ice formation around a horizontal tube in a rectangular vessel

INTRODUCTIONCool thermal energy storage is a technology that shiftselectricity demand from on-peak period to off-peakhours, reducing peak demand. It also reduces electricitybills due to use of electricity for cooling during theperiods when electricity prices are lower (Hasnain,1998). These systems utilize a storage medium to storecooling produced during off-peak hours. The storedenergy is later used during peak hours to meet an airconditioningor process-cooling load. Various mediasuch as chilled water, ice, or phase-change-materialscan be used to store energy (Dorgan and Elleson, 1993).Ice cool thermal energy storage systems are oftenclassified as static or dynamic, according to the way iceis delivered to the storage tank. In dynamic systems, iceis produced outside the storage tank and removed fromthe ice-making surface continuously (Dincer, 2002). Instatic systems, an ice-making pipe is installed in thestorage tank where ice is formed and later melted. Iceon-coilsystems produce ice outside a coil. The ice maybe melted using either an external melting system,which melts ice from the outer side, or an internalmelting system (Dincer, 2002).Understanding of the solid-liquid phase change heattransfer phenomena around cylindrical tubes is neededfor design of efficient ice-on-coil cool thermal energystorage systems. Sasaguchi et al. (1997) studied thesolidification around a single and two vertically spacecylinders in a fixed volume. In their study, pure waterwas used as the working fluid and the non-Boussinesqmodel was utilized to simulate the buoyancy drivenflow. Sasaguchi et al. (1997) investigated effect of theinitial water temperature on the transient cooling ofwater around a cylinder in a rectangular enclosure.Inteman and Kazmierczak (1997) studied heat transferand ice formations deposited upon staggered and an inlinetube bank configuration. The first part of the studyfocuses on the global aspects of the phase changephenomenon and the second part of the study continuesthe analysis by investigating the convectionphenomenon close-up. A review of researches on avariety of water freezing and ice melting problems wasgiven by Fukusako and Yamada (1993). Habeebullah(2007) investigated growth rate of ice on the outside ofcooled copper tubes. Ice formation around anisothermally cooled horizontal cylinder and effect ofnatural convection were studied by Cheng et al. (1988).A similar work on ice formation and heat transfer withwater flow around cooled cylinders was reported byHirata and Matsui (1990). Stampa and Nieckle (2005)studied numerically the growth of an ice layer formedoutside a vertical tube, inside an annular cavity. Ismailet al. (2000) presented a numerical model for thesolidification of phase change material around aradially finned tube with a constant inner walltemperature. In their study, numerical experiments wereperformed to investigate the effects of the number offins, fin thickness, fin material, aspect ratio of tubearrangement and tube wall temperature. Erek et al.(2005) studied numerically and experimentally thesolidification around the horizontal radially finned tubeby considering fully developed velocity profile in thetube. Effect on cold thermal energy storage of heattransfer fluid inlet temperature, flow rate, fin densityand fin size were investigated by Kayansayan and Acar(2006). They experimentally tested a total of seven tubeconfigurations including a bare tube. Milon and Braga(2003) developed an experimental apparatus toinvestigate the supercooling phenomenon of waterinside cylindrical capsules used for cool thermal energystorage process. They investigated supercooling periodof the water and the nucleation temperature for differentheat transfer fluid temperatures. Sparrow and Broadbent(1983) investigated freezing of an initially superheatedor nonsuperheated liquid in a cooled vertical tube.Measurements were made which yielded informationabout the freezing front and the frozen mass, about thevarious energy components extracted from the tube.Sparrow and Bahrami (1980) performed experiments todetermine the natural convection heat flux distributionon the faces of isothermal circumferential fins affixed toa horizontal cooled tube. Sparrow and Hsu (1981)solved the conjugate phase change convection problemwhich results when a coolant passes through a tubesituated in a liquid phase-change medium. It wasreported that, there is generally axial variation of thesolid thickness along the tube because of the increase inthe coolant temperature as it warms up along the length.Although thermal energy storage systems for coolingwere first used commercially in the 1940s and havegaining popularity in the USA and Europe in recentyears, application of these systems are very limited inTurkey. Although there has been an interest in thesesystems in recent years, barriers still exist in Turkeyregarding the application of these systems; such systemsare not well known, they are relatively expensivebecause storage tanks are imported and they arenotorious for troublemaking.Büyükalaca et al. (2003) designed and constructed anice-on-coil type ice storage tank and air-conditionedtwo rooms located inside the Laboratories ofMechanical Engineering Department of ÇukurovaUniversity, Adana, Turkey. General view of the icestorage based air-conditioning system is shown inFigure 1. The system can be operated both under fullstorageand partial storage load leveling operatingstrategies. The experimental system included an aircooledwater chiller, an ice storage tank,interconnecting piping, controllers (two and three wayvalves), measuring equipment, and data collectionsystem. The ethylene glycol-water solution of 30 %concentration by weight was used as the heat transferfluid.338 m long, polyethylene coil pipe with 16 mm outsidediameter and 2 mm thickness was divided into 26 partsand each part (13 m long) was wounded to form a coil.76

Exp.valveCondenserEvaporatorComp.Three-waymixing valveManometerTCBDAIce tankRotometerCirculationpumpGlycolwatertankWaterTest tubeFigure 1. Schematic view of experimental test rigThe coils were placed into a circular polyethylene tankhaving a diameter of 1 m and an overall height of 1.5 m.Charging and discharging tests were carried out todetermine the performance of the system constructed.The experiments of Fertelli et al. (2006 and 2008)showed that mechanisms observed during the iceformation (supercooling, nucleation, and dendritic iceformation) are very important parameters for systemperformance and they should be investigated in detail.A parallel experimental system was constructed tounderstand solid-liquid phase change heat transferphenomena around a tube.In this study, phase change phenomena around ahorizontal tube were investigated both experimentallyand analytically. An experimental investigation wascarried at three different heat transfer fluid inlettemperatures in order to study the dynamic behavior ofice layer formation on tube. A mathematical model wasalso developed the heat transfer process for phasechange phenomenon around a horizontal tube.EXPERIMENTAL SYSTEMAs seen in Figure 1, experimental unit consists of a testtube for ice formation, an air-cooled water chiller unit, apump, piping, control valves and measurement system.To simulate the conditions in ice-on-coil type icestorage tank, the test tube is made of polyethylene pipewith 16 mm outside diameter, 2 mm thickness and 4 mlong. It was placed inside a rectangular vessel of 64 mmwide, 75 mm high and 4000 mm long. The vessel wasbuilt from transparent acryclic plates of 4 mm thick, tobe able to take photographs of solidification processaround the tube. The vessel was filled with tap water toa depth of 50 mm and insulated from all sides includingits cover, by 5 cm thick elastomeric rubber foam layerto reduce the heat transfer through the walls.Ethylene glycol-water solution was used as the heattransfer fluid. It is cooled in the evaporator of the aircooledwater chiller unit and pumped to the test sectionpassing through a valve, which is used for flow rateadjusment, and a rotameter. Temperature of theethylene glycol-water solution entering into the testsection was kept constant at the desired level using acontrol system, which is constituted of motorized twoand three way valves and temperature sensors. Flowrate of the transfer fluid was measured using arotometer located upstream of the test section. Twothermocouples were used to measure the inlet and theoutlet temperatures of the heat transfer fluid. Totally 20thermocouples were replaced into water at 7 differentaxial positions (Figure 2) along the test tube fordetecting variation of the water temperature in thevessel One thermocouple was utilized at eachmeasuring station except the first (x/L=0.1), the middle(x/L=0.47) and the last (x/L=0.86) ones. 8thermocouples were used at the first and the laststations, while 4 thermocouples were used at the middleone (Figure 3) to be able to detect the water temperaturevariation at these cross sections in detail. Ambient airtemperature was also monitored using a thermocouple.77

Waterglycolsolution410 mm 500 mm 500 mm 470 mm 580 mm 480 mm 500 mm 560 mmFigure 2. Measurement stations along test tubeTw4Tw4Tw3Tw3Tw3Tw6 Tw5Tw2Tw1Tw7 Tw813 10 1 1 10 135 14324275Tw2Tw1143242755Tw6 Tw5Tw2Tw1Tw7 Tw813 10 1 1 10 131 8323975a) x/L= 0.1 b) x/L= 0.47 c) x/L= 0.86Figure 3. Temperature measurement locations within the waterAll the thermocouples used were K type, tefloninsulated thermocouples. An ice-bath was employed asthe reference junction for all the thermocouples used.Outputs of the thermocouples were monitoredcontinuosly with a sampling rate of 30 readings/s, usinga computer-controlled data logging system.Due to the experimental nature of the present work, carewas given to the uncertainty in measurements. The fluidflow rate is measured by means of a calibratedrotometer with accuracy of ± 4%. To performcalibration of the rotometer, water was pumped from asupply reservoir to a constant head reservoir. Fromthere, water flowed to a weighing tank through themeasuring section with the rotometer under test. Theflow rate was adjusted with a control valve situated atthe downstream end of the test section. The rotometerwas calibrated by measuring the volume of water thatpasses the rotometer during a certain measuring time.Table 1. Uncertanity and range for different parametersParameter Range UncertainityTube inner radious (mm) 6 ± 0.1 mmTube outer radious (mm) 8 ± 0.1 mmTube length (m) 0 − 4 ± 0.5 cmFlow rate (l/h) 0 − 1000 ± %4Temperature (ºC) -7 to 30 ± 0.3 ºCIce thickness (mm) 0 − 20 ± 0.1 mmHereto the water was collected in the weighing tank,whereby the volume increment was measured andaverage flow rate was computed by dividing the volumeby the time. Temperatures of water and transfer fluidare measured by means of K type teflon insulatedthermocouples with accuracy ± 0.3 ºC. Thethermocouples were calibrated against a platinumresistance thermometer in a stirred liquid bath whosetemperature was adjusted with the help of atemperature-measuring bench. Thickness of the iceformed around the tube is measured with ± 0.1 mm.Table 1 shows the range and uncertainity of parametersfor the experiments. Thickness of the ice formed on thetube was determined using a visual technique. Icethickness was measured at the same stations wheretemperature measurements were made. Smallobservation windows are opened through the insulationmaterial at the measuring stations to be able to make icethickness measurements. Photographs of the ice weretaken with a digital camera (Nikon Coolpix 8700, 8MP,10x optic zoom) from the front and the top of thevessel at each station to be able to detect circumferentialchanges in ice thickness. Different light sources andpositions were tested during the prelimenary experimetsto determine the best lighting arrangement. A traversingmechanism was designed to be able to detect the iceformation along the tube with a short period of timebetween the stations. The time period between any twomeasuring station was 20 s at maximum. Allphotographs were taken in close-up mode with 5 MPresolution. Figure 4 shows two images of solidificationtaken at two different times at the first measuringstation (x/L=0.1) for an inlet temperature of -5.8 ºC,flow rate 900 l/h as an example to the images takenduring the experimets. At the end of experiments, thephotographs taken were transferred to an image analysisprogram (Image-ProPlus 4.5) for post processing. Usingsome of the techniques offerred by the program, brightimages that clearly indicate the boundaries of the iceformed around the tube were obtained. From the imagesprocessed, thickness of the ice was determined for thetop and bottom of the pipe separately (left and rightsides of the pipe for the photographs taken from the topof the pipe) using the pipe diameter as a referencedimension.78

(a)(b)Figure 4. Photographs of the ice formed around test tube at two different times (a:3000, b:11900 s)MATHEMATICAL MODELResults of the experiments carried out revealed that,temperature of the heat transfer fluid flowing throughthe tube (T f ) does not change too much between thetube inlet and exit. Therefore, it can be assumedconstant (Figure 5). Because tube wall is thin andthermal conductivity is not high, the heat transferbetween ice-water surface and fluid flowing in the tubecan be assumed as one-dimensional.Figure 5. Schematic of ice formation processAt an axial position of the tube, the amount of heattransferred per length of the tube (calculated from;.QTeFlowxTPhase Change Material (Water)Te−T=RpfIceRi Ro rQ .x) can be(1)where T e is the temperature of the ice and R p is thermalresistance based on the peripheries, instead of thesurface areas:seswr RoRiR .wp +hiPikwP+wm= 1sesek Pem(2)in which k w and k e are the thermal conductivities of thetube material and ice, respectively. h i is the convectiveheat transfer coefficient inside the tube. The peripheries(P i , P wm and P em ) are obtained from the followingequations:P = 2 π(3)iR i( R R )o iwm= π (4)RoPPem2 −lnR2 π se=⎛ Ro+ sln⎜⎝ RoDefiningR +ies⎞⎟⎠(5)wpo= 1 (6)hiPikwPwmr = R o+ s e(7)andR * = R / R(8)Rp*poppo= 2 π k R R(9)R oepoor * = r /(10)the following equation is obtained from eq. (2):79

** ln rRp= 1+(11)R*poSimilar to the heat transfer from ice-water surface toflowing fluid in the tube, heat is also transferred fromambient to the ice-water surface. The heat gain ( Q & g)per length of the tube can be written as:TaTeQ& −g=(12)RaHere T a and R a are ambient temperature and thermalresistance based on periphery between ambient and icewatersurface, respectively. The outer surface of thevessel is insulated to minimize heat gain ( Q & g) from theambient. R a can be calculated from:Rss1ipa= + + +(13)kiPmikpPmphciPichaPawhere h a is the heat transfer coefficent between ambientand outer surface of the vessel and h ci is heat transfercoefficient between water in the vessel and innersurface of the vessel. s p and k p are the thickness andthermal conductivity of the plexiglas, respectively. P mpis the mean periphery of the plexiglas. P ic and P a are theinner and outer peripheries of the vessel, respectively. s iand k i are the thickness and thermal conductivity of theinsulation. P mi is the mean periphery of the insulatingmaterial. The difference between1Q & xandQ & gis usedfor ice production, because the temperature differencebetween fluid (T f ) and ice-water surface temperature(T e ) is so low that sensible heat storage of ice can beneglected compared with the latent heat of freezing.Therefore, one can write the following equation;( Q − Q & ) dt = ρ h 2 π r dr& (14)xgeewwhere ρeis the density of the ice, h ew is the latent heatof freezing of water and t is the time. Using eqs. (1),(12) and (14), the following equation is obtained:2 π hewr drdt =Te−TfTa−T−R RpUtilizing the following definitions,*e ft t22 π ρehewRoaepo(15)T −T= (16)R* RaRa=RpoTe−TfT −Tae(17)the following dimensionless differential equation (18) isfound:dt*R r dr= (18)* * *p* *1−Rp/ RaBecause of the insulation,compared withas:R * ashould be very big*Rp. Therefore eq. (18) can be rewritten⎛dt⎟ ⎟ ⎞⎝ ⎠Inserting eq. (11) into eq. (18) yields:*R2* * p * *=⎜Rp+ r dr(19)⎜*Ra* 2( r )⎤* *⎡* 1 1 ⎛ 2 ⎞* lndt = ⎢1 + 1 ln rr dr* + ⎜ ⎟*** * 2⎢ RaR+ +poRa RaR ⎥ ⎥ (20)⎣⎝ ⎠po ⎦With the integration of this equation with the boundarycondition,t * = 0 : r * = 1(21)the following equation is obtained:* 2( ln r )*12*12*⎛ ⎞ ⎛ ⎞1 ⎜ r − ⎟ rt = ⎜ ⎟+*+2*⎝ R ⎜ ⎟a ⎠ ⎝ ⎠ 2 RaR⎡ * 2 ⎛ * 1 ⎞ 1 ⎤⎢r⎜ln− ⎟ + ⎥⎣ ⎝r 2 ⎠ 2⎦* 2po1 +2 R*po⎡⎛ 2 ⎞ 1⎢⎜1⎟+ −**⎢⎣⎝ Ra⎠ RaRIce thickness s e can be made dimensionless witheeo*po(22)s * = s / R(23)Therefore, the following equation can be written usingeq. (10):* *r 1+ se= (24)RESULT AND DISCUSSIONSExperimental ResultsExperiments were carried out at a fixed transportmedium flow rate of 900 l/h (corresponds to a Reynoldsnumber of 5318 at the tube inlet) and three differentinlet temperatures (-4.3 ºC, -4.9 ºC and -5.8 ºC). Thetemperature of the water inside the vessel was keptbetween 0 ºC and 1 ºC at the beginning of theexperiments to minimize the natural convection effects.As an example to the water temperature measurements,Figure 6 shows time wise variations of the watertemperatures at the measurement points inside thevessel for a transfer fluid inlet temperature of -5.8 ºC.Sensible heat is extracted from the water inside thevessel and transferred to the transfer medium.⎤⎥⎥⎦80

Temperature ( o C)Temperature ( o C)Temperature ( o C)1,510,50-0,5-1-1,5-2-2,51,510,50-0,5-1-1,5-2-2,51,510,50-0,5-1-1,5-2-2,5-3(a)45 637 80 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000Time (s)(b)0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000Time (s)(c)0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000Time (s)21432145 637 821T1 T2 T3 T4 T5 T6 T7 T8Figure 6. Variation of water temperature with time at three measurement stations for T in = -5.8 ºC (a: x/L = 0.1, b: x/L = 0.47,c: x/L = 0.86)81

10,50x/L=0.10 x/L=0.23 x/L=0.35 x/L=0.4781x/L=0.62 x/L=0.74 x/L=0.86Temperature ( o C)-0,5-1-1,5-2-2,5-30 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000Time (s)Figure 7. Variation of water temperature with time at positions just above the test tube (thermocouple 3) for different axialmeasurement stations (T in = -5.8 ºC)As a result, temperature of the water decreases steadilyat all measurement points except at the bottom of thetube.The phase change process does not start at theequilibrium freezing point of water (0 ºC) andtemperature of the water decreases below 0 ºC. Coolingcontinues until the start of nucleation of ice. Thisprocess is called as supercooling. Supercooled water isin a metastable liquid state. The maximum possiblesupercooling (T max ), which is -38±2 ºC, can be achievedwhen the water is free of any solid particles that cancatalyze ice germ formation (pure water) (Koop, 2004).At about this temperature, freezing happens as ahomogeneous nucleation. When the temperature ishigher than the limit temperature (T max ), heterogeneousnucleation occurs. In heterogeneous nucleation, icegrows from crystals that form in the water on freezingnuclei. The ice formed during this stage is calleddendritic ice. The degree of supercooling temperaturedepends on properties of the solid particles inside thefluid. A liquid below its freezing point will crystallize inthe presence of a seed crystal or nucleus around which acrystal structure can form. However, lacking any suchnucleus, the liquid phase can be maintained all the waydown to the temperature at which crystal homogeneousnucleation occurs.As can be seen from Figure 6, supercooling occurs at allmeasurement points except the bottom of the tube andthe degree of it is between -1.4 ºC and -3 ºC. Oncenucleation occurs, a very thin layer of dendritic iceformation starts at the outer surface of the tube in theentrance region (x/L=0.1), and propagates in the axialdirection. When the dendritic ice first forms, thetemperature rises rapidly to the freezing temperature asa result of the latent heat of fusion of water .Compared with the supercooling process, duration ofthis process is very short, lasting only 1 to 3 s,depending on the degree of supercooling and theposition.Dendritic ice formation is followed by latent heatthermal storage process, during which a relatively thicklayer of ice starts to form around the tube outer surface.During this period, temperature remains at freezingpoint temperature (0 ºC) for a certain period of time.However, when the ice is formed, temperature of the icebegins to fall below 0 ºC, due to heat transfer from thewater inside the vessel to the heat transfer fluid (coolingof ice). Temperature decrease can be clearly observedwhen the measuring point (thermocouple) is covered bythe ice (thermocouples 2, 3, 6 and 7 in Figure 6a).Temperature of the water never decreases down tofreezing temperature (0 ºC) at the bottom of the vessel(thermocouples 1 and 2) during ice formation. Theslight temperature increase at the beginning of theexperiment is due to collection of warm water at thebottom of the vessel.The same processes can also be seen from Figure 7 thatshows the variation of water temperature with time nearthe tube wall (thermocouple 3) at different axialpositions. Since the thermocouple 3 is located veryclose to the tube, it is covered by the ice as soon as iceformation starts. Therefore, temperature stays constantat 0 ºC only for a very short period of time.Variation of the time from the beginning of theexperiment to the start of nucleation (nucleation time)along the pipe length is given in Figure 8 for differentinlet temperatures of heat transfer fluid. As can be seenfrom the figure, nucleation first starts at the entranceregion, which is the coldest region, and propagatesalong the tube.82

Nucleation time (s)35003000250020001500Tglycol= -5.8 ºCTglycol= -4.9 ºCTglycol= -4.3 ºCIce thickness (mm)141210864TopBottomRightLeft10005000 0,2 0,4 0,6 0,8 1Axial distance (x/L)Figure 8. Variation of nucleation time along tube for differenttransfer fluid inlet temperaturesNucleation time increases almost linearly with the axialdistance for all inlet temperatures. It is obvious thatfluid inlet temperature seriously affects the onset of thephase change of water and nucleation time. Nucleationtime increases with the increase of transfer fluid inlettemperature. Increase of inlet temperature from -5.8 ºCto -4.3 ºC almost doubles the nucleation time at all axialpositions along the tube. Due to high fluid inlettemperature, the temperature of water needs more timeto reach the nucleation temperature. Similar results werealso reported by Chen et al. (2000) who conducted aseries of experiments to investigate the supercoolingphenomenon and the effect of inlet coolant temperatureon the nucleation of water inside cylinder.The thickness of the ice formed at the bottom, top, rightand left sides of the tube was determined at differentaxial positions during the experiments. As an exampleto the measurements, Figure 9 shows time wisevariations of the ice thickness at three axial positions(x/L=0.1, 0.47 and 0.86) for a transfer fluid inlettemperature of -5.8 ºC. Photographs of the iceformation from the top of the tube were not suitable forthe post processing because of dense dendritic iceformation at the beginning of the experiments at the topof the tube.Similar trends are observed at all measuring stationsalong the tube. At the beginning of the ice formation,thickness of the ice is almost equal around the tube. Iceformation is then better at the upper part of the tubethan the lower part of the tube. This is because of heattransfer from outside, although the vessel is insulatedproperly. As result of heat gain from outside, warmwater is collected at the bottom and the cold water iscollected at the top. The difference between the top andthe bottom of the tube in terms of the ice thickness isapproximately constant (0.6 mm) after 3000 s. Thethickness of the ice at the right and left sides of the tubeis close to each other.Ice thickness (mm)Ice thickness (mm)201412108642014121086420(a)0 5000 10000 15000Time (s)TopBottomRightLeft(b)0 5000 10000 15000Time (s)TopBottomRightLeft(c)0 5000 10000 15000Time (s)Figure 9. Variation of ice thickness with time for top, bottomand sides of tube for T in = -5.8 ºC (a: x/L= 0.1, b: x/L= 0.47,c: x/L= 0.86)The shape of the ice formation during the experimentsat three axial positions (x/L=0.1, 0.47 and 0.86) isdepicted in Figure 10 for three different transfer fluidinlet temperatures (-4.3 ºC, -4.9 ºC and -5.8 ºC).83

2010(mm)0-10-2020-10x/L=0.10x/L= 0.47 x/L= 0.860(mm)10 20 -20 -10 0(mm)10 20 -20 -10 0(mm)10 20(a)10(mm)0-10x/L=0.10 x/L= 0.47 x/L= 0.86-20 -10 0 10 20 -20 -10 0 10 20-20-10 0 10 20(mm) (mm) (mm)(b)2010(mm)0-10x/L=0.10 x/L= 0.47 x/L= 0.86-20 -10 0 10 20 -20 -10 0 10 20 -20 -10 0 10 20(mm) (mm) (mm)(c)1920s2880s 4140s 5760s6900s9420s11700sFigure 10. Evolution of ice-water interface with time for different transfer fluid inlet temperatures (a: T in = -4.3 °C, b: T in = -4.9°C, c: T in = -5.8 °C)When the ice forms first, it has almost a circular shape.However, the shape is distorted in later stages of iceformation. Ice thickness at the top becomes higher thanthe bottom, right and left, and therefore, the shape takesan oval form.Influence of transfer fluid inlet temperature on icethickness is shown in Figure 11 for 3 different timesduring the experiments (2880 s, 6900 s and 11700 s) at84an axial position of x/L=0.47. With the increase oftransfer fluid inlet temperature, ice thickness decreasesat all points around the tube. The same trend is alsoobserved at other axial positions. As discussed above,ice formation is almost uniform around the tube at theinitial stages of ice formation (2880 s). However laterthe uniformity is distorted and a clear differencebetween the top and bottom is observed at 11700 s.

1412Ice thickness (mm)1086420-6 -5,8 -5,6 -5,4 -5,2 -5 -4,8 -4,6 -4,4 -4,2 -4Transfer fluid inlet temperature ( o C)2880 Top 2880 Bottom 2880 Right 2880 Left6900 Top 6900 Bottom 6900 Right 6900 Left11700 Top 11700 Bottom 11700 Right 11700 LeftFigure 11. Influence of transfer fluid inlet temperature on ice thickness at x/L= 0.47Comparison of Mathematical Model withExperimental ResultsA simple model for the formation of ice around acylindrical tube is given above. Comparison of themodel and experimental results are given in Figure 12for three different transfer fluid inlet temperatures (-4.3,-4.9 and -5.8 ºC) which corresponds to R * = 3. 866 ,R * a= 4.566 and R* a= 5. 238 , respectively. The value of*Rpois 1.912 for the experiments. The experimentalvalues shown in this figure were obtained by averagingice thickness values determined at top, bottom, right andleft of the tube at each axial position. Since the timescale starts from the formation of ice in the model, theexperimentally obtained values at different axialpositions were shifted according to time at which iceformation starts at the axial position of interest.Therefore, time scale in Figure 12 is not the absolutetime scale, rather represents the time passed from thestart of ice formation. The figure shows theexperimental ice thicknesses at all axial positions withshifted time scales.12ExperimentalIce thickness (mm)1086420Model(a)0 5000 10000 15000Time (s)aIce thickness (mm)Ice thickness (mm)1412108642014121086420ExperimentalModel(b)0 5000 10000 15000Time (s)ExperimentalModel(c)0 5000 10000 15000Time (s)Figure 12. Comparison of ice thickness obtained frommeasurements and model for different transfer fluid inlettemperatures (a: T in = -4.3 °C, b: T in = -4.9 °C, c: T in = -5.8°C)85

As can be seen from Figure 12, the agreement betweenthe model and the experimental values is good duringthe ice formation process for all transfer fluid inlettemperatures. The model under predicts slightly icethickness at the beginning of the process and overpredicts at the end.CONCLUSIONIn the present study, phase change phenomena and theeffect of heat transfer fluid temperature on ice formationaround a horizontal tube in a rectangular vessel wereinvestigated both experimentally and analytically.Based on the obtained results our conclusions aresummarized as follows:1. Experimental results show that supercooling occursat all axial measuring stations along the tube andnucleation first starts at the entrance region andpropagates along the tube. Nucleation temperaturesfor all measurement points are between -1.4 ºC and-3 ºC.2. It is obvious that the heat transfer fluid inlettemperature seriously affects the onset of the phasechange of water and nucleation time. With thedecrease of heat transfer fluid inlet temperature,nucleation time decreases. Furthermore nucleationtime increases with the axial distance for all heattransfer fluid inlet temperatures.3. The thickness of the ice formed at the top andbottom sides of tube is different from each other,although the thickness of the ice formed at the rightand left sides is close to each other. Ice thicknessaround tube is substantially influenced by naturalconvection that occurs within the water. With theincrease of transfer fluid inlet temperature, icethickness decreases at all points around the tube.4. Ice thickness evaluated by mathematical modelagrees reasonably with the experimental data for alltransfer fluid inlet temperatures.REFERENCESBüyükalaca O., Yılmaz T., Fertelli A., Ice Storage Air-Conditioning System, Proceedings of the EasternMediterranean and Southeastern Anatolia RegionMachine and System Design-Production Congress (inTurkish), 95-101, Turkey, 2003.Chen S.L., Chen C.L., Thin C.C., Lee T.S., Ke M.C.,An Experimental Investigation of Cold Storage in anEncapsulated Thermal Storage Tank, ExperimentalThermal and Fluid Science 23, 133-144, 2000.Cheng K.C., Inaba H., Gilpin R.R., Effects of NaturalConvection on Ice Formation around an IsothermallyCooled Horizontal Cylinder, Journal of Heat Transfer110, 931-937, 1988.Dincer I., On Thermal Energy Storage Systems andApplications in Buildings, Energy and Buildings 34,377-388, 2002.Dorgan C.E., Elleson J.S., Design Guide for CoolThermal Storage, American Society of HeatingRefrigeration and Air-Conditioning Engineering,Atlanta, 1993.Erek A., Ilken Z., Acar M.A., Experimental andNumerical Investigation of Thermal Energy Storagewith a Finned Tube, International Journal of EnergyResearch 29, 283–301, 2005.Fertelli A., Büyükalaca O., Yılmaz T., Performance ofan Ice Storage Air-Conditioning System, Proceedingsof the Fifth GAP Engineering Congress (in Turkish),273-280, Turkey, 2006.Fertelli A., Air-conditioning System with Ice ThermalEnergy Storage, Ph.D Thesis, Institute of natural andapplied sciences of Çukurova University, Adana, 2008.Fukusako S., Yamada M., Recent Advances in Researchon Water-Freezing and Ice-Melting Problems, Exp.Thermal and Fluid Science 6, 90-105, 1993.Habeebullah B.A., An Experimental Study on IceFormation around Horizontal Long Tubes, InternationalJournal of Refrigeration 30 (5), 789-797, 2007.Hasnain S.M., Review on Sustainable Thermal EnergyStorage Techologies, Part:II Cool Thermal Storage,Energy Conversion &Management 39 (11), 1139-1153,1998.Hirata T., Matsui H., Ice Formation and Heat Transferwith Water Flow around Isothermally Cooled CylindersArranged in a Line, Journal of Heat Transfer 112, 707-713, 1990.Intemann P.A., Kazmierczak M., Heat Transfer and IceFormations Deposited upon Cold Tube BundlesImmersed in Flowing Water - I. Convective analysis,Int. J. Heat Mass Transfer 40 (3), 557-572, 1997.Intemann P.A., Kazmierczak M., Heat Transfer and IceFormations Deposited upon Cold Tube BundlesImmersed in Flowing Water - II. Conjugate analysis,Int. J. Heat Mass Transfer 40 (3), 573-588, 1997.K.A.R I., Henriquuez , Moura L.F.M., Ganzarolli M.M.,Ice Formation around Isothermal Radial Finned Tubes,Energy Conversion & Management 41, 585–605, 2000.Kayansayan N., Acar M.A., Ice Formation around aFinned-Tube Heat Exchanger for Cold Thermal EnergyStorage, International Journal of Thermal Sciences 45,405-418, 2006.86

Koop T., Homogeneous Ice Nucleation in Water andAqueous Solutions, Z. Phys. Chem. 218 (11), 1231-1258, 2004.Milon J.J., Braga S.L., Supercooling Water inCylindrical Capsules, 15. Symposium on themophysicalproperties, Boulder, Colorado, U.S.A., 2003.Sasaguchi K., Kusano K., Viskanta R., A NumericalAnalysis of Solid Liquid Phase Change Heat Transferaround a Single and Two Horizontal, Vertically SpacedCylinders in a Rectangular Cavity, Int. J. Heat MassTransfer 40 (6), 1343-1354, 1997.Sasaguchi K., Kusano K., Kitagawa H., Effect ofDensity Inversion on Cooling of Water around aCylinder in a Rectangular Cavity, Numerical HeatTransferPart A 32, 131-148, 1997.Sparrow E.M., Bahrami P.A., Experiments on NaturalConvection Heat Transfer on the Fins of a FinnedHorizontal Tube, International Journal of Heat andMass Transfer 23, 1555-1560, 1980.Sparrow E.M., Hsu C.F., Analysis of Two-DimensionalFreezing on the Outside of a Coolant-Carrying Tube,International Journal of Heat and Mass Transfer 24(8), 1345-1357, 1981.Sparrow E.M., Broadbent J.A., Freezing in a VerticalTube, Journal of Heat Transfer 105 (2), 217-225, 1983.Stampa C.S., Nieckele A.O., Numerical Study of IceLayer Growth around a Vertical Tube, ThermalEnginering 4 (2), 138-144, 2005.Ahmet FERTELLI was born in Sivas-Turkey in 1975. He graduated from the departmentof mechanical engineering at Çukurova University in 1998. He obtained his MSc degreeat Cumhuriyet University in 2001. He pursued his PhD studies at Çukurova Universityand received PhD degree in 2008. He has been working as a research assistant atCumhuriyet University since 1999. His main research fields are cool thermal energystorage systems and phase change. He is a member of TTMD and MMO.Orhan BÜYÜKAALACA was born in Kaş-Turkey in 1964. He graduated from theMechanical Engineering Department of Çukurova University in 1984. In 1987, heobtained his MSc degree from the same institution. He pursued his PhD studies atManchester University and received PhD degree in 1993. He worked as AssisstantProfessor and Associate Professor between 1994 and 2003 in Çukurova University. Heworked as a visiting researcher at Manchester University in 1993 and 1996. He wasappointed as a Professor in the Mechanical Engineering Department of ÇukurovaUniversity in 2003. He has been the Rector of Osmaniye Korkut Ata University since2008. His main research fields are analysis of meteorological data for heating and airconditioningsystems, energy analysis, heat pumps and turbulent convective heat transfer.He is a member of TTMD and MMO.Alper YILMAZ was born in Tarsus-Turkey in 1975. He graduated from the department ofMechanical Engineering Department of Boğaziçi University in 1997. He pursued his MScand PhD studies at Çukurova University and got his degrees in 1999 and 2004,respectively. He was assigned assistant professor of mechanical engineering in 2006 andsince then he has been working at the same institution. In 2000, he was visiting researcherin Berlin-Germany by DAAD award. His main research interests are conductive andconvective heat transfer. He is a member of TTMD and MMO.87