You are on page 1of 9

# Progettazione Ottica Roncati

info@progettazioneottica.it
www.progettazioneottica.it

Iterative calculation of the heat transfer coefficient
D.Roncati
Progettazione Ottica Roncati, via Panfilio, 17 44121 Ferrara

Aim
The plate temperature of a cooling heat sink is an important parameter that has to be determined with
accuracy. The estimated value depends on its geometrical shape, on the total amount of energy to be
dispersed, and on the air flow. The heat transfer coefficient, h, is the most difficult parameter to be settled.
In this report it is shown a fast and easy iterative method to calculate the h value and later, the
temperature of cooling for heat sink.
Introduction
The heat transfer coefficient or convective coefficient (h), is used in thermodynamics to calculate the heat
transfer typically occurring by convection. simple way to calculate h is to define it through the classical
formula for convection, and compare it with a different definition of h, through dimensionless parameters.
!nfortunately, even if defined by means of different parameters, both the environment and the heat sink
temperature are important to estimate h. n iterative method is then re"uired, by setting an initial value
of the Tp.
Convection heat transfer coefficient

The formula for heat transfer is#
= S (I
p
I
u
) (\$)
%here#
& 'heat transferred, ()s ' %
h ' heat transfer coefficient, %)(m
*
+)
, ' transfer surface, m
*

Tp ' -late temperature, +
Ta ' ir temperature, +

.or convection we use the convection heat transfer coefficient h
c
, %)(m
*
+). different approach is to
define h through the /usselt number /u, which is the ratio between the convective and the conductive
heat transfer#
Nu =
Concctc hcut tuns]c
Conductc hcut tuns]c
= (
c
I)k (*)
%here#
/u ' /usselt number
h
c
' convective heat transfer coefficient
k ' thermal conductivity, %)m+
0 ' characteristic length, m

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

The convection heat transfer coefficient is then defined as following#

c
=
Nuk
L
(1)

The Nusselt number depends on the geometrical shape of the heat sink and on the air flow. .or natural
convection on flat isothermal plate the formula of /a is given in table \$.

Table \$# /usselt number formula.
2ertical fins 3orizontal fins
0aminar flow Nu = u.S9 Ro
0.25
!pward laminar flow Nu = u.S4 Ro
0.25

Turbulent flow Nu = u.14 Ro
0.33
4ownward laminar flow Nu = u.27 Ro
0.25

Turbulent flow Nu = u.14 Ro
0.33

%here#
Ro = 0r Pr (5)
is the Rayleigh number defined in terms of -randtl number (-r) and 6rashof number (6r). If 7a 8 1u
9
the
heat flow is laminar, while if 7a 9 1u
9
the flow is turbulent.

The Grashof number, 6r is defined as following#

0r =
gL
3
[(1
p
-1
c
)
q
2
(:)
%here#
g ' acceleration of gravity ' ;.<\$, m)s
*

0 ' longer side of the fin, m
= ' air thermal e>pansion coefficient. .or gases, is the reciprocal of the temperature in +elvin#
[ =
1
1
c
, \$)+
Tp ' -late temperature, ?@.
Ta ' ir temperature, ?@
A ' air kinematic viscosity, is 1.S
5
at *B ?@. 1.6
5
at 1B ?@.

.or plate temperature, Tp, set a e>pected value. .inally, the Prandtl number, -r is defined as#

Pr =
cp
k
(C)
%here#
D ' air dynamic viscosity, is 1.81
5
at *B ?@. 1.86
5
at 1B ?@.
cp ' air specific heat ' \$BB: ()(+gE+) for dry air
k ' air thermal conductivity ' B.B*C %)(mE+) at *F ?@

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

The total amount of energy leaving a surface as radiant heat depends on the absolute temperature and on
the nature of the surface. nonGblackbody, emits radiant energy from its surface at a rate that is given by#
q = o e S AI
4
= o e S (I
p
4
I
u
4
) (F)
%here o = S.67 1u
-8
w
m
2
K
4
is the ,tefanGHoltzmann constant and I is the emissivity. Jmissivity varies
from B (for non emitting body) to \$ (for blackbody). 2alues of emissivities for aluminum is presented in
Table *. 7eplacing "of J". F in J".\$ we obtain#
o e S (I
p
4
I
u
4
) = S (I
p
I
u
) (<)
nd solving for h obtain the formula for radiant heat transfer coefficient, h
r
#

=
cs(1
p
4
-1
c
4
)
(1
p
-1
c
)
(;)
Table *. Jmissivities of different aluminum surface treatment
luminum 1\$B + :1B + <BB +
-olished B.B5 B.B: B.B<
K>idized B.\$\$ B.\$* B.\$<
*5G,T weathered B.5B B.1* B.*F
nodized (at \$BBB?.) B.;5 B.5* B.CB

.or not negligible radiation the overall heat transfer coefficient is#
=
c
+

(\$B)
In .igure \$ are reported the values of hr as a a function of Tp for emissivity of B.< and Ta ' *B ?@.

.ig. \$# 7adiant heat transfer coefficient for varies temperature plate with ambient temperature *B ?@ and
emissivity B.<.

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

Iterative method

n initial guess value for Tp is set in J". : to determine 6r. Knce calculated the dimensionless parameters
6r, -r, 7a and /u,the value of h is obtained through the J". 1. Hy introducing the value of h in J". \$, a new
value of Tp is given by#
I
p
= I
u
+

hS
(F)

and the obtained Tp can be used to estimate iteratively the proper value of 6r for the particular
e>perimental setGup by using it in the J". :. The obtained value of h is then used again to estimate the 6r
parameters, in an iterative way, that rapidly converges to the correct estimation of h.
Example 1.
0J4 is assembled on a 5B > 5B > C mm aluminum heat sink with four fins 1B mm long and 5 mm large. The
0J4 generate 1 % and the room temperature is *B ?@. %hich is the value of the heat sink temperatureL

Fig. 2 scheme showing the assembled heat sink and energy flow.

\$. 4efine 6rashof number with 0 ' 5B mm, the greater side of fins and assuming a plate temperature of \$BB
?@#

0r =
9,81 u,u4
3

1
27S +2u
(1uu 2u)
(1,S 1u
5
)
2
= 7.62 1u
5

*. 4efine -randtl number#

Pr =
1,81 1u
5
1uuS
u,u26
= u.7

1. 4efine the 7ayleigh number# Ro = 7.62 1u
5
u.7 = S.SS 1u
5
< 1 1u
9
so is laminar flow.

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

.or laminar flow, horizontal fins (flat and isothermal with good appro>imation) and heat upward flow, from
Table \$ the value of /usselt number /u is determined by using the relation#

Nu = u.S4 Ro
0.25
= u.S4 (S.SS 1u
5
)
0,25
= 14.S9

5. .inally, the heat transfer coefficient turns out to be#

c
=
14.S9 u,u26
u,u4
= 9.S
w
m
2
K

:. The cooling area is the sum of lateral fins surface and the plane surface, as shown in .ig. 1.

Fig. 3. J>planation of the dispersion surfaces

The total lateral fins surface is# 4u Su 8 = 96uu mm
2
= u,uu96 m
2
, the plan surface is 4u 4u =
16uu mm
2
= u,uu16 m
2
for a total cooling surface of u,u112 m
2
.

Hy using the value of the cooling surface, it is possible to check the correctness of the Tp that was initially
set to calculate a starting value for 6r (J". :).

I
p
= 2u C +
S w
9.S
w
m
2
K
u,u112 m
2
= 48 C

s evident, the set and the estimated values of Tp are inconsistent. The calculated value of Tp is used in J".
: to estimate a more appropriate value of 6r. %ith Tp '5< ?@, the new parameters are#
0r = 2.67 1u
5
,
Ro = 1.87 1u
5
,
Nu = 11.22,

c
= 7.S
w
m
2
K
.
-r is always B,F./ow with a check the plate temperature result to be I
p
= S7 C.
.urther iterations gets the following values#
\$.
c
= 7.8
w
m
2
K
and I
p
= S4 C.
*.
c
= 7.7
w
m
2
K
and I
p
= SS C.
The last iteration give a values of Tp for both convection heat formula and via /usselt number. The value of

c
= 7.7
w
m
2
K
can be considered.

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

If also the flat side of the heat sink is in free air we must add this other convection which has a different h
value. In the J>ample * such effect is also considered.
Example 2.
@onsider the heat sink in e>ample \$ with heat generated by a 5 > 5 mm 0J4 die set in the middle of the
plane as show in figure 5. 4efine the heat sink temperature, Tp, considering the two different heat transfer
coefficients.

Fig.4. !am"le #ith u"#ard and do#n#ard con\$ection

The flat surface is 4u 4u = 16uu mm
2
. .or the cooling area we must subtract the die area#
16uu - 16 = 1S84 mm
2
= u.uu1S84 m
2
.
The /usselt formula for horizontal flat plate in natural air flow and downward heat flow is#
Nu = u.27 Ro
0.25

%e have already defined the value for 6r, and 7a with 0 ' 5B mm and Tp ' :: ?@ and can calculate /u and
h#
0r = S.SS 1u
5
M Ro = 2.SS 1u
5
,
Nu = S.9SM
c
= S.9
w
m
2
K
.
%ith two different convection coefficient the J".\$ becomes#

= (
1
S
1
+
2
S
2
) (I
p
I
u
) (<)
%here the subscript \$ is for upward convection and * for downward. ,olving for I
p
the e".F becomes#
I
p
= I
u
+

h
1
S
1
+h
2
S
2
(;)
,ubstituting the values we get#
I
p
= 2u +
S
7.7 u.u112uu +S.9 u.uu1S84
= S2 C
.or the iterative method we set Tp ' :* ?@ in e".: to reGcalculate /u and h
c
for upward and downward
convection and set the new /u and h value in e".;.
\$.
1
= 7.S
w
m
2
K
,
2
= S.8
w
m
2
K
, I
p
= SS C

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

*.
1
= 7.6
w
m
2
K
,
2
= S.8
w
m
2
K
, I
p
= SS C
The last iteration give a values of Tp for both convection heat formula and via /usselt number. The value of
.
1
= 7.6
w
m
2
K
and
2
= S.8
w
m
2
K
can be considered.
Validation of the method

The iterative method was applied to define the heat sink temperature of the 0J4 floodlight ,I7IK*5 (a new
product developed by -rogettazione Kttica 7oncati and -aolo @olombani 4esign, see .igure :). The results
of the iterative method were compare with both the thermal simulation (performed with the finite
element analysis software 0I, <.B.B) and the laboratory tests.

Fig.%. &' floodlight ()R)O24
The ,I7IK*5 heat sink is a black anodized aluminum \$;< > \$1* >C mm plate with *B fins \$1* > 15 > 1 mm.
The heat flu> transferred from 0J4s to the heat sink is \$C %att. The laboratory temperature is \$: ?@ and the
proNector is placed with the 0J4s on the downside. Input parameters for the iterative method are#
3eat power, & ' \$C %
mbient temperature, ta ' \$: ?@
6reater fin dimension, 0 ' \$1* mm
air kinematic viscosity, A ' \$.5:E\$B
G:

air dynamic viscosity, D '\$.F<E\$B
G:

s show in .igure 1 the cooling surface, , is e"ual to lateral fin surface (,f) plus plane surface (,p).
S
]
= 1S2 S4 2u 2 = 179S2u mm
2
= u.179S2 m
2

S
p
= 198 1S2 = 261S6 mm
2
= u.u261S6 m
2

S = S
]
+S
p
= u.2uS6S6 m
2

ppling the method for horizontal fins, laminar air flow and upward heat flow we find a heat transfer
convection coefficient,
c
= 4.8
w
m
2
K
and a plate temperature, Tp ' 1\$.*B ?@.
t 1\$.*B ?@ radiation heat transfer is negligible.

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

Fig.*. &' floodlight ()R)O24 3' model for thermal analysis created #ith &)(+ ,.-.-

t first, a 14 model of the heat sink was created with 0I, <.B.B. \$C % flow rate was then applied on the
plane surfaces as well as a convection for cooling the surfaces with a coefficient
c
= 4.8
w
m
2
K
. The result
of the simulation are shown in .ig. F in which the estimated temperature varies from 1B.;B ?@ to 1\$.:C ?@
while the average temperature is 1\$.*1 ?@. The analytic)iterative method based on the /usseltOs formula
for isothermal fin appro>imation was also compared with the previous results (,ee Table 1).

Fig... &' floodlight ()R)O24 thermal analysis done #ith &)(+ ,.-.-
The previous results were compared with a laboratory test of heat sink temperature (\$: ?@ ambient
temperature, horizontal fins with upward heat flow) using an infrared thermal gun with precision :P,
resolution B.\$ ?@. The e>perimental results agree with the simulations within the accuracy of the testing
device (see Table 1).
/able 3. ,I7IK*5 heat sink temperature results.
Iterative method .J with 0I, <.B.B 0aboratory
1\$.* ?@ 1\$.* ?@ 1*.\$ ?@

Progettazione Ottica Roncati
info@progettazioneottica.it
www.progettazioneottica.it

Conclusions.

Iterative method and thermal analysis done with finite element software 0I, given values that match with
difference of B.\$P with a heat sink 14 model precision of :P. The difference of B.; ?@ (*.<P) with the
laboratory run can be e>plained in terms of uncertainty in the determination of &. Indeed, the heat power
0J4 generated is calculated as difference between power absorption and radiant power. The latter is
estimated from comple> computation.
The iterative method can be used to evaluate accurately the temperature value for heat sink or to define
with better precision the heat transfer convection coefficient for finite elements software for thermal
analysis like 0I,. It is worth noting that the /usselt formula tabulated are for rectangular plate and natural
air flow (free convection) and is not valid for other cases like for heat sink with different shape or for forced
convection.