- Email: [email protected]
Visualization experiment on boiling heat transfer and flow characteristics in separated heat pipe system
Accepted Manuscript Visualization experiment on boiling heat transfer and flow characteristics in separated heat pipe system Tao Ding, Han wen Cao, Zhi guang He, Zhen Li PII: DOI: Reference:
S0894-1777(17)30326-6 https://doi.org/10.1016/j.expthermflusci.2017.10.019 ETF 9243
To appear in:
Experimental Thermal and Fluid Science
Received Date: Revised Date: Accepted Date:
12 April 2017 29 September 2017 19 October 2017
Please cite this article as: T. Ding, H. wen Cao, Z. guang He, Z. Li, Visualization experiment on boiling heat transfer and flow characteristics in separated heat pipe system, Experimental Thermal and Fluid Science (2017), doi: https:// doi.org/10.1016/j.expthermflusci.2017.10.019
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Visualization experiment on boiling heat transfer and flow characteristics in separated heat pipe system Tao Ding, Han wen Cao, Zhi guang He, and Zhen Li* Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Engineering Mechanics, Tsinghua University, Beijing 100084, China Tel.: 010-62772918; E-mail: [email protected]
Abstract: This paper presents an experimental study on boiling heat transfer and flow characteristics in a separated heat pipe system. The flow patterns in liquid and vapor pipes are clearly observed, and the boiling heat transfer coefficient is measured. The main aim of this study is to examine the heat transfer characteristics at different filling ratios using a combination of visualization and measurement methods. First, the relationship between the boiling heat transfer coefficient and the heating capacity is analyzed at different filling ratios when the heat pipe system operates normally. Second, the conditions for reaching the heat transfer limits and the phenomena occurring at these limits are analyzed. The analysis of the heat transfer characteristics shows that for a given heating capacity, the boiling heat transfer coefficient in the evaporator remains nearly constant over a wide range of filling ratios. At a given filling ratio, the heat transfer coefficient increases with an increase in the heating capacity. Moreover, the flow pattern at the evaporator exit changes from slug flow to 1
bubble flow. Two types of heat transfer limits, i.e., the flooding limit and the dry-out limit, are analyzed at sufficiently high and low filling ratios. Keywords: separated heat pipe, heat transfer coefficient, flow pattern, visualization, filling ratio
Nomenclature h
Heat transfer coefficient, W/m2 K
Q
Heating capacity, W
qh
Heat flux, W/m2
G
Acceleration due to gravity, m/s2
FR
Filling ratio
ΔT
Temperature difference, °C
T
Temperature, °C
ρ
Density, kg/m3
Cp
Specific heat, J/kg
μ
Viscosity, kg/ms
Pr
Saturation pressure/critical pressure
M
Relative molecular weight
Subscri pt i
inside 2
o
outside
v
vapor
l
liquid
1. Introduction The heat pipe is an efficient and energy-saving heat transfer device. Heat pipes are widely used in several applications, such as energy storage systems [1, 2] and waste heat recovery systems [3], particularly in data centers or cooling systems of IT equipment [4-7]. A thermosiphon is a type of heat pipe that is driven by gravity. Thermosiphons can be broadly classified into two types. The first type is the two-phase closed thermosiphon. This system contains only a single tube. The evaporator and condenser are located at the bottom and top, respectively, of the thermosiphon. The second type is the loop thermosiphon. In this system, the condenser and evaporator are mounted separately and connected by liquid and vapor pipes. The loop thermosiphon is also known as the separated heat pipe system. Because of the specific structure of this heat pipe system, heat can be transferred over long distances. Fig. 1 shows the schematic of a simple separated heat pipe system. This system comprises four components: an evaporator, a condenser, vapor pipe, and liquid pipe. As the condenser is installed above the evaporator, this system can be driven by the gravity of the liquid (or by the buoyancy force of the vapor), instead of being driven by a pump. When there is no overheating or overcooling of the coolant, 3
phase changes occur only inside the heat pipe. Moreover, the heat transfer coefficient of the two-phase flow is higher than that of the single-phase flow. Thus, this heat pipe system has a higher heat transfer ability than a single-phase heat exchanger.
Figure 1. Schematic of heat pipe system After making minor improvements and modification, the separated heat pipe system shown in Fig. 1 can be used for the cooling of a data center [8, 9]. A data center is a facility where IT equipment is centrally stored. The heat generation rate in a data center can be extremely high. If the heat generated is not transferred immediately, the temperature of the IT equipment may increase rapidly, thereby threatening safe operation. In a heat pipe system installed for cooling a data center, the evaporator is installed inside the front and back doors of the rack. Thus, the heat generated by the IT equipment can be immediately absorbed by the nearby parts. A plate-type heat exchanger is used as the condenser (with chilled water used as the heat sink); it is installed in another room because water is prohibited in a data center. Although heat pipe systems are being used for the cooling of data centers, some 4
theoretical problems with these systems need to be addressed. The filling ratio is an important parameter for evaluating the heat transfer ability of the separated heat pipe system. Some studies [10-18] aimed at determining the relationship between the heat transfer ability of a heat pipe and its filling ratio. A few studies analyzed the filling ratio of a separated heat pipe system [10,11,13] and found that with an increase in the filling ratio, the heat transfer capacity increased initially, then attained the maximum value, and subsequently decreased. Ling et al.’s results [12] showed that the heat transfer capacity of the loop thermosiphon system remained constant in a certain range of the filling ratio. Zhang et al. [13] used R22 and R744 as the coolants and found that the optimal filling ratio for R744 was approximately 150%; the heat transfer capacity remains almost constant when the filling ratio is higher than 150%. However, the optimal filling ratio for R22 is a single value, and after this filling ratio was achieved (at higher filling ratios), the heat transfer capacity decreased. Researchers have also investigated the filling ratio of a two-phase closed thermosiphon system [14-18]. Noie [14] reported that the optimal filling ratio is influenced by aspect ratio, where the aspect ratio is defined as the ratio of the section length of the evaporator to the inner diameter of the tube. Payakaruk et al. [15] studied the relationship between the filling ratio and the inclination angle, and concluded that the filling ratio does not affect the heat transfer capacity at different inclination angles. Noie et al. [16] studied the performance of the two-phase closed thermosiphon and found that the heat transfer rate increased with increasing filling 5
ratio. Park et al. [17] studied the effects of fill charge ratio and found that the effects seemed to be negligible in the evaporator. Aly et al. [18] found that the filling ratio has significant effects on the performance of the heat pipe (a two-phase closed thermosiphon). The present study focuses on the performance of a separated heat pipe system. The following aspects are considered in the study. I)
The mechanism of heat transfer at different filling ratios.
II)
The heat transfer coefficient of the evaporator. As the flow in the heat pipe is driven by the buoyancy force, the evaporation that occurs in the separated heat pipe system is different from that occurring in pool boiling and forced flow boiling in the tube. Hence, it is important to study the heat transfer characteristics.
Some studies [8-10, 19, 20] examined the overall performance of the heat pipe system, such as the temperature difference or the thermal resistance in the entire heat transfer process. However, it is difficult to differentiate the heat transfer coefficient of the evaporator from the total thermal resistance (for example, in evaporators, the total thermal resistance includes the air-to-tube and fin convection thermal resistances, heat conduction resistance in the tube and boiling heat transfer thermal resistance). Some studies [21-24] analyzed the heat transfer characteristics of a two-phase closed thermosiphon. Imura et al. [23] established an empirical formula to calculate the heat transfer coefficient of the evaporator. Jafari et al. [24] subsequently used this 6
formula to calculate the boiling heat transfer coefficient. However, there is a clear difference between the two-phase closed thermosiphon and the separated heat pipe system: in the two-phase closed thermosiphon, the vapor and liquid flow in different directions, whereas in the separated heat pipe system, they flow in the same direction. Hence, the heat transfer characteristics of these two systems are different. In the present study, the heat transfer coefficient of the evaporator was directly measured.
2. Experimental equipment and method 2.1 Experimental equipment To overcome the above problems, a visualization experiment was performed. Fig. 2 shows a schematic of the experiment equipment.
Figure 2. Structure of heat pipe system. Left: Overall view; Right: Enlarged view of evaporator
7
The evaporator comprises a copper tube, whose inner and outer diameters are 8 mm and 12 mm, respectively. The length of the evaporator is 151.5 cm (this value is the tube length around which an electric resistance wire is wound; the total tube length is slightly greater than 151.5 cm). The condenser is in the form of a tube in the tube heat exchanger. The evaporator is heated via the electric resistance wire, and the condenser is cooled with chilled water. Sixteen thermocouples are welded by soldering tin on the surface of the outer tube wall. Table 1 lists the positions of the thermocouples.
Table 1. Positions of thermocouples, with reference to the position of the thermocouple at the top considered as the zero point No.
101
102
103
104
105
106
107
108
0
100
198
300
396
500
598
694
109
110
111
112
113
116
115
118
796
894
1007
1097
1197
1302
1404
1501
Position (mm) No. Position (mm)
The electric resistance heating wire is wrapped with an insulation film, and two layers of thin Kapton film are placed between the outer tube wall and the electric resistance wire for providing electrical insulation (the thermocouples are welded by soldering tin on the surface of the outer tube wall and they are also wrapped inside 8
Kapton film), as shown in Fig. 2. The thickness of the Kapton film is approximately 0.13 mm. The electric resistance wire is wrapped with thermal insulation material. Such structure ensures that the heat generated by the wire is transferred to the evaporator. The electric resistance wire is heated using direct current. As the wire is wound tightly into a spiral form around the tube, the heating boundary can be regarded as a constant heat flux boundary. The heat flux may be transferred to the coolant or the external environment. The thermal resistance to the coolant is termed the internal thermal resistance, and that to the external environment is termed the external thermal resistance. The internal thermal resistance is estimated to be approximately only 0.53% of the external thermal resistance. This implies that the heat loss from the evaporator can be ignored. The vapor and liquid pipes are made of glass so that the flow pattern of the coolant can be viewed clearly. In the experiment, R134a and R22 are used as coolants. In the heating circuit, a constant resistance of 0.01 Ω is connected in series, as shown in Fig. 2. Thus, through measurement of the voltage across the constant resistance, the current in the circuit can be calculated using the following equation: current (A) = voltage (V) / 0.01 Ω. The data required for the calculation of the heat transfer coefficient are collected under steady conditions of the heat pipe system. 2.2 Experimental method The heat transfer coefficient is calculated using Eq. (1), where Q is calculated by 9
multiplying the voltage with the current. The voltage of the power supply is measured using a Fluke 15B digital multimeter, whose relative accuracy is 0.5%. The voltage across the resistance (0.01 Ω) is measured using an Agilent 34970A data acquisition unit, whose relative accuracy is approximately 0.3% (the accuracy of this device is actually much higher, but as only three significant digits are considered, the accuracy is estimated as 0.3%). The inner surface area is 0.0381 m2 (A = 3.14 × 0.008 m × 1.515 m). The temperature difference is taken as the difference between the temperature of the inner wall of the tube and the saturation temperature. The inner wall temperature can be calculated using Equation (2), where Ti and To are the inner and outer wall temperatures, respectively, and λ is the thermal conductivity of copper, which is equal to 397 W/mK. Further, ro = 12 mm and ri = 8 mm. The outer wall temperature refers to the average outer surface temperature obtained using the 16 thermocouples. The saturation temperature is calculated using the evaporation pressure. The pressures at the entrance and exit points of the evaporator are measured, and the saturation pressure is taken as the average of these two values.
h=
UI Q = AT AT
（1）
ro ri
（2）
Ti =To
Q 2l
ln
3. Analysis of results 3.1 Variation trend of heat transfer coefficient 10
Fig. 3 shows the experimental results of the heat transfer coefficient for the R22 coolant at a filling ratio of 105%. Here, the heating capacity is in the range of 109–699 W. As shown in Fig. 3, the heat transfer coefficient increases with an increase in the heating capacity. The curve has a parabolic profile. A detailed definition of the filling ratio is provided in section 3.2. Fig. 4 shows the flow pattern at the exit area of the evaporator. When the heating capacity is 109 W (considering the transfer area is 0.0381 m2), the heat flux is 2.86 kW/m2. Two-phase slug flow is observed, in which large bubbles are clearly visible. The results show that the flow velocity is not very high. It can be seen that small bubbles coalesce to form larger ones. However, as can be seen in Fig. 4, the flow pattern at a heating capacity of 699 W (heat flux of 18.35 kW/m2) is more complex than that at 109 W. The vapor generation rate is higher than that at 109 W. Because the vapor generation is higher, the vapor fills the central part of the evaporator and pushes the liquid aside. Moreover, it is observed that the small bubbles move upward. The flow pattern in the inner copper tube cannot be viewed, nor can it be predicted extremely realistically using the flow pattern at the exit point; however, it can still be predicted using the images captured (flow pattern) at the exit point. There are three main reasons for the heat transfer coefficient to be high at high heating capacities. The first reason is that more bubbles are formed when the heat flux is high. The bubbles are formed under two conditions: the first condition is when small sunken holes are present on the heating surface; when the surface is overheated, bubbles will be generated at these small 11
sunken holes. The places where the bubbles are generated are called nucleation sites. The second condition is when the surface is overheated. Overheating increases with an increase in heating capacity. Thus, at low heating capacities, bubbles do not form at some nucleation sites, whereas at high heating capacities, bubbles form at these nucleation sites. Hence, there is an increase in the number of nucleation points. In other words, the bubble formation rate is much higher at high heating capacities. The second reason for the increase in heat transfer coefficient at higher heating capacities is that the two-phase flow can become more uniform with increasing heat flux. The third reason is that the velocity of the two-phase flow increases with increasing heat capacity. The flow pattern in the separated heat pipe can be regarded as a combination of pool boiling and flow boiling. Flow boiling at a low flow speed may enhance the heat transfer coefficient; however, when the flow speed increases continuously, the flow boiling may hinder pool boiling. In fact, in our experiment, this effect is not seen. To sum up, a large number of nucleation sites, more uniform two-phase flow, and high flow speed are the three main reasons for a high heat transfer coefficient at high heat transfer capacities.
12
Heat transfer coefficient (W/m2K)
5000 4500 4000 3500 3000 2500 2000 1500 1000 0
100
200
300
400
500
600
700
800
Heating capacity (W)
Figure 3. Relationship between heating capacity and heat transfer coefficient at a filling ratio of 105% with R22 coolant
109 W
301 W
699 W
Filling ratio: 105%; R22 coolant Figure 4. Flow patterns at the exit point of evaporator with different heating capacities
Some others empirical equation has been used to compare our experiment results. For example, Imura et al. [23] proposed an empirical equation (Equation 3) to predict the heat transfer coefficient of the evaporator of a two-phase closed thermosiphon. In Equation (3), subscripts “l” and “g” indicate liquid and vapor, respectively. 13
he 0.32
0.2 0.4 l0.65l0.3c 0.7 qh pl g ( Pin / Pa )0.3 0.25 0.4 0.1 v r l
(3)
The results of the present study were compared with other researchers’ correlations, such as Rohsenow’s correlation [25,26] (Equation 4), Cooper’s correlation [27] (Equation 5, in fact, this equation has considered the surface roughness, here, the surface roughness is omitted), and Hasna’s correlation [28] (Equation 6). h
c pl q 0.667 hsf Cwl
(
1 l hsf
g ( l v )
) 0.33 (
c pl l
l
) n
(4)
he 55 pr0.12 (-lg pr )0.55 M 0.5 q0.67
(5)
he 7704M 0.5 q0.157 pr 0.12 / ( log pr )0.55
(6)
Out of these equations, Rohsenow’s correlation is applicable for pool boiling, and is obtained by theoretical derivation; in this equation, Cwl is taken as 0.0049, and n = 1.7. hsf means latent heat (J/kg). The average error obtained is 0.43. Cooper’s equation is mainly applicable for the coolant, and the average error obtained is 0.17. Hasna’s equation is derived based on the results for the separated heat pipe system (loop thermosiphon, water used as coolant, sub-atmospheric pressure). The results show that when the heat transfer capacity is 109 W, the error is 67%; however, when the heat transfer capacity is higher than 400 W, the error is less than 8.2%. The average error is 28%. Thus, the results obtained using Hasna’s equation are the closest to the results obtained in our experiment. In fact, the heat flux as per Hasna’s equation is approximately 2–20 W/cm2, while in our experiment, it is approximately 0.26–1.84 W/cm2. Thus, Hasna’s equation is more suitable for the high heat flux condition in our 14
experiment. The experimentally obtained values of the heat transfer coefficient are much higher than the values obtained using the equation proposed by Imura et al. [23]. As can be seen in Table 2, when the heating capacity is 109 W, the difference between these two sets of results is 20.5%. However, when the heating capacity is 699 W, the difference is as high as 42.3%. This is because of the difference in the flow patterns in the loop thermosiphon and the two-phase closed thermosiphon.
Heat transfer coefficient(W/ m2K)
9000 8000
Experiment
7000
Imura Rohsenow
6000
Hasna
5000
Cooper 4000 3000
2000 1000 0 0
100
200
300
400
500
600
700
800
Heat transfer capacity (W)
Figure 5. Comparison of heat transfer coefficients obtained experimentally in the present study and using equation proposed by Imura et al. [23] (R22 coolant, filling ratio: 105%) Equation (3) was derived based on an experiment and analysis of the two-phase closed thermosiphon. In the two-phase closed thermosiphon, the condenser and evaporator are located in the same vertical tube. The vapor and liquid flow in the thermally insulated part; the vapor flows upward, whereas the liquid flows downward. 15
The two-phase coolant in the evaporator remains almost constant, in another word, there is no macro flow for coolant and the coolant do not flow upward or downward. Thus, the heat transfer in the evaporator is more similar to the one in pool boiling. However, in the separated heat pipe system, the vapor and liquid flow in the same direction. The two-phase coolant in the evaporator flows faster than that in the two-phase closed thermosiphon for the same tube diameter and heat flux. Thus, the heat transfer in the evaporator of the separated heat pipe system is more similar to a combination of the heat transfer processes of pool boiling and flow boiling. The differences in the heat transfer patterns lead to different heat transfer characteristics of the separated heat pipe system and the two-phase closed thermosiphon. Table 2. Comparison of heat transfer coefficients obtained experimentally in the present study and using the equation proposed by Imura et al. [23] (R22 coolant, filling ratio: 105%)
Heating capacity (W)
Experimental result (W/m2K)
As per the equation proposed by Imura [23] (W/m2K)
109 206 301 400 498 600 699
1536 2280 2888 3381 3797 4306 4695
1221 1587 1865 2108 2322 2526 2709
Relative error (%)
-20.5 -30.4 -35.4 -37.7 -38.8 -41.3 -42.3
3.2 Relationship between heat transfer coefficient and filling ratio 16
Five different filling ratios are considered in the experiment with the R22 coolant. The filling ratio (FR) is defined under the condition when there is no heat flux (i.e., the heat pipe system is not operational). The filling ratio is defined as the height of the liquid coolant (Hl) (with the entrance of the evaporator considered as the origin) divided by the height of the entire evaporator tube (He, 151.5 cm) (see Equation (7) and Fig. 6). Thus, the filling ratio can be higher than 100%.
FR%
Hl 100% He
(7)
Figure 6. Definition of filling ratio (only a schematic, not to scale) Fig. 7 shows the boiling heat transfer coefficient for R22 at five different filling ratios considered in the experiment. The lowest filling ratio is 65% (which is a very low value). The highest filling ratio of 193% (which is a sufficiently high value) implies that nearly half of the condenser is submerged in the liquid.
17
Heat transfer coefficient (W/m2K)
5000 4500
65%
4000 105% 3500 3000
99%
2500
161%
2000
193%
1500 1000 0
200
400
600
800
1000
Heating capacity (W)
Figure 7. Heat transfer coefficient at different heating capacities and filling ratios with R22 coolant The results show that despite significant variations in the filling ratio, the heat transfer coefficient remains nearly constant. This also implies that if the heat pipe can operate normally (no heat transfer limit is reached), the heat transfer ability of the evaporator remains unchanged.
Figure 8. Schematics of heat pipe system. Left: No heat flux as the system is not 18
operational; Right: System operating normally
The above phenomenon can be explained with the help of the schematics shown in Fig. 8. The figure shows the flow-pattern diagrams of the heat pipe system when it is not operational (view on the left) and when it is operating normally (view on the right). For example, if the filling ratio is sufficiently high so that some part of the condenser is submerged (the filling ratio is much higher than 100%), and when there is sufficient heat flux, bubbles will form. Moreover, because of the viscosity between the vapor and the liquid, the liquid and vapor will flow out of the evaporator and flow into the vapor pipe. With continuous flow of the liquid and vapor into the vapor pipe, the liquid level in the condenser may decrease. Consequently, a larger area of the tube wall of the condenser is exposed, and the condenser operates normally. If the filling ratio is not very high, e.g., lower than 100%, the entire tube wall of the condenser is exposed, and the condenser operates normally. The experimental results lead to the conclusion that the liquid level in the evaporator does not influence the heat transfer coefficient even the filling ratio is sufficiently high (but without reaching the flooding limit). If the filling ratio is sufficiently low, the liquid film cannot cover the evaporator, and the dry-out limit may be reached. If the filling ratio is very high, when the heat pipe starts operating, only a small area of the tube wall of the condenser is exposed, in spite of the liquid and vapor flowing 19
into the vapor pipe. This area is insufficient to condense the vapor leaving the evaporator, and hence the heat pipe system cannot operate normally. We define this operating limit as the flooding limit. However, as long as the heat pipe system operates normally, the liquid level in the evaporator does not influence the heat transfer coefficient.
3.3 Heat transfer limit for heat pipe system: dry-out limit If the filling ratio of the heat pipe is insufficient when the heat flux is very high, the dry-out limit may be reached. Figures 9, 10, and 11 show the temperatures measured by the thermocouples at four points on the outer tube wall at heating capacities of 109.9 W, 493.9 W, and 597.2 W, respectively. The results show that although the filling ratio is 65%, when the heating capacity is 109.9 W, the temperature of the outer tube wall remains constant. Fig. 12(a) shows the flow pattern of the coolant at the evaporator exit point, wherein the external flow is slug flow.
20
19 18.5
101 (C)
106 (C)
112 (C)
118 (C)
Temperature (oC)
18 17.5 17 16.5 16 15.5 15 14.5 14 0
100
200
300
400
500
600
Time (s)
Figure 9. Temperatures measured by thermocouples at four points at a heating capacity of 109.9 W and a filling ratio of 65%, with R134a coolant
34 33.5
101 (C)
106 (C)
112 (C)
118 (C)
Temperature (oC)
33 32.5 32 31.5 31 30.5 30 29.5 29 0
100
200
300
400
500
600
Time (s)
Figure 10. Temperatures measured by thermocouples at four points at a heating capacity of 493.9 W and a filling ratio of 65%, with R134a coolant
21
38.5 38
101 (C)
106 (C)
112 (C)
118 (C)
Temperature (oC)
37.5 37 36.5 36 35.5 35 34.5 34 33.5 0
100
200
300
400
500
600
Time (s)
Figure 11. Temperatures measured by thermocouples at four points at a heating capacity of 597.2 W and a filling ratio of 65%, with R134a coolant
(a) 65%, 109.9 W Two-phase slug flow
(b) 65%, 493.9 W
(c) 65%, 597.2 W
Annular flow
Annular flow
Figure 12. Flow patterns at exit point of evaporator at different heating capacities and a filling ratio of 65%, with R134a coolant
At the heating capacity of 493.9 W, the temperature of the outer tube wall remains nearly constant. Fig. 12(b) shows the annular flow pattern at the exit position. The flow pattern indicates that with the increase in heating capacity, the vapor generation 22
rate is higher than that at 109.9 W. This implies that the vapor has already filled the central part of the pipe. The vapor pushes the liquid aside, which results in the formation of a liquid film on the surface of the inner tube wall. At the heating capacity of 597.2 W (see Fig. 11 for temperature results and Fig. 12(c) for the flow pattern), the temperature at the exit position of the evaporator varies considerably with time, and the variation trend is found to be irregular. This phenomenon is attributed to the flow pattern. Since the heat flux is very high at this heating capacity, the vapor generation rate also increases significantly; thus, the void fraction also increases, as can be seen in Fig. 12. Consequently, the liquid cannot cover the inner tube wall uniformly. In other words, the tube wall of the evaporator cannot be covered completely by the liquid film. Thus, the temperature of the top tube wall fluctuates significantly.
3.4 Heat transfer limit for heat pipe system: flooding limit The term flooding limit has already been explained in section 3.2. Figures 13 and 14 show the temperature distributions when the flooding limit is reached. These results correspond to very high filling ratios. As can be seen in these figures, when the flooding limit is reached, the temperature indicated by thermocouple no. 118 is much lower than those indicated by the other thermocouples. This phenomenon is caused by overcooling of the liquid pipe. As the filling ratio is very high, the bottom of the condenser is completely covered by the liquid coolant. Thus, the liquid coolant 23
temperature is closer to the heat sink temperature. Moreover, the temperatures at these points change cyclically with time because the liquid level in the condenser increases and decreases cyclically. 38 36
Temperature (oC)
34 32 30 101 (C)
28
106 (C)
111 (C)
118 (C)
26 24 0
100
200
300
400
500
600
Time (s)
Figure 13. Temperature distribution at a heating capacity of 102.6 W with R134a coolant and a filling ratio of 293% (the entire system except the horizontal tube at the top is submerged) 23 22.5
Temperature (oC)
22 21.5 21 20.5 20 19.5 19 101 (C)
18.5
106 (C)
111 (C)
118 (C)
18 0
100
200
300
400
500
600
Time (s)
Figure 14. Temperature distribution at a heating capacity of 107.2 W with R134a 24
coolant and a filling ratio of 225% (the entire condenser is submerged) When the flooding limit is reached, the heat pipe operates cyclically. Fig.15 shows the occurrence of such a phenomenon at the exit point of the evaporator. The images in this figure were captured in chronological order. First, no bubbles emerge from the evaporator, as the heating capacity is constant. The bubbles form gradually, and subsequently small bubbles combine to form larger ones. A two-phase slug flow can be observed at the exit point of the evaporator. The uniform small bubbles follow the slug flow. Finally, there are no bubbles coming out. This cycle is repeated. The heat pipe periodically starts and stops operating because of the change in the liquid level in the condenser. When the filling ratio is very high, the entire condenser is filled with liquid, and when the heat flux becomes nonzero, bubbles start forming. The vapor and liquid flow into the vapor pipe, and the liquid level in the condenser decreases. Consequently, a small area is available in the condenser for the vapor to condense, and the heat pipe starts operating. When this happens, the vapor condenses and changes into liquid. Thus, the liquid level in the condenser increases, and the exposed area in the condenser is insufficient for condensation of the vapor leaving the evaporator; therefore, the heat pipe stops operating. The heat pipe operates cyclically. Thus, the temperature of the tube wall changes periodically. In conclusion, three phenomena are caused when the flooding limit is reached. First, the total temperature difference in the heat pipe (i.e., the difference between the tube-wall temperature of the evaporator and the heat sink temperature) 25
increases. Second, overcooling occurs in the liquid pipe. Third, the tube-wall temperature of the evaporator changes periodically.
Figure 15. Flow patterns at exit point of the evaporator at a heating capacity of 102.6 W with R134a coolant and a filling ratio of 293% (the entire heat pipe system except the horizontal tube at the top is filled with the coolant)
3.5 Uncertainty analysis Before an uncertainty analysis is performed, the measurement accuracy of the instruments must be obtained. In this experiment, the temperature is measured using thermocouples, whose accuracy is 0.1 °C after adjustment. (It may be noted that when the heat transfer coefficient is calculated, the accuracy is 0.1 °C. However, Figures 9, 10, 11, 13, and 14 show the variation trends of tube wall temperature when the heat transfer limit is reached. To show the temperature variation trends more clearly, the temperature results obtained using the Agilent equipment are used directly; these 26
temperature values are not adjusted, and the max error is less than 0.5 °C). The pressure is measured using a pressure sensor, whose accuracy is 0.2%. The saturation temperature is calculated using the measured value of pressure; therefore, the accuracy of the saturation temperature can be calculated using the accuracy of the pressure sensor. The uncertainty in the temperature difference is calculated using Equation (8).
y x1 x2 y ln y ln y = ( x1 ) 2 ( x2 ) 2 y x1 x2 (
（8）
x1 2 x2 2 ) ( ) x1 x2 x1 x2
h=
Q AT （9）
h ln h ln h ln h Q A DT 2 = ( Q)2 ( A)2 ( DT ) = ( )2 ( ) 2 ( ) h Q A DT Q A DT
Table 3. Uncertainty under operating conditions of heat pipe system Table 3(a). Results of uncertainty analysis at a heating capacity of 109 W with R22 coolant and a filling ratio of 105% Saturation
Temperature
Heating
Outer
Length of
Heat
Heat
temperature
difference
capacity
diameter
evaporator
transfer
transfer
area
coefficient
0.18%
15.52%
of tube 0.27 °C
15.51%
0.58%
0.17%
0.07%
Table 3(b). Results of uncertainty analysis at a heating capacity of 699 W with R22 27
coolant and a filling ratio of 105% Saturation
Temperature
Heating
Outer
Length of
Heat
Heat
temperature
difference
capacity
diameter
evaporator
transfer
transfer
area
coefficient
0.18%
5.52%
of tube 0.19 °C
5.49%
0.58%
0.17%
0.07%
4. Conclusion a) The boiling heat transfer coefficient increases with an increase in heat flux. Overall, a larger number of nucleation sites, more uniform two-phase flow, and high flow speed are the three reasons for the heat transfer coefficient to be high at high heating capacities. b) If the heat transfer limit is not reached, the heat transfer coefficient of the evaporator does not change with the filling ratio in the filling ratio range of 65–193%. c) The filling ratio is an indirect parameter for characterizing the heat pipe system. The direct parameter influencing the heat pipe system when the heat pipe is operational (i.e., when the heat pipe transfers heat flux) is the liquid level. For example, if the filling ratio is as high as 193% (i.e., when half of the condenser is submerged in liquid), bubbles will form when the heat flux is not zero. Moreover, because of the viscosity between the vapor and the liquid, both the liquid and vapor will flow out of the evaporator and flow into the vapor pipe. Consequently, 28
the liquid level in the condenser decreases, which results in the exposure of a larger area of the tube wall. The condenser can still operate normally under this condition. d) When the heat flux is very high and if the filling ratio is very low, the dry-out limit is reached. On the other hand, if the filling ratio is very high, the flooding limit may be reached. When the dry-out limit is reached, the temperature at the exit point of the evaporator increases and fluctuates considerably because of the overheated vapor. When the flooding limit is reached, the temperature at the entry point of the evaporator is very low because of the overcooled liquid.
Reference [1] Wu X P, Mochizuki M, Mashiko K, et al. Energy conservation approach for data center cooling using heat pipe based cold energy storage system [C]. Semiconductor Thermal Measurement and Management Symposium, 2010. SEMI-THERM 2010. 26th Annual IEEE. IEEE, 2010: 115-122. [2] Azad E. Theoretical and experimental investigation of heat pipe solar collector [J]. Experimental Thermal and Fluid Science, 2008, 32(8): 1666-1672. [3] Noie-Baghban S H, Majideian G R. Waste heat recovery using heat pipe heat exchanger (HPHE) for surgery rooms in hospitals [J]. Applied Thermal Engineering, 2000, 20(14): 1271-1282. [4] Zhou F, Tian X, Ma G Y. Investigation into the energy consumption of a data 29
center with a thermosiphon heat exchanger [J]. Chinese Science Bulletin, 2011, 56: 2185-2190. [5] Qian X D, Li Z, Tian H. Application of Heat Pipe System in Data Center Cooling, 1st ed., Progress in Sustainable Energy Technologies, Vol. II, Springer International Publishing, 2014: 609-620. [6] Chernysheva M A, Yushakova S I, Maydanik Y F. Copper–water loop heat pipes for energy-efficient cooling systems of supercomputers [J]. Energy, 2014, 69: 534-542. [7] Tian H. Research on cooling technology for high heat density data center, Doctoral Dissertation, Tsinghua University, 2012. (in Chinese) [8] Tong Z, Ding T, Li Z, Liu X H. An experimental investigation of an R744 two-phase thermosyphon loop used to cool a data center [J]. Applied Thermal Engineering, 2015, 90: 362-365. [9] Ding T, He Z guang, Hao T, et al. Application of separated heat pipe system in data center cooling [J]. Applied Thermal Engineering, 2016, 109: 207-216. [10] Qian X D, Li Z, Li Z X. Experimental study on data center heat pipe air conditioning system [J]. Journal of Engineering Thermophysics, 2012, 33(7): 1217-1220. (in Chinese) [11] Hu Z, Zhang Z, Jin T, et al. Experimental study on the working fluid filling rates of microchannel evaporator in a special separate type heat pipe [J]. Journal of Refrigeration, 2015, 36(4): 98-102. (in Chinese) 30
[12] Ling L, Zhang Q, Yu Y, et al. Experimental study on the thermal characteristics of micro channel separate heat pipe respect to different filling ratio [J]. Applied Thermal Engineering, 2016, 102: 375-382. [13] Zhang H, Shi Z, Liu K, et al. Experimental and numerical investigation on a CO2 loop thermosyphon for free cooling of data centers [J]. Applied Thermal Engineering, 2017, 111: 1083-1090. [14] Noie S H. Heat transfer characteristics of a two-phase closed thermosiphon [J]. Applied Thermal Engineering, 2005, 25(4): 495-506. [15] Payakaruk T, Terdtoon P, Ritthidech S. Correlations to predict heat transfer characteristics of an inclined closed two-phase thermosyphon at normal operating conditions [J]. Applied Thermal Engineering, 2000, 20(9): 781-790. [16] Noie S H, Sarmasti Emami M R, Khoshnoodi M. Effect of inclination angle and filling ratio on thermal performance of a two-phase closed thermosyphon under normal operating conditions [J]. Heat Transfer Engineering, 2007, 28(4): 365-371. [17] Park Y J, Kang H K, Kim C J. Heat transfer characteristics of a two-phase closed thermosyphon to the fill charge ratio [J]. International Journal of Heat & Mass Transfer, 2002, 45(23): 4655-4661. [18] Aly W I A, Elbalshouny M A, El-Hameed H M A, et al. Thermal performance evaluation of a helically-micro-grooved heat pipe working with water and aqueous Al 2 O 3 nanofluid at different inclination angle and filling ratio [J]. Applied Thermal Engineering, 2016, 110: 1294-1304. 31
[19] Zheng Y, Li Z, Liu X, et al. Retrofit of air-conditioning system in data center using separate heat pipe system [C]. Proceedings of the 8th International Symposium on Heating, Ventilation and Air Conditioning. Springer Berlin Heidelberg, 2014: 685-694. [20] Zhang P, Wang B, Shi W, et al. Experimental investigation on two-phase thermosyphon loop with partially liquid-filled downcomer [J]. Applied Energy, 2015, 160: 10-17. [21] Solomon A B, Mathew A, Ramachandran K, et al. Thermal performance of anodized two phase closed thermosyphon (TPCT) [J]. Experimental Thermal and Fluid Science, 2013, 48: 49-57. [22] Jouhara H, Robinson A J. Experimental investigation of small diameter two-phase closed thermosyphons charged with water, FC-84, FC-77 and FC-3283 [J]. Applied Thermal Engineering, 2010, 30(2): 201-211. [23] Imura H, Kusuda H, Ogata J I, et al. Heat transfer in two-phase closed-type thermosyphons [J]. JSME Transactions, 1979, 45: 712-722. [24] Jafari D, Filippeschi S, Franco A, et al. Unsteady experimental and numerical analysis of a two-phase closed thermosyphon at different filling ratios [J]. Experimental Thermal and Fluid Science, 2017, 81: 164-174. [25] Rohsenow W M. A method of correlating heat transfer data for surface boiling of liquids. Cambridge, Mass.: MIT Division of Industrial Cooporation, 1951. [26] Bergman T L, Incropera F P, DeWitt D P, et al. Fundamentals of heat and mass 32
transfer. John Wiley & Sons, 2011. [27] Cooper M G. Saturation nucleate pool boiling—a simple correlation//Inst. Chem. Eng. Symp. Ser. 1984, 86(2): 785-793. [28] Louahlia-Gualous H, Le Masson S, Chahed A. An experimental study of evaporation and condensation heat transfer coefficients for looped thermosyphon. Applied Thermal Engineering, 2017, 110: 931-940.
Acknowledgments This study was supported by the State Natural Sciences Foundation of China (No. 51376097，No. 51326002).
33
Highlights 1. Boiling heat transfer coefficient at different filling ratio and heating capacity is measured. 2. Flowing pattern in the liquid and vapor pipe is studied. 3. Heat transfer limits of the separated heat pipes system is researched. 4. As long as no heat transfer limits, the boiling heat-transfer coefficient of the evaporator almost does not change with respect to the filling ratio.
34
S0894-1777(17)30326-6 https://doi.org/10.1016/j.expthermflusci.2017.10.019 ETF 9243
To appear in:
Experimental Thermal and Fluid Science
Received Date: Revised Date: Accepted Date:
12 April 2017 29 September 2017 19 October 2017
Please cite this article as: T. Ding, H. wen Cao, Z. guang He, Z. Li, Visualization experiment on boiling heat transfer and flow characteristics in separated heat pipe system, Experimental Thermal and Fluid Science (2017), doi: https:// doi.org/10.1016/j.expthermflusci.2017.10.019
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Visualization experiment on boiling heat transfer and flow characteristics in separated heat pipe system Tao Ding, Han wen Cao, Zhi guang He, and Zhen Li* Key Laboratory for Thermal Science and Power Engineering of Ministry of Education, Department of Engineering Mechanics, Tsinghua University, Beijing 100084, China Tel.: 010-62772918; E-mail: [email protected]
Abstract: This paper presents an experimental study on boiling heat transfer and flow characteristics in a separated heat pipe system. The flow patterns in liquid and vapor pipes are clearly observed, and the boiling heat transfer coefficient is measured. The main aim of this study is to examine the heat transfer characteristics at different filling ratios using a combination of visualization and measurement methods. First, the relationship between the boiling heat transfer coefficient and the heating capacity is analyzed at different filling ratios when the heat pipe system operates normally. Second, the conditions for reaching the heat transfer limits and the phenomena occurring at these limits are analyzed. The analysis of the heat transfer characteristics shows that for a given heating capacity, the boiling heat transfer coefficient in the evaporator remains nearly constant over a wide range of filling ratios. At a given filling ratio, the heat transfer coefficient increases with an increase in the heating capacity. Moreover, the flow pattern at the evaporator exit changes from slug flow to 1
bubble flow. Two types of heat transfer limits, i.e., the flooding limit and the dry-out limit, are analyzed at sufficiently high and low filling ratios. Keywords: separated heat pipe, heat transfer coefficient, flow pattern, visualization, filling ratio
Nomenclature h
Heat transfer coefficient, W/m2 K
Q
Heating capacity, W
qh
Heat flux, W/m2
G
Acceleration due to gravity, m/s2
FR
Filling ratio
ΔT
Temperature difference, °C
T
Temperature, °C
ρ
Density, kg/m3
Cp
Specific heat, J/kg
μ
Viscosity, kg/ms
Pr
Saturation pressure/critical pressure
M
Relative molecular weight
Subscri pt i
inside 2
o
outside
v
vapor
l
liquid
1. Introduction The heat pipe is an efficient and energy-saving heat transfer device. Heat pipes are widely used in several applications, such as energy storage systems [1, 2] and waste heat recovery systems [3], particularly in data centers or cooling systems of IT equipment [4-7]. A thermosiphon is a type of heat pipe that is driven by gravity. Thermosiphons can be broadly classified into two types. The first type is the two-phase closed thermosiphon. This system contains only a single tube. The evaporator and condenser are located at the bottom and top, respectively, of the thermosiphon. The second type is the loop thermosiphon. In this system, the condenser and evaporator are mounted separately and connected by liquid and vapor pipes. The loop thermosiphon is also known as the separated heat pipe system. Because of the specific structure of this heat pipe system, heat can be transferred over long distances. Fig. 1 shows the schematic of a simple separated heat pipe system. This system comprises four components: an evaporator, a condenser, vapor pipe, and liquid pipe. As the condenser is installed above the evaporator, this system can be driven by the gravity of the liquid (or by the buoyancy force of the vapor), instead of being driven by a pump. When there is no overheating or overcooling of the coolant, 3
phase changes occur only inside the heat pipe. Moreover, the heat transfer coefficient of the two-phase flow is higher than that of the single-phase flow. Thus, this heat pipe system has a higher heat transfer ability than a single-phase heat exchanger.
Figure 1. Schematic of heat pipe system After making minor improvements and modification, the separated heat pipe system shown in Fig. 1 can be used for the cooling of a data center [8, 9]. A data center is a facility where IT equipment is centrally stored. The heat generation rate in a data center can be extremely high. If the heat generated is not transferred immediately, the temperature of the IT equipment may increase rapidly, thereby threatening safe operation. In a heat pipe system installed for cooling a data center, the evaporator is installed inside the front and back doors of the rack. Thus, the heat generated by the IT equipment can be immediately absorbed by the nearby parts. A plate-type heat exchanger is used as the condenser (with chilled water used as the heat sink); it is installed in another room because water is prohibited in a data center. Although heat pipe systems are being used for the cooling of data centers, some 4
theoretical problems with these systems need to be addressed. The filling ratio is an important parameter for evaluating the heat transfer ability of the separated heat pipe system. Some studies [10-18] aimed at determining the relationship between the heat transfer ability of a heat pipe and its filling ratio. A few studies analyzed the filling ratio of a separated heat pipe system [10,11,13] and found that with an increase in the filling ratio, the heat transfer capacity increased initially, then attained the maximum value, and subsequently decreased. Ling et al.’s results [12] showed that the heat transfer capacity of the loop thermosiphon system remained constant in a certain range of the filling ratio. Zhang et al. [13] used R22 and R744 as the coolants and found that the optimal filling ratio for R744 was approximately 150%; the heat transfer capacity remains almost constant when the filling ratio is higher than 150%. However, the optimal filling ratio for R22 is a single value, and after this filling ratio was achieved (at higher filling ratios), the heat transfer capacity decreased. Researchers have also investigated the filling ratio of a two-phase closed thermosiphon system [14-18]. Noie [14] reported that the optimal filling ratio is influenced by aspect ratio, where the aspect ratio is defined as the ratio of the section length of the evaporator to the inner diameter of the tube. Payakaruk et al. [15] studied the relationship between the filling ratio and the inclination angle, and concluded that the filling ratio does not affect the heat transfer capacity at different inclination angles. Noie et al. [16] studied the performance of the two-phase closed thermosiphon and found that the heat transfer rate increased with increasing filling 5
ratio. Park et al. [17] studied the effects of fill charge ratio and found that the effects seemed to be negligible in the evaporator. Aly et al. [18] found that the filling ratio has significant effects on the performance of the heat pipe (a two-phase closed thermosiphon). The present study focuses on the performance of a separated heat pipe system. The following aspects are considered in the study. I)
The mechanism of heat transfer at different filling ratios.
II)
The heat transfer coefficient of the evaporator. As the flow in the heat pipe is driven by the buoyancy force, the evaporation that occurs in the separated heat pipe system is different from that occurring in pool boiling and forced flow boiling in the tube. Hence, it is important to study the heat transfer characteristics.
Some studies [8-10, 19, 20] examined the overall performance of the heat pipe system, such as the temperature difference or the thermal resistance in the entire heat transfer process. However, it is difficult to differentiate the heat transfer coefficient of the evaporator from the total thermal resistance (for example, in evaporators, the total thermal resistance includes the air-to-tube and fin convection thermal resistances, heat conduction resistance in the tube and boiling heat transfer thermal resistance). Some studies [21-24] analyzed the heat transfer characteristics of a two-phase closed thermosiphon. Imura et al. [23] established an empirical formula to calculate the heat transfer coefficient of the evaporator. Jafari et al. [24] subsequently used this 6
formula to calculate the boiling heat transfer coefficient. However, there is a clear difference between the two-phase closed thermosiphon and the separated heat pipe system: in the two-phase closed thermosiphon, the vapor and liquid flow in different directions, whereas in the separated heat pipe system, they flow in the same direction. Hence, the heat transfer characteristics of these two systems are different. In the present study, the heat transfer coefficient of the evaporator was directly measured.
2. Experimental equipment and method 2.1 Experimental equipment To overcome the above problems, a visualization experiment was performed. Fig. 2 shows a schematic of the experiment equipment.
Figure 2. Structure of heat pipe system. Left: Overall view; Right: Enlarged view of evaporator
7
The evaporator comprises a copper tube, whose inner and outer diameters are 8 mm and 12 mm, respectively. The length of the evaporator is 151.5 cm (this value is the tube length around which an electric resistance wire is wound; the total tube length is slightly greater than 151.5 cm). The condenser is in the form of a tube in the tube heat exchanger. The evaporator is heated via the electric resistance wire, and the condenser is cooled with chilled water. Sixteen thermocouples are welded by soldering tin on the surface of the outer tube wall. Table 1 lists the positions of the thermocouples.
Table 1. Positions of thermocouples, with reference to the position of the thermocouple at the top considered as the zero point No.
101
102
103
104
105
106
107
108
0
100
198
300
396
500
598
694
109
110
111
112
113
116
115
118
796
894
1007
1097
1197
1302
1404
1501
Position (mm) No. Position (mm)
The electric resistance heating wire is wrapped with an insulation film, and two layers of thin Kapton film are placed between the outer tube wall and the electric resistance wire for providing electrical insulation (the thermocouples are welded by soldering tin on the surface of the outer tube wall and they are also wrapped inside 8
Kapton film), as shown in Fig. 2. The thickness of the Kapton film is approximately 0.13 mm. The electric resistance wire is wrapped with thermal insulation material. Such structure ensures that the heat generated by the wire is transferred to the evaporator. The electric resistance wire is heated using direct current. As the wire is wound tightly into a spiral form around the tube, the heating boundary can be regarded as a constant heat flux boundary. The heat flux may be transferred to the coolant or the external environment. The thermal resistance to the coolant is termed the internal thermal resistance, and that to the external environment is termed the external thermal resistance. The internal thermal resistance is estimated to be approximately only 0.53% of the external thermal resistance. This implies that the heat loss from the evaporator can be ignored. The vapor and liquid pipes are made of glass so that the flow pattern of the coolant can be viewed clearly. In the experiment, R134a and R22 are used as coolants. In the heating circuit, a constant resistance of 0.01 Ω is connected in series, as shown in Fig. 2. Thus, through measurement of the voltage across the constant resistance, the current in the circuit can be calculated using the following equation: current (A) = voltage (V) / 0.01 Ω. The data required for the calculation of the heat transfer coefficient are collected under steady conditions of the heat pipe system. 2.2 Experimental method The heat transfer coefficient is calculated using Eq. (1), where Q is calculated by 9
multiplying the voltage with the current. The voltage of the power supply is measured using a Fluke 15B digital multimeter, whose relative accuracy is 0.5%. The voltage across the resistance (0.01 Ω) is measured using an Agilent 34970A data acquisition unit, whose relative accuracy is approximately 0.3% (the accuracy of this device is actually much higher, but as only three significant digits are considered, the accuracy is estimated as 0.3%). The inner surface area is 0.0381 m2 (A = 3.14 × 0.008 m × 1.515 m). The temperature difference is taken as the difference between the temperature of the inner wall of the tube and the saturation temperature. The inner wall temperature can be calculated using Equation (2), where Ti and To are the inner and outer wall temperatures, respectively, and λ is the thermal conductivity of copper, which is equal to 397 W/mK. Further, ro = 12 mm and ri = 8 mm. The outer wall temperature refers to the average outer surface temperature obtained using the 16 thermocouples. The saturation temperature is calculated using the evaporation pressure. The pressures at the entrance and exit points of the evaporator are measured, and the saturation pressure is taken as the average of these two values.
h=
UI Q = AT AT
（1）
ro ri
（2）
Ti =To
Q 2l
ln
3. Analysis of results 3.1 Variation trend of heat transfer coefficient 10
Fig. 3 shows the experimental results of the heat transfer coefficient for the R22 coolant at a filling ratio of 105%. Here, the heating capacity is in the range of 109–699 W. As shown in Fig. 3, the heat transfer coefficient increases with an increase in the heating capacity. The curve has a parabolic profile. A detailed definition of the filling ratio is provided in section 3.2. Fig. 4 shows the flow pattern at the exit area of the evaporator. When the heating capacity is 109 W (considering the transfer area is 0.0381 m2), the heat flux is 2.86 kW/m2. Two-phase slug flow is observed, in which large bubbles are clearly visible. The results show that the flow velocity is not very high. It can be seen that small bubbles coalesce to form larger ones. However, as can be seen in Fig. 4, the flow pattern at a heating capacity of 699 W (heat flux of 18.35 kW/m2) is more complex than that at 109 W. The vapor generation rate is higher than that at 109 W. Because the vapor generation is higher, the vapor fills the central part of the evaporator and pushes the liquid aside. Moreover, it is observed that the small bubbles move upward. The flow pattern in the inner copper tube cannot be viewed, nor can it be predicted extremely realistically using the flow pattern at the exit point; however, it can still be predicted using the images captured (flow pattern) at the exit point. There are three main reasons for the heat transfer coefficient to be high at high heating capacities. The first reason is that more bubbles are formed when the heat flux is high. The bubbles are formed under two conditions: the first condition is when small sunken holes are present on the heating surface; when the surface is overheated, bubbles will be generated at these small 11
sunken holes. The places where the bubbles are generated are called nucleation sites. The second condition is when the surface is overheated. Overheating increases with an increase in heating capacity. Thus, at low heating capacities, bubbles do not form at some nucleation sites, whereas at high heating capacities, bubbles form at these nucleation sites. Hence, there is an increase in the number of nucleation points. In other words, the bubble formation rate is much higher at high heating capacities. The second reason for the increase in heat transfer coefficient at higher heating capacities is that the two-phase flow can become more uniform with increasing heat flux. The third reason is that the velocity of the two-phase flow increases with increasing heat capacity. The flow pattern in the separated heat pipe can be regarded as a combination of pool boiling and flow boiling. Flow boiling at a low flow speed may enhance the heat transfer coefficient; however, when the flow speed increases continuously, the flow boiling may hinder pool boiling. In fact, in our experiment, this effect is not seen. To sum up, a large number of nucleation sites, more uniform two-phase flow, and high flow speed are the three main reasons for a high heat transfer coefficient at high heat transfer capacities.
12
Heat transfer coefficient (W/m2K)
5000 4500 4000 3500 3000 2500 2000 1500 1000 0
100
200
300
400
500
600
700
800
Heating capacity (W)
Figure 3. Relationship between heating capacity and heat transfer coefficient at a filling ratio of 105% with R22 coolant
109 W
301 W
699 W
Filling ratio: 105%; R22 coolant Figure 4. Flow patterns at the exit point of evaporator with different heating capacities
Some others empirical equation has been used to compare our experiment results. For example, Imura et al. [23] proposed an empirical equation (Equation 3) to predict the heat transfer coefficient of the evaporator of a two-phase closed thermosiphon. In Equation (3), subscripts “l” and “g” indicate liquid and vapor, respectively. 13
he 0.32
0.2 0.4 l0.65l0.3c 0.7 qh pl g ( Pin / Pa )0.3 0.25 0.4 0.1 v r l
(3)
The results of the present study were compared with other researchers’ correlations, such as Rohsenow’s correlation [25,26] (Equation 4), Cooper’s correlation [27] (Equation 5, in fact, this equation has considered the surface roughness, here, the surface roughness is omitted), and Hasna’s correlation [28] (Equation 6). h
c pl q 0.667 hsf Cwl
(
1 l hsf
g ( l v )
) 0.33 (
c pl l
l
) n
(4)
he 55 pr0.12 (-lg pr )0.55 M 0.5 q0.67
(5)
he 7704M 0.5 q0.157 pr 0.12 / ( log pr )0.55
(6)
Out of these equations, Rohsenow’s correlation is applicable for pool boiling, and is obtained by theoretical derivation; in this equation, Cwl is taken as 0.0049, and n = 1.7. hsf means latent heat (J/kg). The average error obtained is 0.43. Cooper’s equation is mainly applicable for the coolant, and the average error obtained is 0.17. Hasna’s equation is derived based on the results for the separated heat pipe system (loop thermosiphon, water used as coolant, sub-atmospheric pressure). The results show that when the heat transfer capacity is 109 W, the error is 67%; however, when the heat transfer capacity is higher than 400 W, the error is less than 8.2%. The average error is 28%. Thus, the results obtained using Hasna’s equation are the closest to the results obtained in our experiment. In fact, the heat flux as per Hasna’s equation is approximately 2–20 W/cm2, while in our experiment, it is approximately 0.26–1.84 W/cm2. Thus, Hasna’s equation is more suitable for the high heat flux condition in our 14
experiment. The experimentally obtained values of the heat transfer coefficient are much higher than the values obtained using the equation proposed by Imura et al. [23]. As can be seen in Table 2, when the heating capacity is 109 W, the difference between these two sets of results is 20.5%. However, when the heating capacity is 699 W, the difference is as high as 42.3%. This is because of the difference in the flow patterns in the loop thermosiphon and the two-phase closed thermosiphon.
Heat transfer coefficient(W/ m2K)
9000 8000
Experiment
7000
Imura Rohsenow
6000
Hasna
5000
Cooper 4000 3000
2000 1000 0 0
100
200
300
400
500
600
700
800
Heat transfer capacity (W)
Figure 5. Comparison of heat transfer coefficients obtained experimentally in the present study and using equation proposed by Imura et al. [23] (R22 coolant, filling ratio: 105%) Equation (3) was derived based on an experiment and analysis of the two-phase closed thermosiphon. In the two-phase closed thermosiphon, the condenser and evaporator are located in the same vertical tube. The vapor and liquid flow in the thermally insulated part; the vapor flows upward, whereas the liquid flows downward. 15
The two-phase coolant in the evaporator remains almost constant, in another word, there is no macro flow for coolant and the coolant do not flow upward or downward. Thus, the heat transfer in the evaporator is more similar to the one in pool boiling. However, in the separated heat pipe system, the vapor and liquid flow in the same direction. The two-phase coolant in the evaporator flows faster than that in the two-phase closed thermosiphon for the same tube diameter and heat flux. Thus, the heat transfer in the evaporator of the separated heat pipe system is more similar to a combination of the heat transfer processes of pool boiling and flow boiling. The differences in the heat transfer patterns lead to different heat transfer characteristics of the separated heat pipe system and the two-phase closed thermosiphon. Table 2. Comparison of heat transfer coefficients obtained experimentally in the present study and using the equation proposed by Imura et al. [23] (R22 coolant, filling ratio: 105%)
Heating capacity (W)
Experimental result (W/m2K)
As per the equation proposed by Imura [23] (W/m2K)
109 206 301 400 498 600 699
1536 2280 2888 3381 3797 4306 4695
1221 1587 1865 2108 2322 2526 2709
Relative error (%)
-20.5 -30.4 -35.4 -37.7 -38.8 -41.3 -42.3
3.2 Relationship between heat transfer coefficient and filling ratio 16
Five different filling ratios are considered in the experiment with the R22 coolant. The filling ratio (FR) is defined under the condition when there is no heat flux (i.e., the heat pipe system is not operational). The filling ratio is defined as the height of the liquid coolant (Hl) (with the entrance of the evaporator considered as the origin) divided by the height of the entire evaporator tube (He, 151.5 cm) (see Equation (7) and Fig. 6). Thus, the filling ratio can be higher than 100%.
FR%
Hl 100% He
(7)
Figure 6. Definition of filling ratio (only a schematic, not to scale) Fig. 7 shows the boiling heat transfer coefficient for R22 at five different filling ratios considered in the experiment. The lowest filling ratio is 65% (which is a very low value). The highest filling ratio of 193% (which is a sufficiently high value) implies that nearly half of the condenser is submerged in the liquid.
17
Heat transfer coefficient (W/m2K)
5000 4500
65%
4000 105% 3500 3000
99%
2500
161%
2000
193%
1500 1000 0
200
400
600
800
1000
Heating capacity (W)
Figure 7. Heat transfer coefficient at different heating capacities and filling ratios with R22 coolant The results show that despite significant variations in the filling ratio, the heat transfer coefficient remains nearly constant. This also implies that if the heat pipe can operate normally (no heat transfer limit is reached), the heat transfer ability of the evaporator remains unchanged.
Figure 8. Schematics of heat pipe system. Left: No heat flux as the system is not 18
operational; Right: System operating normally
The above phenomenon can be explained with the help of the schematics shown in Fig. 8. The figure shows the flow-pattern diagrams of the heat pipe system when it is not operational (view on the left) and when it is operating normally (view on the right). For example, if the filling ratio is sufficiently high so that some part of the condenser is submerged (the filling ratio is much higher than 100%), and when there is sufficient heat flux, bubbles will form. Moreover, because of the viscosity between the vapor and the liquid, the liquid and vapor will flow out of the evaporator and flow into the vapor pipe. With continuous flow of the liquid and vapor into the vapor pipe, the liquid level in the condenser may decrease. Consequently, a larger area of the tube wall of the condenser is exposed, and the condenser operates normally. If the filling ratio is not very high, e.g., lower than 100%, the entire tube wall of the condenser is exposed, and the condenser operates normally. The experimental results lead to the conclusion that the liquid level in the evaporator does not influence the heat transfer coefficient even the filling ratio is sufficiently high (but without reaching the flooding limit). If the filling ratio is sufficiently low, the liquid film cannot cover the evaporator, and the dry-out limit may be reached. If the filling ratio is very high, when the heat pipe starts operating, only a small area of the tube wall of the condenser is exposed, in spite of the liquid and vapor flowing 19
into the vapor pipe. This area is insufficient to condense the vapor leaving the evaporator, and hence the heat pipe system cannot operate normally. We define this operating limit as the flooding limit. However, as long as the heat pipe system operates normally, the liquid level in the evaporator does not influence the heat transfer coefficient.
3.3 Heat transfer limit for heat pipe system: dry-out limit If the filling ratio of the heat pipe is insufficient when the heat flux is very high, the dry-out limit may be reached. Figures 9, 10, and 11 show the temperatures measured by the thermocouples at four points on the outer tube wall at heating capacities of 109.9 W, 493.9 W, and 597.2 W, respectively. The results show that although the filling ratio is 65%, when the heating capacity is 109.9 W, the temperature of the outer tube wall remains constant. Fig. 12(a) shows the flow pattern of the coolant at the evaporator exit point, wherein the external flow is slug flow.
20
19 18.5
101 (C)
106 (C)
112 (C)
118 (C)
Temperature (oC)
18 17.5 17 16.5 16 15.5 15 14.5 14 0
100
200
300
400
500
600
Time (s)
Figure 9. Temperatures measured by thermocouples at four points at a heating capacity of 109.9 W and a filling ratio of 65%, with R134a coolant
34 33.5
101 (C)
106 (C)
112 (C)
118 (C)
Temperature (oC)
33 32.5 32 31.5 31 30.5 30 29.5 29 0
100
200
300
400
500
600
Time (s)
Figure 10. Temperatures measured by thermocouples at four points at a heating capacity of 493.9 W and a filling ratio of 65%, with R134a coolant
21
38.5 38
101 (C)
106 (C)
112 (C)
118 (C)
Temperature (oC)
37.5 37 36.5 36 35.5 35 34.5 34 33.5 0
100
200
300
400
500
600
Time (s)
Figure 11. Temperatures measured by thermocouples at four points at a heating capacity of 597.2 W and a filling ratio of 65%, with R134a coolant
(a) 65%, 109.9 W Two-phase slug flow
(b) 65%, 493.9 W
(c) 65%, 597.2 W
Annular flow
Annular flow
Figure 12. Flow patterns at exit point of evaporator at different heating capacities and a filling ratio of 65%, with R134a coolant
At the heating capacity of 493.9 W, the temperature of the outer tube wall remains nearly constant. Fig. 12(b) shows the annular flow pattern at the exit position. The flow pattern indicates that with the increase in heating capacity, the vapor generation 22
rate is higher than that at 109.9 W. This implies that the vapor has already filled the central part of the pipe. The vapor pushes the liquid aside, which results in the formation of a liquid film on the surface of the inner tube wall. At the heating capacity of 597.2 W (see Fig. 11 for temperature results and Fig. 12(c) for the flow pattern), the temperature at the exit position of the evaporator varies considerably with time, and the variation trend is found to be irregular. This phenomenon is attributed to the flow pattern. Since the heat flux is very high at this heating capacity, the vapor generation rate also increases significantly; thus, the void fraction also increases, as can be seen in Fig. 12. Consequently, the liquid cannot cover the inner tube wall uniformly. In other words, the tube wall of the evaporator cannot be covered completely by the liquid film. Thus, the temperature of the top tube wall fluctuates significantly.
3.4 Heat transfer limit for heat pipe system: flooding limit The term flooding limit has already been explained in section 3.2. Figures 13 and 14 show the temperature distributions when the flooding limit is reached. These results correspond to very high filling ratios. As can be seen in these figures, when the flooding limit is reached, the temperature indicated by thermocouple no. 118 is much lower than those indicated by the other thermocouples. This phenomenon is caused by overcooling of the liquid pipe. As the filling ratio is very high, the bottom of the condenser is completely covered by the liquid coolant. Thus, the liquid coolant 23
temperature is closer to the heat sink temperature. Moreover, the temperatures at these points change cyclically with time because the liquid level in the condenser increases and decreases cyclically. 38 36
Temperature (oC)
34 32 30 101 (C)
28
106 (C)
111 (C)
118 (C)
26 24 0
100
200
300
400
500
600
Time (s)
Figure 13. Temperature distribution at a heating capacity of 102.6 W with R134a coolant and a filling ratio of 293% (the entire system except the horizontal tube at the top is submerged) 23 22.5
Temperature (oC)
22 21.5 21 20.5 20 19.5 19 101 (C)
18.5
106 (C)
111 (C)
118 (C)
18 0
100
200
300
400
500
600
Time (s)
Figure 14. Temperature distribution at a heating capacity of 107.2 W with R134a 24
coolant and a filling ratio of 225% (the entire condenser is submerged) When the flooding limit is reached, the heat pipe operates cyclically. Fig.15 shows the occurrence of such a phenomenon at the exit point of the evaporator. The images in this figure were captured in chronological order. First, no bubbles emerge from the evaporator, as the heating capacity is constant. The bubbles form gradually, and subsequently small bubbles combine to form larger ones. A two-phase slug flow can be observed at the exit point of the evaporator. The uniform small bubbles follow the slug flow. Finally, there are no bubbles coming out. This cycle is repeated. The heat pipe periodically starts and stops operating because of the change in the liquid level in the condenser. When the filling ratio is very high, the entire condenser is filled with liquid, and when the heat flux becomes nonzero, bubbles start forming. The vapor and liquid flow into the vapor pipe, and the liquid level in the condenser decreases. Consequently, a small area is available in the condenser for the vapor to condense, and the heat pipe starts operating. When this happens, the vapor condenses and changes into liquid. Thus, the liquid level in the condenser increases, and the exposed area in the condenser is insufficient for condensation of the vapor leaving the evaporator; therefore, the heat pipe stops operating. The heat pipe operates cyclically. Thus, the temperature of the tube wall changes periodically. In conclusion, three phenomena are caused when the flooding limit is reached. First, the total temperature difference in the heat pipe (i.e., the difference between the tube-wall temperature of the evaporator and the heat sink temperature) 25
increases. Second, overcooling occurs in the liquid pipe. Third, the tube-wall temperature of the evaporator changes periodically.
Figure 15. Flow patterns at exit point of the evaporator at a heating capacity of 102.6 W with R134a coolant and a filling ratio of 293% (the entire heat pipe system except the horizontal tube at the top is filled with the coolant)
3.5 Uncertainty analysis Before an uncertainty analysis is performed, the measurement accuracy of the instruments must be obtained. In this experiment, the temperature is measured using thermocouples, whose accuracy is 0.1 °C after adjustment. (It may be noted that when the heat transfer coefficient is calculated, the accuracy is 0.1 °C. However, Figures 9, 10, 11, 13, and 14 show the variation trends of tube wall temperature when the heat transfer limit is reached. To show the temperature variation trends more clearly, the temperature results obtained using the Agilent equipment are used directly; these 26
temperature values are not adjusted, and the max error is less than 0.5 °C). The pressure is measured using a pressure sensor, whose accuracy is 0.2%. The saturation temperature is calculated using the measured value of pressure; therefore, the accuracy of the saturation temperature can be calculated using the accuracy of the pressure sensor. The uncertainty in the temperature difference is calculated using Equation (8).
y x1 x2 y ln y ln y = ( x1 ) 2 ( x2 ) 2 y x1 x2 (
（8）
x1 2 x2 2 ) ( ) x1 x2 x1 x2
h=
Q AT （9）
h ln h ln h ln h Q A DT 2 = ( Q)2 ( A)2 ( DT ) = ( )2 ( ) 2 ( ) h Q A DT Q A DT
Table 3. Uncertainty under operating conditions of heat pipe system Table 3(a). Results of uncertainty analysis at a heating capacity of 109 W with R22 coolant and a filling ratio of 105% Saturation
Temperature
Heating
Outer
Length of
Heat
Heat
temperature
difference
capacity
diameter
evaporator
transfer
transfer
area
coefficient
0.18%
15.52%
of tube 0.27 °C
15.51%
0.58%
0.17%
0.07%
Table 3(b). Results of uncertainty analysis at a heating capacity of 699 W with R22 27
coolant and a filling ratio of 105% Saturation
Temperature
Heating
Outer
Length of
Heat
Heat
temperature
difference
capacity
diameter
evaporator
transfer
transfer
area
coefficient
0.18%
5.52%
of tube 0.19 °C
5.49%
0.58%
0.17%
0.07%
4. Conclusion a) The boiling heat transfer coefficient increases with an increase in heat flux. Overall, a larger number of nucleation sites, more uniform two-phase flow, and high flow speed are the three reasons for the heat transfer coefficient to be high at high heating capacities. b) If the heat transfer limit is not reached, the heat transfer coefficient of the evaporator does not change with the filling ratio in the filling ratio range of 65–193%. c) The filling ratio is an indirect parameter for characterizing the heat pipe system. The direct parameter influencing the heat pipe system when the heat pipe is operational (i.e., when the heat pipe transfers heat flux) is the liquid level. For example, if the filling ratio is as high as 193% (i.e., when half of the condenser is submerged in liquid), bubbles will form when the heat flux is not zero. Moreover, because of the viscosity between the vapor and the liquid, both the liquid and vapor will flow out of the evaporator and flow into the vapor pipe. Consequently, 28
the liquid level in the condenser decreases, which results in the exposure of a larger area of the tube wall. The condenser can still operate normally under this condition. d) When the heat flux is very high and if the filling ratio is very low, the dry-out limit is reached. On the other hand, if the filling ratio is very high, the flooding limit may be reached. When the dry-out limit is reached, the temperature at the exit point of the evaporator increases and fluctuates considerably because of the overheated vapor. When the flooding limit is reached, the temperature at the entry point of the evaporator is very low because of the overcooled liquid.
Reference [1] Wu X P, Mochizuki M, Mashiko K, et al. Energy conservation approach for data center cooling using heat pipe based cold energy storage system [C]. Semiconductor Thermal Measurement and Management Symposium, 2010. SEMI-THERM 2010. 26th Annual IEEE. IEEE, 2010: 115-122. [2] Azad E. Theoretical and experimental investigation of heat pipe solar collector [J]. Experimental Thermal and Fluid Science, 2008, 32(8): 1666-1672. [3] Noie-Baghban S H, Majideian G R. Waste heat recovery using heat pipe heat exchanger (HPHE) for surgery rooms in hospitals [J]. Applied Thermal Engineering, 2000, 20(14): 1271-1282. [4] Zhou F, Tian X, Ma G Y. Investigation into the energy consumption of a data 29
center with a thermosiphon heat exchanger [J]. Chinese Science Bulletin, 2011, 56: 2185-2190. [5] Qian X D, Li Z, Tian H. Application of Heat Pipe System in Data Center Cooling, 1st ed., Progress in Sustainable Energy Technologies, Vol. II, Springer International Publishing, 2014: 609-620. [6] Chernysheva M A, Yushakova S I, Maydanik Y F. Copper–water loop heat pipes for energy-efficient cooling systems of supercomputers [J]. Energy, 2014, 69: 534-542. [7] Tian H. Research on cooling technology for high heat density data center, Doctoral Dissertation, Tsinghua University, 2012. (in Chinese) [8] Tong Z, Ding T, Li Z, Liu X H. An experimental investigation of an R744 two-phase thermosyphon loop used to cool a data center [J]. Applied Thermal Engineering, 2015, 90: 362-365. [9] Ding T, He Z guang, Hao T, et al. Application of separated heat pipe system in data center cooling [J]. Applied Thermal Engineering, 2016, 109: 207-216. [10] Qian X D, Li Z, Li Z X. Experimental study on data center heat pipe air conditioning system [J]. Journal of Engineering Thermophysics, 2012, 33(7): 1217-1220. (in Chinese) [11] Hu Z, Zhang Z, Jin T, et al. Experimental study on the working fluid filling rates of microchannel evaporator in a special separate type heat pipe [J]. Journal of Refrigeration, 2015, 36(4): 98-102. (in Chinese) 30
[12] Ling L, Zhang Q, Yu Y, et al. Experimental study on the thermal characteristics of micro channel separate heat pipe respect to different filling ratio [J]. Applied Thermal Engineering, 2016, 102: 375-382. [13] Zhang H, Shi Z, Liu K, et al. Experimental and numerical investigation on a CO2 loop thermosyphon for free cooling of data centers [J]. Applied Thermal Engineering, 2017, 111: 1083-1090. [14] Noie S H. Heat transfer characteristics of a two-phase closed thermosiphon [J]. Applied Thermal Engineering, 2005, 25(4): 495-506. [15] Payakaruk T, Terdtoon P, Ritthidech S. Correlations to predict heat transfer characteristics of an inclined closed two-phase thermosyphon at normal operating conditions [J]. Applied Thermal Engineering, 2000, 20(9): 781-790. [16] Noie S H, Sarmasti Emami M R, Khoshnoodi M. Effect of inclination angle and filling ratio on thermal performance of a two-phase closed thermosyphon under normal operating conditions [J]. Heat Transfer Engineering, 2007, 28(4): 365-371. [17] Park Y J, Kang H K, Kim C J. Heat transfer characteristics of a two-phase closed thermosyphon to the fill charge ratio [J]. International Journal of Heat & Mass Transfer, 2002, 45(23): 4655-4661. [18] Aly W I A, Elbalshouny M A, El-Hameed H M A, et al. Thermal performance evaluation of a helically-micro-grooved heat pipe working with water and aqueous Al 2 O 3 nanofluid at different inclination angle and filling ratio [J]. Applied Thermal Engineering, 2016, 110: 1294-1304. 31
[19] Zheng Y, Li Z, Liu X, et al. Retrofit of air-conditioning system in data center using separate heat pipe system [C]. Proceedings of the 8th International Symposium on Heating, Ventilation and Air Conditioning. Springer Berlin Heidelberg, 2014: 685-694. [20] Zhang P, Wang B, Shi W, et al. Experimental investigation on two-phase thermosyphon loop with partially liquid-filled downcomer [J]. Applied Energy, 2015, 160: 10-17. [21] Solomon A B, Mathew A, Ramachandran K, et al. Thermal performance of anodized two phase closed thermosyphon (TPCT) [J]. Experimental Thermal and Fluid Science, 2013, 48: 49-57. [22] Jouhara H, Robinson A J. Experimental investigation of small diameter two-phase closed thermosyphons charged with water, FC-84, FC-77 and FC-3283 [J]. Applied Thermal Engineering, 2010, 30(2): 201-211. [23] Imura H, Kusuda H, Ogata J I, et al. Heat transfer in two-phase closed-type thermosyphons [J]. JSME Transactions, 1979, 45: 712-722. [24] Jafari D, Filippeschi S, Franco A, et al. Unsteady experimental and numerical analysis of a two-phase closed thermosyphon at different filling ratios [J]. Experimental Thermal and Fluid Science, 2017, 81: 164-174. [25] Rohsenow W M. A method of correlating heat transfer data for surface boiling of liquids. Cambridge, Mass.: MIT Division of Industrial Cooporation, 1951. [26] Bergman T L, Incropera F P, DeWitt D P, et al. Fundamentals of heat and mass 32
transfer. John Wiley & Sons, 2011. [27] Cooper M G. Saturation nucleate pool boiling—a simple correlation//Inst. Chem. Eng. Symp. Ser. 1984, 86(2): 785-793. [28] Louahlia-Gualous H, Le Masson S, Chahed A. An experimental study of evaporation and condensation heat transfer coefficients for looped thermosyphon. Applied Thermal Engineering, 2017, 110: 931-940.
Acknowledgments This study was supported by the State Natural Sciences Foundation of China (No. 51376097，No. 51326002).
33
Highlights 1. Boiling heat transfer coefficient at different filling ratio and heating capacity is measured. 2. Flowing pattern in the liquid and vapor pipe is studied. 3. Heat transfer limits of the separated heat pipes system is researched. 4. As long as no heat transfer limits, the boiling heat-transfer coefficient of the evaporator almost does not change with respect to the filling ratio.
34