ASHRAE IJHVAC 9-1-2003 HVAC&R Research《《HVAC&R研究》第9卷 1号 2003年1月》.pdf

1、VOL. 9, NO. 1 HVAC related research should find a home in thisHVAC we cannot sit idly as technologies pass us by.RadermacherED.fm Page 1 Monday, December 16, 2002 10:33 AM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC the ot

2、her two nonintegralfinned tubes and the 1024 fins/m (26 fins/in.) tube performed at 80, 72, and 60% of that of thehighest one. The 3D tube used in the present study is of a similar design (saw-toothed) to thetube used by Webb and Murawski, but significant differences do exist.Experiments on single-t

3、ube condensation in the literature (e.g., Cavallini et al. 1996; Chengand Tao 1994; Cheng et al. 1996; Cheng et al. 1997) reported single-tube condensation heattransfer coefficients on enhanced tubes but none directly with R-134a. Chang et al. (1996)reported results for R-134a on four different enha

4、nced tubes, which included 1024 and 1575fins/m (26 and 40 fins/in.) low-fin tubes and two 3D fin tubes.EXPERIMENTAL APPARATUSThe experimental test facility used in this study was designed to measure shell-side condensa-tion heat transfer coefficients on a horizontal single tube. The major systems of

5、 the test facilityhoB,akfDo-4mrfLN-b g fg()L2-cCpf,fkf-d=hNaReLn=Eckels406.fm Page 6 Friday, December 13, 2002 3:21 PM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC for the saturated pressure of refrigerant, it was less than

6、 2.1 kPa (0.3 psi). For thewater flow rate, the permissible range was 0.5%. Once steady state was achieved, 10 data scanswere taken over the course of 2 min. After anomaly inspection, the average values of the 10scans were used to compute the shell-side heat transfer coefficients.DATA ANALYSISModifi

7、ed Wilson PlotA modified Wilson plot technique (Briggs and Young 1969) was used to determine thein-tube heat transfer correlation. The procedure applies a convection correlation to both theTable 1. Test Tube Geometry SpecificationsTubeFinCount,fins/mDoNominal, mm (in.)DiNominal,mm (in.)DrNominal,mm

8、(in.)FinHeight,mm (in.)AefNominal,m2/m (in2/in)AiNominal,m2/m (in2/in)Smooth 19.1 (0.752) 15.9 (0.626) 19.1 (0.752) 0.04995 (1.97)2D 1575 19.1 (0.752) 14.46 (0.569) 15.9 (0.626) 1.45 (0.057) 0.263 (10.35) 0.0454 (1.79)3D 19.1 (0.752) 15.54 (0.612) 17.07 (0.672) 0.91 (0.036) 0.0488 (1.92)Eckels406.fm

9、 Page 8 Friday, December 13, 2002 3:21 PM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC the 3D tube exhibits the highest heat transferperformance among all the three test tubes, 11.8 times higher than that of the smooth tube

10、 onaverage. It is also interesting to compare the results of the 2D and 3D tubes. The heat transferbenefit of the 3D tube decreases as LMTD increases, if the lowest heat flux is ignored becauseof its high uncertainty. The structures of the fins on the 2D and 3D tube are similar except thatlongitudin

11、al slots are cut in the upper half of the 3D tube fin. This creates a series of pin-typefins around the circumference of the tube. At higher loads, the efficiency of the pin fins mayplay a limiting role, thus offsetting the surface area advantage of the 3D tube.The overall heat transfer coefficients

12、 Uoof all three test tubes as a function of LMTD at a satu-ration temperature of 40C (104F) are presented in Figure 6. For the smooth tube, Uodoes notdrop drastically with increased LMTD despite the relatively sharp drop in howith increased load,as shown in Figure 2. As the heat load of the test sec

13、tion increased, the water flow rate was alsoincreased, which caused a higher water-side heat transfer coefficient, partially offsetting the effectof the drop in ho. For the 2D tube, the overall heat transfer coefficient increased from 4523 to 6740W/(m2K) (796 to 1187 Btu/hft2F) over the test range;

14、for the 3D tube, the overall heat transfercoefficient increased from 5111 to 8835 W/(m2K) (900 to 1556 Btu/hft2F) over its test range.This is obviously different from the phenomenon observed in the smooth tube test. The dominantresistance to heat transfer lies on the water side for the 2D and 3D tub

15、es; thus, increasing thewater-side heat transfer coefficient has a direct effect on Uo. The higher overall performance for theFigure 5. Comparison of Shell-Side Heat Transfer Coefficients as Function of Log-Mean Temperature Difference for All Three Test Tubes at Tsat= 40CEckels406.fm Page 13 Friday,

16、 December 13, 2002 3:21 PM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVACthe measured value of hoexhibits a drop of 34%. The Nusselt correlation overpredicts theshell-side heat transfer coefficient by up to 15% at the heat fl

17、ux of 3155 W/m2(1000 Btu/hft2)and tends to become less as heat flux increases. The average difference between the Nusselt cor-relation and the experimental data is only 5.2%. The Honda model also overpredicts the data, withan average deviation of 21%. The Honda model was primarily developed for tube

18、 bundles, so itis not surprising that it overpredicts the single-tube data with no vapor shear.Figure 6. Comparison of Overall Heat Transfer Coefficients as Function of Log-Mean Temperature Difference for All Three Test Tubes at Tsat= 40CDeviationhexphmodelhexp- 1001nn-=Eckels406.fm Page 14 Friday,

19、December 13, 2002 3:21 PM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVACsize, size distribution, and geometrical characteristics affect the physical properties and practicalapplications of the ice slurry, it is difficult to p

20、rescribe generally valid models and calculationprocedures for designing ice slurry systems.Geometrical characteristics can result from the method of ice generation or from the ice crys-tal history of partial melting and recrystallization. The pressure drop of freshly produced iceslurry was found to

21、be up to 60% higher compared to the same ice slurry after five cycles of par-tial melting and regeneration (Frei and Egolf 2000). Recent findings at the Danish Technologi-cal Institute (DTI) by Nrgrd et al. (2001) have revealed an influence of ice crystal size on heatexchanger performance. At the ou

22、tlet of heat exchangers, the carrier fluid is superheated (i.e.,has a temperature above the freezing point of the mixture), but ice crystals are still present in theTorben M. Hansen is project manager at the Danish Technological Institute, Marija Radosevic is a graduate student atthe University of B

23、elgrade, Michael Kauffeld is project manager at the Danish Technological Institute, and ThomasZwieg is a graduate student at the University of Dresden.Hansen456.fm Page 19 Monday, December 16, 2002 8:23 AM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ash

24、rae.org). Published in HVAC over the 12 to 24 hstorage periods studied, reduced agglomeration was found with low concentration of ethyleneglycol as additive. Antifreeze proteins found in arctic animals are known to prevent crystal growth. Japanesestudies have found that these proteins also prevent r

25、ecrystallization in ice slurry. Other, artificialsubstances (e.g., silicon monomer) have a similar effect and are more stable in ice slurry systems(Inada et al. 2000). The substances create an energy barrier on the particle surface and inhibitcrystal growth. The different additives have a crystal-gr

26、owth-inhibiting effect with both hydro-phobic and hydrophilic ends (Matsumoto et al. 2000).Knowledge of ice slurry geometrical characteristics and time behavior is believed to beimportant for understanding the deviations in results of numerous comparative experimentalinvestigations of thermophysical

27、 properties, pressure drop, heat transfer coefficients, etc. Under-standing crystal size range and growth rate can be useful in designing thermal storage systemsfor maximum ice storage capacity and obtainable meltoff rates, and for developing componentssuch as pumps, control valves, heat exchangers,

28、 and ice concentrators for use with ice slurry.THEORETICAL CONSIDERATIONSThe crystallization mechanisms in ice slurry systems can be divided into three governingdomains:Crystal nucleation (initial formation of ice embryos in the ice generator); includes homoge-neous and heterogeneous nucleationCryst

29、al growth during generation (continued growth of nuclei larger than the criticalradius); includes normal growth, surface nucleation, and growth from imperfectionsGrowth of ice crystals in storage systems (thermal equilibrium)In this investigation, only growth in storage systems is observed. Suspensi

30、on in storage isassumed to be in thermal equilibrium; therefore, thermal and additive concentration boundarylayers differ from those in the other domains.Hansen456.fm Page 20 Monday, December 16, 2002 8:23 AM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.

31、ashrae.org). Published in HVAC subsequent samples were taken after approximately 24, 48, and 96 h of storage.EXPERIMENTAL APPARATUSIn the homogeneous storage investigation, a 35 ft3(1 m3) storage tank with a mixing propellerwas used (Figure 2). The tank was connected to a scraped-surface ice generat

32、or with a separatepump. The ice concentration in the storage was maintained constant at a given set point by on-offcontrol of the ice generator. A Coriolis mass flow meter with an accuracy of 0.6% (Hansen andKauffeld 2001) produced the signal to start and stop the refrigerant compressor, and ice con

33、cen-tration in the storage was maintained within 1% of ice during the entire experimental period of4 days. This approach resulted in new ice being generated throughout the entire period. The iceformed as new ice crystals on the cold wall in the ice generator or by building onto existing crys-tals. T

34、he predominance of a crystallization mechanism influenced time-behavior findings.In the heterogeneous storage study, the lower ice concentration (30%) was built up in thehomogeneously mixed 35 ft3(1 m3) storage tank before the ice slurry was pumped to a 10 ft3(285 L) tank, where it was allowed to se

35、parate. To reach the higher ice concentration of 46%,the ice-free carrier fluid was extracted from the bottom of the tank, pumped through the iceTable 1. Test Parameters in Agglomeration StudyParameter Homogeneous Storage Heterogeneous StorageIce concentration, percentageby weight (low/high)10% 30%

36、30% 46%Additive, percentage byweight10% ethanol 15% propylene glycol10% ethanol 15% propylene glycolStorage condition Closed to air Open to air Closed to air Open to airSurfactant inhibitors(0.15%)Polyoxyethylenesorbitan trioleateNone Polyoxyethylenesorbitan trioleateNoneFigure 2. Schematic Layout o

37、f Experimental ApparatusHansen456.fm Page 23 Monday, December 16, 2002 8:23 AM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC the ice crystals in different layers overlap and must be sep-arated to identify the actual shape an

38、d size of individual ice crystals. Separation was done manu-ally on transparent film (B). After this separation, the crystals might not be located in the sameposition relative to each other, and some might even have been rotated. Therefore, although (B)and (C) differ from (A) in the location and ori

39、entation of the ice crystals, the individual ice crys-tals were the same in all figures. Finally, the image was scanned into a computer where ice crys-tals were filled as solids before digital analysis was performed. Only crystals fully within thepicture frame were included in the analysis.Reliabili

40、tyEnsuring a Sufficient Number of ObservationsTo ensure a sufficient number of ice crystals to provide exhaustive information, the test sizewas checked by applying a “split half” method. From one image, the random geometric infor-mation was divided into two equally sized groups and a hypothesis of t

41、he distribution beingidentical was tested. To determine test size and function, it was assumed that the underlyingpopulation was close to normal N(, 2), with unknown but identical standard deviation. Theassumption of normal distribution is justified from the results shown in Figure 5.If the null hyp

42、othesis (H0: 1= 2) is accepted (i.e., the two samples have the same distribu-tion), then the number of observed particles is maintained. Otherwise, the number of observa-tions is increased until it can be accepted that any analysis of the same number of observationswill lead to the same result.Disti

43、nction Between Joined Ice Crystals and Ice Crystals in Boundary ContactDuring image analysis, it was necessary to distinguish the case when two ice crystals hadjoined to form one single crystal (Figure 6, point A) from the case where two individual iceFigure 4. Steps of Preparation for Digital Image

44、 AnalysisHansen456.fm Page 25 Monday, December 16, 2002 8:23 AM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC the projection ofan individual ice crystal does not reveal the crystals real shape and dimension. If the analysis

45、isused to determine absolute dimensions and description of shape, the lost information must becompensated for with a division of all measures by a view factor, defined by assuming a randomorientation of crystals around any axis with the same probability. For perfectly spherical parti-cles, the proje

46、ction is independent of the orientation of the crystal to the two axes of the viewplan, and the view factor is 1. Any other particle shape will lead to a view factor less than unityand will depend on the actual shape. However, for analyzing crystal growth, it is mainly the rel-ative change of partic

47、le dimensions over time that must be determined. If a large number ofcrystals are involved in the analysis, it may be reasonable to assume that all crystals have ran-dom orientation and that the average view factor of all crystals in one image does not changeover time, which implies that dominant ge

48、ometrical characteristics, such as elongation androundness of the crystals, do not change over time. This assumption can be justified from theresults shown in Figure 10.Figure 5. Example of Change of Size Distribution of Ice CrystalsOver Time Illustrated by Feret Diameter10% Ice, 15% Propylene Glyco

49、l, Homogeneous Mixed Storage, Stabilizer Tween 85 and Air PresentFigure 6. Distinction Between (A) Joined and (B) Individual CrystalsHansen456.fm Page 26 Monday, December 16, 2002 8:23 AM 2003. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC with these two values, various geometri-cal descriptions could be defined.Because th