You are on page 1of 9

Microelectronics Reliability 73 (2017) 97–105

Contents lists available at ScienceDirect

Microelectronics Reliability

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

Experimental and numerical investigation of circular minichannel heat


sinks with various hydraulic diameter for electronic cooling application
Seyed Ebrahim Ghasemi a,⁎, A.A. Ranjbar a, M.J. Hosseini b
a
Department of Mechanical Engineering, Babol University of Technology, Babol, Iran
b
Department of Mechanical Engineering, Golestan University, Gorgan, Iran

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

Article history: This article presents the experimental thermal and hydraulic performances of heat sinks with various channel di-
Received 11 January 2017 ameter for cooling electronic components. A heat sink with the length and width of 60 mm and total height of
Received in revised form 21 April 2017 16 mm fabricated from aluminum material. The heat sink is designed with four circular minichannels and
Accepted 21 April 2017
three different values of hydraulic diameter of channel (D = 4 mm, D = 6 mm and D = 8 mm). The minichannel
Available online xxxx
heat sink is heated with a uniform base heat flux. Also, numerical simulation of the problem is performed using
Keywords:
Finite Volume Method (FVM). Comparing the experimental and numerical results show that numerical results
Electronics cooling are in a good agreement with experimental data. The variation of channel diameter affects the heat transfer
Heat transfer coefficient and pressure drop characteristics of the circular shaped minichannel heat sink. The experimental results show
Thermal resistance that the increase of channel diameter reduces the pressure drop in the heat sink. Also, the minichannel heat
Channel diameter of heat sink sink with a hydraulic diameter of 4 mm has a much lower thermal resistance than the minichannel heat sinks
Pumping power with a hydraulic diameter of 6 mm and 8 mm. Furthermore, the optimization is done to have the maximum
heat transfer coefficient and minimum of pressure drop along the heat sink.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction uses experimental procedures to investigate the thermal performance


of two embedded U-shaped heat pipe and six embedded L-shaped
With the rapid development of electronic technology, electronic ap- heat pipe thermal modules with different fan speeds and heat source
pliances and devices now are always in our daily life. However, as the areas. Numerical investigation was performed to improve the hydro-
components size shrinks, the heat flux per unit area increases dramati- thermal behavior of the pin fin heat sink adding splitter on the back of
cally. The working temperature of the electronic components may ex- the pin by Sajedi et al. [8]. They found that adding the splitter behind
ceed the desired temperature level. Thus, promoting the heat transfer the pins will improve both thermal and hydrodynamics of flow inside
rate and maintaining the die at the desired operating temperature the heat sink by reducing the pressure drop around the pins and reduc-
have played an important role in insuring a reliable operation of elec- ing the thermal resistance of heat sink. Sufian and Abdullah [9] investi-
tronic components. There are a number of methods in electronics gated the enhancement of the heat transfer in high power LEDs by a
cooling, such as jet impingement cooling [1,2] and heat pipe [3–5]. Con- combination of piezoelectric fans and a heat sink. They evaluated the
ventional electronics cooling normally used forced air cooling with heat heat dissipation efficiency of the package of high power LEDs operating
sink showing superiority in terms of unit price, weight and reliability. In undermultiple vibrating fans experimentally and numerically. Thermal
order to design a practical heat sink, some criterions such as a large heat optimization of water heat sink for power converters with tight thermal
transfer rate, a low pressure drop, and a simpler structure should be constraints was analyzed by Cova et al. [10]. Conrad et al. [11] investi-
considered. gated the performance of a pin fin heat sink that was directly attached
Numerical simulations of plate-circular pin-fin heat sinks in thermal to the chip. Their packaging concept exposed the chip directly to the
and hydraulic performance were investigated by Yang an Peng [6]. The cooling fluid. They concluded that plastic strain during operation can
objective of their study was to examine the influence of the configura- be significantly reduced compared to standard non-structured chip con-
tions of pin-fins design on the thermal resistance and the pressure tacts. Ijam et al. [12,13] studied the cooling performance of a
drop of the heat sinks. The configuration and thermal performance of microchannel heat sink using nanofluids. They found that using the
the heat sinks with inserted heat pipes were studied by Wang [7]. He nanofluids such as Al2O3/water, TiO2/water, and SiC/water instead of
water improved the cooling by 2.95–17.32%, 1.88–16.53%, and
⁎ Corresponding author. 7.25–12.43%, respectively. An inverse geometric optimization for
E-mail address: s.ebrahim.ghasemi@stu.nit.ac.ir (S.E. Ghasemi). nanofluid cooled microchannel heat sink considering effects of

http://dx.doi.org/10.1016/j.microrel.2017.04.028
0026-2714/© 2017 Elsevier Ltd. All rights reserved.
98 S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105

temperature-dependent thermophyiscal properties for the water based Design (CCD) is applied to obtain an optimization design of channel di-
Al2O3 nanofluid with 1% particle volume fraction was numerically per- ameter of heat sink which has low thermal resistance and low pressure
formed by Wang et al. [14] under a constant pumping power constraint. drop.
Wu et al. [15] experimentally have studied using Al2O3–water
nanofluid as a coolant for the copper microchannel. They found that
2. Description of the experiment
the heat transfer coefficient and pumping power are higher at low
wall temperature and thermal resistance. Raisi et al. [16] numerically
2.1. Experimental setup
studied the thermal performance of a microchannel cooled with Cu–
water nanofluid and water. They concluded that the heat transfer rate
Fig. 1 demonstrates the flow loop and components that are designed
was affected by particle volume fraction and slip velocity coefficient at
and constructed for the present study.
high Reynolds numbers. The magnetohydrodynamic (MHD) flow of
The experiments of this study consist of a closed loop for fluid circu-
Jeffrey fluid in a circular microchannel using the method of variable
lation, the heat sink test section, a data acquisition system and a pres-
separation was presented by Gao and Jian [17]. Also, the thermal perfor-
sure drop measuring system. The experimental setup with a
mances of a minichannel heat sink were experimentally investigated for
corresponding apparatus is shown schematically in Fig. 2.
cooling of electronics using nanofluid coolant instead of pure water by
A fluid is sent into the loop from a fluid injection tank and is contin-
Sohel et al. [18]. They used the Al2O3-H2O nanofluid including the vol-
uously circulated by a pump. Volumetric flow rate passing through the
ume fraction ranging from 0.10 to 0.25 vol% as a coolant. Their experi-
loop is measured using two calibrated flow meters. The fluid after pass-
mental results showed the higher improvement of the thermal
ing through the flow meter is entered into the plexiglass tank and
performances using nanofluid instead of pure distilled water. Sandeep
passes through the heat sink. A constant thermal bath is installed up-
and Reddy [19] considered a nonlinear thermal radiation, viscous dissi-
stream of the pump in order to control the inlet temperature of the
pation, Joule heating with non-uniform heat source/sink effects for in-
heat sink. The loop flow rate is controlled as the fluid leaves the pump
vestigation the heat transfer nature of electrically conducting MHD
through a by-pass line. The test section of the experimental setup con-
nanofluid flow over a cone and a wedge. The effect of using nanofluid
sists of a heat sink, plexiglass cover plates as insulators, and a heater iso-
as working fluid on thermal performance of solar parabolic trough col-
lated in a block. The heat sink is fabricated from an aluminum square
lector was investigated numerically [20,21]. Ghasemi et al. [22,23] pro-
block using a Computer Numerical Control (CNC) machine. The width,
posed segmental rings for heat transfer enhancement of solar parabolic
length, and height of the heat sink are 60 mm, 60 mm, and 16 mm, re-
trough collector. The effects of segmental rings layouts on the heat
spectively. As shown in Fig. 3, four circular channels with various diam-
transfer and system performance for non-uniform heat flux were
eters of 4 mm, 6 mm and 8 mm have been created in this heat sink.
discussed. The heat transfer and pressure drop characteristics of para-
The assembled heat sink is placed in a plexiglass case in order to pre-
bolic solar collector with solid rings and porous rings were numerically
vent heat transfer from the surrounding of the test section and isolate it
studied using finite volume method by Ghasemi and Ranjbar [24,25].
from the ambient. Because of plexiglass low thermal conductivity, the
Heat transfer characteristics of R600a in smooth and flattened tubes
effect of lateral heat transfer from the sides of the test section is elimi-
were studied experimentally [26–28].
nated. Also, the surfaces of the plexiglass box except the base surface
Thermal and hydraulic performances of two types of fin, namely
of the heat sink are insulated using Styrofoam. Fig. 4 shows the heat
plate and plate-pin, in water-cooled corrugated miniature heat sinks
sink embedded in the plexiglass case with its inlet and outlet.
having triangular, trapezoidal, and sinusoidal shapes were evaluated
In order to provide a constant heat flux to the bottom surface of the
by Khoshvaght-Aliabadi et al. [29]. Their obtained results showed that
heat sink for simulating an electronic chip, a heater block was fabricated
the thermal performance of a corrugated heat sink with plate-pin fins
from the same material as that of the heat sink.
is better than that of a corrugated heat sink with plate fins. Takács
A cartridge heater is placed in the heater block and the produced
et al. [30] presented an enhanced thermal characterization method of
heat flux will be measured using 6 K-type thermocouples. Insulation
microscale heat sink structures, determining relevant partial thermal
of the heater block from surrounding is done by placing it in a thick fi-
resistances from structure functions obtained by thermal transient test-
berboard box filled with slag wool, so that heat transfer interaction
ing. The influence of nanofluids on performance of heat sink for CPU
could only take place at the interface of the heater block and the heat
water cooling was investigated experimentally and numerically by Al-
sink. To improve the heat transfer efficiency, the contact surface be-
Rashed et al. [31]. A satisfactory agreement between experimental and
tween the heat sink and the heater block is covered with a thin layer
calculated values of thermal resistance and conductance was obtained.
of high thermal conductive silicon paste. Two K-type thermocouples
Recently, the effect of CuO nanoparticles in distilled water on heat dissi-
with an accuracy of 0.1 °C have been placed at the inlet and outlet to cal-
pation from electronic components was investigated numerically by
culate the bulk temperature of fluid. Four K-type thermocouples, coated
Ghasemi et al. [32]. They used Computational Fluid Dynamics (CFD)
with a compound of copper powder and thermal silicon paste are em-
simulations to study the rectangular and circular cross-sectional shaped
bedded in the heat sink for measuring the base plate temperature. The
heat sinks. Also, the thermal performance of a triangular shaped
set of thermocouples are connected to a Jumo-706581 data logger
minichannel heat sink heat sink using alumina-water nanofluid as a
which is connected to a computer using a Universal Serial Bus (USB)
coolant with different volume fractions was examined by Ghasemi
cable and all the temperatures can be monitored and recorded simulta-
et al. [33]. They concluded that when the volume fraction of nanoparti-
neously during the experiments. The pressure drop of fluid flowing
cles is increased under the extreme heat flux, the thermal resistance of
through the heat sink is measured using a high-precision differential
the heat sink is decreased.
pressure transmitter (Rosemount-Model 1151GP). It is connected to
The main objective of this paper is to experimentally study the circu-
two holes located on either side of the manifold, as shown in Fig. 4.
lar cross-sectional shaped heat sink, and effect of the channel diameter
of heat sink on the thermo-hydraulic characteristics.
Furthermore, the heat sink is modeled numerically using finite vol- 2.2. Experimental data calculation
ume method. The CFD results of convective heat transfer coefficient
and pressure drop are compared with those of experimental data to val- The heat applied to the bottom surface of the heat sink is calculated
idate the numerical model. Also, the influences of volume flow rate on by the following equation:
the convective heat transfer coefficient, pressure drop, thermal resis-
tance and pumping power are examined experimentally. Finally, Re-
sponse Surface Methodology (RSM) based on Central Composite q ¼ ρQCp ðTout −Tin Þ ð1Þ
S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105 99

Fig. 1. Experimental setup.

where ρ, Q, Cp and q are density (kg/m3), volume flow rate (m3/s), spe- Tmax and Tmin are the maximum and minimum temperatures ob-
cific heat (J/kg·K) and total heating power (W), respectively. served in the heat sink, respectively. Tmin is equal to the inlet tempera-
The average heat transfer coefficient is defined by: ture of the coolant.
Pumping power (PP) which is the result of the pressure drop across
q the channel and volume flow rate (Q) is needed to drive the fluid in
have ¼ ð2Þ
Ah ΔTm channels and is defined by:

where Ah is heat transfer area. The temperature on the bottom side of


PP ¼ Δp  Q ð5Þ
the heat sink is measured using thermocouples placed in the heat sink.
The mean temperature difference between the solid wall and the
fluid (ΔTm) is defined as:

    2.3. Uncertainty analysis


Tw1 þ Tw2 þ Tw3 þ Tw4 T þ Tout
ΔTm ¼ − in ð3Þ
4 2
The uncertainty analysis for the experimental results was carried out
using the procedure proposed by Moffat [34]. The maximum uncer-
where Tw1, Tw2, Tw3 and Tw4 are the temperatures of heat sink base tainties was b6% for all parameters, in the present study.
plate. Also, Tin and Tout are the inlet and outlet fluid temperature
respectively. 3. Numerical simulation
An important parameter in evaluating the thermal performance of
the heat sink is its thermal resistance, which is defined as: 3.1. Geometry

Tmax −Tmin Fig. 5 shows the geometry of the problem that is used for numerical
Rth ¼ ð4Þ
q simulation based on the experimental study.

Fig. 2. Schematic diagram of the experimental apparatus.


100 S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105

Fig. 5. Geometric configuration of the heat sink.

Fig. 3. Heat sinks with various channel diameter. Energy equation for the solid region:

3.2. Governing equations 2 2 2


!
∂ Ts ∂ Ts ∂ Ts
ks þ 2 þ 2 ¼0 ð11Þ
∂x2 ∂y ∂z
To numerical study on the effect of channel diameter on the heat
sink performance, the governing mass, momentum and energy equa-
tions can be written as follows for incompressible, laminar and where Ts is the solid temperature, and ks is the thermal conductivity of
steady-state fluid flow [35]: the solid.
Continuity equation:
3.3. Boundary conditions
∂u ∂v ∂w
þ þ ¼0 ð6Þ Fluid enters the channel at the inlet with the same uniform axial ve-
∂x ∂y ∂z
locity that is specified according to the volume flow rate. The no-slip
boundary condition at the walls is appropriate for this study. For ther-
where u, v, and w are the velocity components in the x, y, and z direc-
mal boundary conditions, it is assumed that the fluid enters the channel
tions, respectively.
with a constant temperature and also a constant heat flux is applied
Momentum equation:
from the bottom, while the top glass layer is considered adiabatic. For
  2 2 2
! the channel outlet, the outflow boundary condition is considered.
∂u ∂u ∂u ∂p ∂ u ∂ u ∂ u
ρf u þ v þw ¼ − þμ f þ þ ð7Þ
∂x ∂y ∂z ∂x ∂x2 ∂y2 ∂z2 3.4. Numerical procedure

  2 2 2
! The geometrical model is created and meshed using commercial
∂v ∂v ∂v ∂p ∂ v ∂ v ∂ v software GAMBIT. In order to reduce the number of grid cells and cut
ρf u þ v þ w ¼ − þμf þ þ ð8Þ
∂x ∂y ∂z ∂y ∂x2 ∂y2 ∂z2 down the computational time, only one channel is modeled and studied.
Fig. 6 shows the geometry of the grid of heat sink with circular cross-
  2 2 2
! section. In order to simplify the mesh generation and cut down the com-
∂w ∂w ∂w ∂p ∂ w ∂ w ∂ w
ρf u þv þw ¼ − þμf þ þ ð9Þ putational time, only one half of the computational domain is
∂x ∂y ∂z ∂z ∂x2 ∂y2 ∂z2
considered.
The governing equations are solved using finite volume method
where ρf and μf are the density and dynamic viscosity of the coolant, re- with Computational Fluid Dynamics (CFD) commercial software
spectively, and p is the coolant pressure. ANSYS FLUENT [36]. The discretization scheme used for pressure is stan-
Energy equation for the coolant: dard. The solution is based on pressure correction method and uses
SIMPLE algorithm. The second order upwind differencing scheme is
  2 2 2
! used for momentum and energy equations. The solution is considered
∂T ∂T ∂T ∂ Tf ∂ Tf ∂ Tf
ρ f Cp; f u f þv f þw f ¼ kf þ þ ð10Þ to be converged sufficiently when the normalized residual of 10−5 for
∂x ∂y ∂z ∂x2 ∂y2 ∂z2
momentum and mass and 10−8 for energy equations.

where Tf is the coolant's temperature, Cp,f is its specific heat, and kf is its
thermal conductivity.

Fig. 4. Embedded heat sink in the plexiglass case. Fig. 6. Grid generation for a heat sink with circular channels.
S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105 101

Table 1 reduces and the velocity increases (in a constant volume flow rate), so
Grid independence study. the heat transfer rate increases and the cooling performance of the elec-
Grid system Mesh cells ΔP (Pa) ΔP-difference (%) tronic chip is improved and the bottom surface temperature of the heat
Grid 1 268,570 53.1506 2.36%
sink decreases. Also, with increasing the volume flow rate, the base tem-
Grid 2 702,930 54.4050 1.48% perature reduces for all diameters of channel. Such a behavior is obvious
Grid 3 1,248,570 55.2102 0.13% due to the fact that with increasing volume flow rate, the forced convec-
Grid 4 1,721,300 55.2819 Baseline tion effect also increases considerably, which in turn, augments the heat
transfer within the fluid in general. These results clearly demonstrate
the ability of the minichannel heat sink to remove enough heat from
3.5. Grid independence study the electronic chip surface.

To check for the independency of the results from the number of grid
points used, a grid-independency study is done by considering the 4.3. Convective heat transfer coefficient
amount of the pressure drop. To do this, different numbers of grid points
are used. The results are shown in Table 1, where the volume flow rate is In the laminar flow regime, the most important parameter influenc-
equal to 0.6 L/min. According to this study, the number of the grid points ing the heat transfer properties in minichannels is the hydraulic diame-
is considered 1,248,570 in the present study. ter. Fig. 9 shows the variation of the average heat transfer coefficient as a
function of volume flow rate for various hydraulic diameter of channels.
4. Results and discussion According to Fig. 9, the average heat transfer coefficient increases with
an increase in volume flow rate. Such a behavior is obviously a direct
4.1. Comparison between experimental and numerical results consequence of an enhancing convection effect under a higher volumet-
ric flow rate. Also, it is obvious that the values of average heat transfer
Fig. 7 compares the experimental and the numerical simulation re- coefficient for 4 mm channel diameter are higher than those of D =
sults for heat transfer coefficient and pressure drop. Numerical results 6 mm and D = 8 mm. Therefore, by decreasing the diameter of channel
show a higher heat transfer coefficient than the experimental results. the average heat transfer coefficient increases.
For both the experimental and numerical results, the heat transfer coef- Because when the channel diameter decreases, the cross-section
ficient increases with an increase in volume flow rate. According to Fig. area reduces and the velocity increases (in a constant volume flow
7a, the maximum deviation of numerical simulation from the experi- rate), so the heat transfer increases. From these results, one may see
mental results for heat transfer coefficient is 6.4%. Also, for pressure that corresponding to a reduction in hydraulic diameter from 8 mm to
drop (Fig. 7b) this deviations is 7%. It can be seen that the numerical 6 mm, the average heat transfer coefficient improves by N5.7%. Also,
simulation results are in good agreement with the experimental data. when the channel diameter reduces from 6 mm to 4 mm, the average
Therefore the present numerical model is reliable and can be used to heat transfer coefficient improves by N3.3%.
study the performance of a minichannel heat sink.

4.2. Temperature of the bottom surface of heat sink 4.4. Pressure drop

The main objective of the cooling system is to minimize the base The variation of pressure drop versus volume flow rate is plotted in
temperature as well as the temperature of the chip. Fig. 8 shows the var- Fig. 10 for three different hydraulic diameters of 4 mm, 6 mm and 8 mm.
iation of bottom surface temperature of the heat sink versus volume From this figure it can be interpreted that by increasing the volume flow
flow rate for different hydraulic diameters of channel. It is clear that rate of 0.3 L/min to 1.5 L/min, the pressure drop is increases for three
using channel with lower diameter has a remarkable effect on the re- configuration of hydraulic diameter. Also, the higher hydraulic diameter
ducing temperature variations. The bottom surface temperature of the of channel shows lower pressure drop. Because when the channel diam-
heat sink increases with increasing hydraulic diameter of channel. Be- eter increases, the cross-section area enhances and the velocity de-
cause when the channel diameter decreases, the cross-section area creases (in a constant volume flow rate), so the pressure drop decreases.

Fig. 7. Comparison between experimental and numerical results for: a) heat transfer coefficient and b) pressure drop.
102 S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105

Fig. 8. Heat sink base temperature versus volume flow rate for different channel diameter. Fig. 10. Pressure drop versus volume flow rate for different channel diameter.

4.5. Thermal resistance flow rate increases. The increase in convective heat transfer coefficient
and thermal dispersion of the higher velocity can be expressed as the
The main principal factor utilized in electronic chip cooling is the main reasons of decrease in convective thermal resistance with devel-
thermal resistance. It is plainly a measure of the resistance of the oping the volume flow rate.
minichannel heat sink to expend the input power. A purpose of using
this parameter is that it minimizes the complexity of local behavior 4.6. Pumping power
and provides an overall metric of the performance. The cooling perfor-
mance of the heat sinks with various channel diameters can be deter- When the coolant passes through the narrow channel of the heat
mined by using the thermal resistance. Based on the obtained sink, a pressure drop occurs. The pumping power is defined as the crite-
experimental data, the effects of volume flow rate and the channel di- rion for measuring the pressure loss between the inlet and the outlet of
ameter on the thermal resistance of the heat sink is depicted in Fig. the heat sink. To overcome this pressure drop, the system needs some
11. From the figure, it is seen that the lower channel diameter repre- extra pumping power. The channel diameter and volume flow rate ef-
sents the lower thermal resistance. This is due to the decreasing hydrau- fect the pumping power for water-cooled minichannel heat sink as
lic diameter producing an increase in cooling performance of the shown in Fig. 12. This figure demonstrates the decreasing in pumping
minichannel heat sink. In this research, 3.4% and 12.2% increases in con- power with the increasing in diameter of channel. Increasing the chan-
vective thermal resistance, respectively, for D = 6 mm and D = 8 mm nel diameter of heat sink reduces the pressure drop across the heat sink,
have been achieved in the highest volume flow rate compared with and this is the main reason for the decrement of pumping power of D =
D = 4 mm. Obviously, thermal resistance decreases when the volume 6 mm and D = 8 mm compared with D = 4 mm. The results show that

Fig. 9. Heat transfer coefficient versus volume flow rate for different channel diameter. Fig. 11. Thermal resistance versus volume flow rate for different channel diameter.
S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105 103

numerical experiments. The difference is in the type of error generated


by the responses. In physical experiments, inaccuracy can be due to
measurement errors while in computer experiments, numerical noise
is a result of incomplete convergence of iterative processes, round-off
errors or the discrete representation of continuous physical phenome-
na. The application of RSM to design optimization is aimed to reduce
the cost of expensive analysis methods (e.g. finite element method or
CFD analysis) and their associated numerical noise. Generally, the struc-
ture of the relationship between the response and the independent var-
iables is unknown. The first step in RSM is to find a suitable
approximation to the true relationship. The most common forms are
low-order polynomials (first or second-order). Second order model
can significantly improve the optimization process when a first-order
model suffers lack of fit due to interaction between variables and surface
curvatures. A general second-order model is defined as [41]:

n n n n
y ¼ a0 þ ∑ ai xi þ ∑ aii x2i þ ∑ ∑ aij xi x j ; ðibjÞ ð12Þ
i¼1 i¼1 i¼1 i¼1

where xi and xj are the design variables and a are the tuning parameters.
CCD or central composite design is one of modules in RSM to obtain the
Fig. 12. Pumping power versus volume flow rate for different channel diameter. points of each factor according to their levels.

there are 8% and 15% decreases in pumping power, respectively, for 5.2. Central composite design (CCD)
D = 6 mm and D = 8 mm compared with D = 4 mm. Lower channel
diameter causes more pressure drop at the heat sink. Therefore, the A Box-Wilson Central Composite Design, commonly called “central
flow requires additional pumping power to compensate this pressure composite design”, contains an imbedded factorial or fractional factorial
drop. Also, higher values of volume flow rate increase the velocity of design with center points that is augmented with a group of ‘star points’
fluid which in turn cause higher pressure drop and hence, an increase that allow estimation of curvature. If the distance from the center of the
in the pumping power. design space to a factorial point is ±1 unit for each factor, the distance
from the center of the design space to a star point is ±α with |α| N 1.
5. Optimization analysis The precise value of α depends on certain properties desired for the de-
sign and on the number of factors involved [41,42]. In CCD technique,
Optimization methods are applied in order to find the optimal solu- optimization is based on a parameter called “desirability”. Desirability
tions in engineering problems, so the optimization analysis is an impor- is an objective function that ranges from zero outside of the limits to
tant step to reach the best designs in practical applications [37–40]. one at the goal. The numerical optimization finds a point that maxi-
mizes the desirability function. The characteristics of a goal may be al-
5.1. Response surface methodology (RSM) tered by adjusting the weight or importance. For several responses
and factors, all goals get combined into one desirability function. The
Response surface methodology (RSM) is a collection of mathemati- goal of optimization is to find a good set of conditions that will
cal and statistical techniques. Originally, RSM was developed to model meet all the goals, not to get to a desirability value of 1.0. Desirability re-
experimental responses and then migrated into the modeling of flects the desirable ranges for each response (di). The simultaneous

Fig. 13. Contour plots for showing the effect of diameter and volume flow rate on: a) heat transfer coefficient and b) pressure drop.
104 S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105

Fig. 14. 3D graph for: a) heat transfer coefficient and b) pressure drop.

objective function is a geometric mean of all transformed responses: Maximum:

 n
1n di ¼ 0; Yi ≤Lowi
1  
D ¼ ðd1  d2  d3  …  dn Þn ¼ ∏ di ð13Þ Yi −Lowi wti
i¼1 di ¼ ; Lowi bYi bHighi ð14Þ
Highi −Lowi
di ¼ 1; Yi ≥Highi
where n is the number of responses in the measure. If any of the re-
sponses or factors falls outside their desirability range, the overall func-
Minimum:
tion becomes zero. For simultaneous optimization each response must
have a low and high value assigned to each goal. On the CCD worksheet,
the “Goal” field for responses must be one of five choices: “none”, “max- di ¼ 1; Yi ≤Lowi
imum”, “minimum”, “target”, or “in range”. In this study, the used goal  wti
Highi −Yi
parameters are maximum (for heat transfer coefficient) and minimum di ¼ ; Lowi bYi bHighi ð15Þ
Highi −Lowi
(for pressure drop) according their definition: di ¼ 0; Yi ≥Highi

where Yi is the ith response value and wt is the weight of that response.
Weights give added emphasis to the goal. Weights N 1 (maximum
weight is 10), give more emphasis to the goal and b1 (minimum weight
is 0.1), give less emphasis to the goal. In this paper, weights for both re-
sponses are considered the same and unit to have a linear function for di.
In this study, as shown in the results and discussion section, channel
diameter and volume flow rate have the most effect on thermo-
hydraulic performances of the heat sink. Because the effect of channel
diameter and volume flow rate was not very clear, an optimization anal-
ysis based on CCD technique is applied. In this optimization two re-
sponses are considered, heat transfer coefficient and pressure drop.
Fig. 13 is obtained from DOE software based on CCD analysis to show
the effects of channel diameter and volume flow rate on heat transfer
coefficient and pressure drop to find the optimum point to have maxi-
mum heat transfer coefficient and minimum pressure drop.
3D graphs for RSM analysis are shown in Fig. 14 which shows that
maximum heat transfer coefficient occurs in the approximately average
levels of factors (volume flow rate and channel diameter) and also min-
imum pressure drop happens in the approximately average levels. Be-
cause we want to reach the maximum heat transfer coefficient as well
as minimum pressure drop, after the CCD analysis, the best heat sink de-
sign is presented. As described in Section 5.2 for CCD technique, optimi-
zation is based on the desirability which high desirability means better
operating condition. Desirability contour for the best case is presented
in Fig. 15 that predicts the desirability for this operating point about
0.650 which is an acceptable optimization. For this value of desirability
the maximum heat transfer coefficient as well as minimum pressure
Fig. 15. Desirability contour for the optimized point. drop occur in Q = 0.92 m3/s and D = 0.541 mm.
S.E. Ghasemi et al. / Microelectronics Reliability 73 (2017) 97–105 105

6. Conclusion [18] M.R. Sohel, S.S. Khaleduzzaman, R. Saidur, A. Hepbasli, M.F.M. Sabri, I.M. Mahbubul,
An experimental investigation of heat transfer enhancement of a minichannel heat
sink using Al2O3–H2O nanofluid, Int. J. Heat Mass Transf. 74 (2014) 164–172.
The main purpose of this study was focused on experimental inves- [19] N. Sandeep, M. Gnaneswara Reddy, Heat transfer of nonlinear radiative magnetohy-
tigation of cooling performance of heat sinks with different hydraulic di- drodynamic Cu-water nanofluid flow over two different geometries, J. Mol. Liq. 225
(2017) 87–94.
ameter of channel. Tests were done for pure water as the coolant with a [20] S.E. Ghasemi, A.A. Ranjbar, A. Ramiar, Numerical investigation of effect of al-water
range of volume flow rates from 0.3 to 1.5 L/min. The thermal and hy- nanofluid on performance of solar parabolic collector, Nano 5 (14) (2013) 100–107.
drodynamic characteristics of the heat sink for various channel diame- [21] S.E. Ghasemi, A.A. Ranjbar, Thermal performance analysis of solar parabolic trough
collector using nanofluid asworking fluid: a CFD modelling study, J. Mol. Liq. 222
ter were presented and discussed in details. Also, Computational Fluid (2016) 159–166.
Dynamics (CFD) simulations were carried out to study the performance [22] S.E. Ghasemi, A.A. Ranjbar, A. Ramiar, Three-dimensional numerical analysis of heat
of heat sink, and the verification of numerical model was examined by transfer characteristics of solar parabolic collector with two segmental rings, J. Math.
Comput. Sci. 7 (2013) 89–100.
comparison with experimental results. Experimental results showed
[23] S.E. Ghasemi, A.A. Ranjbar, A. Ramiar, Numerical study on thermal performance of
that the heat transfer coefficient obtained from minichannel heat sink solar parabolic trough collector, J. Math. Comput. Sci. 7 (2013) 1–12.
with D = 4 mm is higher than that obtained from the minichannel [24] Seyed Ebrahim Ghasemi, Ali Akbar Ranjbar, Thermal efficiency evaluation of solar
heat sink with D = 6 mm and D = 8 mm. According to the measured rings in tubes, Eur. Phys. J. Plus 131 (2016), X. .
[25] Seyed Ebrahim Ghasemi, Ali Akbar Ranjbar, Numerical thermal study on effect of
data, the heat sink with D = 4 mm significantly lowered the heat sink porous rings on performance of solar parabolic trough collector, Appl. Therm. Eng.
base temperature in comparison with other two cases. Furthermore, 118 (2017) 807–816.
by decreasing the channel diameter, the system needs some extra [26] M.A. Akhavan Behabadi, M.K. Sadoughi, Milad Darzi, M. Fakoor Pakdaman, Experi-
mental study on heat transfer characteristics of R600a/POE/CuO nano-refrigerant
pumping power. Finally, a central composite design or CCD optimiza- flow condensation, Exp. Thermal Fluid Sci. 66 (2015) 46–52.
tion analysis was applied on the results to minimize the pressure drop [27] Darzi Milad, M.A. Akhavan-Behabadi, M.K. Sadoughi, Pooyan Razi, Experimental
and maximize the heat transfer rate in the heat sink. study of horizontal flattened tubes performance on condensation of R600a vapor,
International Communications in Heat and Mass Transfer 62 (2015) 18–25.
[28] M.A. Akhavan-Behabadi, M.K. Sadoughi, Milad Darzi, M. Fakoor-Pakdaman, A.
References Abbasi, Simultaneous effects of flattening tube and adding nanoparticles on boiling
heat transfer, J. Thermophys. Heat Transf. 31 (1) (2017) 78–85.
[1] Y.M. Chung, K.H. Luo, Unsteady heat transfer analysis of an impinging jet, J. Heat [29] M. Khoshvaght-Aliabadi, S.M. Hassani, S.H. Mazloumi, Comparison of hydrothermal
Transf. 124 (2002) 1039–1048. performance between plate fins and plate-pin fins subject to nanofluid-cooled cor-
[2] K. Nishino, M. Samada, K. Kasuya, K. Torii, Turbulence statistics in the stagnation re- rugated miniature heat sinks, Microelectron. Reliab. 70 (2017) 84–96.
gion of an axisymmetric impinging jet flow, Int. J. Heat Fluid Flow 17 (1996) [30] G. Takács, P.G. Szabó, Gy. Bognár, Enhanced thermal characterization method of mi-
193–201. croscale heatsink structures, Microelectron. Reliab. 67 (2016) 21–28.
[3] K.S. Kim, M.H. Won, J.W. Kim, B.J. Back, Heat pipe cooling technology for desktop PC [31] Mohsen H. Al-Rashed, Grzegorz Dzido, Mateusz Korpyś, Jacek Smołka, Janusz
CPU, Appl. Therm. Eng. 23 (2003) 1137–1144. Wójcik, Investigation on the CPU nanofluid cooling, Microelectron. Reliab. 63
[4] Y. Wang, K. Vafai, An experimental investigation of the thermal performance of an (2016) 159–165.
asymmetrical flat plate heat pipe, Int. J. Heat Mass Transf. 43 (2000) 2657–2668. [32] Seyed Ebrahim Ghasemi, A.A. Ranjbar, M.J. Hosseini, Numerical study on effect of
[5] Z. Zhao, C.T. Avedisian, Enhancing forced air convection heat transfer from an array CuO-water nanofluid on cooling performance of two different cross-sectional heat
of parallel plate fins using a heat pipe, Int. J. Heat Mass Transf. 40 (13) (1997) sinks, Adv. Powder Technol. (2017), http://dx.doi.org/10.1016/j.apt.2017.03.019.
3135–3147. [33] Seyed Ebrahim Ghasemi, A.A. Ranjbar, M.J. Hosseini, Thermal and hydrodynamic
[6] Yue-Tzu Yang, Huan-Sen Peng, Investigation of planted pin fins for heat transfer en- characteristics of water-based suspensions of Al2O3 nanoparticles in a novel
hancement in plate fin heat sink, Microelectron. Reliab. 49 (2009) 163–169. minichannel heat sink, J. Mol. Liq. 230 (2017) 550–556.
[7] Jung-Chang Wang, U- and L-shaped heat pipes heat sinks for cooling electronic [34] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Fluid Therm.
components employed a least square smoothing method, Microelectron. Reliab. Sci. 1 (1988) 3–17.
54 (2014) 1344–1354. [35] S.V. Patankar, Numerical Heat Transfer and Fluid Flow, Hemisphere, New York,
[8] R. Sajedi, B. Osanloo, F. Talati, M. Taghilou, Splitter plate application on the circular 1980.
and square pin fin heat sinks, Microelectron. Reliab. 62 (2016) 91–101. [36] Fluent Inc., FLUENT 6.3 User's Guide, Fluent Publishing Company, Lebanon, 2007.
[9] S.F. Sufian, M.Z. Abdullah, Heat transfer enhancement of LEDs with a combination of [37] A. Zolfagharian, P. Valipour, S.E. Ghasemi, Fuzzy force learning controller of flexible
piezoelectric fans and a heat sink, Microelectron. Reliab. 68 (2017) 39–50. wiper system, Neural Comput. & Applic. 27 (2016) 483–493.
[10] P. Cova, N. Delmonte, F. Giuliani, M. Citterio, S. Latorre, M. Lazzaroni, A. Lanza, Ther- [38] A. Zolfagharian, A. Noshadi, S.E. Ghasemi, M.Z.Md. Zain, A nonparametric approach
mal optimization of water heat sink for power converters with tight thermal con- using artificial intelligence in vibration and noise reduction of flexible systems,
straints, Microelectron. Reliab. 53 (2013) 1760–1765. Proc IMechE Part C: J Mechanical Engineering Science 228 (8) (2014) 1329–1347.
[11] M. Conrad, A. Diatlov, R.W. De Doncker, Purpose, potential and realization of chip at- [39] A. Zolfagharian, S.E. Ghasemi, M. Imani, A multi-objective, active fuzzy force control-
tached micro-pin fin heat sinks, Microelectron. Reliab. 55 (2015) 1992–1996. ler in control of flexible wiper system, Latin American Journal of Solids and Struc-
[12] A. Ijam, R. Saidur, P. Ganesan, Cooling of minichannel heat sink using nanofluids, Int tures 11 (2014) 1490–1514.
Commun Heat Mass Transfer 39 (2012) 1188–1194. [40] A. Zolfagharian, A. Noshadi, M.R. Khosravani, M.Z.Md. Zain, Unwanted noise and vi-
[13] A. Ijam, R. Saidur, Nanofluid as a coolant for electronic devices (cooling of electronic bration control using finite element analysis and artificial intelligence in flexible
devices), Appl. Therm. Eng. 32 (2012) 76–82. wiper system, Appl. Math. Model. 38 (2014) 2435–2453.
[14] X.D. Wang, B. An, D.J. Lee, Inverse geometric optimization for geometry of [41] N. Aslan, Application of response surface methodology and central composite rotat-
nanofluid-cooled microchannel heat sink, Appl. Therm. Eng. 55 (2013) 87–94. able design for modeling the influence of some operating variables of a multi-
[15] X. Wu, H. Wu, P. Cheng, Pressure drop and heat transfer of Al2O3–H2O nanofluids gravity separator for coal cleaning, Fuel 86 (2007) 769–776.
through silicon microchannels, J. Micromech. Microeng. 19 (2009) 105020. [42] Lei Sun, Chun-Lu Zhang, Evaluation of elliptical finned-tube heat exchanger perfor-
[16] A. Raisi, B. Ghasemi, S. Aminossadati, A numerical study on the forced convection of mance using CFD and response surface methodology, Int. J. Therm. Sci. 75 (2014)
laminar nanofluid in a microchannel with both slip and no-slip conditions, Numer- 45–53.
ical Heat Transfer, Part A 59 (2) (2011) 114–129.
[17] Chonghua Gao, Yongjun Jian, Analytical solution of magnetohydrodynamic flow of
Jeffrey fluid through a circular microchannel, J. Mol. Liq. 211 (2015) 803–811.

You might also like