HEAT D I S S I P A T I O N OF P A R A L L E I . PLATES
BY F R E E CONVECTION
by W. ELENBAAS
Natuurkundig Laboratorium der N.V. Philips' Gloeilampenfabrieken Eindhoven-Holland
Zusammenfassung
Die W~rmeabgabe bei freier Konvektion in Luft yon vertikalen quadra-
tischen parallelen Platten yon h × h cm'- im gegenseitigen Abstand b wird
gernessen. Die W/irmea15gabe wird berechnet iiir den Fall, dass die Platten
in horizontaler Richtung unendlich ausgedehnt sind (G1.39). Die Oberein-
s t i m m u n g ist gut, wenn eine Korrektion angebracht wird, welche berfick-
sichtigt dass die Messungen an quadratischen a n s t a t t an unendlich langen
Platten ausgefiihrt sind.
Wenn die Platten fiber einen Winkel T aus der vertikalen Lage gedreht
we~den, ist fiir nicht zu grosse Werte yon T, g in (39) durch g cos T zu
ersetzen.
Bei gegebener H6he h der Platten und l~bertemperatur 0w wird der
gegenseitige Abstand b,n
berechnet, welcher die gr6sste W/irmeabgabe je
L/ingeneinheit liefert (G1.53). Wird h variiert und der P l a t t e n a b s t a n d je-
wells gleich dem zu jeder H6he geh6renden gfinstigsten Wert bm gew~hlt,
so n i m m t bei Vernachlassigung der Plattendicke die W~rmeabgabe je
Langeneinheit wie %/h zu.
H e r e ~ is t h e h e a t d i s s i p a t i o n per u n i t area, u n i t of t i m e a n d u n i t of
t e m p e r a t u r e difference b e t w e e n t h e h e a t d i s s i p a t i n g b o d y a n d t h e
Physica I X I
2 W. ELENBAAS
• :m ~l
Fig. 1. P r e l i m i n a r y a p p a r a t u s for m e a s u r in g the h e a t dissipation of pa-
rallel plates. The plates are k e p t a distance b a p a r t by the rings R, and are
heated by the h e a t e r H.
Z
I_ o
-I- ~X
Fig. 2. Location of the coordinates. The Y axis passes through o
perpendicular to the X Z plane.
~'U'THERNOPIL E
j
Fig. 3. F i n a l a r r a n g e m e n t for measuring the heat dissipation.
input. The difference between the inputs for b ---- 0 and b = bo is the
amount of energy 10st by the two inner surfaces of the plates at the
distance b0. From this amount only the radiation emerging from the
gap between the plates must be subtracted to obtain the dissipation
by convection, This radiation was measured by means of a thermo-
pile and a Moll-galvanometer. Since we must measure the radiation in
different directions, the plates can be rotated around D (fig. 3) through
certain angles y from the vertical position, whereas the thermopile
remains in the same position. The galvanometer deflection is
observed arising from the sudden reduction of b from bo to 0. For
each of the various values of b0 and 0~ this measurement has to be
carried out for several values of 7. Instead of drawing R o u s s e a u
diagrams, we measured only in a limited number of directions
(y -----25 °, 45 °, 60 °, 72½° and 85°), so-called R u s s e 1 1 angles, the
mean value of which gives the mean spherical radiation.
The thermopile-galvanometer combination was calibrated with an
AI sphere of 10 cm diameter with a heater built inside. The sphere
was either highly polished or blackened with aquadag, and in both
conditions the temperature of the sphere and the galvanometer de-
flection were measured as a function of the input for a known distan-
ce between the thermopile and the centre of the sphere. In fig. 4 the
values of 0= and the galvanometer deflections u are drawn as a func-
tion of the input for the polished as well as for the blackened sphere.
Supposing the energy dissipated by convection to be independent of
the surfsce condition of the sphere, the galvanometer deflection A u
corresponds to the spherically radiated energy A W (fig. 4).
For each pair of plates, the radiation emerging from the gap was
measured as a function of b at 0~ ~- 330 ° only. To obtain the radia-
tion of the gap for smaUer values of 0w the radiation of the highly
polished A1 sphere was measured as a function of temperature and
the way in which the radiation from the gap depended on the temper-
ature was assumed to be the same as that for the sphere. For the
radiation of the highly polished A1 sphere we found:
S = 3.4 10-t5 (T 4~ - - T~) Watt/cm 2 (7)
(verified ~oetween 350 and 650°K).
The exponent is in good agreement with that found by S c h m i d t
and F u r t h m a n n V ) , who give S = 2 . 4 0 I0 -Is (T 4 " 7 3 - T ~ 3)
Watt/cm 2 (verified between 500 and 850°K). We find, however, the
HEAT DISSIPATION OF PARALLEL PLATES 7
/ / d I ~o
L° r U
*u
.I
0 tO. 20 30
"-------W
Fig, 4. C a l i b r a t i o n of t h e t h e r m o p i l e - g a l v a n o m e t e r c o m b i n a t i o n w i t h t h e
a i d of an Al s p h e r e w h i c h is e i t h e r h i g h l y p o l i s h e d or b l a c k e n e d . T e m p e r a -
t u r e e x c e s s 0w a n d g a l v a n o m e t e r d e f l e c t i o n u as a f u n c t i o n of t h e w a t t s
d i s s i p a t e d in b o t h cases. T h e g a l v a n o m e t e r d e f l e c t i o n A u c o r r e s p o n d s t o
t h e s p h e r i c a l l y r a d i a t e d e n e r g y A W.
"° ! ! i
~o i ' / / I
--- '- t--T.7; -T, i
." I/T | i
"'0 . . . . . . -T~- t°..=. L t
~0-- -. . . . . . . .
Fig. 5. R a d i a t i o n e m e r g i n g f r o m t h e g a p b e t w e e n t w o h i g h l y p o l i s h e d AI
p l a t e s as a f u n c t i o n of t h e d i s t a n c e b e t w e e n t h e p l a t e s . T h e p l a t e s are
330°C a b o v e r o o m t e m p e r a t m e (20°C).
8 W. ELENBAAS
Since for the smaller values of 0" the radiation loss decreases rela-
t i v e l y to the c o n v e c t i o n loss, an error in the radiation loss for smaller
values of 0= c a n n o t be serious except p e r h a p s for v e r y small values of b,
because with decreasing b the c o n v e c t i o n loss decreases more r a p i d l y
t h a n the radiation loss.
Fig. 5 gives the radiation as a function of b at 0= = 330°C for the
three plate dimensions. F o r similar gaps the radiation per cm 2 m u s t
be the same, so t h a t , on p l o t t i n g the r a d i a t i o n per cm 2 as a function
of b/h, the lines of fig. 5 m u s t coincide. As fig. 6 shows, this is indeed
the case. There are two more tests. F o r large values of b/h the radia-
tion per c m 2 m u s t be the same as t h a t of the A1 sphere. F o r 0~, ---
330 ° and 20 ° room t e m p e r a t u r e we find for the radiation per cm 2
from (7):
S = 3.4 l0 -15 (6234t - - 2934t) ---- 6.2 10 -2 W a t t / c m 2.
9 w:3~o I
,~.I0 "2 _
t -v
,/ _/
f
o h=5,.95
+ h=f2
L x h=24
I
!
I
0
Fig. 6. The three lines of fig. 5 coincide when the radiation in Watt/cm:
is plotted as a function of gap width, divided bv the length of the sides of
the squar.e plates.
I /x'/1 l
•/..2 /
[oew.332 °
lY
//A
K X /I
/I
,Ow=~ °
4 /
~ bCcm)
TABLE 1
energy ==
? II •
Atmos- dissipa. convec-
input input tion radia- tion b
pheric b (era) for tion [I -- Gr*. Pr Nu* : q
0~ loss X
pressure (Watt) 0w = 132 ° b e t w e e n ( W a t t s ) h
g ~.v
mm Hg) the plates (Watts)
..o I"e II
(Watts)
756.4 0.915 79.6 134.5 78.2 34.7 = 1.55 33.2 8.75 20.9 2.37 315 230 2.26
,o
I ,~J.--f'T - I AT o
i//r i.,,,~/
LIL/lll ...L= • :
i i i.k
IN
g
/~ f2xl3e'ma' I:+.0w=333
Ow:f32 °
e w:,l¥ °
~.Ow:fO °
111 II
I II II 1 II
IJ/ I A
r//
r x
ca o,s ---~b(c, 2 4
+-~ [~-~
a(a.~ + ~ +av- - ~ -a~) + -a~.
~ + -a~~.- + ~-~-
a~.] (8,,~
12 w. ELENBAAS
p~+p u-~+v-~+w-~ = - - ~ +
+~- ~+~+-~ +-~+ ~- a:~] (sb)
v - ~ + v ~-~-+v-~-+ ~- ----~-~-+
av ow
+ N + ~ -) + -~ + ~ + - ~ + g (ooo--p) (80
• I
i
/'~_,~-~-I .I '/'I . s w=+....,,,,.,
w/~"¢
/ .
"o 0 w = 3 2 5 °
i
x Ow,.f35 °
34rB~cn~,
+0w=50 °
. O w,: fO °
I /AI I
~ I / / i i:
i ;
. b(cm)
pep '/b' ~
(a2o a2o
= k ~ ~.X2 "t- ~ ' ~ ] (12)
0 = 0~ ,1 - - e - P ~ c o s ~ - (14)
and
a2 - - x 2 ~x'l, ~ Too + 0
w = s ~.2 + r cos 2--dqF] Too ( 1 5)
irhe factor (Too + O)]Too in (15) causes (13) to be satisfied. (14) and
(15) satisfy the b o u n d a r y conditions and for a constant value of
x(]x] < a), 0 increases with z to reach asymptotically the value 0,, if
h is large enough.
(14) and (15) are introduced in ( l l ) and (12) and the values of p, s
and r are so chosen t h a t (1 I) and (12) are satisfied at x = 0 and
X ~ -3c a.
F r o m (11) at x = ± a we find"
( 3nr 2s ] Too + 0~ 2~, ( 3nr + 2s I 0~,
e-#, +
0 =- - - ~., k8a2 + -~ / Too Too \ 4a -~1 ~
+ p.g0. (16)
Too
(index w means t h a t the quantities are taken at surface temperature).
14 w. ELEN BA.AS
p2 + 3p~ (r + s) p 3 ~ - o, (20)
4.0o 16a 2
wkile the remaining part of (19) gives:
Pog 8o a 2
s = 2.0o(Too + 0o)" (21)
Eq. (21) is of the same form as (18). It is less rigorous, since 0o and
to a smaller extent .0o and Po are functions of z.
E q u a t i o n (12) gives) for x = ± a
0=0
and for x = 0
poo c~ (r + s) 0~ p e-P" = Xo 0~ ~ e-P" - - Xo 0~ p2 e-P" (22)
we have to deal with gases (for mon-atomic gases Pr = 2/3, for dia-
tomic gases 0,73 ; in general Pr = (n + 2)/(n + 4~) if n = the n u m b e r
of degrees of freedom of the gasmolecules).
As can be shown, the t e r m p2 is small compared with the other
terms of (20) and (23). We will, therefore, neglect p2 in these equa-
tions. This means t h a t the terms azw/az2 and M0/az2 in equations (11)
and (12) are heglected. Using (23) to preserve the dependence on Pr,
we find with (17), (18), a = -.}b and replacing ~]o b y v]., which causes
only a small error:
P -- b¢g p~ 0. Pr (24)
Q= X~.=.dz=X=0.2aj bp ( l - - e - P * ) . (25)
0 0
By the definition of a:
<2 ~- ~ 0. h. (26)
from which"
g pw 0w
---- 2 ~ T------~( ~ - - '~)" (36)
Per cm width in the Y direction the mass of the escaping air is'
+a
f p~ w d x and this mass is 0~ higher in temperature t h a n at the mo-
ment of flowing in. The energy Q dissipated per cm width in the Y
direction and per plate is therefore :
a
Q - ~ h O~ --- cp O w f p , , . w d x
0
ted in fig. 10a, whereas for large values of b the distributions will be
as indicated in fig. 10b. If in fig. 10a b becomes infinite and the form
of the 0 c u r v e remains the same, the 0 curve will be perpendicular to
the wall and no energy can be given off b y the latter. We shall, there-
fore, have to alter (32) and as this alteration must not influence the
case of small values of b, we must change the exponent. In order to
obtain a c o n s t a n t value for ~, when b ~ oo, we.change (32) into:
_ [ --/ c,h y,I c,
1 b Gr* P r 1 - - e ~0Gr..P~, j (33)
Nu* 24 h '
............................... ~W
L 0
--8 "-'-~'X 8
_b
FI~. 10. T h e t e m p e r a t u r e e x c e s s 0 o v e r t h e t e m p e r a t u r e of t h e s u r r o u n d i n g s
a n d t h e v e l o c i t y w, w i t h n a r r o w g a p s (a) a n d w i t h w i d e g a p s (b).
(36) has the same form.as (5) where the linear dimension in Nu a n d
Gr is also the height h of the vertical plate.
Comparison of (36)and (5)gives: c~/24 -~ 0,6 from which cl = 35.
(33) becomes therefore:
Since (37) is rigorously valid for small values of b and will hold
for large values of b so long as (5) is correct, c2 or cs m a y be chosen in
such a way, t h a t the intermediate region is correctly described b y
(37).
(37) shows t h a t the predicted form/5 of (6) m a y be more specially
written as:
Nu* ---- / (Gr*. Pr b/h) (6a)
Until now we have assumed the plates to be in a vertical position.
If the plates m a k e an angle ~, with the vertical direction, we expect
t h a t in (37) we must replace g b y g cos y, since' only this component
of g is active for the flow of the m e d i u m between the plates. For large
values of y discrepancies m u s t be expected, because for y ---- 90 ° the
formula gives no heat dissipation at all, whereas for ~, ---- 90 ° and b
large, the dissipation is equal to the mean of those of the upper and
lower sides of a horizontal plate (which is of the same order as the
dissipation of a vertical plate). For ¥ = 90 ° and b very small, one
expects the heat dissipation to be roughly proportional to the area
4 bh, so t h a t a becomes proportional to b/h and c~ b/), proportional to
b2/h. If, for constant values of h, Nu* is plotted as a function of
(b/h) G r * . Pr cos y (which is proportional to b4/h), we expect there-
fore for small values of b a slope of ½ (Fig. 13). If the dissipation were
proportional to b2/h~, the lines would coincide for different values of
h; now one expects the height of the lines to be proportional t o . h - t
which for y ---- 85 ° is in agreement with the measurements (fig. 13).
OXl i I#II
,,, ,I ,,[i]) ,,,,,,
ill i i111"~¢1- I
;~
~.¢',,,,,,fill ~,llll , 1111111 I
~,
IIIIIIIIIIII I
-~
I111111t1111
,,,, ,,111 11I tl I Iit ] 0.5
o
I I I 11 IIl/i~ ' [ I ~ I II 3 ,.~ ,.~ • 4
I I II!1 1 I 117-',I ,s,~:~,l" I I ~'i'I IIII ~.~o ~.~ ~o z:,o ~.~o ~ z,o 5.,o ~,
I I lill I I I I llll I i~rll i i i -J.'lil i iiii
nnS.!, I I 111 I I I I IIII L,4illl I .ll'l 1111 I III III11
'~" 1111 I I I t I-.~, I l l I.;~'~ ! ~,~
llllll i I Il~!lllll tA I
[lllll 1/1 I I llll,,<'4"lllll I I -,~w I I I III11 (~o ~,u -- *ttu/#t] ~ut ._r-i/
i FI a
~, I
I. . 1¼11
l i l l y
... /I I L~ Ik, llll.
III I II
. . . _. , ,a, , ~ .
Ol-I
~ . square plates.
I I I I-"I | l~<i-~, ' [ ~ ~ b. c o r r e c t i o n applied in such a way that the measure-
axT051 I l il I li'~l- I I Pr=--P-P-P-P-P-P~- I ' i ' 'iii m e n t s h o l d for p l a t e s t h a t a r e i n f i n i t e l v l o n g in t h e
~r~ I I A/'~ , x~, ~q03 • I Y d i r e c t i o n . T h e line is d r a w n a c c o r c l i n g t o (39).
',,,~1/I / I I I III111111 / " ' • - •
I III I= i T h e s c a l e of b is s h i f t e d a f a c t o r 5 w i t h r e s p e c t t o
I III I I I 1 I I 1 IIII IIIII I I III0~ th~tofa.
20 W. ELENBAAS
3) For large values of (b/h) Gr*. Pr the slope of the line must be ~.
4) For.(b/h) Gr*. P r - - l, Nu* must be 1/24 = 0.0415 (the e
power in (37) m a y be neglected here).
5) For (b/h) Gr*. Pr = l04, Nu* must be 6.0.
In fig. 1 l a all measurements of figs. 7, 8 and 9 are plotted on a
double logarithmic scale. The points are indeed scattered about one
single line. At small values of (b/h) Gr*. Pr the deviations are, how-
ever, rather large. Conditions 3) and 5) are satisfied. The slope at
small values of (b/h).Gr*. Pr is, however, somewhat too small and
the Nu* value for (b/h) Gr*. Pr = 1 is larger than 1/24.
The cause of this discrepancy lies in the fact that the calculation is
done for plates which are infinitely lon.g in the Y direction, whereas
the experiments are done with square plates. In order to take this
into account, we shall correct the measurements so that they hold for
plates infinitely long in the Y direction. V~e assume that small ver-
tical strips near the edges of the plate (at y = ~ ½h) lose energy in
the same w a y as the free vertical plate (eq. (5)). We estimate the
width of these strips to be ½b. Each square plate of h2 cm 2 has, there-
fore, an area of bh cm 2, losing energy according to (5), whereas the
rest loses energy according to (37). As we are only concerned with a
correction occurring for small values of b, we may omit the e power in
(37). Denoting the energy dissipation of the square plates b y Qo and
the energy dissipation of the infinitely long plates by Qoo we find"
Qo 24 x 0.6 ~ , T'& b
q : Qoo - 1 + h,i, b2 g,i, p~, 0'~' Pr'/, h
7.35 -~-" b
= l + h,lob2.p~ , 0 ~ , h (38)
(the last expression holding for air and roomtemperature by substi-
tuting - - Pr = 0.73 - - and Too = 293°K).
T A B L E II
Oa,(°C)
I0 1.87 I0 -4 1.16 ~ lO-~ 1.15 lO- t
50 2.04 ! .03 0.46 s
I00 2.26 0.90 0.40
180 2.57 0.748 I 0.41
330 3.10 0.568 i 0.52
I
h2f2 '
h-.~4!
0w5o0-3o" \\\! ew=,o
\ %..
N 1
o q,s o
,, b (crn) ,~b(cm)
Fig. 12. Correction factors q, by which the heat dissipation valid for
square plates, must be divided to obtain the heat dissipation of plates
infinitely long in the Y direction.
Nu* = ~2-4
- -h Gr*. Pr 1 - - e bG-~:
(39)
with N u * = ~bb. Gr* t
b3g~O=" Pr = cp ~____w_~
X= ~}2Too ' ),~
TABLE III
I0 I 1
50 3.62 1.49
I00 5.05 1.76
180 5.62 2.02
330 5.05 2.30
we m a y write (39) as
ol I IIIII - -~"
l !
i
g~
/ J :
! I ,
i~u I i I i iiii
I I ~ I I I L~f
~esI ~- : I lYlL, • I iiii u
] I11 iiii
] I I [ Jill
I I I I
~. r ~"~'=85 ° ,~4x24cnP I III
iiii
0,o, :o,~.~L~-~-- " i " '"' I I I
must be a maximum.
F r o m (46) it follows that the gap width bm giving the highest
energy dissipation, satisfies the equation :
bm 105 / e-~/(b'm bm
3 b,,, + d + b 4 - ~ ] = 3 b,. + d " (47)
105 ~ e _ ~ l ( b , A ) = 2 (48)
2 + b AJ
from which
b~ A = b~ g f~ ~0~, P r / h ~ = 46. (49)
In fig. ( 11 b) this corresponds to the point P.
As can easily be calculated, the tangent in P has a slope of ½. On
the line S in fig. 1 l b, 0¢ b is, therefore, proportional to b2, so that Gtis
proi~rtional to b. Since the total-energy loss is proportional to
n ~ = L a/b, the total energy-loss is constant on S. The departure of
the curve from S in the vertical direction gives, therefore, the devia-
tion from the greatest cooling. For 10% less cooling (b/h) G r * . P r
m a y differ b y nearly a factor 3 from the value at P. This corresponds
to a deviation of 3 0 % of b from bt.
In order to solve (47), taking into account the thickness d of the
plates, we write:
bm = bl + a (50)
t~
5 fO ~0 50 tO0
,,, h(cm)
Fig. 14. Optimum distance between the plates diminished by 0.,3 of the
plate thickness d as a function of the height h for different values of
temperature excess 0w over room temperature for vertical plates in air
of atmospheric pressure.
given h. If now, we vary h, and for each value of h choose the value of
b equal to bin, we m a y regard the energy dissipation as a function of
h. In neglecting the thickness d of the plates and the heat-loss of the
two outer surfaces, the heat-loss over the length L per cm of width
in the Y direction is according to (45) proportional to od~/bm. Intro-
ducing (53) with d 0 in (39) we find Nu*
= ~b,,/k~ constant = =
\ -%
= fOOcm
t~
,=50cm
!
h=fOcn
_____.---~
h=-5cm!
2o 5o -.tocJ aoo
REFERENCE~