International Journal of Heat and Mass Transfer 101 (2016) 656–666

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Enhancement thermodynamic performance of microchannel heat sink

by using a novel multi-nozzle structure
Ngoctan Tran a, Yaw-Jen Chang a,⇑, Jyh-tong Teng a, Thanhtrung Dang b, Ralph Greif c
a
Department of Mechanical Engineering, Chung Yuan Christian University, Chung-Li City, Taiwan
b
Department of Heat and Refrigeration Technology, Ho Chi Minh City University of Technical Education, Hochiminh City, Viet Nam
c
Department of Mechanical Engineering, University of California at Berkeley, CA 94720, USA

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

Article history: In the present study, a novel multi-nozzle microchannel heat sink (MN-MCHS) was proposed. The chan-
Received 26 January 2016 nel length, channel aspect ratio, rib width, pumping power, and heat flux were numerically investigated
Received in revised form 7 April 2016 in detail. It was found that the MN-MCHS with a shorter channel length not only could significantly
Accepted 30 April 2016
improve the temperature uniformity on the bottom wall and thermodynamic performance index, but
it also could significantly reduce the overall thermal resistance. With the decrease in the channel length
from 10 mm to 1 mm, the temperature uniformity was enhanced by approximately 10 times, the overall
Keywords:
thermal resistance improved 62% and the pressure drop was reduced approximately 12 times. For all
Channel length
Multi nozzle microchannel heat sink
cases in this study, the optimal structure of the MN-MCHS could dissipate a heat flux up to 1300 W/
High heat flux cm2, and kept the temperature rising above the inlet coolant temperature under 77.5 °C. In addition, at
the same pumping power, it could improve the overall thermal resistance up to 62% and 47.3% compared
to that of the single layer MCHS and the double layer MCHS, respectively. This structure of MN-MCHS is
really a promising structure of MCHS because it can improve thermal performance and reduce the pres-
sure drop by optimizing its geometric dimensions.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction
With the rapid development of scientific technology, air-cooling
heat sinks (ACHS) have been gradually replaced by fluid-cooling
heat sinks (FCHS) because the higher heat dissipation demands
of electronic devices in a smaller volume have gradually exceeded
the heat dissipation capacity of the ACHSs. A typification for FCHSs
is the microchannel heat sink (MCHS) that was first proposed by
Tuckerman and Pease [1] in the early 1980s. Since then, many
investigations on MCHS have been conducted in many aspects,
and a great number of studies have focused on single-layer and
parallel microchannel heat sinks (SL-P-MCHS). Yu et al. [2] and
Zhang et al. [3,4] reported investigations on the fractal-like MCHS.
They found that the fractal tree-like MCHS had a much higher heat
transfer coefficient than that of the straight microchannels, but it
consumed a much higher pumping power. Azizi et al. [5] and Neb-
bati et al. [6] reported investigations on convection heat transfer of
the MCHSs using Cu–water and Al2O3–water nanofluids, respec-
⇑ Corresponding author. Tel.: +886 3 265 4307; fax: +886 3 265 4399.
E-mail addresses: ngoctantran73@gmail.com (N. Tran), justin@cycu.edu.tw
(Y.-J. Chang), jtteng1@gmail.com (J.-t. Teng), trungdang@hcmute.edu.vn (T. Dang),
greif@berkeley.edu (R. Greif).
tively. Azizi et al. [5] found that nanofluids not only could increase
the average and local Nusselt numbers but could also decrease the
bottom surface temperature. Sakanova et al. [7] reported a study
on wavy MCHS using nanofluids. Their research revealed that the
wavy MCHS could improve heat transfer performance compared
to that of the straight MCHS. Especially, the wavy MCHS would
work more efficiently than with the nanofluids. Leng et al. [8] stud-
ied porous fins MCHS, and they reported that the pressure drop on
the porous fins MCHS could decrease remarkably compared to that
of the solid fins. Chu et al. [9] studied the triangle channel shape,
and Wong et al. [10] studied the triangle fin shape. They reported
that a high temperature gradient existed in the regions between
the inlet and outlet of the triangle channel [9]. The heat transfer
rate increased when the rib width or rib height increased, but it
decreased when the rib length increased [10]. Square and circle
channel shapes were investigated by Normah et al. [11]. They con-
cluded that at the same hydraulic diameter and pumping power,
the thermal resistance of the circle MCHS was lower than that of
the square MCHS. The thermal performance of MCHS with internal
Y-shaped bifurcations was much better than that of the rectangu-
lar MCHS, as reported by Xie et al. [12]. Although the high heat flux
and small thermal resistance of SL-P-MCHS were proven in pub-
lished literature, the nonuniform bottom-wall temperature and

http://dx.doi.org/10.1016/j.ijheatmasstransfer.2016.04.111
0017-9310/Ó 2016 Elsevier Ltd. All rights reserved.
 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666
Nomenclature
ACHS air-cooling heat sinks
cp specific heat at constant pressure (J/kg K)
Bth bottom thickness (m)
Dc channel depth (m)
FCHS fluid-cooling heat sinks
Ht top cover height (m)
Hs substrate height (m)
Lc channel length (m)
Lm length of computational model (m)
Ln nozzle length (m)
Ls substrate length (m)
Lt top cover length (m)
M mass flow rate (kg/s)
MCHS microchannel heat sink
Nc channel number of a column
Nr channel number of a row
Nt total number of the channels on the heat sink
DP pressure drop (Pa)
Q dissipated power (W)
q heat flux (W/cm2)
R thermal resistance (°C/W)
high pressure dropped on the channels of the SL-P-MCHSs were
still barriers for wide application. To overcome the challenge of
nonuniform bottom-wall temperatures of the heat sink, Vafai and
Zhu [13] first proposed a concept of double-layer MCHS (DL-
MCHS) in 1999. Eight years later, it was reintroduced by Wei
et al. [14] with the thermal resistance improved up to 0.09 °C/
Wcm2. Recently, DL-MCHSs were continuously investigated by
other groups of authors such as Hung et al. [15–17], Leng et al.
[18–20], Lin et al. [21], Rajabifar et al. [22], Wu et al. [23], and Zhai
et al. [24]. The channel inlets of the upper layer of a DL-MCHS were
designed on the top of the channel outlets of the lower layer;
therefore, the temperature in the lower-layer-outlet region would
be reduced compared to that of the single layer leading to the tem-
perature on the bottom wall being more uniform. Sakanova et al.
[25] reported an optimization and comparison of double-layer
and double-side micro-channel heat sinks with nanofluid as the
coolant. They revealed that the thermal resistance of the heat sink
with sandwich structure and counter flows was much smaller than
those of the single-layer and the double-layer with unidirectional
or counter flow. A truncated structure of the upper layer channel
of the DL-MCHS was proposed by Leng et al. [19]. This truncation
could improve the temperature uniformity on the bottom wall.
Although the channel length of the upper layer was shorter than
the original, the channel length of the lower layer was not changed
compared to the original. This indicates that with the same
hydraulic diameter and channel length, the pressure drop of the
lower-layer channels of the DL-MCHS was approximately the same
as that of the single layer MCHS. This means that the DL-MCHS
could not reduce the pressure drop compared to that of the single
layer. Furthermore, Zhai et al. [24] suggested that DL-MCHSs
should not be used for cooling microelectronic equipment under
a small volumetric flow rate due to a larger irreversibility. In
2012, Boteler et al. [26] were first to propose the concept of a man-
ifold microchannel heat sink (M-MCHS). Following that proposi-
tion, the investigations on the M-MCHS were continued by other
groups of authors [27–29]. They reported that the M-MCHS could
significantly reduce the pressure drop and improve the tempera-
ture uniformity on the bottom wall due to the truncation of the
path of the coolant. Yin et al. [30] proposed a structure of heat sink
with AIN-base micro-channels in direct bond copper. They
reported that the thermal resistance of heat sink with the proposed
657
T temperature (°C)
Wc channel width (m)
Wm width of computational model (m)
Wr rib width (m)
Ws substrate width (m)
Greek symbols
a channel aspect ratio
d ratio between rib width and channel width
g thermodynamic performance index
l dynamic viscosity of the coolant (kg/m s)
q density of the coolant (kg/m3)
X pumping power (W)
Subscripts
bt,max bottom maximum
i inlet
o outlet
T total
w wall
structure decreased by 15% and 80% compared with that of the
conventional packaged structures with Cu-based MCHS being
bonded to the direct bond copper by solder or thermal interface
material, respectively. Based on the foundations of the reviewed
literature, a novel multi-nozzle microchannel heat sink (MN-
MCHS) is presented in the present work to improve the thermal
performance of the heat sink, reducing the pressure drop along
the channel and enhancing the uniformity of the bottom wall’s
temperature.
2. Methodology
2.1. Geometric design of a novel multi-nozzle microchannel heat sink
The schematic of a novel multi-nozzle MCHS is presented in
Fig. 1a. The blue and red parts in Fig. 1a are the inlet and outlet
channel nozzles, respectively. The gray part on the upper layer is
the top cover, and the lower gray part is the substrate. With this
structure of a heat sink, the channels can be arrayed on the sub-
strate following both row and column directions. The number of
channels can be calculated for a row, Nr, a column, Nc, or a total
number of the channels on the heat sink, Nt, respectively, as
follows:
Nr ¼ ðW s  W r Þ=ðW c þ W r Þ ð1Þ
Nc ¼ ðLs  W r Þ=ðLc þ W r Þ ð2Þ
Nt ¼ Nr  Nc ð3Þ
where Ws, Ls, Wr, Wc and Lc are the substrate width, substrate
length, rib width, channel width, and channel length, respectively.
Fig. 1b presents a unit cell of the MN-MCHS which includes parts
of the MN-MCHS such as the substrate, channel, top cover, inlet
and outlet. This unit cell would be used as a computational model
for analysis. Fig. 1c presents the dimensions of the MN-MCHS.
Fig. 1d presents the dimensions of the unit cell. With the original
parallel MCHS, supposedly, the flow is laminar; the coolant will
flow along the channel from the channel inlet to the channel outlet
without changing direction. However, in the MN-MCHS, the inlet
and outlet nozzles are located on the top of the channel, as shown
in Fig. 1b; therefore, the flow direction of the coolant will be
658 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666

Inlet manifolds Outlet

Top cover
Inlet

Unit cell

Substrate
Channel
Outlet manifolds
Heat flux at bottom wall

a) Substrate, channels and top-cover b) A unit cell for analysis

Unit cell

c) Dimensions of the multi nozzle microchannel heat sink

Straight channel

Channel in MN-MCHS

d) Dimensions of the model e) flow directions

Fig. 1. A novel multi-nozzles microchannel heat sink.

changed when it flows from the inlet to the outlet, as seen in Fig. 1e.
The mean velocity of the coolant along the channel from the inlet to
the outlet will be constant if the cross-sectional area of the channel
equals that of the inlet and the outlet. Therefore, the nozzle length,
Ln, must be equal to the channel depth. The function of the top cover
is not only to cover the channel but also to separate the inlet and
outlet nozzle of the channel. Therefore, the top cover length must
always exist, and its length, Lt, can be calculated by Lt ¼ Lc  2Ln .
The substrate height, Hs, is fixed, but the channel depth, Dc, is vari-
able in the investigative process leading to the values of the bottom
thickness, Bth , to be changed depending on the channel depth, and
could be calculated by:
Bth ¼ Hs  Dc :
2.2. Mathematic model
In the present work, the conjugate 3D flow and heat transfer
models of the CFD-ACE+ software package were employed to ana-
lyze the flow and heat transfer characteristics of the MN-MCHS.
The computational model was employed, including the parts of
the substrate, channel, inlet, outlet and top cover, as presented in
Fig. 1b; and its dimensions are shown in Fig. 1d. To simplify the
analytic process, assumptions for the model were as follows:
(1) Flow is assumed to be three-dimensional, single phase,
steady state, incompressible and laminar.
(2) Thermophysical properties of fluid and solid are assumed to
ð4Þ be temperature independent.
 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666 659

(3) Gravitational force is neglected, and there is no internal heat qw ¼ Mcp ðT 0  T i Þ=ðLm W m Þ; ð12Þ
generation within the model.
where qw is the heat flux applied on the bottom wall of the model,
(4) The walls of the channel have a no-slip condition for velocity
M is the mass flow rate of the coolant passed through the channel,
and temperature.
cp is the specific heat of the coolant at a constant pressure, T 0 is the
outlet water temperature, and T i is the inlet water temperature. Lm
For a steady state condition and incompressible flow,
and W m are the length and width of the computational model,
@V
@t
¼ 0; @T
@t
¼ 0; @p
@t
¼ 0, and @@tq ¼ 0; for a no-slip condition,
respectively. In order to guarantee that the simulation results were
uwall = vwall = wwall = 0. Based on the above assumptions, the independent on the grid scheme, three different grid schemes were
reviewed literature [17,25,28,29], and the equation for conserva- tested on various models. The numbers of the grid points in the x-,
tion of energy given by the CFD-ACE+ software package [31], the y-, z-direction of a 100 lm3 of the computational domain were: (I)
governing equation for computing the heat transfer and fluid flow 2  5  5, (II) 4  10  10 and (III) 6  15  15. The difference for
in this study can be rewritten as follows: the maximum temperature on the bottom wall between grid (I)
Continuity equation and grid (II) was 0.27 °C; it reduced to 0.12 °C for grid (II) and grid
r~
V ¼0 ð5Þ (III). Therefore, to save computational efforts without a loss in accu-
racy, grid (II) was adopted for analysis. To verify simulation results,
Momentum equations a comparison between the inlet and outlet water temperatures of
1 l the simulation results obtained in this study and that of the exper-
~
V  r~
V ¼  rp þ r2 ~
V ð6Þ imental results reported by Lee et al. [35] was conducted. The com-
q q
parison is shown in Fig. 2a. Good agreement between the two sets
Energy equation for the fluid of results was achieved with a maximum derivation of 5%. Other
comparisons between the simulated results obtained in this study
~ kf
V  rT f ¼  r2 T f ð7Þ and the experimental results reported by Tuckerman [1] for maxi-
qf cp;f mum temperature and overall thermal resistance are presented in
Energy equation for solid Fig. 2b. Good agreement between the two results was also achieved
with the percentage errors for maximum temperature less than
k s r2 T s ¼ 0 ð8Þ 1.8%, and that for thermal resistance less than 2%. Temperature pro-
The boundary conditions for the model are assumed as follows: files, which were simulated based on the three experiments done in
[1], are presented in Fig. 2c. The rise in temperature, for the three
Inlet: u = 0, v = 0, w ¼ w0 , T ¼ T 0 . profiles in Fig. 2c, was similar to that reported in [1]. In addition,
Outlet: p ¼ p0 . all results which were reported in this study were verified by Eq.
Bottom wall: p ¼ pw . (12) with a maximum percentage error always less than 3%. The
Other outer walls and symmetric interface: rT s ¼ 0. important tests for creating meshes and the good agreement
between numerical and experimental results in the comparisons
Based on the propositions by Hung et al. [17], and Knight et al. indicate that the simulation results obtained in this study could
[32,33], Wang et al. [34] the inlet velocity depending on pumping be used to accurately predict the flow and heat transfer character-
power in this study can be calculated as follows: istics of the heat sinks.
!0:5
4a X 2.4. Work description
w0 ¼ ; ð9Þ
2lð4:7ð1 þ aÞ2 þ 19:64ð1 þ a2 ÞÞ Nt Lc
In this study, five parameters including a ¼ Dc =W c , d ¼ W r =W c ,
where a, X, and l are channel aspect ratio ða ¼ Dc =W c Þ, pumping Lc , pumping power X and heat flux qw, were individually investi-
power, and dynamic viscosity of water, respectively. gated. First, the three geometric parameters of the MN-MCHS,
The overall thermal resistance of the heat sink can be calculated including a, d and Lc , were examined in detail based on the follow-
by Tuckerman and Pease [1], Hung et al. [17], Leng et al. [19] and ing given parameters: constant heat flux, qw = 200 W/cm2, the
Zhai et al. [24], pumping power, X = 0.1 W, top cover height, Ht = 100 lm, channel
width, Wc = 50 lm, substrate width, Ws = 10 mm, substrate length,
T max  T min T bt;max  T water-inlet
RT ¼ ¼ ð10Þ Ls = 10 mm and substrate height, Hs = 0.4 mm. After examination of
Q Q the three geometric parameters was finished, a minimum thermal
and thermodynamic performance index can be calculated by resistance value, and a maximum thermodynamic performance
index in the obtained results were found; therefore, a locally opti-
RT T max  T min
g¼ ¼ ; ð11Þ mal value of the three geometric parameters was defined. Then,
DP Q DP the locally optimal MN-MCHS was employed to examine the
where T max and T min are the maximum and minimum temperature effects of the heat flux and pumping power on the thermodynamic
obtained in the MCHS, respectively. Q is dissipated power, and DP is parameters of the MN-MCHS.
pressure drop of the heat sink.
3. Results and discussions
2.3. Model validation
3.1. A rib width examination
Based on the assumptions in Section 2.2, all outer walls except
the bottom wall of the computational model are thermally insu- To examine the effects of the rib width on the flow and heat
lated or symmetrical (rT s ¼ 0); there is no internal heat genera- transfer characteristics of MN-MCHS, the two base-line cases listed
tion within the model; and the heat transfer process is a steady in Table 1 and the given parameters stated in Section 2.4 were
state. The energy from the heat which is dissipated from the bot- employed. The ratio between the rib width and the channel width,
tom wall must be equal to the energy necessary for raising the d ¼ W r =W c , altered from 1 to 6 leading to the rib width increasing
coolant temperature, so the following equation must be satisfied: from 50 lm to 300 lm. Silicon and poly(methyl methacrylate)
660 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666

(a) (b)

(c)
Fig. 2. Comparisons between numerical and experimental results.

Table 1 Outlet: p0 ¼ 0 (compared to atmosphere pressure).

The parameters of two base-line cases. The rest of the outer walls: rT s ¼ 0.
Base-line No. Nt Lc (mm) d a wo (m/s)
1 942 1 1 6 1.247
Fig. 3a presents an overall thermal resistance RT versus the ratio
2 159 4 2 3 2.015 between rib width and channel width, d. The results showed that
the RT increased with the increase of the d. It indicated that the
MN-MCHS worked more efficiently than with a smaller rib width.
This is due to the increase in the rib width leading to the decrease
in the total channel number [15] and total heat transfer area. The
Table 2
Thermophysical properties of the fluid and solid materials (at T = 26.85 °C). overall thermal resistance of the MN-MCHS with a base-line 2
increased faster and was found to be much higher than that of
Materials q (kg/m3) cp (J/kg K) k (W/m K) l (kg/ms)
base-line 1. This indicated that the overall thermal resistance of
Silicon 2329 702 148 – the MN-MCHS was not only affected by the rib width, but it was
PMMA 1420 1960 0.19 –
also affected by other mentioned geometric parameters. This will
Water 997 4179 0.613 0.000855
be discussed in more depth in the next sections. Fig. 3b presents
a thermodynamic performance index, g, versus the ratio, d. The
results show that with the increase of ratio, d, the thermodynamic
(PMMA) were utilized as the substrate and the top cover materials, performance index decreased rapidly with the base-line 1, and it
respectively. Water was applied as the coolant. Their thermophys- decreased more slowly with the base-line 2. The thermodynamic
ical properties are listed in Table 2 [1,19]. The boundary conditions performance indexes of base-line 1 were always much higher than
for the computational models were assumed as follows: those of base-line 2. Fig. 3c presents temperature profiles of the
models with the d of 1, 3 and 6 for both base-line 1 and 2. Observ-
Bottom wall: qw = 200 W/cm2. ing Fig. 3c, it is easy to find that with the same base-line, the smal-
Pumping power: X = 0.1 W. ler the d: the lower the bottom wall temperature. This proved the
Inlet: Ti = T0 = 23 °C; w ¼ w0 , (velocities, w0 , were calculated smaller overall thermal resistance at the smaller ratio, d, as men-
by using Eq. (9)). tioned in Fig. 3a. In addition, the bottom wall temperatures of
 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666 661

(a) (b)

(c)

(d)
Fig. 3. The effects of rid width on thermodynamic properties.

the models with base-line 2 were much higher than those of base- 1. This explains why the thermal resistance of the models with
line 1. The differences among three bottom wall temperatures of base-line 2 is always higher and grows faster than that of
the models with base-line 2 were greater than those of base-line base-line 1, as mentioned in Fig. 3a. The maximum bottom wall
662 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666
temperature of each model was always found at the end region of
the model as shown in Fig. 3c. Fig. 3d presents the pressure profiles
of the models with the d of 1, 3 and 6 for both base-line 1 and 2.
Fig. 3d shows that at the same base-line, the pressure drop
increased with the increase in d. Base-line 2’s pressure drops were
always greater than those of base-line 1. In this case, although it
was at the same hydraulic diameter of each single base-line, and
the same pumping power for both base-lines 1 and 2, the models
with a smaller ratio, d, obtained lower pressure drops due to their
total channel number being more than that of the greater ratio, d,
leading to the mean velocity of the channel with greater ratio, d,
being faster than that of the smaller ratio, d, (Eq. (9)). Also Eq. (9)
indicates that the mean velocity in the channel with a smaller
channel aspect ratio, a, is higher than that of the greater channel
aspect ratio. This is one of the reasons for the higher pressure drop
of the models with base-line 2 compared to that of base-line 1; it
also is one of the reasons for a higher thermodynamic performance
index of base-line 1 compared to that of base-line 2, as shown in
Fig. 3b.
3.2. A channel aspect ratio examination
To examine the effects of the channel aspect ratio on thermo-
dynamic properties of the MN-MCHS, the channel width was
fixed at 50 lm, and the channel depth was varied from 50 lm
to 300 lm leading to the channel aspect ratio, a ¼ Dc =W c , varying
from 1 to 6. The two base-lines, materials and boundary
conditions, as mentioned in Section 3.1, were employed for the
analysis.
Fig. 4a presents the overall thermal resistance, RT , of the
MN-MCHS versus the channel aspect ratio, a, for both base-
lines 1 and 2. The results show that with the increase in a, RT
decreased. It fell rapidly with the a 6 4, however, it decreased
more slowly with the a > 4. This indicates that the channels with
the greater channel aspect ratio at a fixed channel width worked
more efficiently than those of the smaller channel aspect ratio.
The reason was that the heat transfer area of the channel with
a greater channel aspect ratio was greater than that of the smaller
channel aspect ratio. Besides, the bottom thickness, Bth, also was
reduced ðBth ¼ Hs  Dc Þ by the increase in a, leading to a smaller
thermal resistance between the bottom wall of the channel and
the bottom wall of the model. The results also show that the RT
of the MN-MCHS with base-line 1 was smaller than that of
base-line 2 at the same channel aspect ratio. It also decreased
more slowly than that of base-line 2. The causes of this are due
to the effect of the rib width, as discussed in Section 3.1, and
the effect of the channel length which will be discussed in the
next section. Fig. 4b presents the thermodynamic performance
index, g, versus the channel aspect ratio, a, for both base-lines
1 and 2. The results show that g increased with the increase in
a. This indicates that the channel with a bigger channel aspect
ratio worked more efficiently not only for the thermal perfor-
mance but also for the dynamic performance. Fig. 4c presents
the y-cut temperature profiles of the MN-MCHS with channel
aspect ratios of 1, 3 and 5 for both base-lines 1 and 2. The profiles
show that the bottom wall temperatures of the models with
greater channel aspect ratios were smaller and more uniform
than those of the smaller channel aspect ratios [36]. Furthermore,
the bottom wall temperatures of the models with base-line 2
were much higher than those of base-line 1, and the differences
among three of them were also greater than that of base-line 1.
This was due to the influence of the heat transfer area and other
geometric parameters, as discussed above. Fig. 4d presents the
pressure profiles of the MN-MCHS with channel aspect ratios of
1, 3 and 5 for both base-lines 1 and 2. The profiles show that with
the decrease in channel aspect ratios, the pressure drop increased
quickly for the models with base-line 2; however, they increased
only slightly with base-line 1. This was due to the decrease in
hydraulic diameter caused by the decrease in the channel aspect
ratio. In addition, the pressure drops of models with base-line 2
were much greater than those with base-line 1. The differences
in the pressure drop among of the three models with base-line
2 were greater than those with base-line 1. The cause of this
was the effect of the channel length which will be discussed in
Section 3.3.
3.3. A channel length examination
In this section, channel lengths from 1 mm to 10 mm were sim-
ulated to examine the effects of the channel length on thermody-
namic properties of the MN-MCHS. The two base-lines, materials
and boundary conditions, as mentioned in the Section 3.1, were
employed for the studies.
Fig. 5a presents the overall thermal resistance, RT, versus the
channel length for both base-lines 1 and 2. The results show that
the RT increased nearly linearly with the increase in channel length.
This was due to the fact that the time for the coolant to flow into
the shorter channel was shorter than that for the longer channel;
therefore, the coolant’s rise in temperature in the shorter channel
was less than that for the longer channel. In addition, the rise in
temperature at the bottom wall depends on the rise in temperature
of the coolant [8]. Besides, when the channel length increased, the
number of channels decreased (Eqs. (2) and (3)) leading to the
increase of RT [15] due to the decrease in the heat transfer area.
For all studied ranges of the channel length, a maximum RT of
0.35 °C/W was found at a maximum channel length of 10 mm with
base-line 2, and a minimum RT of 0.059 °C/W was found at a min-
imum channel length of 1 mm with base-line 1. Fig. 5b presents
the thermodynamic performance index, g, versus the channel
length. The results show that the g reached its peak at the shortest
channel length. At a channel length from 1 mm to 4 mm, the g fell
rapidly but the g was reduced more slowly with a longer channel
length. The reason for this was that when the channel length
decreased, the decrease of the overall thermal resistance was
slower than that of the pressure drop. The effects of channel aspect
ratio and channel length on the pressure drop also explain the
higher value and the faster decrease of the thermodynamic perfor-
mance index of base-line 1 compared to that of base-line 2, pre-
sented in Fig. 4b. Furthermore, at the same channel length, the RT
of the MN-MCHS with base-line 1 was always smaller than that
of base-line 2 due to the effects of the channel aspect ratio and
the rib width, as discussed in previous sections. Fig. 5c presents
the temperature profiles of the models with channel lengths of
1 mm, 3 mm and 5 mm for both base-lines 1 and 2. The results
show that the bottom wall temperatures of the models with a
shorter channel length were lower and more uniform than those
for longer channels [26,27]. More specifically, with a channel
length of 1 mm, the differences between the maximum and mini-
mum temperatures on the bottom wall were 2.39 °C and 2.56 °C
for the MN-MCHS with base-lines 1 and 2, respectively. However,
with a channel length of 5 mm, the differences went up approxi-
mately 5–10 times that of the channel length of 1 mm for base-
lines 1 and 2, respectively. This indicates that channel length plays
a very important role in controlling the uniformity and the rise of
the bottom wall temperature of the MN-MCHS. The channel aspect
ratio clearly manifested its role on the uniformity and the rise of
bottom wall temperature with the longer channel length; however,
its role gradually decreased with the shorter channel length. Fig. 5d
presents the pressure profiles of the models with channel lengths
of 1 mm, 3 mm and 5 mm for both base-lines 1 and 2. The results
show that the pressure drop increased significantly when the
channel length increased [26,37]. Specifically, it grew faster with
 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666 663

Fig. 4. The effects of a on thermodynamic properties.

the smaller channel aspect ratio. The reason for this was the smal- Based on examinations in Sections 3.1–3.3, it was found that the
ler hydraulic diameter by the smaller channel aspect ratio, and the maximum temperature of the bottom walls was always found at
increase of velocity by the lesser channel number with the longer the end region of the model [29]. The same minimum overall ther-
channel length at the same pumping power. mal resistance of 0.059 °C/W and maximum thermodynamic per-
664 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666

(a) (b)

(c)

(d)
Fig. 5. The effects of channel length on thermal and thermodynamic performance.

formance index, g, of 319.5 W/cm2 Pa were found in the examina-
tions of the rib width, channel aspect ratio and channel length.
Based on the minimum overall thermal resistance and the maxi-
mum thermodynamic performance index, an optimal geometric
structure of the MN-MCHS for the pumping power of 0.1 W was
achieved with a rib width of 50 lm (d = 1), a channel depth of
300 lm ða ¼ 6Þ, a channel length of 1 mm (Lc = 1 mm), a substrate
length of 10 mm (Ls = 10 mm), a substrate width of 10 mm
(Ws = 10 mm), a substrate height of 0.4 mm (Hs = 0.4 mm) and a
channel number of 942 channels (Nt = 942 channels).
 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666 665

3.4. A heat flux and pumping power examination
To examine the effects of the heat flux and pumping power on
the overall thermal resistance, RT, and the thermodynamic perfor-
mance index, g, the optimal geometric structure of the MN-MCHS
was employed for analysis. Different heat fluxes in the range from
100 W/cm2 to 1300 W/cm2 were studied with a pumping power of
0.1 W. Different pumping powers in the range from 0.05 W to 2 W
were studied in detail with a heat flux of 500 W/cm2.
Fig. 6a presents a maximum temperature on the bottom wall,
Tmax, and an overall thermal resistance of the MN-MCHS versus
heat fluxes. The results show that the overall thermal resistances
of the MN-MCHS were independent of the heat flux. However,
the maximum temperatures on the bottom wall were nearly
increased linearly with the increase of the heat flux. It was found
that the optimal geometric structure of the MN-MCHS could dissi-
pate a heat flux up to 1300 W/cm2 and maintain a maximum tem-
perature on the bottom wall under 100.5 °C by using a pumping
power of 0.1 W.
Fig. 6b presents an overall thermal resistance and thermody-
namic performance index versus the pumping power. It was found
that the overall thermal resistance of the MN-MCHS decreased
with the increase of the pumping power due to the increase in
the mass flow rate [7,21,38]. At pumping powers less than 0.5 W,
the overall thermal resistance decreased rapidly with the increase
in pumping power [39]. However, it decreased more slowly when
120
100
80
Tmax, oC
the pumping power became greater than 0.5 W. For all cases in the
present study, the lowest thermal resistance of the optimal geo-
metric structure of the MN-MCHS was achieved by 0.034 °C/W at
a pumping power of 2 W which was an improvement of up to
62% compared to that of the No. 3 experiment with a pumping
power of 2.24 W reported in [1]. At a pumping power of 0.05 W,
the overall thermal resistance was achieved by 0.069 °C/W which
was an improvement of 32.35 %, and 47.3 % compared to those of
the double layer MCHS reported by Leng et al. [19] and Lin et al.
[21], respectively. The thermodynamic performance index fell
rapidly with the increase in pumping power from 0.05 W to
0.5 W, and then, it was reduced more slowly with the increase in
pumping power from 0.5 W to 2 W. This indicates that the MN-
MCHS worked more efficiently here than with the pumping power
less than 0.5 W.
4. Conclusions
In this study, a novel multi-nozzle microchannel heat sink (MN-
MCHS) was proposed. A silicon plate of 10  10  0.4 mm was
used as a fixed substrate, and water was applied as the coolant.
The microchannels could be arrayed up to 942 channels on the
MN-MCHS with a base area of 1 cm2. The effects of the channel
depth, rib width, channel length, pumping power and heat flux
on the flow and heat transfer characteristics of the MN-MCHS were
numerically investigated in detail. Based on the present investi-
gated results, conclusions can be summarized as follows:
0.1
1. At the same pumping power, the rib width not only affected the
overall thermal resistance, but it also affected the thermody-
0.08
namic performance index. The overall thermal resistance and
the thermodynamic performance index can be improved by
0.06 the decrease in rib width.
RT, oC/W

60 2. With a fixed channel width, the channel aspect ratio was

0.04 affected on both flow and heat transfer characteristics. Overall
40 Tmax thermal resistance and thermodynamic performance indexes
improved significantly with the increase in the channel aspect
RT 0.02
20 ratio.
3. The channel length played a very important role in controlling
0 0 the uniformity and the rise of temperature on the bottom wall
0 200 400 600 800 1000 1200 1400 of the heat sink. With the decrease of the channel length from
qw, W/cm2 10 mm to 1 mm, the overall thermal resistance could be
improved up to 62%, and the pressure drop could be decreased
by approximately 12 times.
a) Tmax, and RT versus qw 4. The overall thermal resistance of the MN-MCHS was not
affected by the heat flux. However, the maximum temperature
0.1 1400 on the bottom wall increased with the increase of the heat flux.
5. At the same pumping power of 0.05 W, the overall thermal
1200
0.08 resistance of the MN-MCHS improved up to 32.35% and 47.3%
RT 1000 compared to that of the double layer MCHS reported by Leng
et al. [19] and Lin et al. [21].
, W/cm2 Pa
RT, oC/W

0.06 800 6. The MN-MCHS with a channel length of 1 mm, a channel width
of 50 lm, a channel depth of 300 lm, a rib width of 50 lm and a
0.04 600
channel number of 942 channels, could dissipate a heat flux up
400 to 1300 W/cm2 and kept the rise in temperature on the inlet
0.02 coolant temperature under 77.5 °C.
200

0 0
0 0.5 1 1.5 2 References
,W
[1] D.B. Tuckerman, R.F.W. Pease, High-performance heat sinking for VLSI.pdf,
IEEE Electron Device Lett. EDL-2 (1981) 126–129.
[2] X.F. Yu, C.P. Zhang, J.T. Teng, S.Y. Huang, S.P. Jin, Y.F. Lian, et al., A study on the
b) RT, and versus pumping power hydraulic and thermal characteristics in fractal tree-like microchannels by
numerical and experimental methods, Int. J. Heat Mass Transfer 55 (2012)
Fig. 6. RT, Tmax and g versus qw and X. 7499–7507.
666 N. Tran et al. / International Journal of Heat and Mass Transfer 101 (2016) 656–666
[3] C.P. Zhang, Y.F. Lian, C.H. Hsu, J.T. Teng, S. Liu, Y.J. Chang, et al., Investigations of
thermal and flow behavior of bifurcations and bends in fractal-like
microchannel networks: secondary flow and recirculation flow, Int. J. Heat
Mass Transfer 85 (2015) 723–731.
[4] C.P. Zhang, Y.F. Lian, X.F. Yu, W. Liu, J.T. Teng, T.T. Xu, et al., Numerical and
experimental studies on laminar hydrodynamic and thermal characteristics in
fractal-like microchannel networks. Part B: investigations on the performances
of pressure drop and heat transfer, Int. J. Heat Mass Transfer 66 (2013) 939–
947.
[5] Z. Azizi, A. Alamdari, M.R. Malayeri, Convective heat transfer of Cu–water
nanofluid in a cylindrical microchannel heat sink, Energy Convers. Manage.
101 (2015) 515–524.
[6] R. Nebbati, M. Kadja, Study of forced convection of a nanofluid used as a heat
carrier in a microchannel heat sink, Energy Proc. 74 (2015) 633–642.
[7] A. Sakanova, C.C. Keian, J. Zhao, Performance improvements of microchannel
heat sink using wavy channel and nanofluids, Int. J. Heat Mass Transfer 89
(2015) 59–74.
[8] C. Leng, X.-D. Wang, T.-H. Wang, W.-M. Yan, Fluid flow and heat transfer in
microchannel heat sink based on porous fin design concept, Int. Commun. Heat
Mass Transfer 65 (2015) 52–57.
[9] J.C. Chu, J.T. Teng, R. Greif, Heat transfer for water flow in triangular silicon
microchannels, J. Therm. Sci. Technol. 3 (2008) 410–420.
[10] K.-C. Wong, J.-H. Lee, Investigation of thermal performance of microchannel
heat sink with triangular ribs in the transverse microchambers, Int. Commun.
Heat Mass Transfer 65 (2015) 103–110.
[11] G.-M. Normah, J.-T. Oh, N.B. Chien, K.-I. Choi, A. Robiah, Comparison of the
optimized thermal performance of square and circular ammonia-cooled
microchannel heat sink with genetic algorithm, Energy Convers. Manage.
102 (2015) 59–65.
[12] G. Xie, H. Shen, C.-C. Wang, Parametric study on thermal performance of
microchannel heat sinks with internal vertical Y-shaped bifurcations, Int. J.
Heat Mass Transfer 90 (2015) 948–958.
[13] K. Vafai, L. Zhu, Analysis of two-layered micro channel heat sink concept in
electronic cooling, Int. J. Heat Mass Transfer 42 (1999) 2287–2297.
[14] X. Wei, Y. Joshi, M.K. Patterson, Experimental and numerical study of a stacked
microchannel heat sink for liquid cooling of microelectronic devices, J. Heat
Transfer 129 (2007) 1432.
[15] T.-C. Hung, W.-M. Yan, Enhancement of thermal performance in double-
layered microchannel heat sink with nanofluids, Int. J. Heat Mass Transfer 55
(2012) 3225–3238.
[16] T.-C. Hung, W.-M. Yan, W.-P. Li, Analysis of heat transfer characteristics of
double-layered microchannel heat sink, Int. J. Heat Mass Transfer 55 (2012)
3090–3099.
[17] T.-C. Hung, W.-M. Yan, X.-D. Wang, Y.-X. Huang, Optimal design of geometric
parameters of double-layered microchannel heat sinks, Int. J. Heat Mass
Transfer 55 (2012) 3262–3272.
[18] C. Leng, X.-D. Wang, T.-H. Wang, An improved design of double-layered
microchannel heat sink with truncated top channels, Appl. Therm. Eng. 79
(2015) 54–62.
[19] C. Leng, X.-D. Wang, T.-H. Wang, W.-M. Yan, Multi-parameter optimization of
flow and heat transfer for a novel double-layered microchannel heat sink, Int.
J. Heat Mass Transfer 84 (2015) 359–369.
[20] C. Leng, X.-D. Wang, T.-H. Wang, W.-M. Yan, Optimization of thermal
resistance and bottom wall temperature uniformity for double-layered
microchannel heat sink, Energy Convers. Manage. 93 (2015) 141–150.
[21] L. Lin, Y.-Y. Chen, X.-X. Zhang, X.-D. Wang, Optimization of geometry and flow
rate distribution for double-layer microchannel heat sink, Int. J. Therm. Sci. 78
(2014) 158–168.
[22] B. Rajabifar, Enhancement of the performance of a double layered
microchannel heatsink using PCM slurry and nanofluid coolants, Int. J. Heat
Mass Transfer 88 (2015) 627–635.
[23] J.M. Wu, J.Y. Zhao, K.J. Tseng, Parametric study on the performance of double-
layered microchannels heat sink, Energy Convers. Manage. 80 (2014) 550–560.
[24] Y.L. Zhai, G.D. Xia, X.F. Liu, J. Wang, Characteristics of entropy generation and
heat transfer in double-layered micro heat sinks with complex structure,
Energy Convers. Manage. 103 (2015) 477–486.
[25] A. Sakanova, S. Yin, J. Zhao, J.M. Wu, K.C. Leong, Optimization and comparison
of double-layer and double-side micro-channel heat sinks with nanofluid for
power electronics cooling, Appl. Therm. Eng. 65 (2014) 124–134.
[26] L. Boteler, N. Jankowski, P. McCluskey, B. Morgan, Numerical investigation and
sensitivity analysis of manifold microchannel coolers, Int. J. Heat Mass
Transfer 55 (2012) 7698–7708.
[27] S. Sarangi, K.K. Bodla, S.V. Garimella, J.Y. Murthy, Manifold microchannel heat
sink design using optimization under uncertainty, Int. J. Heat Mass Transfer 69
(2014) 92–105.
[28] C.S. Sharma, M.K. Tiwari, B. Michel, D. Poulikakos, Thermofluidics and
energetics of a manifold microchannel heat sink for electronics with
recovered hot water as working fluid, Int. J. Heat Mass Transfer 58 (2013)
135–151.
[29] Y. Yue, S.K. Mohammadian, Y. Zhang, Analysis of performances of a manifold
microchannel heat sink with nanofluids, Int. J. Therm. Sci. 89 (2015) 305–313.
[30] S. Yin, K.J. Tseng, J. Zhao, Design of AlN-based micro-channel heat sink in direct
bond copper for power electronics packaging, Appl. Therm. Eng. 52 (2013)
120–129.
[31] E-Group, Advanced CFD-ACE + V2008.2 Modules Manual, ESI-Group
document, vol. 1, October 2008, pp. 35–64.
[32] R.W. Knight, J.S. Goodling, B.E. Gross, Optimal thermal design of air cooled
forced convection finned heat sink-experimental verification, IEEE Intersociety
Conference on Thermal Phenomena 5 (1992).
[33] R.W. Knight, D.J. Hall, J.S. Goodling, a.R.C. Jaeger, Heat sink optimization with
application to microchannels, IEEE Trans. Compon. Hybrids Manuf. Technol. 15
(1992) 832–842.
[34] Z.-H. Wang, X.-D. Wang, W.-M. Yan, Y.-Y. Duan, D.-J. Lee, J.-L. Xu, Multi-
parameters optimization for microchannel heat sink using inverse problem
method, Int. J. Heat Mass Transfer 54 (2011) 2811–2819.
[35] P.S. Lee, J.C. Ho, H. Xue, Experimental study on laminar heat transfer in
microchannel heat sink, Eighth Intersociety Conference on Thermal and
Thermomechanical Phenomena in Electronic Systems 6 (2002) 379–386.
[36] V. Leela Vinodhan, K.S. Rajan, Fine-tuning width and aspect ratio of an
improved microchannel heat sink for energy-efficient thermal management,
Energy Convers. Manage. 105 (2015) 986–994.
[37] S.G. Kandlikar, S. Garimella, D.Q. Li, S. Colin, M.R. King, Single-phase liquid flow
in microchannels and minichannels, in: Heat Transfer and Fluid Flow in
Minichannels and Microchannels, Elsevier, 2006, pp. 87–136.
[38] P.S. Lee, J.C. Ho, H. Xue, Experimental study on laminar heat transfer in
microchannel heat sink, Eighth Intersociety Conference on Thermal and
Thermomechanical Phenomena in Electronic Systems 02 (2002) 379–386.
[39] G. Xia, D. Ma, Y. Zhai, Y. Li, R. Liu, M. Du, Experimental and numerical study of
fluid flow and heat transfer characteristics in microchannel heat sink with
complex structure, Energy Convers. Manage. 105 (2015) 848–857.