Experimental investigation for sequential triangular

International Communications in Heat and Mass Transfer 77 (2016) 104–115

Contents lists available at ScienceDirect

International Communications in Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ichmt

Experimental investigation for sequential triangular double-layered microchannel heat sink with nanofluids☆ Hamdi E. Ahmed a,b,⁎, M.I. Ahmed a, Islam M.F. Seder a, B.H. Salman c,d a

Department of Mechanical Engineering, International Islamic University Malaysia, 50728, Gombak, Malaysia Department of Mechanical Engineering, University of Anbar, Ramadi 31001, Iraq Department of Mechanical Engineering, University of Nevada, Las Vegas, NV 89154, USA d DNV GL, Engineering Department, Las Vegas, NV 89146, USA b c

a r t i c l e

i n f o

Available online 21 June 2016 Keywords: Double-layer Microchannel heat sink Sequential triangular Nanofluids

a b s t r a c t This study examined new innovative design of aluminum rectangular and triangular double-layered microchannel heat sink (RDLMCHS) and (TDLMCHS), respectively, using Al2O3–H2O and SiO2–H2O nanofluids. A series of experimental runs for different channel dimensions, different nanoparticles concentrations and types and several pumping powers showed excellent hydrothermal performance for DLMCHS over traditional single-layer (SLMCHS). The results showed that the sequential TDLMCHS provided a 27.4% reduction in the wall temperature comparing with RDLMCHS and has better temperature uniformity across the channel length with less than 2 °C. Sequential TDLMCHS provided 16.6% total thermal resistance lesser than the RDLMCHS at low pumping power and the given geometry parameters. Pressure drop observation showed no significant differences between the two designs. In addition, larger number of channels and smaller fin thickness referred less thermal resistance rather than only increasing the pumping power. Higher nanoparticle concentration showed better thermal stability for both nanofluids than pure water. The Al2O3–H2O nanofluid (0.9 vol.%) showed best performance with the temperature difference of 1.6 °C and lowest thermal resistance of 0.13 °C/W·m2. © 2016 Published by Elsevier Ltd.

1. Introduction Thermal management is a vital part to maintain the functionality and durability of any successful engineering device. As the world goes toward smaller in size, cheaper devises, lighter in weight, and higher in efficiency, microelectronic mechanical systems enable an innovative cooling design to be used in miniature designs, especially electronic applications. Enhancing the performance of electronic chips requires utilizing more resistors which causes massive heat generated in these devices. This pushed researchers to adapt more efficient cooling system. In 1981, Tuckerman and Pease [1] utilized the concept of micro heat exchanger in electronic cooling, opening the door to a new mechanism in dissipating heat rather than the traditional low-thermal capacity of air heat sink. Another important factor in increasing the thermal performance of the cooling system is adapting a proper coolant. It has been proved that different fluids have different flow properties that strongly affect the overall performances. Recently, nanofluids have been used in some engineering applications in order to enhance the thermal conductivity of the based-fluids [2]. Heat transfer coefficient can be improved ☆ Communicated by W.J. Minkowycz. ⁎ Corresponding author at: Department of Mechanical Engineering, International Islamic University Malaysia, 50728, Gombak, Malaysia. E-mail address: [email protected] (H.E. Ahmed).

http://dx.doi.org/10.1016/j.icheatmasstransfer.2016.06.010 0735-1933/© 2016 Published by Elsevier Ltd.

by increasing the ratio of channel surface area to channel volume and also the properties of fluids decide the abilities of fluid to remove heat by influencing the heat transfer coefficient. Thus, improvement in overall heat transfer performance must be concentrated on dimensions of the channel and type of fluids used in heat exchanger. Two main challenging obstacles are still faced in this filed. First, the high pressure drop between the inlet and outlet of the channel which imposes using high pumping power. Second, the large difference in temperature distribution along the channel which increase thermal stresses in the cooled-electronic elements thus reduces the electrical performance due to the electrical–thermal instability and thermal breakdown. This pushed engineers to move from traditional singlelayer microchannel to double or multi-layer microchannel. The noncircular geometry of microchannels are widely used such as rectangle, trapezoid, triangle, and diamond. Moreover, width of channel and width of fin are among critical parameters that are usually investigated since they strongly affect the thermals conductivity and convection coefficient of heat transfer. Double-layered microchannel heat sink (DLMCHS) was suggested by Vafai and Zhu [3] to represent an innovative design that adds more features for cooling system in miniature devices where the pressure drop was deceased dramatically and the heat was better distributed along the MCHS length. Gunnasegaran et al., [4] showed that the smallest hydraulic diameter (Dh) of the MCHS provided better

H.E. Ahmed et al. / International Communications in Heat and Mass Transfer 77 (2016) 104–115

Symbols A Cp Dh DW f g h H I k L, l M N Nu PP Pr ΔP q Q R R Re RWP T t u V vol W x, y, z

area, m2 Specific heat capacity, J/kgK hydraulic diameter, m distilled water Fanning friction factor gravitational acceleration, m/s2 convection heat transfer coefficient, W/m2·°C channel height, m current, A thermal conductivity, W/m·°C channel Length, mm molecular weight, g number of channel Nusselt number pumping power, W Prandtl number pressure drop, Pa heat flux, W/m2 heat transfer rate, W thermal resistance, 'C/W.m2 variable Reynolds number rectangular winglet pair temperature, K thickness, m velocity, m/s voltage, v volume fraction weight, g 3D Cartesian coordinate

Greek symbols ρ density, kg/m3 μ dynamic viscosity, kg/m·s Subscripts avg average bf base fluid c channel eff effective f fin f fluid in inlet m mean nf nanofluid out outlet p nanoparticle s surface tot total w water

uniformities in heat transfer coefficient and temperature distribution and better performance of pressure drop and friction factor. It was also shown that the higher heat transfer coefficients were for rectangular channels, followed by trapezoidal and then triangular channels. Alfaryjat et al. [5] concluded that hexagonal MCHS was the best channel shape for the heat transfer coefficient and pressure drop compared to circular and rhombus channels. However, rhombus cross-section MCHS was the best channel in terms of temperature, friction factor, and thermal resistance. Hung et al. [6] and Hung and Yan [7] illustrated that the number of channel, channel upper and lower width ratio, channel lower and

105

lower AR were crucial parameters for heat transfer augmentation. Thus, optimal values of the geometric parameters could be obtained to reach the lowest thermal resistance. Hung et al. [8] showed that DLMCHS provided higher heat transfer by about 6.3% than SLMCHS with significant decrease in pressure drops. One critical drawback of SLMCHS was the dramatic rise in temperature along the microchannels due to relatively small amount of fluids that take heat generated from the electronic chips, thus the coolant experiences a massive temperature increment. Lin et al. [9] found that channels number, coolant velocity in the lower channel, bottom channel height, and vertical rib affected the values of pumping powers of DLMCHS. Hung et al. [10] emphasized that the optimal design depends on the pumping power of MCHS. Sakanova et al. [11] reported that DLMCHS caused a reduction of 15% in thermal resistance compared to SLMCHS. Moreover, better uniformity for temperature distribution was obtained for DLMCHS leading to enhance the performance of the electronic chips and semiconductor devices. Wei et al. [12] showed that DLMCHS provided larger flow passages than SLMCHS, thus the penalty of pressure drop was dramatically reduced. The thermal resistance was also reduced as the flow rate increased inside the lower-layer microchannels than through the upper layer. Wang et al. [13] emphasized that as the pumping power increased, the heat transfer was remarkably improved. The surface roughness affects the values of the friction factors even under laminar flow condition as Weilin et al. [14] and Wu and Cheng [15] stated. They displayed the dependency of Nusselt number and friction factor on geometric shape of microchannel. Comprehensive review, done by Liu and Garimella [16] ascertained that conventional correlations provide consistent prediction for low Reynolds number regime while Lee and Garimella [17] found that the measured Nusselt number agrees with prediction ones over the whole length of the micro-channel. Thus, the correlations of heat transfer coefficient, which can be found by Nusselt number using Eq. (6). Nusselt number derived by Shah and London [18] and Kays and Crawford [19] is 2 Nu ¼ 4 2:22

31=3 −0:33 !3 l þ ð−0:02 þ 8:31 GÞ3 5 Re Dh Pr

ð1Þ

Sakanova et al. [20] tested the application of nanofluids in wavy channel structure having different wavy amplitudes and wavelengths. They found that the wavy microchannels outperformed the straight microchannels thermally using water as coolant. Moreover, this thermal improvement was enhanced when water was replaced by nanofluid. Hung et al. [21] reported that by using base fluids (having lower dynamic viscosity such as water) and substrate materials (having high thermal conductivity) enhanced the thermal performance of the MCHS. Nanofluids enhanced the thermal performance of MCHS better than water particularly when the volume fraction increased. Ho et al. [22] and Ho and Chen [23] recorded a slight increase in the friction factor of alumina oxide nanofluids flow in copper MCHS compared to water. Besides, a significant enhancement in heat transfer rate (lower thermal resistance and wall temperature) was registered with nanofluids. Mohammed et al. [24] recorded an increase in both the heat transfer coefficient and wall shear stress when the volume fraction increased whereas a slight increase in the pressure drop across the MCHS was monitored compared to the base fluid. Yang et al. [25] found that the thermal resistance of trapezoidal MCHS was smaller compared when water was used, which decreased as the particle concentration and Re number increased. While a small increase in the pressure drop was recorded with increasing the concentration. Chen and Ding [26] reported that the effect of fluid inertia caused a reduction in the total thermal resistance and the temperature difference between the channel wall and nanofluid phase. Halelfadl et al. [27] observed that nanofluids reduced the total thermal resistance and enhanced significantly the thermal performances of a rectangular MCHS

106


at high temperatures. Tokit et al. [28] pointed out that highest thermal enhancement of interrupted rectangular MCHS was registered with Al2O3, followed by CuO, and finally for SiO2. This enhancement was increased with nanoparticle concentrations and reduced with nanoparticles diameter. Farsad et al. [29] presented that Al2O3–water (8 vol.%) nanofluid in zigzag MCHS offered the highest heat transfer compared with CuO– H2O and Cu–H2O nanofluids. Kalteh [30] investigated numerically the effect of Al2O3, CuO, Cu, Fe, Au, Ag, TiO2, SiO2 and diamond dispersed in ethylene–glycol, water and engine oil base fluid on heat transfer and pressure drop characteristics. They reported that the pressure drop in MCHS was almost equal and the heat transfer depended on the type of nanofluid. Lee and Mudawar [31] and Wu et al. [32] proved that nanoparticles affected significantly the thermal resistance of MCHS. Pantzali et al. [33] found that nanofluid removed the same amount of energy by using a significantly lower flow rate and required less pumping power. Rafati et al. [34] stated that the heat transfer of computer microchip was greater when nanofluid used due to the reduction in the operating temperature of processor. Wu et al. [35] observed that Al2O3–water nanofluid in a trapezoidal MCHS experienced almost the same pressure drop with remarkable enhancement in the heat transfer compared to water. This enhancement in the heat transfer was observed higher at higher nanoparticle volume fraction. From the above open literatures, it can be said that channel configuration and coolant type are crucial optimization factor in providing maximum heat transfer in MCHS. Triangle, trapezoid and rectangle are widely adapted in microchannel. While single-layer microchannel heat sink (SLMCHS) has a serious limitation in providing uniform temperature across the channel length and imposing high pressure drop. Double-layer seems to be the futuristic configuration of MCHS. However, DLMCHS is still in immature level of understanding for its thermal and hydrodynamic performance. Although, there are some interesting numerical studies for double-layered MCHS, the experimental study side that represents a significant validation method is still ambiguous and few works have been achieved. Thus, the objective of adapting DLMCHS is to reduce the undesired temperature variation and minimize the pressure drop in the coolant flow by changing the structure and geometry of the microchannels. However, this paper focused on rectangular and triangular DLMCHS having rounded corners using Al2O3–H2O and SiO2–H2O (40 nm in diameter and 0.3, 0.6 and 0.9 vol.%) with Reynolds numbers ranges from 50 to 300. 2. Experimental set-up The experimental cooling system is a closed–loop fluid flow circuit. It consists of condenser (radiator), base block, fluid pump, fluid tubes, and flowmeter and reservoir as shown in Fig. 1(a) and (b). The base block includes a substrate microchannel, electrical heater, Pyrex-glass insulator and several shapes and sizes of DLMCHS. A DLMCHS having an area of 25 × 25 mm2 was fabricated with a semicircular-corner shape having 30 slots. A minimum 100 μm wire diameter (that represents a slot width) was considered. The pump has adjustable meter to provide varied volumetric flow rates depending on the test requirement. After the hot fluid leaving the test section, the fluid was cooled down by the heat exchanger then it was stored in the reservoir. The inlet and outlet manifold of the test section has an identical geometrical shape. The base block size was made with size of 130 × 75 × 20 mm3. An aluminum substrate microchannel was directly attached to the electrical heat source which provides temperature range from 20 to 160 °C. Then, the heat was transferred from the hot microchannel fins to the relatively cooled-flowing fluid. The microchannel and the heat source were covered by an Acrylic glass (160 °C melting point) to prevent heat loss from the base block to surroundings. The place of microchannel and heat source was carried out in the Acrylic glass block as a cube shape groove with a cross-

section of 25 × 25 mm2 (length and width), respectively. This groove was machined carefully to get a precise edges and perfect angles at four corners of cubic groove so the microchannel and the heat source were precisely inserted as shown in Fig. 1(c). The inlet and outlet manifold were drilled in the Acrylic glass from top of base block. The small distance between the inlet and the gate of microchannel results in negligible heat transfer losses. The internal surfaces of the groove were especially treated to get a high degree of smoothness and to make sure that there was no chips or burrs that may disturb the flow. The inlet and outlet of manifolds were identical in size (50 × 25 × 2 mm3). Flexible tubes were connected perpendicularly to the base block by two threaded-holes in the inlet and outlet manifold having 0.40 mm in diameter. Pressure transducer (OMEGA Engineering INC, model PXM6000MC) was connected to the inlet and outlet manifold to measure the difference in pressure drop in the microchannel. Similarly, two thermocouples were inserted in the middle height of the inlet and outlet manifolds to read the bulk fluid temperature. A schematic of MCHS with main dimensions is shown in Fig. 1(d). The Wire-cut EDM machine produced a rounded-end corner for the rectangular shape while end-mill process generated almost a 90° corners. The shape of the channel corner has an influence on the heat transfer performances, however, for high AR of rectangular channels, considering the rounded-end corner channel as a rectangular shape is acceptable and caused insignificant effect on behavior of the fluid. The advantage of rounded-end corner channel is easy to be fabricated and it consumes less time in micro-fabrication. Fig. 2(a)–(d) illustrates the scanning electron microscope (SEM) image for the RDLMCHS and TDLMCHS after connecting and attaching with the substrate. The processes of fabrication of DLMCHS have been followed according to that published by Seder et al. [36]. In the case where the microchannels were attached without connection mechanism or welding to the heat source, small gap may cause high thermal resistance and degrades the overall heat transfer rate. Lu et al. [37] utilized bonding process where upper heating station was carried to bond the base with micro-heat exchanger. However, in this work, high thermal compounds (20 W/cm2) was used for this duty and excellent results were obtained. It is worth to mention that the minimum fin thickness obtained by Wire-cut EDM machine was 150 μm. For surface properties, surface roughness measuring machine was utilized to measure the value of surface roughness. A non-contact optical profiler system allows an accurate measurement and wide ranges of surface conditions. A super-clean reference surface was required to reduce potential errors. High degree of accuracy was conducted to make sure the smoothness of internal surfaces of microchannel using noncontact optical profiler system. Surface finishing with an approximate ± 0.025 mm accuracy was achieved. Scanning electron microscope (SEM) was employed to the analysis of the burr formation throughout a series of images that show top, bottom and entrance views of upper and lower layer microchannel. Although the primary result showed that the top burr was the largest in the size, the entrance and bottom burr formation have higher influence in hydrothermal performance of the DLMCHS. Thus, more intention was given to the entrance sides. It is essential to mention that the depth and width of the micro-slot cut were ranged from 200 μm to 500 μm as recommended by Hung et al. [6] and [8], and Sakanova et al. [11]. The number of channels that is listed in Table 1 represents one of single layer (lower or upper layer). The total area of rectangular and triangular channels was kept constant in order compare between their performances. Where R and T represent the rectangular and triangular channel, respectively. The aluminum DLMCHS having size of (25 × 25 × 2) mm3 was covered by thin aluminum flat-plate with thickness of 2 mm from both upper and down sides. The grooves in the substrate and the covers were forming the microchannels as illustrated in Fig. 1(e). Thus, the electrical heater (sandwich) was attached on the outer surface of the lower cover plate exactly. The plate and the electrical heat source,


107

Fig. 1. (a) Scheme diagram and (b) photograph of the experimental set-up; (1) data logger, (2) fluid pump, (3) power supply, (4) reservoir, (5) radiator, (6) cooling fans, (7) pressure transducer, (8) base block, (9) battery, (10) voltage and (11) AC power source, (c) electrical heater inserted inside Acrylic glass block, (d) base block with pressure transducer and inlet manifold, (e) side view of upper portion of DLMCHS, and (f) thermocouples installation.

which provides uniform heat flux, were having the same circumference dimensions (length and width) of bottom of the DLMCHS, so they can be fitted inside the groove in the Pyrex glass perfectly. Four holes (0.36 mm in diameter) were drilled into the side wall of the Acrylic glass to install four thermocouples (Type-K) with a 0.33 mm probe diameter. Four thermocouples were inserted at the bottom of microchannel (lower cover plate) in arrangement of 5 mm distance between each two thermocouples to measure the distribution of temperature along the axial direction of the DLMCHS. The probe of the thermocouples was inserted 3 mm in depth of the cover plate as illustrated in Fig. 1(f). High conductive glue was used to make sure that the thermocouples were properly attached to the bottom of the microchannel. The upper and bottom sections of Acrylic glass were fixed by drilling 12 holes, and connected by screws, nuts, and steel washers at the circumference of the block. In addition, epoxy was used to seal the contact edges to avoid any leak. The fluid was forced by pump drive (Cole Parmer, USA) with a variable angular speed from (0 to 5000 r.p.m). The fluid pump was

connected to the closed-loop system through the inlet and outlet diameter. The pump was adjustable to control the volumetric flow rate of the fluid according to test requirement. The fluid was circulated from a reservoir through a 6.35 mm diameter flexible-plastic tube. The hot fluid left the microchannel toward the heat exchanger for cooling until getting the required inlet ambient temperature. A multiple-data acquisition system (ECOLOG, Taiwan), which was connected to a PC, was used to record the pressure drop and temperature at the inlet, outlet and along the microchannel length as well as the electrical heater temperature. Voltage regulator (224 V, China) was used to regulate the alternating current for the heat source and other devices. The current and voltage were measured by using multimeter (G W INSTEK). The range of the current was from 0.0 A to 2 A with an accuracy of 0.001 A, while the range of voltage was from 0 V to 600 V with accuracy of 0.1 V. The flow rate of nanofluids was measured by using stopwatch having 0.01 s accuracy and scale beaker with 0.01 l scale. The flow rate was calculated by measuring an amount of nanofluid in a certain time.

108


Fig. 2. SEM of (a) rectangular microchannel, (b) rounded-end microchannel, (c) rectangular DLMCHS, and (d) zigzag TDLMCHS.

This process was performed for each test many times to ensure that the fluid flow was kept constant. While in case of water, a normal rotameters was adapted to record the water flow rate. The system has been left until getting the steady-state condition at which the measured pressure difference and temperature remain constant. In addition, the ambient temperature of the room was kept constant at (Tin = 25 °C) throughout experiments. Measurement devices need to be checked and compared with standard devices and adjusted, calibration, so the accuracy of the measuring instruments was guaranteed. The main measurement devices were the thermocouples, pressure transducer and data acquisition system. Data Acquisition System having the type of (Brainchild Electronic Co., Ltd). 2.1. Nanofluids preparation The Al2O3 and SiO2 nanoparticles were dispersed in distilled water to synthesize the required nanofluid. It was left in the Ultrasonic Bath Cleaner (Fisher Scientific model, S30H, 280 W, Germany) for 12 h. Ultrasonic bath guaranteed a homogeneous dispersion for nanoparticles throughout the base fluid since it has huge transducer areas and tanks that produce a high powered ultrasonic intensity throughout the entire nanofluid. It was observed that there was no sedimentation for all types of nanofluids and for all concentrations. The prepared samples are left for several days to make sure of the nanofluid stability. By using the Table 1 Geometric parameters of the DLMCHS.

following equation from the paper of Sundar and Sharma [38] and by knowing the density of water and nanoparticle, there is only one unknown, the weight of the nanoparticle powder. Wp ¼

ρp ∅ Ww ð1−∅Þ ρw

ð2Þ

where Ww and Wp represent the weight of the water and powder, respectively. After knowing the required amount of nanoparticle, Sartorius balance was used to obtain this amount. Table 2 shows the sample of prepared nanofluid with corresponding weight and volume fraction. The thermophysical properties of the nanofluids (effective thermal conductivity, effective viscosity, and effective density) were measured at room temperature which represents the inlet fluid temperature for all tests. The wall heat flux was ranged from 0.0 to 272.0 kW/m2 and the volumetric flow rate was ranged from 1.1 × 10−4 to 5.5 × 10−4 m3/min. 2.2. Data reduction The heat supplied to the electrical heater was estimated by Q PW ¼ I V

ð3Þ

However, the total amount of heat absorbed by the fluid in the DLMCHS was found by _ p ðT out −T in Þ Q out ¼ mc

Channel shape

Wf (μm)

Wb (μm)

H (mm)

Ach (μm2)

Atotal (mm2)

L (mm)

Dh (μm)

N

T T T T R R R

150 250 350 450 652 652 654

455 455 455 455 350 350 350

1.67 1.67 1.67 1.67 0.575 0.575 0.575

375.75 375.75 375.75 375.75 187 187 187

18.036 16.533 15.03 13.527 18.036 16.533 15.03

25 25 25 25 25 25 25

340 340 340 400 400 400 400

24 22 20 18 48 44 40

ð4Þ

Table 2 Percentage amount of nanoparticle dispersed in (1000 g) of distill water, all weight amounts in grams. Al2O3

SiO2 Nanofluid

0.3%

0.6%

0.9%

0.3%

0.6%

0.9%

Weight

6.6424

13.3251

20.0479

11.9865

24.0455

36.1774


_ is the mass flow rate which is found by m _ ¼ ρAu, Cp is the spewhere, m cific heat of the fluid, Tout and Tin are the outlet and inlet channel temperature, respectively. Heat transfer by convection to the fluid was calculated by Q out ¼ hA T s −T f havg

_ p ðT out −T in Þ mc ¼ A T s −T f

T s ¼ T tc3

q d − eff ks

ð5Þ

ð7Þ

Secondly, Nusselt number where the ratio of convection heat transfer from the solid fins to the conductivity of the working fluid was calculated as: Nuavg ¼

havg Dh k

ð8Þ

where Dh is the hydraulic diameter. The symbol k is the thermal conductivity taken as an effective thermal conductivity in the cases of nanofluids. Then, friction factor, which is a function of physical feature of the channels was found by f ¼

2 Dh ΔP ρ L u2m

ð9Þ

ΔP is the different in pressure between the inlet and outlet of the heat sink, and L is the length of the heat sink. The symbol of um represents the mean velocity of the fluid. While Reynolds number was calculated as ρ u m Dh Re ¼ μ

ð10Þ

The average temperate of the fluid was found by Tf ¼

T out −T in 2

ð11Þ

The net heat loss that has been removed by the fluid was found by qloss ¼ qheater −qout

length; mass flow rate; specific heat (Twall − Tinlet), and (Twall − Toutlet). While, friction factor consists of the uncertainties from width, height, length, density, mass flow rate and pressure drop. The error value of any variable (R) was calculated from the following equation: " 2 #1=2 Xn ∂R U x i i¼1 ∂x i

UR ¼ ∓ ð6Þ

ð12Þ

1 L þ hA f ks As

3. Result and discussion The influence of geometrical parameters especially number of channel and fin thickness is explained. The effect of utilized water, Al2O3–H2O, SiO2–H2O nanofluids were studied at different operating conditions such as Reynold number, pumping power, and different nanoparticle volume concentrations. Validation of the results were included, thus allowing to compare the present study with previous works. 3.1. Influence of geometrical shape The performance of RDLMCHS and TDLMCHS was compared under same operating conditions where distilled-water was the flowing

ð13Þ

Number of channel for lower or upper layer of DLMCHS was N¼

L Wc þ W f

ð14Þ

2.3. Uncertainties and error analysis Construction of the cooling system and collection data from the test block were given special care to obtain accurate dimensions and results. Uncertainty of Nusselt number, Reynolds number, and friction factor were the main parameters to be evaluated. As these parameters consist of several experimental variables, the Kline and McClintock method (Holman [39]) was suitable to conduct the uncertainties analysis. Nusselt number consists of the uncertainties from width; height;

ð15Þ

After derivation, that was omitted for brevity, the maximum uncertainties of Nusselt number, Reynolds number, and friction factor were 2.83%, 3.77%, and 4.25%, respectively. Analysis of heat losses/gains with the surrounding was evaluated at less than 3% of the given heat by the heater which was considered negligible as shown in Fig. 3. Uncertainties analysis related to the wattmeter and pressure transducers was less than 0.6% and 2.5%, respectively. Because the wall temperatures of the DLMCHS were not measured directly for difficulty reasons, the underside base-metal temperature was rather measured. The distance (thickness) between the measured and the calculated points was 2 mm, which was relatively much smaller than the total length and the width of DLMCHS (25 × 25 mm2), meaning that the gradients of the temperature between the base-metal and the microchannel is greater than the gradients of the temperature in length and width path. For this reason, the thermal conduction occurred in the thickness path only was considered here. However, thermocouple measurement errors were estimated at less than 0.4 K. Thus, the measured temperature by thermocouples could be used as the wall temperature of DLMCHS. Whereas the errors in pressure drop readings did not exceed 1%.

where qheater is the power supply to the electrical heater. Total thermal resistance was calculated by summing both parts of heat transfer mechanism (convection and conduction) Rtot ¼

109

Fig. 3. Electrical power supplied to the heater and heat received by the fluid.

110


coolant. Three samples for each shape having same channel surface area were analyzed as explained earlier. First, the distribution of base temperatures across the channel length was illustrated in Fig. 4(a). At constant heat flux input (115 W) and lowest pumping power (0.1 W), RDLMCHS outperformed TDLMCHS under same surface area condition (i.e., same channels number). Increasing number of channel (N) caused a slightly reducing in microchannel wall temperature for TDLMCHS and provided more temperature uniformity along channel length. The percentage of reduction in wall temperature exceeded 27.4% when number of channel increased from 20 to 24 channels. This significant reduction was attributed to the decrement in the fin thickness (solid part of the microchannel) at the expense of increasing channel surface area. Therefore, the convection heat transfer coefficient (h) was increased. In all cases, the maximum temperature of the channel wall was reported after the middle of channel, matching with the result obtained by Wu et al. [40]. Enlarging the pumping power (PP) from 0.1 to 0.5 W influenced the temperature distribution as shown in Fig. 4(b). The maximum wall temperature was located before the middle of the channel. The variations of temperature exhibited a less difference which led to more thermal stability. For rectangular shape, 44 channels provided less than 2 °C temperature difference which was considered an excellent value for electronics chips. Fig. 5(a) shows the effect of channel number on the total thermal resistance for both channels shapes. However, the lowest overall thermal resistance was only associated with increasing number of channel for triangular shape. Smaller number of triangular channels might degrade the heat transfer rate as shown in the case of N = 20. It can be also ascribed to the effect of fin thickness. Increasing the number of channel at a given total channel width (25 mm) increased the convection heat transfer coefficient since the total surface area of the channels was increased. For rectangular shape, increasing number of channels yielded in increasing the thermal resistance because thermal conductivity became very small as the fin thickness decreased (increasing number of channel was at the expanse of reducing fin thickness). Therefore, it was concluded that there was a converse behavior between rectangular and triangular shapes, although they provided the same pumping power and total surface area. It is believed that the relatively high thickness layer between upper and lower channels in rectangular configuration worsened the heat transfer rate and gave the advantage to zigzag triangular channel. The effect of geometry was confirmed to have great influence on overall heat transfer. Escalating pumping power from 0.1 to 0.5 W showed enhancement in overall heat transfer rate for all samples studied. This enhancement was more obvious in

rectangular shape where thermal resistance of (0.16 °C/Wm2) was obtained as shown in Fig. 5(a). On other hand, the relation between Reynolds number and Nusselt number is shown in Fig. 5(b). The growing trend in Nu number as the Reynolds number increased revealed the significant of increasing flow velocity inside the microchannel since the viscosity of the water and hydrodynamic diameter were almost fixed. For the very small range of Reynolds number (Re b 250), increasing the Reynolds number had almost the same effect for both channel shapes. The low range of Reynolds number limited the augmentation of Nu number; however, it consumed less pumping power. In the experiment, it was avoided using more than 0.5 W to keep up with the objective of enhancing the performance with reasonable increase in pumping power. At 0.5 W pumping power case, Reynolds number of 230 resulted in highest value of Nu number, (1.75) for 40-channels rectangular samples. To analyze the hydrodynamic performances, pressure drop was plotted against pumping power available in the gear pump as shown in Fig. 5(c). As the pumping power increased, the pressure drop trend was slightly increased for both channel shapes. At low pumping power (PP b 0.3), it was apparent that zigzag triangular shape suffered more pressure drop than rectangular shape when both have same number of channels, nevertheless, triangular shape returned to behave better in higher pumping power (PP N 0.5). Rectangular geometry behaved oppositely to triangular one where it experienced a less pressure drop at low pumping power. Increasing number of channel imposed more exist ways for the flow to pass the microchannel and, hence less pressure drop was obtained. 3.2. Zigzag triangular DLMCHS This section focuses only on triangular shape channels as it was new adapted modified DLMCHS. The channel number was varied at a given total microchannel width of 25 mm and fixed heat flux of 115 W. It should be mentioned that the variation in channel number caused change in fin thickness of the channel, thus changes the overall heat transfer characteristics. The effect of channel number was studied throughout manufacturing and testing four different DLMCHS. The designed DLMCHS had 18, 20, 22 and 24 channels. The cross section area of each channel was kept constant for all samples; therefore increasing the channel number was at the expense of decreasing the fin thickness. Fig. 6(a) shows the trends of base temperature distribution across the channel length for water and (0.6%) SiO2–H2O nanofluid under pumping power of 0.3 W,

Fig. 4. Substrate temperature distribution along the length of rectangular and triangular DLMCHS, (a) PP = 0.1 W, and (b) PP = 0.5 W.


111

Fig. 5. Effect of (a) PP on Rtotal, (b) Reynolds number on Nusselt number, and (c) PP on pressure drop, for rectangular and triangular DLMCHS.

compared with the results of Lei et al. [41]. The figure reveals the great uniformity in temperature variation across the channel length for all the samples tested, N = 18–22. The highest difference in temperature between the inlet and outlet tips of the channel was recorded for the case of N = 22 where water was the coolant fluid. Similar temperature trends were obtained with changing the number of channel, disclosing the independency of the channel number and fin thickness on temperature variation. Comparing the zigzag triangular shape (current work) with rectangular shape done by Wu et al. [40] revealed the significant advantages of using triangular DLMHHS in providing better temperature distribution which was a key role in stability of the electronic parts. Thus, less thermal stress was obtained across the electronics chips and longer device life was expected. For cases where N = 18, 20, and 24, the maximum temperature was recorded after the middle of the channel length which was considered one of the advantages of DLMCHS over single layer as shown in literature. This result was completely matched with numerical work done by Hung et al. [6] and Hung et al. [7]. Fig. 6(b) shows another crucial parameter to overall performance, which is the thermal resistance versus the pumping power. It was shown that increasing pumping power reduced the total thermal

resistance regardless to the number of channel which agreed with almost all previous works shown in the open literature. However, at lager number of channels (smaller fin thickness channels), the pumping power provided no much significant influence to the total thermal resistance. Using distilled-water, the thermal resistance decrement for DLMCHS with 24 channels was calculated to be 10% when the pumping power increased from 0.1 W to 0.5 W, while N = 18, the decrement was observed to be 15% under same conditions. The importance of this result revealed that one can adapt larger number of channel and smaller fin thickness to have less power consumption. It is worth to mention that the percentage of thermal resistance reduction contradicted with Hung et al. [10] who obtained a 30% numerically under same operating conditions. Fig. 6(c) illustrates the pressure drop across the inlet and outlet of the microchannel was plotted against the pumping power available in the gear pump for several channels. At the same pumping power, larger number of channels provided a less press drop. The DLMCHS with N = 24 showed a reduction in the pressure drop of 13% less than the DLMCHS having N = 18. Increasing the pumping power influenced the pressure drop in similar trend for all samples tested indicating that pressure drop depended on configuration of the channels more

112


Fig. 6. (a) Substrate temperature distribution along the DLMCHS compared to Lie et al. [41], and (b) effect of PP on Rtotal, for different channels numbers at 0.0 vol.% and 0.6 vol.% SiO3–H2O nanofluids, (c) Effect of PP on pressure drop in DLMCHS for different channels numbers.

than their numbers especially when cross section of all the channels was the same. 3.3. Influence of nanofluid on temperature uniformity along the channel Two nanoparticle types; Al2O3 and SiO2 were immersed in distilled water with different concentrations. Fig. 7(a) shows the base temperature variation along the length of DLMCHS where different coolant fluids were utilized. Generally, zigzag Triangular DLMCHS provided a unique temperature variation. The difference in temperature along the channel length was very small and did not exceed (5 °C) in highest value. Fixing the pumping power at 0.3 W and measuring the temperature at its distance from the inlet (5, 10, 15, and 20 mm) showed that Al2O3–H2O (0.9 vol.%) nanofluid experienced the best performance in terms of base thermal stability since the temperature difference was in 1.5 °C range. Increasing the nanoparticle concentration in water caused a better thermal stability for both nanofluids studied here than pure water. A comparison was also carried out between the temperature distribution of the SLMCHS of the current work with the results of Bhattacharya et al. [42]. In Bhattacharya's work, the RSLMCHS provided a big difference between beginning and final base temperature around 7 °C. Considering that the nanofluid is a homogenous fluid,

Kalteh et al. [43] stated that nanofluid caused an increase in the thermal conductivity of the flow, thus increasing the overall performance. Therefore, it was concluded that Zigzag TDLMCHS granted an excellent thermal stability for the cooled devices. As it has been recorded that utilizing nanofluid was associated with heat transfer enhancement, thermal resistance was calculated for several types and concentrations of nanofluid as shown in Fig. 7(b). At fixing pumping power of 0.1 W and channel number of 24, an increase in the volume fraction yielded a decrease in the total thermal resistance for both nanofluids. The alumina oxide nanofluid (0.9 vol.%) showed the lowest thermal resistance up to 0.13 °C/W·m2. The superiority of alumina oxide nanofluid over silica oxide nanofluid and pure water was ascribed to the high effective conductivity and lowest viscosity compared to other coolants. The numerical work of Hung et al. [7] for RDLMCHS showed a thermal resistance about 0.097 °C/W·m2 for same pumping power. Direct comparison with Hung work provided 25% deviation with the current experimental work which was considered a high value, however, in the current experiment, the microchannel has a different shape and hydraulic diameter. Enhancement in overall thermal performance was more observed in nanofluids than pure water. Increasing pumping power from 0.1 W to 0. 5 W caused a reduction in the total thermal resistance by about 10% and


113

Fig. 7. (a) Substrate temperature distribution along the zigzag TDLMCHS at PP = 0.3 W compared with the literature, (b) Rtotal with pumping power at PP = 0.3 W, and (c) Nusselt number with Reynolds number compared with literature, for different volume fractions and nanofluids, (d) Effect of volume fraction on the pressure drop in DLMCHS for different pumping powers.

19.91% for water and silica oxide nanofluid (0.6 vol.%), respectively, whereas the pressure drop was slightly bigger in all nanofluid cases. The effect of Reynolds number was also examined in this study, (30 ≤ Re ≤ 200). Fig. 7(c) shows the change in Nusselt number. It was obvious that the raising of Reynolds number was guaranteed by increasing the pumping power, thus the flow velocity increased leading to an increment in Nusselt number. The physical understanding of Reynold number that showed that increasing Reynolds number was due to the inertial forces at the expense of almost fixed viscous forces. This increment caused an increase in Nusselt number. Laminar flow regime was dominant in all tests nanofluid the maximum Reynolds number was 200. Obtaining higher Reynolds number to get higher Nusselt number requires higher pumping power which was energy consumable factor and not adapted in this work. For alumina oxide nanofluid and at Re = 200, the convection heat transfer coefficient was calculated to be 2100 W/°C·m2, which produced only (0.13 °C/W·m2). The current results were compared with the experimental results of Jung et al. [44] and an explicit deviation is obvious. Numerically, Hung et al. [7] obtained (0.065 °C/W·m2) for the total thermal resistance. However, this value, for Hung, was obtained at PP of 0.9 W. The only unexplained result was that silica oxide nanofluid (0.3 vol.%) had smaller Nusselt number and Rtotal than water. At hydrodynamic level, pressure drop between the inlet and outlets versus the nanoparticles concentrations is shown in Fig. 7(d). This figure

indicates that the pressure drop has taken place when the nanoparticle concentration increased. For each pumping power, increasing the nanoparticle concentration from 0% to 0.9% for both nanofluid caused a slight increase in the pressure drop. At pumping power of 0.5 W, alumina oxide nanofluid sowed 16.1% pressure drop increase when the volume fraction increased from (0–0.9%). This simply was attributed to the higher density of the nanofluid which could reach to 1023.4 kg/m3 at 0.9 vol.%. It is worth to mention that this investigation avoided using high nanoparticle concentration, for its limitations in microchannel blockage and avoiding malfunction of the gear pump. Indeed, Hung et al. [7], numerically expanded the nanoparticle concentration up to 5% for Al2O3– H2O nanofluid, showed that, at (0.1 W) pumping power, exceeding 1 vol.% degraded the thermal performance. They attributed this degradation to the higher viscosity associated with high nanoparticle concentration, thus less flow velocity was available to the microchannel. It is important to state that a researcher has to deal carefully with alumina oxide nanofluid and more studies about the stability should be held.

4. Conclusion The essence target of the present work is to examine different channel shapes of DLMCHS and shows its hydrothermal performance. Two

114


different nanofluids with different volume fractions and pumping powers were investigated. It was shown that: • RDLMCHS outperformed TDLMCHS in removing heat regardless of channel number; however, TDLMCHS had better temperature uniformity across the channel length. Enlarging the pumping power changed the location of the maximum temperature and made it close to the inlet side. • Superiority of zigzag TDLMCHS over rectangular one in reducing Rtotal was experienced at low pumping power. The lowest Rtotal was only associated with increasing number of the channels for triangular shape. For rectangular channel, increasing number of channel yielded an increase in Rtotal. Pressure drop was lesser for rectangular channel at low pumping power. Variations of geometric parameters of zigzag TDLMCHS proved its massive influence on overall heat transfer behavior. The trend of temperature with changing the number of channel (N) disclosed the independency of N and fin thickness on temperature variation. Moreover, as N deceased, the Rtotal increased. The cooling system exhibited diminution in overall thermal performances and the same are going to Reynold number. • Increasing pumping power (PP) reduced the Rtotal and then increment in h regardless to the channel number. The reduction of R for DLMCHS having 24 channels was found around 10% when the PP increased from 0.1 W to 0.5 W. TDLMCHS with N = 24 yielded a reduction in the pressure drop about 13% less than that of N = 18. The pressure drop depended on configuration of the channels more than their numbers especially when cross section of all the channels was the same. • Nanofluid showed an excellent thermal stability for the cooled devices. The Al2O3–H2O (0.9 vol.%) nanofluid provided lower R than that having 0.3 vol.%. Generally, nanofluids exhibited the superiority over pure water and SiO2–H2O was the best.

The suggested shape has proved its excellent thermal performance over the existing ones in open literature. The present results could improve the cooling of electronic chips where high thermal stability was guaranteed. Less pumping power could consume less energy leading to improve the efficiency of cooling system. Acknowledgments This work was supported by International Islamic University Malaysia (IIUM) (Grant No. FRGS 11–043–0192). The financial assistance, nanofluids synthesis, and test sections fabrications vitally supported by the Department of Mechanical Engineering are very gratefully acknowledged. References [1] D.B. Tuckerman, R.F.W. Pease, High-performance heat sinking for VLSI, Electron Device Lett., IEEE 2 (5) (1981) 126–129. [2] S.U.S. Chol, Enhancing thermal conductivity of fluids with nanoparticles, ASMEPublications-Fed 231 (1995) 99–106. [3] K. Vafai, L. Zhu, Analysis of two-layered micro-channel heat sink concept in electronic cooling, Int. J. Heat Mass Transf. 42 (12) (1999) 2287–2297. [4] P. Gunnasegaran, H.A. Mohammed, N.H. Shuaib, R. Saidur, The effect of geometrical parameters on heat transfer characteristics of microchannels heat sink with different shapes, Int. Commun. Heat Mass Transfer 37 (8) (2010) 1078–1086. [5] A.A. Alfaryjat, H.A. Mohammed, N.M. Adam, M.K.A. Ariffin, M.I. Najafabadi, Influence of geometrical parameters of hexagonal, circular, and rhombus microchannel heat sinks on the thermohydraulic characteristics, Int. Commun. Heat Mass Transfer 52 (2014) 121–131. [6] T. Hung, W. Yan, W. Li, Analysis of heat transfer characteristics of double-layered microchannel heat sink, Int. J. Heat Mass Transf. 55 (11) (2012) 3090–3099. [7] T. Hung, W. Yan, Enhancement of thermal performance in double-layered microchannel heat sink with nanofluids, Int. J. Heat Mass Transf. 55 (11) (2012) 3225–3238. [8] T. Hung, W. Yan, X. Wang, Y. Huang, Optimal design of geometric parameters of double-layered microchannel heat sinks, Int. J. Heat Mass Transf. 55 (11) (2012) 3262–3272.

[9] C. Lin, Y. Chen, X. Zhang, X. Wang, Optimization of geometry and flow rate distribution for double-layer microchannel heat sink, Int. J. Therm. Sci. 78 (2014) 158–168. [10] T. Hung, T. Sheu, W. Yan, Optimal thermal design of microchannel heat sinks with different geometric configurations, Int. Commun. Heat Mass Transfer 39 (10) (2012) 1572–1577. [11] 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 (1) (2014) 124–134. [12] 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 Transf. 129 (10) (2007) 1432–1444. [13] X. Wang, B. An, L. Lin, D. Lee, Inverse geometric optimization for geometry of nanofluid-cooled microchannel heat sink, Appl. Therm. Eng. 55 (1) (2013) 87–94. [14] Q. Weilin, M. Mala, L. Dongqing, Pressure-driven water flows in trapezoidal silicon microchannels, Int. J. Heat Mass Transf. 43 (3) (2000) 353–364. [15] H.Y. Wu, P. Cheng, An experimental study of convective heat transfer in silicon microchannels with different surface conditions, Int. J. Heat Mass Transf. 46 (14) (2003) 2547–2556. [16] D. Liu, S.V. Garimella, Investigation of liquid flow in microchannels, Eighth AIAA/ ASME Joint Thermophysics and Heat Transfer Conference, St. Louis, MO, 2002. [17] P. Lee, S.V. Garimella, Experimental investigation of heat transfer in microchannels, ASME Summer Heat Transfer Conference, Las Vegas, 2003. [18] R.K. Shah, A.L. London, Laminar flow forced convection in ducts, Adv. Heat Transfer (Suppl. 1) (1978). [19] W.M. Kays, M.E. Crawford, Convective Heat and Mass Transfer, third ed. McGrawHill, 1993. [20] A. Sakanova, C.C. Keian, J. Zhao, Performance improvements of microchannel heat sink using wavy channel and nanofluids, Int. J. Heat Mass Transf. 89 (2015) 59–74. [21] T.C. Hung, W.M. Yan, X.D. Wang, C.Y. Chang, Heat transfer enhancement in microchannel heat sinks using nanofluids, Int. J. Heat Mass Transf. 55 (9–10) (2012) 2559–2570. [22] C.J. Ho, L.C. Wei, Z.W. Li, An experimental investigation of forced convective cooling performance of a microchannel heat sink with Al2O3/water nanofluid, Appl. Therm. Eng. 30 (2–3) (2010) 96–103. [23] C.J. Ho, W.C. Chen, An experimental study on thermal performance of Al2O3/water nanofluid in a minichannel heat sink, Appl. Therm. Eng. 50 (2013) 516–522. [24] H.A. Mohammed, P. Gunnasegaran, N.H. Shuaib, Heat transfer in rectangular microchannels heat sink using nanofluids, Int. Commun. Heat Mass Transfer 37 (10) (2010) 1496–1503. [25] Y. Yang, K. Tsai, Y. Wang, S. Lin, Numerical study of microchannel heat sink performance using nanofluids, Int. Commun. Heat Mass Transfer 57 (2014) 27–35. [26] C. Chen, C. Ding, Study on the thermal behavior and cooling performance of a nanofluid-cooled microchannel heat sink, Int. J. Therm. Sci. 50 (3) (2011) 378–384. [27] S. Halelfadl, A. Mohammed, N. Mohd-ghazali, T. Maré, P. Estellé, R. Ahmad, Optimization of thermal performances and pressure drop of rectangular microchannel heat sink using aqueous carbon nanotubes based nanofluid, Appl. Therm. Eng. 62 (2) (2014) 492–499. [28] E.M. Tokit, H.A. Mohammed, M.Z. Yusoff, Thermal performance of optimized interrupted microchannel heat sink (IMCHS) using nanofluids, Int. Commun. Heat Mass Transfer 39 (10) (2012) 1595–1604. [29] E. Farsad, S.P. Abbasi, M.S. Zabihi, J. Sabbaghzadeh, Numerical simulation of heat transfer in a micro channel heat sinks using nanofluids, Heat Mass Transf. 47 (4) (2011) 479–490. [30] M. Kalteh, Investigating the effect of various nanoparticle and base liquid types on the nanofluids heat and fluid flow in a microchannel, Appl. Math. Model. 37 (18) (2013) 8600–8609. [31] J. Lee, I. Mudawar, Assessment of the effectiveness of nanofluids for single-phase and two-phase heat transfer in micro-channels, Int. J. Heat Mass Transf. 50 (3) (2007) 452–463. [32] X. Wu, H. Wu, P. Cheng, Pressure drop and heat transfer of Al2O3–H2O nanofluids through silicon microchannels, J. Micromech. Microeng. 19 (10) (2009) 105–120. [33] M.N. Pantzali, A.G. Kanaris, K.D. Antoniadis, A.A. Mouza, S.V. Paras, Effect of nanofluids on the performance of a miniature plate heat exchanger with modulated surface, Int. J. Heat Fluid Flow 30 (4) (2009) 691–699. [34] M. Rafati, A.A. Hamidi, N.M. Shariati, Application of nanofluids in computer cooling systems (heat transfer performance of nanofluids), Appl. Therm. Eng. 45 (2012) 9–14. [35] X. Wu, W.U. Huiying, Q.U. Jian, P. Zheng, Flow and heat transfer characteristics of nanofluids in silicon chip microchannels, J. Chem. Ind. Eng. (Chin.) 9 (2008) 006. [36] I.M.F. Seder, M.I. Ahmed, S. Ahmed, M.N.A. Hawlader, Feasibility of double-layer microchannel fabrication at low speed micro end-mill and wire-cut EDM machines, Aust. J. Basic Appl. Sci. 8 (15) (2014) 211–217. [37] B. Lu, W.J. Meng, F. Mei, Experimental investigation of Cu-based, double-layered, microchannel heat exchangers, J. Micromech. Microeng. 23 (3) (2013) 035017. [38] L.S. Sundar, K.V. Sharma, Thermal conductivity enhancement of nanoparticles in distilled water, Int. J. Nanopart. 1 (1) (2008) 66–77. [39] J. Holman, Experimental Methods for Engineers, seventh ed. McGraw-Hill, 2001. [40] J.M. Wu, J.Y. Zhao, K.J. Tseng, Parametric study on the performance of doublelayered microchannels heat sink, Energy Convers. Manag. 80 (2014) 550–560. [41] N. Lei, P. Skandakumaran, A. Ortega, Experiments and modeling of multilayer copper minichannel heat sinks in single-phase flow, Paper Presented at the Thermal

H.E. Ahmed et al. / International Communications in Heat and Mass Transfer 77 (2016) 104–115 and Thermomechanical Phenomena in Electronics Systems, (2006). ITHERM'06. The Tenth Intersociety Conference on, 2006. [42] P. Bhattacharya, A.N. Samanta, S. Chakraborty, Numerical study of conjugate heat transfer in rectangular microchannel heat sink with Al2O3/H2O nanofluid, Heat Mass Transf. 45 (10) (2009) 1323–1333.

115

[43] M. Kalteh, A. Abbassi, M. Saffar-Avval, A. Frijns, A. Darhuber, J. Harting, Experimental and numerical investigation of nanofluid forced convection inside a wide microchannel heat sink, Appl. Therm. Eng. 36 (2012) 260–268. [44] J. Jung, H. Oh, H. Kwak, Forced convective heat transfer of nanofluids in microchannels, Int. J. Heat Mass Transf. 52 (1) (2009) 466–472.