t i m u m value of r is therefore b e t t e r for smal-
ler values of ~ Re, so t h a t among the above considered shapes of cross-"
sections the equilateral triangle i s the best one. The ratio of this
dissipation tb t h a t with the square form is (1,4.2/13.3)17,= 1.02
and lies therefore within the range of possible errors and is w i t h o u t
significance in practice. The cooling w i t h squares is according to (24)
19 % better t h a n with parallel plates, if for b o t h the o p t i m u m value
of r is chosen.
According to (22) ~,.rm/X~ is proportional to (+ Re) '/,. As rm is pro-
portional to (qJ Re) 'h (eq. 21) we find ~,,, independent of + Re, so t h a t
under the same circumstances, the energy dissipation per cm z of
surface at the o p t i m u m value of r is the same for all shapes of cross-
section.
As m a y be seen from (24) the h e a t dissipation at the o p t i m u m
value of r increases as 0~,. As was also found in I the h e a t dissipation
increases only as h 'h. F r o m (24) we find for the max. h e a t dissipation
per cm 2 of the square cross-section in air at atmospheric pressure
and room t e m p e r a t u r e for h = 50 cm and 0~ = 100°C: 1.25 cal/cm 2.
sec.
If the tubes make an angle 7 with the vertical, we expect t h a t as
long as y is not too large (see I) the above formulae will retain their
v a l i d i t y if we replace g b y g cos y.

Received June 25th)1942. Eindhoven, May 6th, 1942.

874 T H E D I S S I P A T I O N OF H E A T BY F R E E CONVECTION

R E F E RENCES

I) W. E l e n b a a s , Physicag, 1, 1942.
2) E. S c h m i d t undW. Beckmann, Techn. lVIech, u n d T h e r m o d y n , l , 3 4 1 , 3 9
1930.
3) Handb. der exp. Phys. IV 4, p. 146.
4) J. B o u s s i n e s q ~ J. d e m a t h , p u r e t a p p l . ( 2 ) 13,377, 1868.
5) F.C. L e a a n d A . G. T a d r o s , phil. M a g . ( 7 ) , l l , 1235, 1931.