CHAPTER 1 INTRODUCTION
1.2 Literature Review
In what follows the literature relevant to the present study is briefly reviewed. Pool boiling heat transfer is a process of vigorous heat transfer resulting from latent heat exchange associated with liquid-to-vapor phase change in a quiescent liquid. Nukiyama [2] conducted a pioneering pool boiling experiment in 1934 and arranged the experimental heat transfer data as a form of the wall superheat versus the heat flux, which is known as the “boiling curve”
today. After that, the pool boiling heat transfer research has received considerable attention.
The state of the art cooling technologies for handling heat dissipation in microelectronic equipments have been developed extensively over the past 30 years. Several products were released including Air-Cooled Modules, High Thermal Conduction Modules, and Liquid-Cooled Modules, as discussed by Bar-Cohen
In an early attempt to improve pool boiling heat transfer by using a micro-configured surface, Miller et al. [4] found that vapor retention could be a function of the scale and geometry of the micro-configurations. They pointed out that the relation between the stability
[3].
of the potential nucleation sites and the micro-configuration size and geometry required further investigation, so that the size and the site density of the cavities could be optimized for boiling heat transfer enhancement.
A few studies have been carried out to examine the influences of the surface fabricated microstructures on the pool boiling heat transfer. These include boiling of FC-72 on micro-porous surfaces with particle coating tested by Chang and You [5], adding microporous pin-fins and coating particles to the surface in the mean time investigated by Rainey and You [6] and by Rainey et al. [7], and fabricating micro-pin-fins and submicron-scale roughness on the surfaces by Honda et al. [8] and Wei et al. [9]. The study of Rainey and You [6] and Rainey et al. [7] concluded that the microporous coating can significantly enhance the boiling heat transfer performance over the pin-finned surfaces. In examining the pool boiling on the micro-pin-fin surfaces, Honda et al. [8] and Wei et al. [9] noted that the boiling curves were characterized by that a very small increase in the wall superheat can cause a large increase in the heat flux. And increasing the fin height was found to provide better heat transfer in the nucleate boiling regime and result in a higher critical heat flux. Anderson and Mudawar [10]
reported that microstructures in the forms of fins, studs, grooves and vapor-trapping cavities on the boiling surface significantly shifted the boiling curve toward lower superheats while increasing the incipience excursion. Their results also suggest that the maximum boiling heat flux is a function of surface geometry and orientation but independent of the initial conditions, surface roughness, or the presence of large artificial cavities. Intending to augment boiling heat transfer, O’Connor and You [11] painted silver flakes on the boiling surface. Their experimental data show that the incipience boiling superheats are 70-85% lower and the nucleate boiling superheats are 70-80% lower than the bare surface. Besides the critical heat flux is increased by 109%. O’Connor et al. [12] then compared two methods of generating surface microstructures, “spraying” and “painting”, for pool boiling heat transfer enhancement.
They noted that the incipient boiling superheat has 33-55% reduction for the sprayed alumina
and 63-85% reduction for the painted diamond. The enhancement in the critical heat flux can be up to 47% for the sprayed alumina and 103% for the painted diamond microstructures.
Chang and You [13] further studied the effects of coating different sizes of the diamond particles on the pool boiling performances. They classified the coating thickness into two groups. For coatings thinner than 100μm, increasing the coating thickness would generate a higher active nucleation density. But for coatings thicker than 100μm, a further increase in the coating thickness does not always enhance the pool boiling heat transfer. They attributed this result to higher impedance for liquid-vapor exchange channels and higher thermal resistance for the thicker coating. Jung and Kwak [14] investigated the effects of submicron-scale roughness on the subcooled boiling heat transfer for a boiling surface anodized in DMF (dimethylforamide) and HF (hydrofluoric acid). Both surface treatments were found to increase the effective boiling area and serve for increasing the nucleation sites and hence show considerable enhancement in the boiling heat transfer. The critical heat flux also increases linearly. Honda and Wei [15] reviewed recent advances in enhancing boiling heat transfer from electronic components immersed in dielectric liquids through the use of surface microstructures and concluded that most of the surface microstructures were effective in decreasing the wall superheat at the boiling incipience. The nucleate boiling heat transfer also can be improved and the critical heat flux is raised. Rainey and You [16] and Rainey et al. [17]
respectively studied the effects of the orientation and pressure on the pool boiling heat transfer from microporous surface. Their data show that nucleate boiling performance increases slightly for the surface inclined from 0ο(horizontal) to 45ο and then decreases for the inclination angle from 90ο to 180ο
Chou et al. [18] arranged several grooved patterns on surfaces intending to enhance boiling heat transfer of distilled water. Their experimental data reveal that the radial grooved
. Moreover, for the plain and microporous surfaces increases in boiling performance and critical heat flux and decrease in the incipience wall superheat were noted as the pressure increased.
pattern has the best enhanced boiling heat transfer performance and the spiral or concentric grooved pattern has poorer boiling heat transfer coefficient. The worst performance is noted for the grid or the spotted grooved pattern. All grooved patterns they investigated have better heat transfer performance than the plain surface and the denser groove is better than the sparser one for the same patterns.
Hasegawa et al. [19] covered a heat pipe with a woven screen to investigate the associated boiling characteristics and burnout phenomena. Their results disclose that the additional screen produces two opposite effects of inhibiting and enhancing the boiling heat transfer. Tsay et al. [20] explored pool boiling heat transfer enhancement by covering the boiling surface with a screen in distilled water. They found that the screen coverage could raise bubble generation frequency and enhance the boiling heat transfer. But the screen can also cover some nucleation sites and hence may retard the boiling heat transfer. They also noted that the boiling heat transfer decreased at lowering the liquid level. They concluded that covering the heated surface with a screen can augment the pool boiling heat transfer if the mesh size is comparable with the bubble departure diameter. In boiling of methanol and HFE-7100, Liu et al. [21] pointed out that placing a fine mesh layer on the boiling surface enhances nucleate boiling heat transfer at low wall superheat (∆T<10K) but an opposite trend results at a high superheat (∆T>10K). They also reported that the heat transfer in nucleate boiling always becomes worse with a coarse mesh on the boiling surface when compared with that on a smooth surface. Moreover, Franco et al. [22] used dielectric refrigerant R141b to investigate enhancement in the boiling heat transfer performance by covering the heated surface with wire meshes. The boiling heat transfer coefficient was noted to increase significantly, especially at relatively low heat fluxes. They also found that the wire mesh coverage on the heating surface results in slower transition to steady film boiling. In studying the effects of the wall superheat and the mesh layer covering on boiling heat transfer, Kurihara and Myers [23] tested several working fluids including water, acetone, n-hexane, carbon
tetrachloride, and carbon disulfide. They found that active nucleation sites on the heating plate increased due to the mesh covering and the boiling heat transfer coefficient was proportional to the one-third power of the bubble column numbers at high numbers.
1.3 Objective of Present Study
The above literature review clearly reveals that considerable works have been carried out in the past to investigate the enhancement in the pool boiling heat transfer over a surface by using the surface microstructures such as roughness, micro-pin-fins, mesh screens, and particle coating. All these microstructures are fixed firmly onto the boiling surface. In this study, an experimental study is conducted to explore the possible enhancement in the FC-72 pool boiling heat transfer by placing flexible and movable fine wires above the boiling surface. The wires are loosely fixed at their two ends on the surface and hence are allowed to move adjacent to the heating surface during the boiling processes to some degree. This movement of the wires adjacent to the boiling surface is expected to greatly affect the bubble dynamics near the surface and hence the boiling heat transfer from the surface. Besides, in the present study we will also examine the associated bubble behavior in the boiling flow by visualizing the flow. Both the possible saturated and subcooled pool boiling heat transfer enhancement will be examined.
7
CHAPTER 2
EXPERIMENTAL APPARATUS AND PROCEDURES
A schematic arrangement of the experimental apparatus for the present investigation of the pool boiling heat transfer enhancement by flexible strings is shown in Fig. 2.1. The experimental system includes a main test chamber, a test heater assembly, and other auxiliary parts such as a D.C. power supply, a data acquisition unit and a high-speed photographic unit. The working fluid, FC-72, is a highly wetting dielectric fluorocarbon liquid produced by 3M Industrial Chemical Products Division, which has been considered as a good candidate fluid for liquid immersion cooling applications. It is chemically stable, dielectric, and has a relatively low boiling point (Tsat=56°C at atmospheric pressure). Some thermophysical properties of FC-72 are given in Table 2.1.
2.1 Main Test Chamber
The main test chamber is a hermetic stainless steel pressure vessel of 205mm in height and 216mm in diameter. An internal water condenser is installed inside the chamber and connects with a thermostat (LAUDA RK20) to maintain the bulk temperature of the working fluid in the chamber at the preset level. The maximum cooling power of the thermostat is 200W (at 20°C). We further use an external temperature controller (FENWAL MYSPEC Digital Temperature Controller) to control the bulk temperature of FC-72 in the test chamber with an accuracy of
±0.1°C. Besides, a cartridge heater is located near the bottom of the test chamber to provide additional heating during the degassing process. In order to prevent the heat loss from the vessel to the ambient, a superlon layer of 10-mm thick is wrapped
8
around the chamber. Moreover, a pressure transducer with an operating range of 0-980kPa is located at the gate valve to measure the pressure of the work fluid.
Meanwhile, the working fluid temperature is measured by two resistance temperature detectors (RTDs) located at the gate valve and at a selected location 5cm above the bottom surface of the chamber with a calibrated accuracy of ± 0.1°C. An auxiliary tank of 10-liter liquid FC-72 is placed right above the test vessel and it is only used for subcooled pool boiling experiment to prevent regassing of the working fluid after degassing. A pressure transducer and a RTD are placed in the auxiliary tank to measure the internal gas pressure and liquid temperature. In addition, a test heater assembly is mounted to a stainless steel shelf to fix the Teflon substrate. The working fluid is maintained at approximately 80mm above the heated surface in the experiment.
2.2 Test Heater Assembly
A schematic of the test heater assembly is shown in Fig. 2.2. The assembly consists mainly of a film heater and is adhered to a square copper block with epoxy Omegabond 200 . The heater supplies the required power input to the copper block.
The copper block is flush mounted onto a much larger Teflon block. Liquid FC-72 boils on the upper surface of the copper block and its side is 10-mm long. More specifically, the copper block is heated by the D.C. current delivered from the film heater adhering to the lower surface of the copper block. Besides, three calibrated copper-constantan thermocouples (T-type) with a calibrated accuracy of ±0.2°C are installed at selected locations in the copper block right below the boiling surface.
They are used for the control and determination of the boiling surface temperature.
The detailed locations of the thermocouples are shown in Fig. 2.3. Note that the whole
9
copper block is inserted into a Teflon block which serves as a heat insulator (kT ≈0.35W/m⋅K) intending to reduce the heat loss from the lateral and bottom surfaces of the block to the ambient.
2.3 Installation of Strings on Boiling Surface
The strings are fixed on the upper surface of the copper block by paste at their ends, as schematically shown in Fig. 2.4. The nylon threads are chosen in the present experiment because of its good flexibility. Specifically, each the nylon thread has the same mean diameter. The nylon strings are installed uniformly at the same pitch and height above the heating surface. Besides, the vertical distance between the strings and boiling surface will be varied in the test to investigate its effects on the pool boiling heat transfer. Moreover, the looseness of the strings measured by their length relative to the length of the boiling surface on the boiling heat transfer will be investigated. The measured data apparently will be compared with that of a bare heating surface (without the presence of the strings).
2.4 DC Power Supply
The power generated in the film heater in the test heater assembly is provided by a programmable D.C. power supply (Chroma 6203-15). It offers a maximum D.C.
power of 300W for an output voltage of 15V and an output current of 20A. The power input to the copper block is transmitted through a GPIB interface to a personal computer. In order to measure the D.C. current, a precision ammeter (KYORITSU A.C./D.C. DIGITAL CLAMP METER) is arranged in series connection with the electric circuit. Besides, a YOKOGAWA data recorder is used to measure the voltage drop across the test heater assembly. All the voltage, current and power measurement
10
devices are calibrated by a YOKOGAWA WT200 power meter according to the Center of Measurement Standards in Industrial Technology Research Institute of Taiwan.
2.5 Data Acquisition
A 30-channel YOKOGAWA data recorder (MX-100) combined with a personal computer is used to acquire and process the data from various transducers. All signals detected from the T-type thermocouples, RTDs, pressure transducer, ammeter, data recorders and power meter are all collected and converted by the internal calibration equations in the computer during the data acquisition.
2.6 Experimental Procedures
Prior to putting all the devices and components for the experimental system together, the boiling surface is polished by fine sand paper (Number 3000) and cleaned by alcohol. In each test, we need to remove the non-condensable gases in the empty test chamber by running a vacuum pump for about 15 minutes and then fill the FC-72 liquid into the test chamber. Next, the FC-72 liquid in the test chamber is heated to the saturation state by employing a digital temperature controller and cartridge heater. Moreover, the FC-72 liquid is boiled vigorously for 2 hours to remove the dissolved noncondensible gases in it. After the working fluid pressure and temperature stabilize to one atmosphere and at the saturation state, we turn on the test heater. The imposed heat flux on the boiling surface is adjusted by controlling the electric current delivered to the heater from the D.C. power supply. Upon reaching the statistical state, we begin collecting the required heat transfer data and visualizing the boiling activity.
11
Table 2.1 Thermophysical properties of FC-72.
Properties at 25οC FC-72
Appearance Clear, colorless
Average Molecular Weight 338
Boiling Point (1atm) 56°C
Pour Point (1atm) -90°C
Estimated Critical Temperature 449K
Estimated Critical Pressure 1.83 × 106 Pa
Vapor Pressure 3.09 × 104 Pa
Latent Heat of Vaporization hfg 88 J/g (at normal boiling point)
Liquid Density ρ 1680 kg/m3
Absolute Viscosity µ 6.4× 10-3 poises ; 6.4× 10-4 kg/m∙s Kinematic Viscosity ν 3.8 × 10-3 stokes ; 3.8 × 10-7 m2/ s Liquid Specific Heat cp 1100 J/kg∙°C
Liquid Thermal Conductivity k 0.057 W/m∙°C Coefficient of Expansion β 0.00156 /°C
Surface Tension σ 10 dynes/cm ; 10-2 N/m
12
To Degassing Tank and Drain
Computer
Boiling Surface FC-72
Digital
To Degassing Tank and Drain
Computer
Boiling Surface FC-72
Digital
Fig. 2.1 Schematic diagram of the test apparatus.
13
Embedded Copper Block
Electric Film Heater
Copper Surface
Embedded Copper Block
Electric Film Heater
Copper Surface
Teflon Surface
Perspective view
57
Conductive Pastes
Fig. 2.2 Schematic diagram of the test heater assembly (not to scale).
14
Copper Surface 5
Copper Block
Electric Film Heater Conductive Pastes
Copper Surface 5
Copper Block
Electric Film Heater Conductive Pastes
Fig. 2.3 Locations of three thermocouples in the copper block and one thermocouple below the heater (not to scale).
15
Teflon substrate Flush-mounted Copper Block
(unit:mm)
Strings
62
57 30
Teflon substrate Flush-mounted Copper Block
(unit:mm)
Strings
62
57 30
Fig. 2.4 Schematic diagram of placing strings on heating surface (not to scale).
CHAPTER 3
DATA REDUCTION
3.1 Boiling Heat Transfer Coefficient
The space-average boiling heat transfer coefficient over the upper surface of the heated square copper block at long time when the flow is at a statistical state is defined as
sat superheat defined as the difference between the average surface temperature and the saturated temperature of FC-72. The average heated surface temperature is estimated from the measured average temperature from the thermocouples installed at locations near the upper surface of the copper block according to the steady-state one-dimensional conduction heat transfer. Specifically,
(3.2)
where TCu
k
= the average measured temperature from the thermocouples (°C)
Cu
δ
= the thermal conductivity of copper (W/m·K)
= the vertical distance between the thermocouple tips and the upper surface of the copper block (m)
The total power input Qt to the copper block can be obtained from the voltage drop across the film heater in the test heater assembly and the current passing through it,
V I
Qt = ⋅ (3.3) where
Qt
I = electric current passing through the film heater (Amp.)
= total power input to the upper surface of the copper block (W)
V = voltage drop across the film heater (Volts)
In fact, the Teflon insulator cannot completely prevent the heat loss from the surfaces of the copper block. Heat loss across the insulator does exist, mainly from the lateral sides of the copper block and heater and from the bottom of the heater. The heat loss is estimated by one-dimensional heat conduction in the Teflon insulator and convection from the insulator surface to the ambient based on a model schematically shown in Fig. 3.1. Thus, we have
: the ambient temperature (°C)
5 , T6 , T7
k
: the average measured temperatures at the measured locations inside the Teflon insulator, as schematically shown in Fig. 3.1
T
L
: thermal conductivity of the Teflon insulator (W/m·K)
5 , L6 , L7
A
: shortest distances between locations #5, #6, #7 and the insulator surfaces (m)
T,5 , AT,6 , AT,7 : bottom and lateral surface areas of the Teflon block
hi : estimated natural convection heat transfer coefficient from the Teflon block surfaces to the surroundings by correlations from Incropera et al. [24].
(W/m2
h5: estimated from NuL =0.27RaL1/4 for the bottom surface of the Teflon block.
: estimated from for the lateral surfaces of
the Teflon Block.
Finally, the net imposed input heat flux to the upper surface of copper square can be evaluated from the relation
Cu
where ACu is the area of the upper surface of the copper block.
3.2 Uncertainty Analysis
An uncertainty analysis is carried out here to estimate the uncertainty levels in the experiment. Kline and McClintock [25] proposed a formula for evaluating the uncertainty in the result F as a function of independent variables, X1, X2,
X3···Xn
F=F (X ,
1 ,X2, X3···Xn
The absolute uncertainty of F is expressed as
)
and the relative uncertainty of F is
2
If F =X1aX2bX3c... , then the relative uncertainty is
∂ and ∂ are, respectively, the sensitivity coefficient and uncertainty Xi
level associated with the variableX . The values of the uncertainty intervalsi ∂ are Xi obtained by a root-mean-square combination of the precision uncertainty of the instruments and the unsteadiness uncertainty, as recommended by Moffat [26]. The
level associated with the variableX . The values of the uncertainty intervalsi ∂ are Xi obtained by a root-mean-square combination of the precision uncertainty of the instruments and the unsteadiness uncertainty, as recommended by Moffat [26]. The