A STUDY OF IN-TUBE EVAPORATION HEAT TRANSFER OF CARBON DIOXIDE

A STUDY OF IN-TUBE EVAPORATION HEAT TRANSFER OF CARBON DIOXIDE Nitin N. Saant *, Min Soo Kim **, W. Vance Payne *, Piotr A. Domanski * and Yun Wook Hang ** * NIST MS 8631, Gaithersburg, MD USA 0899 Phone: 301-975-6663 Fax: 301-975-8973 Email: vance.payne@nist.gov ** School of Mechanical and Aerospace Engineering Seoul National University, Seoul 151-74, KOREA ABSTRACT The evaporative heat transfer characteristics of carbon dioxide (CO ) ere measured in a fluid-heated, 8.0 mm ID smooth stainle steel tube. The test apparatus as a concentric tube heat exchanger heated by ater in the annulus ith CO in the central tube. Data ere collected for temperatures of 5 o C and 10 o C; ma fluxes of 50 kg/m s, 500 kg/m s, and 650 kg/m s; and heat fluxes of 4 kw/m to 58 kw/m. The average heat transfer coefficient as determined ith the overall UA equation for the annular configuration. All properties ere calculated using REFPROP 7.0 (Lemmon et. al. 00). Results are presented and compared to existing correlations for the to-phase average evaporative heat transfer coefficient. INTRODUCTION Environmental concerns and regulations in the last to decades have generated the need for extensive research to identify viable and environmentally friendly refrigerant alternatives. As a result, a significant focus has been placed on so-called natural refrigerants (e.g., CO, ammonia, hydrocarbons, air, ater, etc.). Carbon dioxide offers benefits such as non-toxicity, non-flammability, easy availability, lo price, no need of recycling, and compactne of components. It has favorable transport properties (lo viscosity and high thermal conductivity), hich combine to improve the heat transfer characteristics. The past several years have noticed an active research on evaporative heat transfer studies related to CO. Knudsen and Jansen (1997) concluded that the predicted heat transfer coefficient of CO is approximately half of the coefficient predicted by the correlation of Shah (198). In their study on CO in microchannel tubes, Pettersen et al. (000) predicted a strongly decreasing heat transfer coefficient from a certain vapor fraction upards. Pettersen looked at six existing correlations and all of them overpredicted the measured values. Zhao et al. (000) concluded that the boiling heat transfer coefficients of CO increased ith increasing heat flux and/or ma flux. Zhao recommended Gungor and Winterton s correlation due to its mean deviation of le than 10 % from his data. Olson and Allen (1998) measured the heat transfer coefficient in heated, turbulent, supercritical CO floing in a horizontal tube. The supercritical fluid had enhanced heat transfer at lo heat flux and degraded heat transfer at high heat flux. In this study, the average evaporative heat transfer characteristics of to-phase CO are measured in a concentric, ater heated test section. The experimental results are compared to predictions from six published correlations. 1. EXPERIMENTAL SET UP Figure 1a shos the schematic of the test apparatus. The experimental setup consisted of an evaporator (test section), condenser/subcooler, saturator, preheater, and magnetic gear pump. In the concentric type test section, the inner tube as stainle steel (type 304) ith an inner diameter of 8.0 mm and an outer diameter of 9.5 mm. The outer tube as copper ith an inner and outer diameter of 1.5 mm and 15.9 mm respectively. This arrangement produced an annulus hydraulic diameter of 3.0 mm. Carbon dioxide floed in the inner tube and ater, hich heated the CO, floed in the International Congre of Refrigeration 003, Washington, D.C. 1

annulus. The condenser/subcooler and the preheater ere controlled by a secondary heat transfer fluid (HTF) loop. This HTF as a mixture of ethylene-glycol/ater (30 %/70 % by ma). Spacers 1 1 11 6 8 9 10 3 4 5 7 T: Thermocouple P: Preure transducer 1: Main test subsection : Preheater 3: Condenser 4. Subcooler 5. Saturator 6: Coriolis flo meter 7: Magnetic gear pump 8: CO cylinder 9: Syringe pump 10: Vacuum pump 11: Heating element 1: R- cooling system Thermopiles Figure 1b: Schematic of subsection Figure 1a: Experimental test facility The evaporator (test section) as made up of five, ell insulated subsections. Carbon dioxide as charged into the system ith a 500 ml syringe pump, hich also controlled the operating preure. A variable-speed motor on the magnetic gear pump controlled the flo rate of CO. The heating as controlled ith the help of a solid-state controlled rectifier (SCR). All data presented in this paper ere taken in subsection 1 because the combined uncertainty for the to-phase evaporative heat transfer coefficient as smaller in subsection 1 than in the other subsections. The uncertainty analysis is given in section 3. 1.1 Instrumentation Each subsection as instrumented ith preure transducers on the refrigerant side and a ten-junction thermopile on the annulus (ater) side. The thermocouple at the outlet of each CO subsection also acted as the inlet thermocouple for the next subsection. Measurements included the ma flo rate and inlet and outlet temperature for CO and ater at each subsection as ell as the preheater and condenser/subcooler. In addition, the CO inlet preure and the CO differential preure ere measured. The temperature of CO as measured ith type-t thermocouples. A ten-junction thermopile as placed at the entrance and exit of each subsection in a U-bend to measure the annulus temperature change, as shon in Figure 1b. The ma flo rate of CO and ater for the test section as measured using coriolis ma flo meters, and that for the ethylene-glycol/ater mixture in the preheater as measured ith a turbine meter calibrated ith a coriolis meter. The flo meters, preure transducers, thermocouples, thermopiles, and aociated instrumentation ere calibrated. Table 1 shos the range and 95 % relative uncertainty of the installed instrumentation. International Congre of Refrigeration 003, Washington, D.C.

Table 1: Instrumentation range and uncertainty Instrument Range 95 % Relative Uncertainty T-type thermocouples ( C) 0 to 100 0.0 10 Junction Thermopiles ( C) 0.5 to 5 0.1 Evaporator Preure Transducer (kpa absolute) 0 to 0684 1.0 % of full scale Differential Preure Transducer (kpa) 0 to 17 1.0 % of full scale Evaporator Water Ma Flo in Annulus (kg/h) 0 to 544.1 % of full scale CO Ma Flo (kg/h} 0 to 544.1 % of full scale. EXPERIMENTAL AND ANALYTICAL STUDY.1 Annulus, Water-Side, Single-Phase Tests Experiments ere first conducted to establish the average single-phase convective heat transfer coefficient for ater in the annulus. For this purpose, single-phase HFC-134a as used in the inner tube since its heat transfer characteristics ere ell knon and an abundance of experimental data ere available. The energy balance beteen ater and HFC- 134a did not exceed 1.0 % for all tests considered. The Gnielinski (1976) correlation for single-phase turbulent flo inside a smooth tube as used to calculate the average convective heat transfer coefficient of HFC-134a. The pertinent smooth tube, single-phase equations are presented in Equation 1. The calculated single-phase heat transfer coefficient for HFC-134a as used in the UA equation to solve for the ater-side average single-phase convective heat transfer coefficient. Temperature ( o C) 60 50 40 30 0 10 HFC-134a Water [ 0.790 ( ln( Re) 1.64) ] 1 f =, f ( Re 1000) Pr Nu = 8 f + 3 1.07 1.7 Pr 1 8 (1) The single-phase tests had HFC-134a ma fluxes in the range of 1000 to 1545 kg/(m s) and heat fluxes in the range of 4 to 59 kw/m (based on the 8 mm tube inner diameter). The inlet hot ater temperature, the HFC-134a flo rate, and the ater flo rate ere the three major independent parameters, hich alloed for establishing the conditions for these single-phase experiments. 0 0 1 3 4 5 Section Figure : Temperature profiles for a single-phase test ith HFC-134a ith parallel flo configuration Single-phase experiments ere carried out ith a parallel-flo configuration. Figure shos a typical temperature profile for single-phase parallel flo tests ith HFC-134a. The number of the subsection represents the X-axis, 0-1 being the first subsection, 1- the second subsection, etc. The overall conductance equation, hich is a sum of individual thermal resistances of HFC-134a, ater, and tube all separating the to fluids, is used to determine the aterside heat transfer coefficient (h ). International Congre of Refrigeration 003, Washington, D.C. 3

1 UA ln D r D LMTD 1 i 1 = = + + () Q h πd L πk L h πd L r i h The only unknon in Equation () is h. The Nuelt number on the aterside is then represented by the folloing Dittus-Boelter (1930) type equation. Equation (3) is regreed using the calculated values of h to find the values of constants C and a as 0.00599 and 0.91, respectively. Equation (3) represents the final correlation for ater-side singlephase average Nuelt number. hdh a 1/ 3 Nu. Re Pr (3) = = C k Figure 3 shos a plot of percent difference beteen the experimental and calculated values for Nuelt number against the Reynolds number of ater for the first subsection of the evaporator tube. Water Nuelt Number Percent Error (Calc-Meas)/Meas *100 15 10 5 0-5 -10-15 5000 7000 9000 11000 13000 Water Annulus Reynolds Number Figure 3: Comparison of experimental and calculated Nuelt numbers for single-phase tests ith HFC-. Refrigerant-Side,To-Phase Tests ith CO To-phase CO tests ere conducted in a similar manner as HFC-134a single-phase tests and covered the same ranges of ma and heat fluxes of CO. Carbon dioxide of 99.99 % purity as used during the tests. An energy balance on the preheater section of the test rig alloed the calculation of the test section entrance ma quality. Subcooled CO entered the preheater and as heated until it as to-phase at the preheater exit. Preure and temperature at the inlet and exit of the test section alloed the saturation temperature of the CO to be verified. Using thermopile temperature change, inlet/exit ater temperatures, ater ma flo rate, and h, Equation () as solved for the refrigerant-side heat transfer coefficient, h rφ. 3. COMPARISON TO EXISTING CORRELATIONS Chen (1966) developed a to-phase flo boiling correlation here the heat transfer coefficient as divided into a nucleate pool boiling coefficient and a bulk convective coefficient. Different correlations for the to-phase enhancement factor ere selected by using a form of the Lockhart-Martinelli turbulent-turbulent to-phase multiplier (Χ tt ). Bennet and Chen (1980) modified the Chen correlation using a larger data set. They implemented a ne correlation for the nucleate pool boiling suppreion factor. The to-phase enhancement factor as still a function of the Lockhart-Martinelli to-phase multiplier. Shah (198) examined data from many sources to develop a set of equations to determine the average convective heat transfer coefficient for to-phase flo. Shah described the boiling regimes in terms of nucleate boiling regions, convective boiling regions, and bubble suppreion regions. He utilized a convection number (Co, very similar to Χ tt ), a boiling number (Bo), and a liquid Froude number (Fr l ) in his correlation. Gungor and Winterton (1986) developed a correlation using a nucleate pool boiling suppreion factor and a bulk convective enhancement factor in a manner similar to Chen. Their expreion for the pool boiling convective heat transfer coefficient included reduced preure, refrigerant molecular eight, and heat flux. Liu and Winterton (1991) International Congre of Refrigeration 003, Washington, D.C. 4

modified the Gungor and Winterton (1986) correlation by changing the expreions used to calculate the bulk convective enhancement factor and the nucleate pool boiling suppreion factor. They also determined the final overall convective heat transfer coefficient by adding the convective and pool boiling contributions in quadrature. Radermacher and Hang (1997) modified the Bennett and Chen (1980) to correlate the data of Bredesen et al. (1997). The Bredsen et al. data as taken only at -10 C ith ma flux beteen 00 and 400 kg/m and heat flux beteen 3 and 9 kw/m. Figure 4 shos the percent difference beteen the predicted and measured heat transfer coefficient values for carbon dioxide flo boiling inside the smooth tube. The Gungor and Winterton (1986) correlation predicts 48 % of the data to ithin ± 50 % of the measured value ith 97 % of the data underpredicted. The Bennett and Chen (1980) correlation predicts 47 % of the data to ithin ± 50 % of the measured value ith 69 % of the data underpredicted. The Radermacher and Hang (1997) correlation predicts 4 % of the data to ithin ± 50 % of the ith 34 % of the data underpredicted. The remaining correlations (Shah, Liu and Winterton, and Chen) presented in Figure 4 predict le than 40 % of the data to ithin ± 50 % of the measured value. (Predicted - Measured)/Measured * 100 Measured To-Phase Tube-Side Heat Transfer Coefficient, [W/(m K)] 0 5000 10000 15000 0000 5000 30000 35000 40000 45000 50000 55000 60000 1100 1050 1000 950 900 Bennett-Chen 850 800 Chen 750 Gungor-Winterton 700 650 Shah 600 550 Liu-Winterton 500 450 Radermacher-Hang 400 350 300 50 00 150 100 50 0-50 -100-150 Figure 4: Percent difference in measured and predicted average to-phase convective heat transfer coefficient for CO International Congre of Refrigeration 003, Washington, D.C. 5

4. UNCERTAINTY ANALYSIS The uncertainty on the ater-side average convective heat transfer coefficient as calculated on the basis of measured uncertainties of temperature, preure and ma flo rates. From Eq. (1) e understand: h = f ( hr, k, Q, Dr, Di, L, LMTD) (4) The uncertainty in h may be found by folloing the propagation of errors ithin Eq. () by summing the error (taken as 95 % confidence interval) of all the individual terms as shon belo in Eq. (5). E f f + + + h = Eh + Ek EQ ED ED + EL + ELMTD r ri h r k Q D D r L LMTD The uncertainties on other parameters such as heat flux, ma flux, heat transfer coefficient of CO, etc. can be determined in a similar ay. Table lists measured and calculated quantities and their relative uncertainties for the 95 % confidence limits. Table : Single-phase ater-side propagation of error and relative uncertainty in h (95 % confidence interval) Parameter Range Uncertainty (%) Single-phase HFC-134a, average convective heat transfer coefficient from Gnielinski (1976), Equation (3), h (W/m 000 to 900 10.3 % K) Water-side heat transfer rate, Q (W) 1700 to 500 5.6 % to 7.5 % Log-mean temperature difference beteen ater and single-phase HFC-134a, LMTD ( C) 5.3 to 9.7 0.75 % to 0.89 % Water-side average convective heat transfer coefficient, h 3800 to 6100 18 % to 31 % With h knon from the single-phase tests ith HFC-134a, CO as circulated through the test section at various inlet qualities and heat fluxes. The UA equation (Eq. ()) as solved for h r and used to determine the propagation of error for the refrigerant-side average convective heat transfer coefficient. The resulting form of the equation is shon by Equation (6). hr φ = f ( h, k, Q, Dr, D, L, LMTD) (6) The propagation of error as calculated as in Equation (5) ith h replaced by h rφ. Table 3 summarizes the neceary uncertainty terms and the final calculation of the uncertainty of h rφ. Table 3: To-phase CO average convective heat transfer coefficient relative uncertainty (95 % confidence interval) Parameter Uncertainty Range (%) Single-phase ater-side average heat transfer coefficient, h (W/m K) 3986 to 7 8 to 50 Water-side heat transfer rate, Q (W) 1000 to 400 5 to 9 Log-mean temperature difference beteen ater and CO, LMTD ( C) 6 to 18 1.0 to 3.5 To-phase CO, average convective heat transfer coefficient, h rφ (W/m K) 500 to 54400 14 to 55 (5) 4. CONCLUSIONS This investigation examined 71 data points for evaporative flo boiling of CO in a fluid-heated, 8 mm ID smooth tube. Six existing correlations ere examined for their fit to the data collected. The Gungor and Winterton (1986) correlation as the best predictor of the measured values, predicting 48 % of the data to ithin ± 50 % of the measured value. International Congre of Refrigeration 003, Washington, D.C. 6

Belo a heat transfer coefficient value of 5000 W/m K, Radermacher and Hang (1997), Bennett and Chen (1980) and Chen (1966) correlations significantly overpredicted the measured values. At higher heat transfer coefficient values all correlations underpredicted the measured value by as much as 90 %. NOMENCLATURE A Heat transfer area, m Bo Boiling number, q C Co D E Fr G h i k L LMTD Nu Pr Constant Gi fg Convection number, 1 x x Diameter, m Uncertainty Froude number, G ρ g f i fg 0.8 ρ g ρ f Ma flux, kg/m s Heat transfer coefficient, W/m K Enthalpy, J/kg Thermal conductivity, W/m K Length, m Logarithmic Mean Temperature Difference, C Nuelt number Prandtl number 0.5 Q Heat Transfer Rate, W Re Reynolds number, G D µ T Temperature, o C U Overall heat transfer coeff., W/m K Xtt Lockhart Martinelli to-phase multiplier Subscripts f fluid or liquid g vapor or gas fg gas minus fluid property h hydraulic i inner/inlet o outer/outlet r refrigerant/root stainle steel ater rφ to-phase refrigerant ACKNOWLEDGEMENTS This study as sponsored by the US. Department of Energy, Contract Number DE-AI01-97EE3775 Esher Keller manager, and NIST. The authors acknoledge Drs. M. A. Kedzierski and J. S. Bron for their advice. REFERENCES Bennett, D. L., and Chen, J. C., 1980, "Forced convective boiling in vertical tubes for saturated pure components and binary mixtures", AIChE Journal, Vol. 6, pp. 454-461. Bredesen, A. M. et al., 1997, Studies in CO heat exchangers and heat transfer, Proceedings of IIR Heat Pump Centre / IIR Workshop on CO Technology in Refrigeration, Heat Pump and Air-Conditioning Systems, Trondheim, Noray. Chen, J. C., 1966, "A correlation for boiling heat transfer to saturated fluids in vertical flo", I & EC Proce Design and Development, Vol. 5. No. 3, pp. 3-339. Dittus, F. W., and Boelter, L. M. K., 1930, Heat transfer in automobile radiators of the tubular type, University of California Publications of Engineering, USA, Vol., pp. 443 461. Gnielinski, V., 1976, Ne equations for heat and ma transfer in turbulent pipe and channel flo, International Chemical Engineering, Vol. 16, No., pp. 359 368. International Congre of Refrigeration 003, Washington, D.C. 7

Gungor, K. E., and Winterton, R. H. S., 1986, "A general correlation for flo boiling in tubes and annuli", Int. J. Heat Ma Transfer, Vol. 9, No. 3, pp. 351-358. Knudsen, H. J., and Jensen, P. H., 1997, Heat transfer coefficient for boiling carbon dioxide, Workshop Proceedings, CO Technology in Refrigeration, Heat Pump and Air Conditioning Systems, Trondheim, Noray, pp. 319 38. Liu, Z., and Winterton, R. H. S., 1991, "A general correlation for saturated and subcooled flo boiling in tubes and annuli, based on a nucleate pool boiling equation", Int. J. Heat Ma Transfer, Vol. 34, No. 11, pp. 759-766. Lemmon, E.W., McLinden, M.O. and Huber, M. L., 00. NIST reference fluid thermodynamic and transport properties REFPROP (Version 7.0), National Institute of Standards and Technology, Gaithersburg, Maryland. Olson, D., and Allen, D., 1998, Heat transfer in turbulent supercritical carbon dioxide floing in a heated horizontal tube, NISTIR 634. Pettersen, J., Rieberer, R., and Munkejord, S. T., 000, Heat transfer and preure drop characteristics of evaporating carbon dioxide in microchannel tubes, Proceedings of 4 th IIR-Gustav Lorentzen Conference on Natural Working Fluids, Purdue University, Indiana, USA, pp. 107-114. Shah, M. M., 198, "Chart correlation for saturated boiling heat transfer: equations and further studies", ASHRAE Transactions, Vol. 88, No. 1, pp. 185-196. Zhao, Y., Ohadi, M. M., Deiatoun, S. V., Molki, M., and Darabi, J., 000, Forced convection boiling heat transfer of CO in horizontal tubes, Proceedings of 5th ASME/JSME Joint Thermal Engineering Conference, San Diego, California, USA. International Congre of Refrigeration 003, Washington, D.C. 8