Nucleate boiling heat transfer coefficients of halogenated refrigerants up to critical heat fluxes

Transcription

1 Nucleate boiling heat transfer coefficients of halogenated refrigerants up to critical heat fluxes K-J Park 1, D Jung 1, and SEShim 2 1 Department of Mechanical Engineering, Inha University, Incheon, Republic of Korea 2 Department of Chemical Engineering, Inha University, Incheon, Republic of Korea The manuscript was received on 11 September 2008 and was accepted after revision for publication on 11 November DOI: / JMES Abstract: In this work, nucleate pool boiling heat transfer coefficients (HTCs) of five refrigerants of differing vapour pressures are measured on a horizontal, smooth copper surface of mm. The tested refrigerants are R123, R152a, R134a, R22, and R32 and HTCs are taken from 10 kw/m 2 to the critical heat flux (CHF) of each refrigerant. Wall and fluid temperatures are measured directly by thermocouples located underneath the test surface and in the liquid pool, respectively. Test results show that nucleate pool boiling HTCs of halogenated refrigerants increase as the heat flux and vapour pressure increase. This typical trend is maintained even at high heat fluxes above 200 kw/m 2. Zuber s prediction equation for CHF is quite accurate showing a maximum deviation of 21 per cent for all refrigerants tested. For all refrigerants, Stephan and Abdelsalam s well-known correlation underpredicted nucleate boiling HTC data up to the CHF with an average deviation of 21.3 per cent, while Cooper s correlation overpredicted the data with an average deviation of 14.2 per cent. On the other hand, Gorenflo s and Jung et al. s correlations showed 5.8 and 6.4 per cent deviations, respectively, in the entire nucleate boiling range up to the CHF. Keywords: nucleate pool boiling, heat transfer coefficients, refrigerants, critical heat flux, electronic cooling 1 INTRODUCTION Nucleate boiling can transfer considerably large amounts of heat due to the bubble formation by latent heat transfer and strong convection caused by bubbles as compared to a single-phase heat transfer. In nucleate boiling, as the surface is superheated, bubbles are formed vigorously on the surface and a complex thermo-fluid phenomenon occurs with vigorous mixing of the liquid and vapour phases. This phenomenon requires heat and, hence, is affected greatly by the thermophysical properties of working fluids. For the past few decades, nucleate boiling heat transfer has been studied in laboratories mainly experimentally with a small-scale heat transfer test section due to the complex relationship between the surface and fluid combination, phase change, random fluid motion, and so on. Corresponding author: Department of Mechanical Engineering, Inha University, Incheon , Republic of Korea. dsjung@inha.ac.kr JMES1356 IMechE 2009 For boiler design in power plants, in general, nucleate boiling data in the entire nucleate boiling of water up to the critical heat flux (CHF) are needed. On the other hand, for evaporator design in refrigeration and air-conditioning equipment, nucleate boiling data at low heat fluxes of normally 50 kw/m 2 are needed for various refrigerants. Because of this trend and requirement, it is difficult to find nucleate boiling data of many refrigerants at high heat fluxes up to the CHF. Recently, as the information technology has developed rapidly, special attention has been paid to more efficient electronic cooling methods for the dissipation of significantly large amounts of heat from computers and microprocessors. For this application, both conventional and new halocarbon refrigerants can be used as working fluids and nucleate boiling data are needed for these refrigerants up to the CHF. For the past decades, heat exchanger industry has been mainly dependent upon certain heat transfer correlations to calculate the amount of heat dissipated during pool boiling. In view of this, nucleate boiling heat transfer correlations are very important to design engineers in the related industry. To develop Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science

2 1416 K-J Park, D Jung, andseshim such important nucleate boiling heat transfer correlations, complex thermo-fluid interactions are to be considered in bubble microlayers and natural convection at low heat fluxes [1, 2]. For the past few decades, boiling heat transfer experiments have been carried out extensively and a significant progress was made to the understanding of the fundamental processes. Consequently, many nucleate boiling heat transfer correlations were developed based on the experimental data. Most of them, however, have inherent limitations and hence can only be applied to certain combinations of surface and fluid, geometry of the surface, surface roughness, working pressures, heat fluxes, and so on. In 1980, Stephan and Abdelsalam presented a pool boiling correlation based on 5000 data obtained by various researchers with different geometries and fluids via a regression analysis [3]. They developed four separate correlations for water, hydrocarbons, low temperature fluids, and refrigerants and also a single correlation for all fluids considered. The individual correlation for each group showed a mean deviation of per cent, while the single correlation showed a mean deviation of 22.3 per cent for all fluids data. Their correlation has been used in many applications due to its simplicity in use. It, however, remains questionable whether the correlation is valid near the CHF for various fluids. In 1984, Cooper presented a simpler correlation based on extensive data [4]. His correlation was developed based on a thermodynamic corresponding state principle utilizing reduced properties. Cooper fixed the exponent to the heat flux regardless of the fluid and derived a single correlation for various fluids using reduced physical properties representing fluid characteristics [4]. Cooper s correlation has also been widely used in heat exchanger fields due to its simplicity and relatively good accuracy. Like Stephan and Abdelsalam s correlation [3], it is not known whether Cooper s correlation is good at high heat fluxes up to the CHF. In 1984, Gorenflo derived an accurate boiling correlation utilizing the functional relationship between the heat transfer coefficients (HTCs) and reference heat flux [5]. The dataset he used for the correlation, however, was limited to heat fluxes of <100 kw/m 2 and hence the correlation is usually used to predict the HTCs of halocarbon refrigerants. In 2004, Gorenflo et al. measured HTCs of six HFC refrigerants of R32, R125, R134a, R143a, R152a, and R227a and two hydrocarbons of propane and iso-butane in the heat flux range of up to 100 kw/m 2 [6]. They compared the measured data with Gorenflo s correlation and proved its accuracy once again. In 2003, Jung et al. measured pool boiling HTCs of eight halocarbon refrigerants of various vapour pressures (R123, R11, R142b, R134a, R12, R22, R125, and R32) at 7 C in the heat flux range of <100 kw/m 2 [7]. They used the measured data to develop another simple yet accurate pool boiling correlation that is basically based on the combination of Stephan and Abdelsalam s [3] and Cooper s [4] methods and correlations. Later, they also measured pool boiling HTCs of four hydrocarbons of R1270, R290, R600a, R600, and RE170. They also modified their correlations slightly to accommodate hydrocarbons and RE170 data. Thus the generated correlation showed a mean deviation of 5.3 per cent for all refrigerants under the heat flux range of <100 kw/m 2 [8]. Thome [9] and Gorenflo [10] summarized well the past works in this area and interested readers may refer to them. Nowadays, CFCs and HCFCs are being regulated due to their harmful environmental effect on ozone layer depletion and green house warming [11]. Due to the global environmental protection movement, it has become necessary to change the conventional halocarbon refrigerants to more environmentally safe ones. It is known that for refrigerants, Stephan and Abdelsalam s [3] and Cooper s [4] correlations need to be updated since they were derived mainly based on old data of conventional CFCs and HCFCs [8]. Even though Gorenflo s [5] correlation works well for conventional refrigerants at low heat fluxes, its accuracy at high heat fluxes needs to be checked with both old and new refrigerants data. With rapid development of the information technology, electronic components are miniaturized and integrated in large scales. Accordingly, a highly efficient heat removal mechanism needs to be developed for such components as microprocessors and small computers. Considering this trend, the heat removal rate of 300 kw/m 2 is necessary for the next-generation microprocessors and electronic parts to keep the system safe and reliable for long-term use. For these high heat flux applications, conventional single-phase cooling or low heat flux two-phase cooling mechanisms is not adequate. Nucleate boiling heat transfer at high heat fluxes even near the CHF is one of the possible ways to remove that much heat for such applications. It is well known that as the heat flux passes beyond the CHF, the boiling mode is suddenly changed from nucleate to film boiling. This transition usually takes place in such a short time, typically less than a second, that a physical burn out of the heat transfer surface results. This happens due to the impeded heat transfer through the vapour film on the surface. Therefore, for the protection of the system, it is necessary to develop means to increase the CHF. For this, first of all, it is necessary to take nucleate pool boiling data up to the CHF and compare the high heat flux boiling data with conventional correlations. In this study, nucleate boiling heat transfer data are measured for five halocarbon refrigerants of various vapour pressures (R123, R152a, R134a, R22, and R32) up to the CHF and are compared with well-known Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1356 IMechE 2009

3 Nucleate boiling heat transfer coefficients of halogenated refrigerants 1417 correlations with an electronic cooling application in mind. 2 EXPERIMENTS 2.1 Experimental apparatus Figure 1 shows the schematic of the experimental apparatus for nucleate pool boiling heat transfer tests with high-vapour pressure refrigerants. The apparatus is composed of mainly the boiling vessel and external condenser. The hermetically sealed boiling vessel was manufactured with a 170 mm long stainless-steel pipe of 120 mm diameter and flanges at both ends. The vapour generated by the test heater in the vessel was condensed in the external condenser and the condensate was circulated to the bottom of the boiling vessel via gravitation as shown in Fig. 1. The cooling water needed in the condenser was supplied by an independent precision chiller. For low-vapour pressure fluids, sometimes it was difficult to lower the pool temperature to the desired value. Hence, a small copper tube coil was placed in the vessel through which cold water from the chiller was passed for adjusting the pool temperature. Also, a cartridge heater was installed at the bottom of the vessel to adjust the temperature for initial heating. Since the main test heater used in this study was very small, heat loss to the surrounding may cause a significant uncertainty in the measurements. And hence, the boiling vessel, the external condenser, and all connecting pipes were thoroughly insulated with 20 mm thick insulation. 2.2 Heat transfer test section In this study, a small flat plate heat transfer test section was manufactured to measure the nucleate pool boiling data up to the CHF. Figure 2 shows the details of the test section. The test section was composed of a copper plate block and a heater supplying heat to the surface. The heater was a commercial square flat plate uniform heat generating resistor of 20 ohm ( mm, CGI company, model: CCR ) and could generate up to 3800 kw/m 2.A4mm thick square copper plate of the same size as the heater was machined since the heater itself could not be used as the heat transfer surface. At the early stage of the project, the heater and the copper plate block were bonded together using an epoxy of high thermal conductivity (Omega Bond 200, k 1.4 W/m/K). But later, the two were bonded together by a silver solder. As will be seen later, there is a significant deviation between heat transfer data obtained by the two different methods due to a difference in thermal contact resistance. As shown in Fig. 2, four holes (1.0 mm diameter, 5.0 mm length) were machined within the copper plate, 2.0 mm away from the actual heat transfer surface with equal intervals and fine T-type (copper constantan) thermocouples were inserted to these holes to measure directly the surface temperatures. Then, the holes were filled with a silver solder for uniform heating in the copper plate block. For the heat from the heater to be transmitted upwardly to the copper plate block, a plastic insulation block ( mm) was made with a very low thermal conductivity nylon. On the upper portion of the insulation block, a 5.0 mm deep rectangular section of 18.0 mm by 15.0 mm was machined to house the heater and copper plate block assembly. A 15.0 mm long hole of 13.0 mm diameter was machined in one side of the insulation block to accommodate a stainless-steel pipe. Finally, the test heater assembly was put in the insulation block, electrical Fig. 1 Schematic of the pool boiling test facility JMES1356 IMechE 2009 Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science

4 1418 K-J Park, D Jung, andseshim Fig. 2 Details of the test heater wires were carefully connected to the heater, and the wires and four thermocouple wires were led out of the insulation block through the stainless-steel pipe. Thus, these wires were not in direct contact with the working fluids. Finally, another epoxy that does not react with refrigerants was applied to the gap between the heater assembly and stainless-steel pipe and insulating block. The electrical wires were connected to a DC power supply (Agilent model 6030A, 200V, 17A). 2.3 Experimental details Heat transfer performance of the boiling surface tends to degrade over time due to a fouling effect [12]. Therefore, it is important to maintain uniform surface conditions for all tests to generate a reliable dataset for various refrigerants. For this purpose, the surface of the heat transfer section was cleaned with No emery paper and then cleansed with acetone whenever a refrigerant was changed. For a fair comparison, HTCs were measured under the same steady-state condition at the same pool temperature of 7 C for all refrigerants tested in this study, which was accomplished mainly through the use of an external chiller shown in Fig. 1. To measure the pool temperatures and pressure, two T-type thermocouples and a precision pressure transducer were mounted in the liquid and vapour spaces in the vessel, respectively, as shown in Fig. 1. All thermocouples used in this study were T-type and were calibrated against a temperature calibrator of 0.01 C accuracy. On the other hand, the pressure transducer was calibrated against a pressure calibrator of 0.1 kpa accuracy. The power input to the heater was determined with the help of a shunt resistor (the Yokogawa model , 50 mv, 20 A). The experimental procedure for a given refrigerant was as follows. 1. Nitrogen was charged to the refrigerant loop up to 2000 kpa with some halogenated refrigerants to check with a halogen detector if there was any leak. 2. A vacuum pump was turned on for few hours to evacuate the system thoroughly and the refrigerant was charged to the system up to 50 mm higher than the top of the heat transfer block. 3. After 1 h, power to the test heater was initiated and the heat flux was increased to 10 kw/m 2 gradually. The data were taken under steady state at 7 Cfrom 10 kw/m 2 to the CHF. The heat flux was increased with an interval of 10 up to 200 kw/m 2. Beyond 200 kw/m 2, the heat flux was increased with an interval of kw/m 2 up to the CHF. 4. Refrigerant was changed and the same procedures of steps 1 to 3 were repeated after the surface was cleaned as described earlier. Before measurements, the CHFs of the refrigerants used in this study were calculated using Zuber s correlation [13]. During the tests for each refrigerant, much care was exercised as the heat flux approached the theoretically calculated CHF. It was observed that as the heat flux approached the burn out point, the fluid motion was unstable and all of a sudden the surface temperature sharply increased. To prevent the physical burn out from happening, the surface temperature was scanned continuously and the power supply to the heater was shut off as soon as the surface temperature exceeded 60 C. Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1356 IMechE 2009

5 Nucleate boiling heat transfer coefficients of halogenated refrigerants Data reduction In this study, the HTC was determined by h = (q/a) (T s T l ) where h, q, A, T s, and T l are the HTC (W/m 2 /K), power input to the heater (W), heat transfer area (m 2 ), average surface, and liquid temperatures ( C), respectively. As mentioned earlier, the actual surface of the boiling block is 2.0 mm away from the thermocouple holes and, hence, the surface temperature T s in equation (1) needs to be modified from the measured average surface temperature T t by applying a one-dimensional heat conduction equation as follows T s = T t q ( ) L (2) A k (1) where T t, L, and k are the measured average wall temperature by thermocouples ( C), distance from the hole to the surface (m), and thermal conductivity of the test section (W/m/K), respectively. Since the heat transfer block was made of copper, the temperature compensation term in equation (2), (T t T s ), was small at low heat fluxes. However, at high heat fluxes beyond 200 kw/m 2, the term was >1 C and hence was important in HTC determination. The measurement uncertainties were estimated by the method suggested by Kline and McClintock [14] and turned out to be <5.4 per cent at all heat fluxes. In pool boiling heat transfer, the repeatability is very important and hence many measurements were taken repeatedly with an interval of 1 week to 1 month for many fluids to check the repeatability. Overall, the repeatability was always within 5 per cent, which was within the measurement uncertainties. 3 RESULTS AND DISCUSSION In this study, nucleate pool boiling heat transfer measurements were carried out with five halogenated pure refrigerants of various vapour pressures. All data were taken at the pool temperature of 7 C on a small horizontal square copper plate at heat fluxes from 10 kw/m 2 to the CHF. For reference, Table 1 lists some properties of these refrigerants calculated by REFPROP program [15]. 3.1 Effect of thermal contact resistance between the heater and the surface Before presenting the data, it is important to examine the deviation in data caused by the difference in bonding the heater to the surface. As mentioned above, both an epoxy and silver solder were used to bond the heater and copper plate. Figure 3 shows the deviation between the two methods. As seen in Fig. 3, the HTCs obtained from the heating assembly with a silver soldering are consistently per cent higher than those from the one with an epoxy bonding. Of course, the difference was caused by thermal contact resistance. The results indicate that special care must be taken in manufacturing heater specimen for many experiments to simulate actual heating processes. Hence, in this section, only the data obtained from the heating assembly with a silver soldering are presented. 3.2 CHF of refrigerants As mentioned above, nucleate boiling heat transfer data at high heat fluxes are hard to be found for halogenated refrigerants. For electronic cooling, however, high heat flux boiling data are needed for such organic fluids as halocarbon refrigerants. For the pool boiling of pure fluids (especially for water), Zuber s correlation [13] has been utilized extensively over the past few decades q CHF, Zuber = π 24 h fgρ 1/2 g [gσ(ρ f ρ g )] 1/4 (3) Table 2 shows the calculated CHFs by Zuber s correlation and measured ones in this study for all refrigerants tested. As seen in Table 2, Zuber s correlation predicted the CHF quite accurately for such refrigerants of R123, R152a, and R134a that have low and medium vapour pressures as listed in Table 1. For high-vapour pressure fluids such as R22 and R32, Zuber s correlation overpredicted the CHF by per cent. As seen in equation (3), Zuber s correlation is greatly influenced by the saturated vapour density. For high-vapour pressure fluids of R22 and R32, the saturated vapour densities are relatively higher Table 1 Some properties of the tested refrigerants at 7 C p sat ρ f ρ g k f k g µ f µ g h fg σ Refrigerants (kpa) (kg/m 3 ) (kg/m 3 ) (W/m/K) (W/m/K) 10 6 (Pa s) 10 6 (Pa s) (kj/kg) (N/m) Low pressure R Medium pressure R152a R134a High pressure R R JMES1356 IMechE 2009 Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science

6 1420 K-J Park, D Jung, andseshim Fig. 3 Comparison of HTCs between epoxy bonding and silver soldering Fig. 4 Comparison of test results with Jung et al. s data [7] up to 100 kw/m 2 Table 2 Predicted and experimental CHFs for refrigerants at 7 C q CHF, Zuber Experimental Refrigerants (kw/m 2 ) data (kw/m 2 ) R R152a R134a R R than those of R123, R152a, and R134a as shown in Table 1. Accordingly, Zuber s correlation overpredicted the CHF for high-vapour pressure fluids up to 21 per cent. For the past few decades, pool boiling heat transfer experiments at high heat fluxes up to the CHF were carried out mainly for water in power plant applications. Since the vapour pressure of water is very low, Zuber s correlation has worked fine to estimate the CHF of water accurately. Nucleate boiling heat transfer correlations predict the data typically with per cent deviation. Therefore, 20 per cent overprediction of Zuber s correlation for the CHF of high-vapour pressure refrigerants is acceptable. 3.3 Comparison with the data from a horizontal cylinder To check the reliability of the present data, a comparison was made against the previous data measured in our laboratory. In the past, Jung et al. [7, 8] measured pool boiling HTCs of more than 13 halocarbon and hydrocarbon refrigerants at 7 C from a 250-mm-long horizontal copper cylinder of 19 mm outside diameter. A cartridge heater was inserted to the copper cylinder and data were obtained in the heat flux range of kw/m 2. Figure 4 shows the comparison of the present data obtained from a small square copper plate with the previous data obtained from a long copper cylinder in the heat flux range of kw/m 2.As seen in Fig. 4, the present data agree well with the data from the horizontal cylinder with a mean deviation of 5 per cent. This comparison indirectly substantiates the reliability of the present data. At the same time, the results show that boiling HTCs are affected little by the geometry as long as the surface condition is not altered. This shows the importance of the consistent measurements in generating reliable datasets in pool boiling. 3.4 Nucleate boiling HTCs of refrigerants up to the CHF Figure 5 shows the nucleate boiling HTCs of five refrigerants up to the CHF. For a given heat flux, HTCs increased with the vapour pressure. Thus, R32, the highest vapour pressure refrigerant, showed the highest HTCs, whereas, R123, the lowest vapour pressure refrigerant, showed the lowest HTCs among the tested refrigerants. In fact, this is a typical trend in pool boiling at low heat fluxes and is observed in the entire range of heat fluxes for all refrigerants tested in this study. These results show that even at high heat fluxes up to the CHF, HTCs are greatly influenced by the vapour pressure as they are at low heat fluxes, typically observed in normal air-conditioning and refrigeration applications. This also indicates that most of the heat transfer prediction correlations for refrigerants, based Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1356 IMechE 2009

7 Nucleate boiling heat transfer coefficients of halogenated refrigerants 1421 Fig. 5 Nucleate boiling HTCs of refrigerants up to CHF Fig. 7 Comparison of Cooper s correlation [4] against experimental data Fig. 6 Comparison of Stephan and Abdelsalam s correlation [3] against experimental data Fig. 8 Comparison of Gorenflo s correlation [5] against experimental data on the data at low heat fluxes, might work well even at high heat fluxes. 3.5 Comparison with existing correlations After confirming the present data, a comparison is further extended to compare them with existing correlations. Figures 6 to 9 and Table 3 show the comparison of the present data with the Stephan and Abdelsalam [3], Cooper [4], Gorenflo [5], and Jung et al. [8] correlations. Table 4 summarizes these correlations for reference. JMES1356 IMechE 2009 In the heat flux range of <100 kw/m 2, Stephan and Abdelsalam s correlation underpredicted the present data by 22.8 per cent, whereas Cooper s correlation overpredicted them with 19.4 per cent deviation. On the other hand, Gorenflo s and Jung et al. s correlations predicted the present data very well, showing mean deviations of 5.4 and 5.1 per cent, respectively. In the entire heat flux range up to the CHF, a similar trend was observed. Stephan and Abdelsalam s correlation underpredicted the present data by 21.3 per cent, whereas Cooper s correlation overpredicted them with 14.2 per cent deviation. On the other hand, Gorenflo s Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science

8 1422 K-J Park, D Jung, andseshim Fig. 9 Comparison of Jung et al. s correlation [8] against experimental data and Jung et al. s correlations showed mean deviations of 5.8 and 6.4 per cent, respectively. Therefore, Stephan and Abdelsalam s and Cooper s correlations have been used extensively in the prediction of nucleate boiling HTCs of many fluids. Their correlations, however, show somewhat greater deviations due partly to the old database they used to develop those correlations. Also, these correlations are not accurate for halogenated refrigerants at high heat fluxes up to the CHF where electronic cooling is applied. Even though Gorenflo s and Jung et al. s correlations were developed based on similar conventional refrigerants data, their correlations predicted the present data up to the CHF with good accuracy. One of the main reasons for this difference among the existing correlations is the way the exponent of the heat flux term is expressed, which is present in most of the pool boiling correlations. As seen in Table 4, for Stephan and Abdelsalam s and Cooper s correlations, the exponents of the heat flux term are fixed regardless of the working fluids. Accordingly, the effect of various Table 3 Deviations of various correlations against the present data Stephan and Abdelsalam [3] Cooper [4] Gorenflo [5] Jung et al. [8] Range Refrigerants Average Mean Average Mean Average Mean Average Mean kw/m 2 R R152a R134a R R All All R R152a R134a R R All Average deviation = 1 n [ ] (hcal h exp ) 100 Mean deviation = 1 n [ ] (hcal h exp ) 100 ABS n h exp n h exp 1 1 Table 4 Some well-known correlations for predicting nucleate boiling HTCs Author Stephan and Abdelsalam [3], 1980 Equation h = 207 k [ ] ( ) ( ) f (q/a)db ρg νf where D b = β[2σ /g(ρ f ρ g )] 0.5, D b k f T sat ρ f α f β = 35 (contact angle) ( q ) 0.67 Cooper [4], 1984 h = 90 M 0.5 pr n A ( log 10 p r) 0.55 where n = log 10 Rp, Rp (surface roughness (µm)) Gorenflo [5], 1984 Jung et al. [8], 2004 ( ) h q n(pr) ( ) Ra 2/15 = F(p r ) where h = h 0 for p h 0 q 0 Ra = 0.1, q r0 0 = 20 kw/m 2, Ra 0 = 0.4 µm, 0 copper n(p r ) = p 0.3 r, F(p r ) = 1.2p 0.27 r + h = 41.4 k f D b [ (q/a)db k f T sat ( ) p r 1 p r ] C ( ( log p 10 r) ρ ) 0.53 g where C = (1 p r ) 1.33 ρ f Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1356 IMechE 2009

9 Nucleate boiling heat transfer coefficients of halogenated refrigerants 1423 physical properties associated with the heat flux is not taken into consideration in their rigid form of correlations. On the other hand, in Gorenflo s and Jung et al. s correlations, that effect is taken into account by correlating the exponent using reduced pressure as a variable. Due to this basic difference, Gorenflo s and Jung et al. s correlations can accurately predict the nucleate boiling data in the entire heat flux range up to the CHF. In 2004, Gorenflo et al. [6] made an excellent analysis on seven well-known pool boiling correlations including four correlations discussed in this study. The interested reader is referred to their excellent work for details. The present results show that in the design of electronic cooling devices at high heat fluxes, Gorenflo s and Jung et al s correlations are good for predicting the HTCs of halocarbon refrigerants in the entire heat flux range up to the CHF. 4 CONCLUSIONS In this study, the nucleate boiling HTCs of five pure halogenated refrigerants of R123, R152a, R134a, R22, and R32 were measured at the liquid pool temperatureof7 C. An experimental apparatus was designed and manufactured by which the boiling data can be obtained in the entire heat flux range up to the CHF. The test section was made of a small square copper plate of mm and data were taken from 10 kw/m 2 to the CHF in the increasing order of heat flux. Based on the test results and comparison with existing correlations, the following conclusions can be drawn. 1. For halocarbon refrigerants, the measured HTCs increased with both heat flux and vapour pressure as observed in the published data at low heat fluxes of <100 kw/m 2. As the heat flux increased up to the CHF, the same trend continued. 2. Zuber s correlation [13] predicted the CHF of all fluids tested within a deviation of 21 per cent. Especially, the deviation increased as the vapour pressure increased. 3. In the entire heat flux range up to the CHF, Stephan and Abdelsalam s correlation [3] underpredicted the present data by 21.3 per cent, whereas Cooper s correlation [4] overpredicted them with 14.2 per cent deviation. On the other hand, Gorenflo s [5] and Jung et al. s correlations [8] showed mean deviations of 5.8 and 6.4 per cent, respectively. 4. The method of bonding the heater and test surface together had a great impact on thermal contact resistance. Test results showed that metallic bonding was far better than the epoxy bonding. 5. The pool boiling HTCs are less affected by the geometry as long as the surface condition, especially the roughness, was carefully maintained at the same level. JMES1356 IMechE 2009 ACKNOWLEDGEMENT This work was supported by the Korea Science and Engineering Foundation grant funded by the Korean government (MOST) (No. R ). REFERENCES 1 Hsu, Y. Y. and Graham, R. W. Transport processes in boiling and two-phase system, 1976 (Hemisphere Publishing Company, Washington, DC). 2 Van Stralen, S. J. D. The growth rate of vapor bubbles in superheated pure liquids and binary mixtures. Int. J. Heat Mass Transf., 1968, 11, Stephan, K. and Abdelsalam, M. Heat transfer correlations for natural convection boiling. Int. J. Heat Mass Transf., 1980, 23, Cooper, M. G. Correlations for nucleate boiling formulation using reduced properties. Physicochem. Hydrodyn., 1982, 3, Gorenflo, D. Pool boiling. In VDI heat atlas, 1984, ch. Ha (VDI-Verlag, Dusseldorf). 6 Gorenflo, D., Chandra, U., Kottoff, S., and Luke, A. Influence of thermophysical properties on pool boiling heat transfer of refrigerants. Int. J. Refrig., 2004, 27, Jung, D., Kim, Y., Ko, Y., and Song, K. Nucleate boiling heat transfer coefficients of pure halogenated refrigerants. Int. J. Refrig., 2003, 26, Jung, D., Lee, H., Bae, D., and Oho, S. Nucleate boiling heat transfer coefficients of flammable refrigerants. Int. J. Refrig., 2004, 27, Thome, J. R. Boiling of new refrigerants: a state-of-the-art review. Int. J. Refrig., 1996, 19, Gorenflo, D. State of the art pool boiling heat transfer of new refrigerants. Int. J. Refrig., 2001, 24, Molina, M. J. and Rowland, F. S. Stratospheric sink for chlorofluoromethanes: chlorine atom catalyzed destruction of ozones. Nature, 1974, 249, Webb, R. L. Principles of enhanced heat transfer, 1994 (John Wiley & Sons. Inc., New York). 13 Zuber, N. Hydrodynamic aspects of boiling heat transfer. AEC report no. AECU-4439, Physics and Mathematics, Kline, S. J. and McClintock, F. A. Describing uncertainties in single-sample experiments. Mech. Eng., 1953, 75, Lemmon, E. W., Huber, M. L., and McLinden, M. O. NIST reference fluid thermodynamics and transport properties. REFPROP version 8.0, 2007 (National Institute of Standards and Technology, Gaithersburg, Maryland, USA). APPENDIX Notation A heat transfer area (m 2 ) C constant or exponent CHF critical heat flux D diameter (m) Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science

10 1424 K-J Park, D Jung, andseshim g gravitational acceleration (m/s 2 ) h heat transfer coefficient (W/m 2 /K) h fg heat of evaporation (kj/kg) k thermal conductivity (W/m/K) L test section thickness (m) M molecular weight (kg/kmol) n exponent p pressure (kpa) q heat transfer rate (W) q heat flux (kw/m 2 ) Ra standardized roughness parameters Rp surface roughness (µm) T temperature (K or C) α thermal diffusivity (m 2 /s) β contact angle ( ) µ viscosity (Pa s) ν kinematic viscosity (m 2 /s) ρ density (kg/m 3 ) σ surface tension (N/m) Subscripts 0 reference state for Gorenflo s correlation b bubble cal calculated values ep epoxy exp experimental values f saturated liquid g saturated vapour l liquid r reduced property s surface of test section sat saturation ss silver solder t thermocouple Proc. IMechE Vol. 223 Part C: J. Mechanical Engineering Science JMES1356 IMechE 2009