International Journal of Heat and Mass Transfer 54 (2011) 5602–5611

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ijhmt

Experimental study on liquid flow and heat transfer in micro square pin fin heat sink
Minghou Liu ⇑, Dong Liu, Sheng Xu, Yiliang Chen
Department of Thermal Science and Energy Engineering, University of Science and Technology of China (USTC), Hefei, Anhui 230027, PR China

a r t i c l e i n f o a b s t r a c t

Article history: The heat sinks, with the total heat transfer area of 20  20 mm2 and an array of 625 staggered micro
Received 26 January 2011 square pin fins of 559  559 lm2 or 445  445 lm2 cross section by 3 mm height were fabricated on a
Received in revised form 5 July 2011 copper test section. Using deionized water as coolant liquid, the flow and heat transfer performance of
Available online 16 August 2011
the high pin fins were studied with the Reynolds number ranging from 60 to 800. For the
445  445 lm2 cross section pin fin heat sink, the heat dissipation could reach 2.83  106 W/m2 at the
Keywords: flow rate of 57.225 L/h and the surface temperature of 73.4 °C. The experimental data also showed that
Micro fin
the pressure drop and the average Nusselt number increased with the fin Reynolds number. The heat
Heat transfer
Liquid cooling
resistance decreased with, with its decreasing rate inversely proportional to, the pressure drop. Since four
Single phase chosen previous correlations overestimate the flow and heat transfer performance of the present sinks,
we also proposed two new correlations for the average Nusselt number and pressure drop prediction.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Marques and Kelly [14] fabricated a micro pin fin heat exchanger
and tested its performance experimentally. The results showed
With the rapid development of semiconductor industry, the that the micro fin heat exchangers offered the potential to control
scale of heat dissipation rate from electronic devices increases the surface temperature in high heat flux applications more effec-
drastically. Nowadays, the average power dissipation is about tively. Kosar et al. [15] experimentally investigated the pressure
100 W/cm2, and the maximum local power dissipation could reach drop and friction factors in different aspect ratio micro pin fins
500 W/cm2 [1], which is unreachable by traditional direct air cool- and proposed a modified friction factor correlation. Peles et al.
ing methods due to the maximum heat dissipation capability lim- [16] theoretically and experimentally studied the heat transfer
itation. Since Tuckerman and Pease [2] proposed the first micro and pressure drop phenomena over a bank of micro pin fins with
channel heat sinks and exhibited the large heat dissipation ability a focus on the effect of geometrical and thermo-hydraulic param-
of micro channels, numerous investigations have been conducted eters on the total thermal resistance. They found that the micro
on the convection heat transfer in micro channels [3–7]. Due to pin fin heat radiator could take away 790 W/cm2 when the pres-
the small width and height, most flow patterns in micro channels sure drop was two atmospheric and the surface wall temperature
were laminar. The mass transfer between the layers could be ig- increment was 30.7 °C. Compared with the micro channels, the
nored so the heat transfer mode in the fluid interior is considered heat transfer performance of micro fins was better while the pres-
as heat conduction only. When the channel length is short or the sure drop was much bigger. Siu-ho et al. [17] experimentally inves-
flow velocity is high, there would not be enough time for interior tigated the pressure drop and heat transfer characteristics of a
fluid heat transfer. Some investigations have employed different single-phase copper micro pin fin heat sink. The results indicated
methods to increase the interior layers mixing and heat transfer that the micro pin fin heat sink could be very effective at meeting
[6,8]. the needs of high-heat-flux electronic cooling applications. The
Micro scale pin fins (staggered or in-line) are the common previous friction factor and heat transfer correlations might not
geometry used to increase the surface area and the passage flow be usable for situations beyond their original ranges of validity.
turbulence. A number of researchers have studied the heat transfer Therefore, new predictive tools specifically tailored for the sin-
as well as the flow friction in these areas and presented a lot of gle-phase flow and heat transfer in the micro pin fin heat sink
experimental data and relational formulas [9–13]. According to might be required. Qu et al. [18–20] compared thermal-hydraulic
the results of these formulations, the average heat transfer coeffi- performance of a single-phase micro pin fin heat sink against a mi-
cient in short fins is larger than in long tubes. Using LIGA (Lithog- cro channel heat sink. He found that the micro-channel heat sink
raphy Electroforming Micro Molding) micromachining process, had a higher convection thermal resistance with a lower pressure
⇑ Corresponding author. Tel.: +86 551 3603127. drop at high cooling flow rates. In their study, the experiments
E-mail address: [email protected] (M. Liu).
were conducted at 30 °C and 60 °C coolant inlet temperature and

0017-9310/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2011.07.013
 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611 5603

Nomenclature

A area of heat surface (m2) ST transverse distance of pin fins (m)

A1 bottom areas of heat sink (m2) T temperature (°C)
cp specific heat (J/kg K) u flow velocity (m s1)
Dc equivalent diameter of a fin pitch (m) W width (m)
De equivalent diameter of a fin size (m) Wc channel width (m)
f friction factor Wfin fin width (m)
H height of pin fin (m) Dp pressure drop (Pa)
h heat transfer coefficient (W m2 K1)
k thermal conductivity (W m1 K1) Greek symbols
L length of pin fin region (m) mfin fin efficiency coefficient
l distance between the thermocouple and down surface gfin fin efficiency
of heat sink (m) l viscosity (Pa s)
m mass flux (kg s1) n fin/channel ratio
N row number of pin fins p 3.1415
Nt the number of pin fins q density (kg m3)
Nu Nusselt number
p pressure drop (Pa) subscripts
pd inlet/outlet pressure drop (Pa) w surface
Pfin Perimeter of micro pin fin cross-section (m) cu copper
Pr Prandtl number f fluid
Q total heat dissipation (W) fin square fin
q heat flux (W m2) i thermocouple (i = 1–4)
Qf volume flow rate (L/h) in inlet
Re Reynolds number based on the fin size m average
Rec Reynolds number based on the channel size max maximum
Rth heat resistance (W °C1) min minimum
SL streamwise distance between pin fins (m) out outlet

with six mass velocities for each inlet temperature to test the char- liquid singe-phase flow in intermediate pin-fins (pin-fin height-
acters of micro pin fin. Their data showed that the pressure drop to-diameter ratio of 6) with 559  559 lm2 or 445  445 lm2
was over predicted by previous correlations, and the average cross section by 3 mm height; 2, to develop new heat transfer
Nusselt number increased with the average Reynolds number. and pressure drop correlations for the micro pin fins and to com-
They developed new correlations to predict the pressure drop pare with the previous.
and the heat transfer in the micro pin fin heat sink, based on the
thermal performance, hydraulic performance, and the cost of
manufacturing. 2. Description of the experimental setup and test section
Jasperson et al. [21] also compared the micro channel and micro
pin fin heat sinks. It is found that the average convection thermal 2.1. Experimental setup
resistance decreased with the flow rate for the micro pin fin heat
sink, while it was almost independent on the flow rate in the micro The schematic diagram of experimental setup, which consists of
channel heat sink. The pressure drop in the micro pin fin heat sink a cooling system, a heating and control unit, a date collecting &
was larger than that in the micro channel heat sink when the flow processing unit, is shown in Fig. 1. In the cooling system, a low
rate was fixed. temperature sink was used to keep the inlet temperature of the
An effective design and a performance assessment of a micro experimental section at a fixed value. Deionized water, which
pin fin heat sink require a fundamental understanding of the fluid was driven from the low temperature sink, flowed through a ripple
flow and heat transfer aspects of micro pin fins arrays. Most exist- damper and a 50 lm filter, and then was shunted into two parts by
ing studies were empirical due to the complex nature of the fluid three way valves. The one part passed though the flowmeter, and
flow and heat transfer in micro pin fins. Therefore, many correc- later flew into the micro pin fin test section. After taking away
tions were proposed to provide valuable insight into their flow the heat dissipation, it mixed with the other part and returned to
and heat transfer aspects of micro pin fins. However, no one correc- the low temperature sink.
tion is conclusive. The heating and control unit was designed to heat the test sec-
The heat transfer and pressure drop performance of pin fins tion. The key part of this unit was the heat source simulator. The
could be affected by many factors. However, pin fin height-to- heat source simulator was made of a pure copper block and heated
diameter ratio, operating condition might be the most important by eight 300W power cartridge heaters which were inserted from
ones. It is clear that, for micro pin fins, the larger the flow rate, the button of the copper block (Fig. 2). Those cartridge heaters
the lower convection thermal resistance. In engineering applica- were controlled by an adjustable power meter. The top of copper
tions, in sake of higher heat dissipation, a larger flow rate and high- block was trapped to 20  20  20 mm cube, while the lower part
er power pump are needed. As we known, the pressure drop would of copper is 80 mm in diameter and 110 mm in height. To mini-
be smaller if we get higher micro pin fins. This paper expands the mize heat loss and to keep the 1-D heat transfer characteristic in
micro pin fin studies by investigating the pressure drop and heat the direction perpendicular to the top surface, all the surfaces other
transfer in staggered micro square high pin fins. The objectives than the top one were insulated by Aspen Aerogels insulation
are: 1, to provide new heat transfer and pressure drop data for (thermal conductivity, k = 0.012 W/m K). This heating and control
5604 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611

2.2. Micro pin fin heat sink test section

The micro pin fins heat sink (shown in Fig. 3), which consisted
of a cover plate, a housing, a micro pin fins heat sink and a 1-D con-
duction cube, had a top area of 2  2cm2. Two types of square mi-
cro pin fins heat sinks with an array of 625 staggered micro square
pin fins in 3 mm height were fabricated on the top of pure coppers.
The top view of the micro square pin fins is shown on the left side
of Fig. 3. The dimension of micro-fin-pin was measured by SEM
(scanning electron microscopy). Fig. 4 shows the SEM pictures of
the two heat sinks. The dimension based on the SEM measure-
ments are given in Table 1. Below the micro square pin fin heat
sink, four type-T thermocouples, which were named T1–T4, were
uniformly inserted into the cube along the flow direction. On the
bottom side of the micro pin fin heat sink, a 1-D conduction cube
was made to reduce the temperature errors. Before being fabri-
cated, the heater was numerically studied using a three-dimen-
sional heat diffusion model with the software FLUENT. It was
Fig. 1. Schematic diagram of experimental setup.
found that the heater exhibited a good one-dimensional heat
transfer characteristic [22].
The cover plate was also made by pure copper to avoid the fluid
leakage due to different material thermal stresses. The housing
contained both inlet and outlet tanks of square micro pin fin arrays.
Two type-T thermocouples were fixed in the inlet and outlet tank
for temperature collection. Two pressure transducers were placed
on the inlet/outlet tubes for pressure data measurement. The cover
plate and the housing were bolted on the top of the micro pin fins
to form a closed flow passage for the water. To make the experi-
mental data more accurately, the surface of square micro pin fins
heat sink was also insulated by Aspen Aerogels insulation (thermal
conductivity, k = 0.012 W/m k).

3. Data reduction and uncertainty analysis

3.1. Flow friction

According to the structures of micro pin fin heat sink, the max-

imum flow velocity in fin channel is obtained by

Qf
Fig. 2. Heat source simulator. umax ¼ ; ð1Þ
Amin

where Amin is the minimum transverse flow area of square mi-

unit can supply the different constant heat flux by shift adjustable cro pin fins which was calculated by:
power meter.
The data collecting & processing unit, composed of pressure W c = sin 45
transducers, thermocouples, an Agilent 34970A digital acquisi- Amin ¼ WH : ð2Þ
2ST
tion/control system, flowmeters and a personal computer, was
adopted to collect the temperature and pressure data in the micro The Reynolds number based on the minimum channel width is
pin fins heat sink. given by:

Fig. 3. The structure of square micro pin fin heat sink.

 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611 5605

Fig. 4. SEM pictures of two square micro pin fins arrays. (a) 559  559 lm2 cross section (type 1) (b) 445  445 lm2 cross section (type 2).

Table 1
Dimensions of micro square pin fin heat sink (Unit: mm).

No. Length (L) Width (W) Height (H) C width (Wc) F width (Wfin) ST SL
Type 1 20 20 3 0.241 0.559 0.5657 0.5657
Type 2 20 20 3 0.355 0.445 0.5657 0.5657

qumax Dc Q cp mðT out  T in Þ

Rec ¼ : ð3Þ q¼ ¼ : ð11Þ
lf A1 A1

Based on Eq. (3), the Reynolds numbers are the same for the two The average temperature of water and the temperature of chan-
heat sinks when the flow rate is fixed. nel bottom are defined as:
The minimum channel width is determined by T in þ T out
Tm ¼ ð12Þ
2HW c =sin45 2
Dc ¼ : ð4Þ
H þ W c =sin45 and
The pressure transducers are set on the inlet/outlet of heat T 1 þ T 2 þ T 3 þ T 4 ql
Tw ¼  : ð13Þ
sinks, the pressure drop measured by these transducers is com- 4 kcu
posed of the inlet/outlet pressure loss (pd) and channel pressure
The average heat transfer coefficient of micro square pin fins is
drop (Dp). Therefore, the channel pressure drop is calculated by:
evaluated from:
Dp ¼ pin  pout  pd ; ð5Þ
Q
hav e ¼ ; ð14Þ
where the inlet/outlet pressure drop is obtained by: ðA1  Nt W 2fin þ gfin Nt Afin ÞðT w  T m Þ
pd ¼ 0:15qu2in þ 0:1qu2out ; ð6Þ
where gfin represents the fin efficiency:
Qf
uin ¼ ; ð7Þ tanhðmfin Hfin Þ
0:0022 p gfin ¼ ð15Þ
mfin Hfin
Qf
uout ¼ : ð8Þ and mfin is the fin efficiency coefficient:
HW
The flow friction is calculated by [23]:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
hav e P fin
mfin ¼ ð16Þ
Dp kcu W 2fin
f ¼ qu2max
ð9Þ
Nð 2
Þ
The Reynolds number and Nusselt number based on the fin
dimension are calculated by the following equations respectively:
N is the row number of pin fins.
The maximum ratio of fin/channel is: qumax De
Re ¼ ; ð17Þ
2ST lf
n¼ :
W c = sin 45
hav e De
Nuav e ¼ : ð18Þ
3.2. Heat transfer kf

The fin dimension is obtained by:

The heat dissipation by the micro square-pin-fins is related to pffiffiffi
the temperature difference between inlet and outlet, which were De ¼ 2W fin : ð19Þ
measured by thermocouples.
The heat resistance of heat sink is composed of two parts: the
Q ¼ cp mðT out  T in Þ ð10Þ conduction resistance through the copper and the effective convec-
tion resistance. In this study, those two parts are combined to gain
and the heat flux is expressed as following: the total resistance:
5606 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611

Table 2 5
Experimental uncertainties.
4
Parameters Uncertainty (%) Parameters Uncertainty (%) Type 1
3 Type 2
Fluid temperature 2 Flow rate 4 developed flow
Equivalent diameter 2 Flow velocity 6.6
Heat quantity 6.3 Friction resistance 12.8 2
Re 6.9 f Rec 9.4
Nuave 10.6

f
1

Tw  Tm
Rth ¼ : ð20Þ
Q

3.3. Uncertainty analysis

100 200 300 400 500600
All measurements in the experiment have their maximum er- Rec
rors. The data acquired by the thermocouples were compared with
a precision mercury thermometer with ±0.1 °C rated accuracy in Fig. 6. The friction factor vs Rec number.

the range of 0–300 °C, and the maximum error is 2%. The rotameter
was calibrated by weighing on an electronic scale with an accuracy
of 0.1 g before the experiment, and its maximum error is 4%. The channel heat sinks. However, the disturbed effect makes the pres-
accuracy of the channel width was measured by SEM and the sure drop characteristics in micro square pin fin heat sink similar
max error is ±0.001 mm. to the disturbed flow. Comparison between the type 1 and type 2
According to the traditional methods of error analysis, the heat sinks results shows that the pressure drop is inversely propor-
uncertainty can be calculated by [24]: tional to the size of the fin-pitch for a fixed Reynolds number (Rec),
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi which is due to the fact that a smaller fin-pitch results in a faster
u N 
uX @f 2 flow velocity, and thus a stronger vortex and a higher pressure
ry ¼ t r2xi : ð21Þ
@ðxi Þ drop. For square cylinder cross flow, the vorticities across wake
i1
are affected by streamwise velocity distribution. The magnitude
For example, the uncertainty of Nuave, determined by the uncer- of the vorticity peak largely depends on the cylinder width and free
tainties of Q, A1, Wfin, Afin, Tw, Tm, De using Eqs. (14) and (18), is stream velocity magnitude, thus, a faster flow velocity results in a
10.6%. stronger vortex. On the other hand, the drag due to the square cyl-
The details of measurement errors are listed in Table 2. inder is proportional to velocity square, thus, a faster flow velocity
results in a higher pressure drop.
According to the definition, when the flow rate Qf are kept the
4. Results and discussions
same for both heat sinks, the Rec remains the same, while the flow
friction factors are different due to the different geometry of micro
4.1. Flow friction
pin fins. Therefore, the relationships between flow friction f and Rec
for the two heat sinks are different as shown in Fig. 6. It also shows
Fig. 5 shows relations between the pressure drop and the Rec
that the laminar turbulent transition with respect to the Rec num-
number of two micro pin fin heat sinks, which are qualitatively dif-
ber occurs at above 300, earlier than that in the convention tube
ferent from that of the Poiseuille flow in tubes. When the fluid
flow, due to the staggered arrays structures in the heat sinks which
flows through heat sink, it flows across fins in the heat sinks, lead-
cause the fluid mixing between channels and the vortex evolving
ing to the mixing of the fluid between channels. The vortex is also
behind the fins more intensively and frequently. When Rec is less
evolving behind the fins, resulting in a more complex characteris-
than 300, the flow friction factors of the two micro pin fin heat
tics of the pressure drop in micro fin heat sinks than in micro
sinks in Fig. 6 have a similar pattern, decreasing with the Rec, for
the full development laminar flow. After Rec crosses 300, they tend
to be constants. It is very interesting to find that the friction factor
5 of type 2 is smaller than that of type 1 for Rec less than 200 and lar-
ger for Rec greater than 200. For laminar flow and evenly distrib-
Type 1
Type 2 uted pin fins, the friction factor can be given as 180 ð e Þ0:4 , where
Re 1e
4
e is porosity [16]. As expected, the friction factor decreases with
increasing of Reynolds number and porosity. For turbulent flow,
ΔP(kPa)

3 Daloglu [25] found that there exists a critical spacing (the distance
between two cylinder is about 1  2d, where d is cylinder diame-
ter) at which the pressure drop takes a minimum value, which is
2
even less than the pressure drop obtained for the case of in the
channel with single cylinder. In this study, the ratio of the neigh-
1 boring pin fins distance and pin fin hydraulic diameter is about
1. Probably, this is the main reason that the friction factor of type
2 greater than that of type 1 when Rec is larger than 200.
0
0 100 200 300 400 500 To predict the characteristics of flow friction in present micro
Rec square pin fin heat sinks, two correlations are proposed:

Fig. 5. The pressure drop vs Rec number. f ¼ 43:322Re0:9116

c n0:9362 ð0 < Rec < 300Þ ð22Þ
 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611 5607

1 3.5E+06
o
Type 1 T w=30 C
o
Type 1 Type 1 T w=50 C
Type 2
3E+06 Type 1
o
T w=70 C
Type 2 T w=30 oC
0.5 Type 2 T w=50 oC
2.5E+06 Type 2
o
T w=70 C

+30%

q(W/m )
MRE(%)

2E+06

2
0
1.5E+06
-30%
Corre 1 Corre 2 1E+06
-0.5
500000

-1 0
0 100 200 300 400 500 0 10 20 30 40 50 60
Rec Q f (L/H)

Fig. 7. The mean relative error along the Rec number. Fig. 9. The heat flux vs flow rate.

and transfer coefficients for a constant flow rate in the single phase
heat transfer and, thus, a stronger heat dissipation under the con-
f ¼ 1:6361Re0:01076
c n0:94496 ð300 < Rec < 550Þ; ð23Þ stant flow rate for a bigger temperature difference between the
with the first one denoted as Corre 1 and the second one Corre 2. heat surface and the fluid. The heat dissipation of 2.50  106 W/
The mean relative error (MRE), which is used to estimate the er- m2 was observed with the flow rate of 57.225 L/h and the surface
ror of the two correlations with respect to the experimental data, is temperature of 57.1 °C for the type 1 heat sink. Under the same
defined as: flow rate condition, the heat dissipation was about 2.83  106 W/
m2 with the surface temperature of 73.4 °C for the type 2 heat sink.
1 X jfav e;exp  fav e;pred j These observations suggest that a micro square pin fin heat sink
MRE ¼  100%; ð24Þ
M fav e;exp could have a huge cooling capacity and therefore the potential to
meet the demand of high power density electric devices. Due to
where M is the total number of data points. Fig. 7 shows the com- larger heat transfer surface, it is found that heat dissipation for
parison of the experimental data and two correlations. It is found heat sink 1 is higher than heat sink 2.
that the maximum errors of these two correlations are 30.5% and The relation between the heat dissipation q and the flow rate Qf
10.4%, respectively. In this study, the pin fin height-to-width ratio for fixed surface temperatures in the type 1 and type 2 heat sinks is
ranges from 6–8, thus, the flow in the heat sink is highly three- shown in Fig. 9. For a fixed surface temperature of a heat sink,
dimensional. On the basis of first order channel ratio, the present increasing the flow rate enhances the heat dissipation, which im-
friction factor correlations may have pin fin height-to-width ratio plies the increase of the temperature gradient along fins and thus
limitation. reduces the fin efficiency. Therefore, the increasing rate of the heat
dissipation with respect to the flow rate decreases. Given the same
4.2. Heat transfer surface temperature and total flow rate for both heat sinks, the
type 1 heat sink is supposed to have a larger velocity than the type
The variations of q with Tw for different flow rates in the type 2 heat sink, which implies a larger heat transfer coefficient accord-
1and type 2 heat sinks are shown in Fig. 8. The heat dissipation q ing to the classical heat transfer theory, therefore, a larger heat dis-
increases with the surface temperature Tw for a fixed flow rate sipation. This is the reason for the difference of the heat dissipation
since the classical heat transfer theory predicts a constant heat between the type 1and type 2 heat sinks increases with the flow

3.0E+06
Type 1 Q f=5.693L/H 16
Type 1 Q f=19.075L/H 14
Type 1 Q f=57.225L/H
2.5E+06 Type 2 Q f=5.693L/H 12 Type 1
Type 2 Q f=19.075L/H Type 2
Type 2 Q f=57.225L/H 10
2.0E+06
8
q(W/m2)

Nuave

1.5E+06 6

1.0E+06
4

5.0E+05

0.0E+00
20 40 60 80 2
o
200 400 600800
Tw( C) Re

Fig. 8. The heat flux vs surface temperature. Fig. 10. The Nusselt number vs Reynolds number.
5608 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611

rate and the surface temperature, corroborating the finding about to the laminar to turbulent transition with the Rec number above
the heat transfer coefficient as shown in Fig. 8. 300 in the two heat sinks. For the laminar flow (Re less than
Fig. 10 plots the variation of Nuave with the Reynolds number on 300), the larger channel size of the type 1 heat sink, compared to
a log-log scale for the type 1 and type 2 heat sinks. It shows that the type 2, results in a smaller flow velocity for a fixed Reynolds
Nuave increases with Reynolds number for both heat sinks, which number and thus a smaller heat transfer coefficient, which in turn
might be described by power lines of Nuave / Rem. The Nuave of leads to a smaller Nuave. As the Reynolds number increases to the
the type 2 heat sink is larger than that in the type 1 when the Rey- range of turbulence, the formation of coherent structures may be
nolds number is small (less than 300). However, as the Reynolds shed and advected downstream. The shed structures are primarily
number increases, they tend to be identical. This is probably due responsible for wake mixing properties. Thus, enhanced fluid

Table 3
Correlations for heat transfer in pin fin heat sinks.

No. Reference Fin types Height/width ratio Fluid type Reynolds number Average Nusselt number MRE (%)
1 Zukauskas et al. [9] Convention – Air 10–100 Nuav e ¼ 0:9Re Pr0:36
0:4 0:25 136.6
f ;av e ðPrf ;av e =Prw;av e Þ
2 Short and Raad [12] Convention <7.5 Air >1000 Nuav e ¼
1=3
0:76Re0:33 Prf ;av e ðSL =de Þ0:16 ðST =de Þ0:2 ðH=de Þ0:11 24.1
3 Kosar et al. [15] Micro 4.1 R-123 130–320 121.6
0:99 0:25
Nuav e ¼ 0:0432Re Pr0:21
f ;av e ðPrf ;av e =Prw;av e Þ

4 Qu and Siu-Ho [19] Micro 3.35 Water 45.9–179.6 0:25 41.4

Nuav e ¼ 0:0241Re0:953 Prf0:36
;av e ðPrf ;av e =Prw;av e Þ
5 Present study Micro 6–8 Water 0–800 Nuav e ¼ 0:143Re
1=3
0:615
Prf ;av e 11.0
6 Present study Micro 6–8 Water 0-800 Nuav e ¼ 0:1245Re0:6106
Pr0:36 0:25 11.2
f ;av e ðPrf ;av e =Prw;av e Þ

50 50
40 40 +20%

30 30

20 -20% -20%
20
Nu ave,pred
Nuave,pred

10 10

Type 1 Type 1
Type 2 Type 2
Zukauskas[10] Zukauskas[10]
+20%
Correlation 1 Correlation 3
MRE=136.6% MRE=121.6%

10 20 30 40 50 10 20 30 40 50
Nuave,exp Nuave,exp

(a) (c)
50 50
40 +20% 40 +20%

30 30

20 -20% -20%
20
Nuave,pred

Nuave,pred

10 10

Type 1 Type 1
Type 2 Type 2
Zukauskas[10] Zukauskas[10]

Correlation 2 Correlation 4
MRE=24.1% MRE=41.4%

10 20 30 40 50 10 20 30 40 50
Nuave,exp Nuave,exp

(b) (d)

Fig. 11. Comparison of the average Nusselt number data with predictions of previous correlation.
 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611 5609

mixing and the coherent structures effect impel the heat transfer coef-
ficient increase. The Nuave of those two heat sinks tend to be equal. 15 +20%
In the classic heat transfer theory [26], the Nuave number is a Type 1
Type 2
function of the Reynolds number and the Prandtl number when
10
the shape of apparatus is fixed, written as: -20%
m
Nuav e ¼ cRe Prnf;av e ð25Þ

Nuave,pred
and
5
Nuav e ¼ cRem Prnf;av e ðPrf ;av e =Prw;av e Þl : ð26Þ

Prf,ave in Eqs. (25) and (26) is the water Prandtl number at Tm,
and Prw,ave in Eq. (26) is water Prandtl number at Tw. Correlation 5
MRE=11.0%
There have been lots of correlations for the single-phase heat
transfer in pin-fin arrays. Four previous typical correlations are se-
5 10 15
lected and summarized in Table 3. the correlation 1–2 were based
Nuave,exp
on the heat transfer data from conventional pin fin arrays, and the
correlation 3–4 were based on the heat transfer data from micro Fig. 12. Comparison of the average Nusselt number data with predictions of
pin fin arrays. correlation 5.
To compare the previous correlations with the present data, the
average relative error (MRE) of Eqs. (25) and (26) are calculated as

1 X jNuav e;exp  Nuav e;pred j +20%

MRE ¼  100% ð27Þ 15
M Nuav e;exp
Type 1
Type 2
where M is the total number of data points.
10
Present Nuave data are compared with previous predictions in
-20%
Fig. 11(a)–(d). It is found that the MRE of the correlation 1-4 with
respect to the present data ranges from 23.1%–136.6% as listed on Nuave,pred
Table 3. It should be noted that the correlations from Short et al.
[12] and Zukauskas et al. [9] are for air, while from Kosar et al. 5
[15] is for R-123. It is very interesting that Short’s [12] correlation,
which is for air flow at Re > 1000 flow condition, gives minimum
MRE. The deviations increase with the Nuave number for the corre- Correlation 6
lations of the convention pin fin arrays, however, decrease for MRE=11.2%
those of the micro pin fin arrays. The relative variation in the slopes
of the correlations in Table 3 and Fig. 11 may be due to one dimen- 5 10 15
sional correlations for a highly three dimensional flow. The flow, Nuave,exp
thus heat transfer performance, is not only affected by pin fin con-
figuration, but also by height-to-width ratio and longitudinal and Fig. 13. Comparison of the average Nusselt number data with predictions of
transverse pitches of fin pin. As early in 1972, Zukauskas et al. correlation 6.
[9] found that increase of m (in Eq. (25)) with a constant longitudi-
nal and a decreasing transverse pitch. Therefore, it is concluded
and l in Short’s [12] correlations are both 1/3, while they are 0.36
that the m in correlations developed for highly three dimensional
and 0.25 in Zukauskas et al. [9] and Qu and Siu-Huo’s [19] correla-
flows should be a function of height-to-width ratio and longitudi-
tions. In our study, we assume n is 1/3 in Eq. (25), while the n and l
nal and transverse pitches of fin pin.
in Eq. (26) are 0.36 and 0.25, respectively. The two new correla-
From Table 3 and Fig. 11, it is found that the previous heat
tions are
transfer correlations may not be able to fit the present data of
the micro square pin fin heat sinks. The reasons for poor prediction Nuav e ¼ 0:143Re0:615 Prf1=3 ð28Þ
;av e
performance of previous heat transfer correlations include: (1) Ef-
fect of fluids property. Though the use of the Prandtl number im- and
plies a more general application, it does not guarantee that these
0:25
correlations apply to general gas or liquid flows; (2) Effect of geo- Nuav e ¼ 0:1245Re0:6106 Pr0:36
f ;av e ðPrf ;av e =Prw;av e Þ : ð29Þ
metric configuration and pin fin spacing. Both geometric configura-
tion and pin fin spacing affect fluid flow inside heat sinks. (3) The The two correlations are also listed on Table 3 for comparison.
height-to-width ratio of diamond shaped pin fins in this study is Fig. 12 and Fig. 13 compare the new correlations with respect to
about 6–8, the flow is highly three dimensional. That may be the the present data, which have the MRE of 11.0% and 11.2%, respec-
main reason for different critical Reynolds and heat transfer perfor- tively, acceptable for experiments.
mance. Thus, the discrepancy between the previous correlation
predictions and the present data is not necessarily related to the 4.3. Heat resistance
weaknesses of the correlations themselves but maybe to the
geometric parameters and operating conditions. Therefore, we To predict the performance of heat sinks, the variation of heat
propose two new heat transfer correlations to fit these types resistance with heat dissipation of the type 1 and type 2 heat sinks
micro-square pin fin heat radiator‘s heat transfer performance. are shown in Fig. 14. It is found that when the flow rate was
In present study, the working fluid is water. Therefore, the 47.688 L/h, the heat resistances are nearly fixed values, invariant
range of prandtl number is not wide enough to change the expo- with respect to the heat dissipation. However, distinguishable dif-
nents n and l of Eqs. (25) and (26). As shown in Table 3. The n ference between the heat resistances of the two heat sinks are
5610 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611

1.8 the coolant liquid convection resistance. When the pressure drop
Type 1
is less than 1.1 kPa, due to weak convection, the conduction resis-
Type 2 tance dominates. Hence, the heat resistance of the type 2 is smal-
1.6 Type 1 FIT Q f=47.688L/H
Type 2 FIT
ler. When Rec increases, the convection resistance becomes
1# MRE=6.7% dominant. As a result, the effective resistance of type 1 becomes
1.4 2# MRE=4.6%
smaller.
R th( C/W)

It should be pointed out that the heat resistance is much bigger

1.2 than that of Prasher et al. [23]. This is due to the difference in the
o

pin-fin heights. The pin height in our study is 3 mm, larger than
0.2–0.31 mm in Prasher’s study. The higher pin height, the lower
1
the fin efficiency gf. Hence, the enhancement in the heat transfer
for micro pin fin is not necessary due to the increase of the area.
0.8 As Prasher said, micro pins do more than just increasing the area.

0.6
0 1E+06 2E+06 3E+06 4E+06 5. Conclusions
Q(W)

Fig. 14. The heat resistance vs the heat dissipation. In this study, two micro staggered square high pin fin heat sinks
with different channel sizes were fabricated. Using deionized
water as working fluid, the performance of pressure drop and heat
4 transfer in staggered square long micro pin fins were experimen-
tally studied. The main conclusions include:

1) For both heat sinks, the pressure drop increased with the Rec
3 Type 1 number. The flow friction factor transition phenomenon
Type 2
appeared at Rec  300.
2) Both heat sinks exhibited huge heat dissipation capability.
R th ( C/W)

The experimental data showed that, for the type 2 heat sink,
o

2 the heat dissipation could reach 2.83  106 W/m2 at the flow
rate of 57.225 L/h and the surface temperature of 73.4 °C,
and, therefore, meet the demand of high power heat
removal. The heat dissipation increased with the flow rate
1 for a fixed surface temperature while the increasing rate
decreased with the flow rate.
3) The Nusselt number increased with the fin Reynolds num-
1 2 3 4 5 ber. For both heat sinks, the heat transfer was over predicted
Δ P(kPa) by the previous correlations. Therefore, we proposed new
Fig. 15. The heat resistance vs the pressure drop.
correlations for the average Nusselt number prediction.
The Nusselt number varies as Re0.61.
4) The heat resistance decreased with the pressure drop. The
observed. This is due to the same reason mentioned above, i.e., a deceleration rate was faster for the small pressure drop
smaller width of flow path in the type 1 resulting a larger flow and slower for the large pressure drop. The type 1 heat sink
velocity for a given same flow rate. has larger heat resistances for small pressure drop (del-
Therefore, assuming the heat resistance is invariant with re- ta_P < 1.1 kPa), but smaller ones for large pressure drop (del-
spect to the heat dissipation, the heat resistance is a function of ta_P > 1.1 kPa), than the type 2 heat sink.
flow rate only. It is clear that the increase of the flow rate corre-
sponds to an increase of pressure drop. To evaluate the heat trans-
fer and pressure drop performance, the variation of heat resistance
with pressure drop of different flow rate in type 1and type 2 heat Acknowledgments
sinks is presented in Fig. 15.
For both heat sinks, the heat resistance decreases with the in- The authors gratefully acknowledge the support of the CAS Spe-
crease of the pressure drop. The decreasing rate is inversely pro- cial Grant for Postgraduate Research, Innovation and Practice.
portional to the pressure drop. The type 1 heat sink has larger
heat resistances for small pressure drop (delta_P < 1.1 kPa), but
References
smaller ones for large pressure drop (delta_P > 1.1 kPa), than the
type 2 heat sink. [1] A. Allan, D. Edenfeld, Technology Roadmap for Semiconductors, in: The
As shown in Table 1, the type 1 heat sink has a flow path width International Technology Roadmap for Semiconductors, 2001.
[2] D.B. Tuckerman, R.F.W. Pease, High performance heat sinking for VLSI, IEEE
(Wc) of 0.241 mm and the pin fins width (Wfin) of 0.559 mm, while
Electr. Device Lett. EDL-2 (5) (1981) 126–129.
the flow path and the pin fins width of the type 2 heat sink are [3] S.V. Garimella, C.B. Sobhan, Transport in microchannels – a critical review 13
0.335 mm and 0.445 mm, respectively. For the two heat sinks, (13) (2003) 1–50.
[4] G.L. Morini, Single-phase convective heat transfer in microchannels: a review
the pressure drop of 1.1 kPa corresponds to the Reynolds number
of experimental results, Int. J. Therm. Sci. 43 (7) (2004) 631–651.
of 102 for the type 1 and 164 for the type 2 (based on length scale [5] I. Hassan, P. Phutthavong, M. Abdelgawad, Microchannel heat sinks: an
Wc), respectively. This implies, with the same pressure drop, the overview of the state-of-the-art, Microscale Thermophys. Eng. 8 (2004) 183–
velocity of coolant liquid in the type 2 heat sinks is faster due to 205.
[6] J.L. Xu, Y.H. Gan, D.C. Zhang, X.H. Li, Microscale heat transfer enhancement
the smaller flow path width (Wc). In this study, the heat resistance using thermal boundary layer redeveloping concept, Int. J. Heat Mass Transfer
is an effective one, which includes the conduction resistance and 48 (9) (2005) 1662–1674.
 M. Liu et al. / International Journal of Heat and Mass Transfer 54 (2011) 5602–5611 5611

[7] J.-T. Liu, X.-F. Peng, W.-M. Yan, Numerical study of fluid flow and heat transfer [18] W. Qu, Comparison of thermal-hydraulic performance of singe-phase micro-
in microchannel cooling passages, Int. J. Heat Mass Transfer 50 (9-10) (2007) pin-fin and micro-channel heat sinks, in: Thermal and Thermomechanical
1855–1864. Phenomena in Electronic Systems, 2008. ITHERM 2008. 11th Intersociety
[8] R. Muwanga, I. Hassan, Flow and heat transfer in a cross-linked silicon Conference on, 2008, pp. 105–112.
microchannel heat sink, J. Thermophys. Heat Transfer 22 (3) (2008) 333–341. [19] W. Qu, A. Siu-Ho, Liquid single-phase flow in an array of micro-pin-fins–part I:
[9] A. Zukauskas, P.H. James, Thomas F. Irvine Jr., Heat transfer from tubes in heat transfer characteristics, J. Heat Transfer 130 (12) (2008) 122402–122411.
crossflow, in: Advances in Heat Transfer, Elsevier, 1972, pp. 93–160. [20] W. Qu, A. Siu-Ho, Liquid single-phase flow in an array of micro-pin-fins–Part II:
[10] J. Armstrong, D. Winstanley, A review of staggered array pin fin heat transfer pressure drop characteristics, J. Heat Transfer 130 (12) (2008) 124501–
for turbine cooling applications, J. Turbomach. 110 (1) (1988) 94–103. 124504.
[11] B.E. Short Jr., P.E. Raad, D.C. Price, Performance of pin fin cast aluminum [21] B.A. Jasperson, J. Yongho, K.T. Turner, F.E. Pfefferkorn, Q. Weilin, Comparison of
coldwalls, J. Thermophys. Heat Transfer 16 (3) (2002) 389–396. micro-pin-fin and microchannel heat sinks considering thermal-hydraulic
[12] B.E. Short Jr., P.E. Raad, D.C. Price, Performance of pin fin cast aluminum performance and manufacturability, IEEE Trans. Components Packag. Technol.
coldwalls, Part 2: Colburn j-factor correlations, J. Thermophys. Heat Transfer 33 (1) (2010) 148–160.
16 (3) (2002) 397–403. [22] Y.Q. Wang, M.H. Liu, D. Liu, K. Xu, Yiliang Chen, Experimental study on the
[13] K.A. Moores, Y.K. Joshi, Effect of tip clearance on the thermal and effects of spray inclination on water spray cooling performance in non-boiling
hydrodynamic performance of a shrouded pin fin array, J. Heat Transfer 125 regime, Exp. Therm. Fluid Sci. 34 (2010) 933–942.
(6) (2003) 999–1006. [23] R.S. Prasher, J. Dirner, J.-Y. Chang, A. Myers, D. Chau, D. He, S. Prstic, Nusselt
[14] C. Marques, K.W. Kelly, Fabrication and performance of a pin fin micro heat number and friction factor of staggered arrays of lowaspect ratiomicropin-fins
exchanger, J. Heat Transfer 126 (3) (2004) 434–444. under cross flow for water as fluid, J. Heat Transfer 129 (2) (2007) 141–153.
[15] A. Kosar, C. Mishra, Y. Peles, Laminar flow across a bank of low aspect ratio [24] R.B. Abernethy, R.P. Benedict, R.B. Dowdell, ASME measurement uncertainty,
micro pin fins, J. Fluid. Eng. 127 (3) (2005) 419–430. ASME J. Fluids Eng. 107 (1985) 161–164.
[16] Y. Peles, A. Kosar, C. Mishra, C.-J. Kuo, B. Schneider, Forced convective heat [25] A. Daloglu, Pressure Dro in a channel with cylinders in tandem arrangement,
transfer across a pin fin micro heat sink, Int. J. Heat Mass Transfer 48 (17) Int. Commun. Heat Mass Transfer 35 (2008) 76–83.
(2005) 3615–3627. [26] F.P. Incrperea, D.P. Dewitt, T.L. Bergman, A.S. Lavine, Fundamentals of Heat and
[17] A. Siu-Ho, W. Qu, F. Pfefferkorn, Experimental study of pressure drop and heat Mass Transfer, 6th ed., John Wiley and Sons Inc, New York, 2006.
transfer in a single-phase micropin-fin heat sink, J. Electron. Packag. 129 (4)
(2007) 479–487.