CONJUGATE HEAT TRANSFER OF JET IMPINGEMENT

ON SHORT FINS WITH DIFFERENT SHAPES

X. C. Li
Department of Mechanical Engineering
Lamar University
Beaumont, TX 77710

ABSTRACT Greek
H turbulence dissipation rate (m2/s3)
This study examines numerically details of the flow and
temperature fields of heat sinks with short plate fins cooled by O heat conductivity (W/m-K)
impinging jet. The main focus is on the effect of fin shapes on P dynamic viscosity (kg/m-s)
the heat sink performance. Conjugate heat transfer between Q kinematic viscosity (m2/s)
airflow convection and conduction inside the fin and base is U density (kg/m3)
considered. Three different shapes under study are rectangular, W stress tensor (kg/m-s2)
round-headed and elliptic, while the fin height keeps relatively
short. Reynolds number or jet velocity is spanned to cover Subscript
both laminar and turbulent flows. Distributed heat transfer j jet flow
coefficient is presented and compared. It is observed that the t turbulent
fin shapes affect largely both the heat transfer and pressure w wall
drop, especially of the short fin cases. Based on the same fin
height, the total heat transfer with rectangular fins is 1.4 times INTRODUCTION
that with elliptic fins in one of the cases studied, while the Extended surfaces (fins) and impinging jets have commonly
surface area ratio is 1.1. For longer fins, the shape effect on been used to enhance heat transfer in many applications, for
heat transfer becomes relatively less important than that on example, electronic cooling and gas turbine cooling. In
pressure drop. The results of this paper can help in design of electronic thermal management, heat sinks are designed to
heat sinks with jet impingement, which are commonly used in take advantage of the combined effect of fins and jet
electronic cooling. impingement such as jets impinging on an array of pin fins or
plate fins. Significant studies have been focused on the
KEYWORDS: Jet impingement, heat sink, plate fin thermal resistance, pressure drop, and the parametric effect of
Reynolds number, fin thickness, density and height. Different
NOMENCLATURE correlations based on one-dimensional heat conduction and
B distance between fins (m) experimental data are developed to predict the heat sink
b fin width (m) performance. Efforts to optimize the heat sink design have
k turbulence kinetic energy (m2/s2) also been made by using both experimental and numerical
H height of computational domain (m) methods. Some related research is summarized as follows.
h1 fin height (m)
h2 base thickness (m) Experimental Study on Plate-Fin Heat Sinks with Jet
h convective heat transfer coefficient (W/m2-K) Impingement
l1 half width of the jet (m) Jang and Kim [1] experimentally investigated the fluid flow
l2 confined wall length (m) and heat transfer characteristics of a microchannel heat sink
Nu Nusselt number (hDh/O) subject to an impinging jet. The cooling performance of a
P pressure (N/m2) microchannel heat sink subject to an impinging jet was shown
Pr Prandtl number superior to that of the microchannel heat sink subject to a
q” heat flux (W/m2) parallel flow. Duan and Muzychka [2] measured cooling of
Re Reynolds number (uDh/Q  plate fin heat sinks with various impingement inlet widths, fin
S source term spacings, fin heights, and airflow velocities. Lin et al. [3]
T temperature (K) conducted a series of experimental studies on transient- and
u streamwise velocity component (m/s) steady - state cooling performance from the horizontally
v spanwise velocity component (m/s) heated heat sinks with a confined slot jet impingement. The
x, y, z coordinates (m) transient heat transfer behaviors such as the temperature

978-1-4244-1701-8/08/$25.00 ©2008 IEEE.

49 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded
distribution, local and average Nusselt numbers on the plate over 1500 W/m2K at a pressure drop of less than 100 N/m2.
fin heat sinks were explored. Sansoucy et al. [4] presented an Bhopte [12] numerically examined the effect of an impinging
experimental study on a parallel flat plate heat sink under a mixed convection air jet on the heat transfer rate of a heat sink
turbulent impinging air jet. The jet was discharged from a of parallel flat plates. The heat transfer rates from the bare
sharp-edged nozzle in the nozzle plate. Wu et al. [5] studied heat source surface were compared to the ones obtained with
experimentally the thermal performance of slot jet impinging the heat sink to determine the overall performance of the heat
onto fully- and partially-confined heat sinks with plain-plate sink in an impingement configuration. Using a plate-fin heat
fins. The parameters under study included the ratio of heat sink as the target plate, Lou et al. [13] simulated the conjugate
sink height to jet separation distance, the ratio of heat sink heat transfer under a confined impinging jet. Effects of
height to nozzle width and the jet Reynolds number. geometric parameters such as fin number, fin height, and fin-
Theoretical models based on a developing channel flow with to-spacing ratio were examined by using dielectric fluid FC-72
an impinging effect were proposed. Li and Chen [6] employed as the fluid medium. The concept of an effective Nusselt
the infrared thermography technique to study the plate-fin heat number was introduced to compute the effectiveness of heat
sinks with confined impinging jet. The results showed that the sinks with different fin numbers.
thermal resistance of the heat sink apparently decreased as the
Reynolds number increased; however, the decreasing rate of Numerical Simulation on Pin-Fin Heat Sinks with Jet
the thermal resistance declined with the increase of the Impingement
Reynolds number. An increase of the fin width reduced the Tien and Huang [14] studied numerically the feasibility and
thermal resistance initially but the thermal resistance rose effectiveness of air jet impingement cooling on pin-fin heat
sharply when the fin width was larger than a certain value. sinks applied in personal computers. Effects of fin height, fin
width, base plate thickness of the heat sink, and ratio of the
Experimental Study on Pin-Fin Heat Sinks with Jet vertical spacing between the nozzle and the heat sink to jet
Impingement diameter were discussed in detail. When the fin width
Brignon and Garimella [7] tested a variety of nozzles to increased from 4.13 to 5.3 mm, the performance by jet
characterize and optimize the performance of confined impingement was improved significantly. The cooling effect
impinging air jets used in conjunction with a pin-fin heat sink. was further enhanced when the fin height and the vertical
They found that the enhancement factors relative to a bare spacing between the nozzle and the heat sink increased. Yang
surface were in the range of 2.8-9.7 with the largest value and Peng [15] presented the simulation of pin fin heat sinks
being obtained for a single nozzle of 12.7mm in diameter. El- with confined impingement cooling. The parameters included
Sheikh and Garimella [8] experimentally investigated the Reynolds number, fin width, fin height, and number of fins. A
enhancement of heat transfer from a discrete heat source using porous model was adopted to approach the heat sink. The
multiple jet impingement of air in a confined arrangement. A results showed that increasing the Reynolds number reduced
variety of pin-fin heat sinks were mounted on the heat source. the thermal resistance. Increasing the fin width or height could
Compared to the unpinned ones, the total fin effectiveness for also decrease the thermal resistance. The effect of fin
the pinned heat sinks was in the range of 3 to 6. Kondo et al. geometry at a low Re number was more significant than that at
[9] optimized pin-fin heat sinks for impingement cooling of a high Re number.
electronic components by using a semi-empirical zonal model
to determine the thermal resistance as well as pressure drop. Current Study
The model was validated by experimental results. Issa and This study applies numerical simulation to examine the details
Ortega [10] experimentally studied the flow and heat transfer of flow and temperature fields of plate fin heat sinks with jet
behavior due to free air jet impingement on heat sinks with pin impingement. Based on the literature review above, few
fins. Flow velocities were varied from 2 to 20 m/s and the tip studies have been done on the effect of fin shapes. Therefore,
clearance ratios ranged from 0 to 1. The overall base-to- the main focus of this study is on the effect of fin shapes on
ambient thermal resistance decreased with increasing the Re the heat sink performance. Three different shapes under study
number, pin density, and pin diameter. are rectangular, round-headed and elliptic. The characteristics
of 3D temperature field inside the fins and solid base are also
Numerical Simulation on Plate-Fin Heat Sinks with Jet explored. It is observed that the fin shape affects largely both
Impingement the heat transfer and pressure drop, especially of the short fin
While experimental studies can effectively obtain the overall cases. When the fin height-to-thickness increases, the shape
thermal performance and pressure drop, the largely distributed effect on heat transfer becomes relatively less important than
heat transfer coefficient on the fin and base surfaces can be that on pressure drop. The results of this paper can help in
more conveniently simulated with numerical methods. Sathe design of heat sinks with jet impingement
et al. [11] presented the performance of air-cooled heat sink. A
rectangular jet impinged on a set of parallel fins and then PROBLEMS UNDER CONSIDERATION
turned into cross-flow. They found that pressure drop was
As discussed earlier, there are generally two types of heat
reduced by cutting the fins in the central impingement zone
sinks with jet impingement: One is with parallel plate fins, and
with less reduction of heat transfer. A combination of fin
the other is with pin fins. In this study the plate fin heat sink
thickness of 0.5 mm and channel gaps of 0.8 mm with
is considered. Figure 1 shows the geometric schemes of this
appropriate central cut-out yielded heat transfer coefficients

50 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded
type of heat sink. The slot jet (or round jet in some cases) the heat sink bottom surface for simplification. Table 1 lists
impinges from the top to the plate fins and exits from two the key parameter for the baseline case and the variation. Note
sides into the surroundings. The sink base is connected to the that the fin height-to-width ratio is 3 for the baseline case,
electronic modules to dissipate the heat to the cooling airflow. which is considered as a short fin
The top can be either open or confined, which is considered as
confined in this study. The performance depends on the jet Table 1 Parameters and cases under study
velocity, fin height, fin width, number of fins, and the base
thickness. In practice, there usually is a small gap between the Baseline Variation
fin tip and jet plate. Elliptic
Fin shape Rectangular
Round-tipped
Fin height, h1 (m) 0.006
y
Base height, h2 (m) 0.002
x Jet
Total height, H (m) 0.01
z
Outlet Fin width, b (m) 0.002 0.001
Distance between fins, B (m) 0.004 0.002
Jet velocity, vj (m/s) 10 1, 2, 15, 20
Jet temperature, Tj (K) 300
Jet width (half), l1 (m) 0.01
Base
Confined wall length, l2 (m) 0.015

Figure 1 Geometric scheme of a heat sink of parallel plate fins

cooled by an impinging jet To estimate the Reynolds number and thus the flow regime, a
length scale needs to be selected from the multiple dimensions
The heat sink in Fig. 1 is symmetric in both the x- and z- in this problem. Since the key flow is mostly confined
directions unless the flow or thermal boundary conditions are between the plates, the hydraulic diameter of the channel cross
applied differently. Figure 2 shows the symbolic dimension of section is applied, which is consistent with other studies [2].
the cases under study. Three fin shapes are considered, i.e. The hydraulic diameter for the baseline case is calculated as
rectangular fins, elliptic fins and fin with round tip. All the 0.0034m (three wet sides), which gives a Reynolds number of
fins share the same height and base width, which leads to a 2,348 when the jet velocity of 10 m/s is used. The average
slightly different surface area and thus different volume or channel velocity is larger than the average jet velocity.
weight. The total height in the y-direction is H, and heights Therefore, the flow lies in the margin of laminar and turbulent
for the fin and base are h1 and h2, respectively. The fin has a flows. Note that the Reynolds number will be much larger if
thickness of b, and the distance between the fins is B. the hydraulic diameter of the jet inlet is used as the length
Considering only half of the heat sink in the x-direction, the scale. A turbulence model is needed for the simulation. As to
impinging slot jet has a width of l1 and the rest of the top the low velocity cases (1 and 2 m/s), the flow will essentially
surface is the confined wall (l2). keep as laminar.

MODELING AND NUMERICAL PROCEDURE

As mentioned above, numerical simulation is adopted in this
study to explore details of the flow and temperature fields so
H that the thermal performance of heat sinks can be examined.
b h1
Symmetric The numerical model and procedure are briefly documented
plane
h2 below, which can be referred in many other studies

Governing Equations
B B B The 3D time-averaged steady-state Navier-Stokes equations as
well as the equations for mass and energy transport are solved
l1 l2 in this study. These equations can be given as
w (1)
u 0
wx i
i
Symmetric
plane Outlet
w * wP w (2)
u i u j g j    ij -  u' i u' j
Figure 2 Schemes of different fins and symbolic dimensions wx i wx j wx i

w w § wT · (3)
In this study, the air jet with constant properties impinges on c p u i T ¨¨  - c p u' i T' ¸¸  μ
wx i wx i © w x i ¹
the heat sink vertically with a uniform inlet velocity and
temperature. Constant heat flux or temperature is applied to where Wij is the symmetric stress tensor defined as

51 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded
 § wu j wu i 2 wu k · (4) number is higher than the specified value. The turbulent
 ij μ¨    ij ¸
¨ wx w x 3 wx k ¸ viscosities calculated from these two regions are blended to
© i j ¹
smoothen the transition.
P) is the viscous dissipation and O is the heat conductivity.
Notice that the terms of U u'i u'j and Ucpu'i T' represent the Boundary Conditions for Turbulence
Reynolds stresses and turbulent heat fluxes, respectively, In case the flow is turbulent, the inlet condition of turbulence
which need to be modeled for turbulent flows. kinetic energy and dissipation rate need to be specified. In this
study, the turbulence intensity is kept as 4% for all the cases,
Turbulence Model which is considered reasonable for flow exiting from a fan.
The k-H model is one of the most robust turbulence models.
While many turbulence models have been studied for an Meshing and Simulation Procedures
impinging jet flow [16-17], there is no solid approval to show The major meshing characteristics are shown in Fig. 3. Both
one is much better than the other. Sine this study is not structured and unstructured grids are applied while in the
seeking the best turbulence model, the standard k-H model region with curved faces the unstructured grids are mainly
with enhanced wall function is adopted. Based on the used. As also shown in Fig. 3, nonuniform grids are adopted
Boussinesq hypothesis, it relates the Reynolds stresses to the and boundary layer grids are generated to capture the details
mean velocity as close to the heat transfer surface.
§ wu wu j · 2
- u'i u' j μt ¨ i  ¸  kij (6)
¨ wx j wx i ¸ 3
© ¹
where k is the turbulent kinetic energy, and Pt is the turbulent
viscosity given by
μ t C μ k 2 /  (7)
where CP is a constant and H is the dissipation rate. The
equations for the turbulent kinetic energy (k) and the (a) Rectangular fin (b) Fin with round tip (c) Elliptic fin
dissipation rate (H) are:
w w ª§ μ · wk º
u i k «¨¨ μ  t ¸¸ »  G k   . (8)
wx i wx i ¬«©  k ¹ wx i ¼»

w w ª§ μ · w º  2
u i  «¨¨ μ  t ¸¸ »  C1 G k  C 2  . (9)
wx i wx i ¬«©   ¹ wx i ¼» k k

The term Gk is the generation of turbulence kinetic energy due

to the mean velocity gradients. The turbulent heat flux can be (d) Grids in symmetric planes in the z-direction
modeled with the turbulent heat conductivity (Ot).
wT μ t wT
Figure 3 Meshes of Selected Sections
c p u'i T'  t c p , (10)
wx i Prt wx i
Although it is not the same for different shapes of fins, the
The constants C1H, C2H, CP, Vk, and VH used are: C1H= 1.44, C2H total number of cells in this study is about 800,000. Grid
= 1.92, CP = 0.09, Vk = 1.0, VH=1.3 [18]. The turbulence dependence has been tested for all the cases, and one of the
Prandtl number, Prt, is set to 0.85. Note that these constants examples is shown in Table 2 as well as Fig. 4. It is observed
have widely been used, but they may need to be “tuned” for that when the grid number increases from 793,000 to
specific physics such as accelerated flow, decelerated flow, 1,244,000, the distributions of local heat fluxes are nearly the
separated flow, low-Reynolds number flow, curved flow, same except in some small regions where flow separation
rotating flow, etc. Since the focus of this study is not on occurs as documented later. As to the overall performance
finding the best turbulence model for the problem under listed in the table, increasing the grid leads to a change of
consideration these constants are conserved. 1.5% for heat transfer and 0.4% for pressure drop. Therefore,
it is proved that the results are independent of the meshes.
Enhanced Wall Function – The above k-H model is mainly
valid for high Reynolds number fully turbulent flow. In the Table 2 Overall performance with different grid systems
region close to the wall, special treatment such as wall
functions is needed. For the present study, the whole domain Cases Coarse Grid Fine Grid
is separated into a viscosity-affected region and a fully
Number of cells (in Thousand) 793 1, 244
turbulent region by defining a turbulent Reynolds number. In
the wall region with the turbulent Reynolds number lower than Total heat transfer (W) 2.453 2.489
a specified value, the one-equation model of Wolfstein [19] is Pressure drop (N/m )2
1340 1345
used. The standard k-H model is used if the turbulent Re

52 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded
 20000 Jet Temperature
Local heat flux along the elliptic fin surface
18000
x=0.001 m
16000
Coarse grid
x=0.02 m
Heat flux (W/m )

14000
2

Fine grid
Coarse grid
12000
Fine grid
10000
8000

6000
4000

2000 Heat flux

0
0.002 0.003 0.004 0.005 0.006 0.007 0.008 323 325K
y (m) 0 30,000 W/m2

Figure 4 Grid impendence study (Local heat flux) Figure 5 Temperature and heat flux distributions over the fin
and base surfaces

The software package Fluent (v. 6.3.26) from Fluent, Inc. is 35 m/s
adopted. The simulation uses the segregated solver, which
employs an implicit pressure-correction scheme [20]. The
standard SIMPLE algorithm is used to couple the pressure and
velocity. The second order upwind scheme is selected for
spatial discretization of the convective terms. A converged
result renders at least a mass residual of 10-4, energy residual
of 10-6, momentum and turbulence kinetic energy residuals of
10-5. These residuals are the summation of the imbalance for x=0 Outlet
each cell, scaled by a representative value of the flow rate.
Typically, 1000 to 2000 iterations are needed to obtain a
converged result, which takes about 4~8 hours on a dual-core
PC with a CPU of 3.4-GHz Pentium (R). 0

Figure 6 Velocity field on the symmetric plane and outlet

RESULTS AND DISCUSSION
Figure 6 shows the velocity field on the outlet plant and the
Case 1: Rectangular fin – Baseline case symmetric plane (x=0). For each plane, the left half gives the
Figure 5 shows the temperature distribution over the fin and velocity vector in the y-z plane and the right half shows the
base surfaces. It is reasonable to see that the temperature in total velocity magnitude. Apparently, the overall velocity at
the region close to the fin tip and jet inlet is the lowest. The the outlet is much higher than that in the symmetric plane.
base surface has a quite uniform temperature due to the high Furthermore, the outlet velocity close to the base is higher due
heat conductivity of the aluminum as fin material. In addition, to the impinging effect. In addition, the velocity components
the temperature gradient for the entire fin and base is small – in the y-z direction, which represent the secondary flow as
only 2K, compared to the difference of 25K between the shown by the length of vectors, is much smaller at the outlet
cooling air and the base bottom surface. This can explain that than that on the symmetric plane. It is interesting to see that
the heat transfer estimation based on isothermal surface gives the secondary flow close to the fin side surface is upwards,
a good result in some studies. The temperature difference which indicates a reverse flow or circulation occurs, although
increases when the fin is tall and the heat conductivity is low. it becomes weaker when moving downstream to the outlet.
Figure 5 also shows the distribution of heat flux. The high The plausible reason for the reverse secondary flow close to
heat flux on the base bottom surface is balanced by the low the fin surface is due to the flow acceleration at the sharp
heat flux on the fin and base surfaces exposed to convection. corner of the fin tip. Flow acceleration results in low pressure
Except the base bottom surface, it is observed that the high and then flow separation on the fin surface. As shown later,
heat flux, which represents high heat transfer coefficient, the flow separation is weakened because of the smoother tip.
occurs at the fin top surface close to the impinging jet, the fin
side surface close to the jet or close to the base. The heat flux More details of local heat transfer coefficient are shown in
on the base surface is also high. This feature of heat transfer Fig. 7. Note that theoretically the heat transfer coefficient is
can be reasoned out by the flow field as shown in Fig. 6. defined as
h q" /(Tw  Tj ) (11)

53 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded
where q” is the heat flux, Tw is the local wall temperature, and Figure 9 presents the local heat transfer of elliptic fins. Note
Tj is the cooling jet temperature. To obtain the overall thermal there does not exist a clear top surface for elliptic fins but only
performance more conveniently, the heat transfer coefficient a continuous side surface from the bottom to the tip. The
may also be defined as. coordinate, l, is the curved line length from the fin bottom.
h overall q" /(Tb  Tj ) (12) Compared to the case in Fig. 7, the heat transfer coefficient of
the elliptic fins is low on both the base surface and the most
where q” is the heat flux, Tb is the temperature of base bottom area of the fin side surface. The stagnation line on the top of
surface. This definition may not reflect the actual heat transfer the elliptic fin shows the highest heat transfer coefficient.
rate locally, but Tb - Tj represents the potential of temperature Flow separation produces a minimum heat transfer point,
difference of heat transfer. As seen earlier, the temperature which is at about half the fin height. Note the characteristics
field inside the base and fin is quite uniform, which means the of the fins with round tip lie between cases in Figs. 7 and 8.
h values given by the two definitions are very close. The first
definition is used for the data in Fig. 7. On the top surface of
the fin, it is observed that the heat transfer coefficient is much
higher in the impinging region than the downstream region. In
the impinging region, the heat transfer coefficient close to the
edge is extremely high. In the downstream region, the heat

Fin with round tip

transfer coefficient becomes more uniform. Over the fin side

Rectangular fin
surface along the direction from the base to the fin tip the heat

Elliptic fin
transfer coefficient increases first and then decreases. In the
impinging region, the heat transfer coefficient will increase
again to create another peak while in the downstream region it
will continuously decrease. The heat transfer coefficient on x=0 x=0 x=0
the base surface increases sharply from the fin to the channel
symmetric line along the z-direction in both impinging and 0 20 m/s
downstream regions due to the high velocity in the middle.
Figure 8 Heat transfer coefficient with mean temperature used
1200
as the reference temperature
Base

Fin top Fin sidewall

1000 1000
Rectangular fin
800 x=0.001 m
800
h (W/m K)

x=0.02 m
2

600
h (W/m K)

x=0.001 m 600
2

Elliptic fin
400 x=0.02 m x=0.001 m
400
x=0.02 m
200

200
0
-0.002 0.000 0.002 0.004 0.006 0.008 Base Fin sidewall
z (m) y (m) 0
0.000 0.002 0.003 0.005 0.007 0.010
Figure 7 Local heat transfer coefficient distribution for the z l
rectangular fin
Figure 9 Local heat transfer coefficient distribution for the
elliptic fin
Case 2: Fin with round tip and elliptic fin
The flow and heat transfer of elliptic fins and fins with round Comparison of the overall performance is given in Table 3. It
tip are simulated. Figure 8 compares the velocity distribution is seen that the total heat transfer rate decreases from 2.117 to
on the symmetric plane in the x-direction. It is seen that both 1.511 W when the rectangular fins are replaced with elliptic
the size and location of separation region change significantly. fins. The total heat transfer surface area and total weight also
The maximum velocity, which occurs near the sharp corner of decrease but with a smaller percentage. However, the pressure
the rectangular fins, decreases when the other two shapes are loss decreases significantly, similar to the heat transfer. Note
used. Actually, the maximum velocity for the elliptic fins is that the ratios of heat transfer area and total weight is closely
about the same as jet inlet velocity, which means there is no related to the design of heat sink base. Based on these data for
significant acceleration. the short fins the designer may want to take the advantage of
high heat flux by sacrificing the pressure drop.

54 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded
 Table 3 Overall performance with different shapes of fins Table 4 Overall performance with different fin thickness
Rectangular Fins Elliptic Fins Ratio Rectangular Fins Elliptic Fins Ratio
Heat transfer area q (W) 2.117 1.511 1.40
(Fin & base) (m2) 0.0004 0.000361 1.11 b=2 mm
'P (Pa) 490.0 355.0 1.38
Total weight (kg) 0.00136 0.001184 1.15
q (W) 2.350 2.051 1.15
Total heat rate (W) 2.117 1.511 1.40 b=1 mm
'P (Pa) 606.3 411.5 1.47
Pressure loss (Pascal) 490 355 1.38

Validation of Simulation Results

Case 3: Effect of Reynolds number Numerical simulation serves as the primary tool in this study
To examine the behaviors of different shapes of fins under to explore the performance of plate-fin heat sink with different
other operating conditions, Figure 10 shows the effect of shapes. Although the commercial software package has been
Reynolds number on the overall heat transfer rate, which is proved accurate at least for problems as simple as in this
proportional to the average heat transfer coefficient or Nusselt study, validation is also performed for the current numerical
number. Clearly the rectangular fins produce the highest heat models and settings. Since there is no case in the literature
flux for all the Reynolds numbers in the range of both laminar that is exactly the same as studied here, a simplified model for
and turbulent flows, followed by the fins with round tip and a heat sink flow is considered and Fig. 11 shows the geometric
then the elliptic fins. Not shown in the figure, the pressure dimensions and flow settings. It is basically an impinging jet
drop ratio between different shape fins keeps almost the same into a narrow channel with isothermal wall. The average heat
when the Reynolds number changes. transfer rate can be calculated as [2]
Nu D h (0.54 - 0.1s/L)/ L* (13)
4.0
where L* (L/2)/(D h ReD Pr) and Dh is the channel hydraulic
3.5 h

Rectangular fin diameter. The average channel velocity is used in Reynolds

Total Heat Rate (W)

3.0 number. Another correlation for Nu number is given in [21].

Fin with round tip
2.5 Nu [(6.05x -0.22
t ) -3/4  (0.20x -1.05
t ) -3/4 ] -4/3 (14)
2.0 *
where x t L/(Dh ReDh ) , a similar parameter as L .
1.5

1.0 Elliptic fin s

0.5 B
0.0 Symmetric Inlet
0 1000 2000 3000 4000 5000 plane
Re

Figure 10 Effect of Reynolds number on overall heat flux Outlet

H
L
Case 4: Effect of fin height or thickness
The effect of fin relative height is studied by scaling down the
dimension in the z-direction while keeping all the ratios in the
x- and y-directions. Reducing the fin thickness from 2mm to
1mm gives a height to width ratio of 6. Compared to the Figure 11 Impinging jet flow in a narrow channel
previous cases, it is observed that total heat flux increases
although the pressure drop also increases. Furthermore, the The results are compared to the values given by correlations
difference of heat transfer between the cases with rectangular above. While numerical simulation in this study gives a
fins and elliptic fins becomes smaller but the ratio of pressure Nusselt number of 1.82, the values from the equations above
drop is almost the same. Therefore, it can be concluded that are 1.90 and 1.52, respectively. It can be concluded that the
for fins with larger height-to-width ratio the elliptic fins can numerical results are consistent with those studies.
provide a better performance. Further reducing the thickness
of fins and the distance between the fins may result in flow CONCLUSIONS
redistribution in the gap and flow channel between the fins,
and eventually result in a low heat flux with high pressure With the main focus on the effect of fin shapes this study
drop. Optimized design is needed to consider both the heat examines numerically details of the flow and temperature
transfer and pressure drop, which can be a further study but fields of heat sinks with plate fins cooled by jet impingement.
not detailed in this study. Three different shapes under study are rectangular, round-
headed and elliptic, while the fin height is relatively short.
Based on the simulation results, it is concluded that

55 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded
x The temperature inside the fin and base is quite uniform. [10] Issa, J. S. and Ortega, A., “Experimental measurements of
The secondary flow close to the fin sidewall is upwards. the flow and heat transfer of a square jet impinging on an
Heat transfer on the top of fin surface and base surface is array of square pin fins,” ASME J. of Electronic
strong. Packaging, v. 128, pp. 61-70, 2006
[11] Sathe, S.B., Sammakia, B.G., Wong, A.C., and Mahaney,
x The fin shapes affect largely both the secondary flow and H.V., “Numerical study of a high-performance air-cooled
flow separation. As a result, the heat transfer and pressure impingement heat sink,” Proceedings of the 30th National
drop, especially of the short fin cases, are affected by the Heat Transfer Conference. Part 1, Portland, OR, USA,
fin shapes. Aug 6-8, 1995
[12] Bhopte, S., Alshuqairi, M. S.; Agonafer, D., and Refai-
x The effect of fin shapes on the performance of heat sink is Ahmed, G. “Mixed convection of impinging air cooling
consistent for all the Reynolds number under study, over heat sink in telecom system application,” ASME J.
including both laminar and turbulent flows. of Electronic Packaging, v. 126, pp. 519-523, 2004
[13] Lou, Z.Q., Yap, C., and Mujumdar, A.S, “Numerical
x For longer fins or fins with smaller thickness, the shape investigation of laminar impinging jet heat transfer to
effect on heat transfer becomes relatively less important finned heat sinks,” Proceedings of the ASME Summer
than that on pressure drop. Heat Transfer Conference, v. 4, pp. 577-582, 2005
[14] Tien, H. C. and Huang, W. D, “Simulation and
assessment of air impingement cooling on squared pin-fin
REFERENCES heat sinks applied in personal computers,” J. of Marine
[1] Jang, S. P. and Kim, S. J., “Fluid flow and thermal Science and Technology, v 13, pp. 20-27, 2005
characteristics for a microchannel heat sink subject to an [15] Yang, Y. -T. and Peng, H.-S., “Numerical study of pin-fin
impinging jet,” Advances in Electronic Packaging 2003, heat sinks in a turbulent jet impinging channel,” Applied
v. 2, pp. 263-270, 2003 Simulation and Modeling 2007, Palma De Mallorca,
[2] Duan, Z. and Muzychka, Y.S., “Experimental Spain, Aug. 29 –31, 2007
investigation of heat transfer in impingement air cooled [16] Chuang, S., Chen, M., Lii, S., and Tai, F., “Numerical
plate fin heat sinks,” ASME J. of Electronic Packaging, v simulation of twin-jet impingement on a flat plate coupled
128, pp. 412-418, 2006 with cross-flow,” Int. J. for Numerical Methods in Fluids,
[3] Lin, T.W., Wu, M.C., Liu, L.K., Fang, C.J, and Hung, v. 14, pp. 459-475, 1992
Y.H., “Cooling performance of using a confined slot jet [17] Hofmann, H., Martin, H., and Kind, M., “Numerical
impinging onto heated heat sinks,” ASME J. of Electronic simulation of heat transfer from an impinging jet to a flat
Packaging, v. 128, pp. 82-91, 2006 plate,” Chemical Engineering and Technology, v. 27, pp.
[4] Sansoucy, E., Oosthuizen, P. H., and Refai-Ahmed, G., 27-30, 2004
“An experimental study of the enhancement of air-cooling [18] Launder, B. E. and Spalding, D. B., Lectures in
limits for telecom/datacom heat sink applications using an Mathematical Models of Turbulence, Academic Press,
impinging air jet,” ASME J. of Electronic Packaging, v London, England, 1972.
128, pp. 166-171, 2006 [19] Wolfstein, M., “The velocity and temperature distribution
[5] Wu, M.C., Lee, C.Y., Peng, C.H., Fang, C.J., and Hung, of one-dimensional flow with turbulence augmentation
Y.H., “Cooling performance of plain-plate-fin heat sinks and pressure gradient,” Int. J. Heat Mass Transfer, v. 12,
with slot jet impingement,” ASME Int. Mechanical pp. 301-318, 1969
Engineering Congress and Exposition, IMECE2006, [20] Fluent Manual, Version 6.3.26, Fluent, Inc., 2006
Chicago, IL, United States, Nov 5-10, 2006 [21] Biber, C.R., “Pressure drop and heat transfer in an
[6] Li, H.-Y. and Chen, K.-Y., “Thermal performance of isothermal channel with impinging flow,” IEEE
plate-fin heat sinks under confined impinging jet Transactions, Components, Packaging, and Manufacturing
conditions,” Int. J. of Heat and Mass Transfer, v. 50, pp. Technology, Part A, v. 20, pp. 458 – 462, 1997
1963-1970, 2007
[7] Brignoni, L.A. and Garimella, S.V., “Experimental
optimization of confined air jet impingement on a pin fin
heat sink,” IEEE Transactions on Components and
Packaging Technologies, v. 22, pp. 399-404, 1999
[8] El-Sheikh, H. A. and Garimella, S. V., “Heat transfer
from pin-fin heat sinks under multiple impinging jets,”
Transactions on Advanced Packaging, v 23, pp. 113-120,
2000
[9] Kondo, Y., Matsushima, H., and Komatsu, T.,
“Optimization of pin-fin heat sinks for impingement
cooling of electronic packages,” ASME J. of Electronic
Packaging, v. 122, pp. 240-246, 2000

56 on May 20,2025 at 08:19:18 UTC from IEEE Xplore. Restrictions apply.

Authorized licensed use limited to: Qilu Institute of Technology. Downloaded