Scribd Logo
Upload

Welcome to Scribd!

  • One Dimensional

    Uploaded by

    Lucas Santos de Oliveira
    8 views10 pages

    Original Title

    one dimensional

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    8 views10 pages

    One Dimensional

    Uploaded by

    Lucas Santos de Oliveira

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 10
    SOLID-LIQUID PHASE-CHANGE HEAT TRANSFER AND INTERFACE MOTION IN MATERIALS COOLED OR HEATED FROM ABOVE OR BELOW IN. W. Hate, JR* and R. Viskanta Heat Transfer Laboratory Schoo! of Mechanical Engineering, Puidue University, Went Lafayette, IN 47907, USA. (Received 2 March 1979 and in reisd frm 20 July 1979) Abstract ~ Solid-liqud phasechange heat wansfer has been studied experimentally and analytically in several diferent materials (eg stacie ai, sodium phosphate dodecahydrate, sodium Solfate decahydrate and n-octadccane which have been suguesed as canidates for latent eat-of fasion thermal energy storae ‘materials. Solid-ligud interface motion during freezing and melting from above as well as below has been determined in a rectangular test ell suitable fr photographic observations. Comparison of experimental ata for moctadecane with predictions based on Neumann and other analyses which aceount for natural ‘convection heat transfer at fhe sli hud interface show that natural convection tn the liqud mist be accounted inthe prediction of phase change boundary mation for unstable situations which arse during ‘melting from below and sliifcation fom above parameter in equation (43) defined as (CirRarPr specific heat; Fo, Fourier number, at/H*; H,) eight of test cell; heat transfer coefficient; ‘Aly, latent heat of fusion; A” dretaal eonductivity Nu, Nusselt number defined as hs/kys Pr, Prandtl number, 44/hv; Ra, Rayleigh number, gBATs*/9a,; Ste,, Stefan number for solidification, eT, ~ T.W Ay: Stey Stefan number for melting, «(7 ~ T;Mhy: 5, melt or solid layer thickness; S*, dimensionless melt or solid layer thickness, sis T, temperature; times 8, distance from cooled or heated boundary Greek symbols 4 thermal diffusivity, k/pe B, thermal expansion eoefcient, T, dimensionless parameter, "[i(H ~ sWi) (HCH ~ 3) (yk) [Ts THVT, ~ Tells 1. fatio of densities, apy: ‘S, thermal boundary layer thickness: ©, dimensionless temperature. (TT) = TH) 4, constant defined by equation (10) or equa- tion (13): % kinematic viseosity & dimensionless distance, »/H; ‘Presently at NASA Lyndon B, Johnson Space Center, Houston. TX 77058. USA pe dencitys dimensionless time, Ste: Fos 4, shape function defined by equation (23). Subscripts f refers to fusion; 1 refers to liquid; 0, refers to initial; sealers to soli vw, refers to wall intRopuCTION Hear conduction problems involving solid-tliquid phase change are of interest in a wide range of technologies and geophysics. The class of problems involving melting and freezing, generally referred to at “moving boundary” problems, have been the subject of ‘numerous theoretical investigations of various degree ‘of complexity and rigor. The methods of solution used in the analysis of heat conduction problems involving phase change included exact, integral, variational, Perturbation, purely numerical and other methods. Extensive literature reviews are available [1-3], and the classical Neumann problem is described in various heat conduction books [4]. Experimental studies of transient heat conduction involving. solid-liquid phase transformations and effects of buoyancy have been relatively few (5-7] Density differences between the solid and liquid phases and buoyancy forces produced by density differences due to temperature variations inthe liquid ‘may produce convective motions in the liquid. These and other effects such as superheating ofthe liquid (or subcooling of solid), physical property dependence on temperature have received litle theoretical and even ‘much ess experimental attention. The purpose ofthis study was to determine the effect of heat source/sink 2 ‘orientation on heat transfer and solid-liquid interface motion during phase change. Specifically, meting and freezing from below and above are studied in a rectangular test cell which allows for photographic ‘observation of the phase-change boundary. The effect of buoyancy on the one-dimensional rcting and freezing of water has been studied experi- ‘mentally and analytically [5-8]. Thomas and West- water [9] verified the numerical method of Murray ‘and Landis [10] using their experimental data for n ‘ctadecane. The model was extended (o include nat- ural convection in the liquid by Rager and Westwnter [5] and was verified with their data for water. The velocity ofthe drift due to the density differences of the two phases [1] during meling and freezing was also accounted for in the numerical model [7] and was ‘checked using the data for water, Studies of convective instabilities in the melting of horizontal ie layer [11, 12] have also been conducted, Unfortunately, the density inversion of water is unique ta thie substance, and therefore, the results and conclusions about the effects of buoyancy and density changes during phase ‘ansformation cannot be generalized. Melting and solidification of parafins have also been studied in connection with spacecraft thermal control [13, 14] ‘and thermal energy storage [15]. but the elect of the hheat source/sink orientation on heat transfer and phase change howndary has nat received much attene tion. Its still common practice in analysis of latent- heat-offusion thermal energy storage [15-17] to nieglet the effets of convection during Soll-liguid phase change. ‘This paper describes a combined experimental and analytical study of heat transfer during melting and soliditication from a horizontal plate (eg. heat source /sink) facing either upward or downward. A number of different experiments using stearic acid, sodium phosphate dodecahydrate, sodium sulfate decahydrate find mroctadecane as test substances have been per formed, The work was motivated by the need to gain a more complete understanding of heat transfer pro- cesses during solid~liquid phase change in connection with latentcheat-offusion thermal energy storage EXPERIMENTS Test apparatus ‘The melting and freezing experiments were perfor- ‘med in a rectangular test cell which allowed for photographic observation of the solid- liquid interface motion in the material during phase transformation ‘The materials chosen for the study were required to have acceptable light transmission for visual obser- vation, eg. clear liquid phase and a reasonably opaque solid phase to give a sharp definition of the interface ‘A schematic diagram of the test apparatus is shown in Fig. 1. The test cell where melting and freezing took place was placed inside a temperature control chamber ‘which was maintained as close to the fusion tempera: tute ofthe test material as possible without initiating the meltingyfreezing process, This was accomplished with two heat exchangers which were installed in the chamber through which water from a constant tem perature hath was cireulated The tw fies of the chamber were made of glass with removable insulation placed on the windows to allow for photographic observation, In turn, the entire chamber was covered with @ Sem thick removable styrofoam insulation which had to be lifted off in order to photograph the solid-liquid interface position, Two copper heat exchangers formed the top and bottom walls of the test cell and served ae nearly isothermal heat sources. The vertical walls as well as the front and back faces of the cell were made of plexiglass. The inside dimensions ofthe est cell were 89 em high, by 14.6cm wide by 22cm deep. The depth of the test cell was purposely made much smaller than the height or the width in order t0 minimize the rnonuniformities in the melting front position per pendicular to the faces of the cell. Provisions were ‘made for filing the test cell with a liquid and to add additional liquid to completely fill the test cell to account for contraction during die initial soldi cation, Provisions were also made at oneend of the test cell to allow for liquid expansion during melting Fifleen chromel-alumel thermocouples were in- stalled at different positions in the test cell. The thermocouple wires extending from the measuring Junctions were run parallel tothe heated surface along ‘a small diameter rod which supported up to three ‘thermocouples. Nine thermocouples were installed in ‘each heat exchanger. Small diameter holes were drilled into the copper block, and the thermocouples were soldered very close tothe surface ofthe heat exchangers. Test procedure ‘Care was taken when filling the test cell withthe test ‘material to ensure that no bubbles were trapped in the ‘material. This was accomplished by vibrating the cell during the filing with the phase change material and its solidification In the melting experiments with the material in- inally in the solid phase, water near the fusion temperature was circulated through the heater blocks ofthe test cell and of the chamber, and the material was preheated to as close to the melting temperature as possible, This was dune in order to eliminate the initial ie 5 all Fic. 1 Schematic diagram ofthe test arrangement ‘so subcooling ofthe solid as one ofthe parameters which influences the melting. Depending on the particular ‘material being tested, it took from 4 to 18h to reach the sired conditions. After the material reached a steady, nearly uniform temperature just below its melting point, the temperature of the water circulated through the heater blocks ofthe test cell was raised toa desired level above the fusion temperature and was ‘maintained there for the duration of the experiment By closing or opening control valves in the flow circuit the bottom, the top, or both could be heated for a particular text ‘At the beginning ofeach experiment, and at regular interval thereafter, photographs were taken and other ‘visual observations were made. Photographing and visual observations required that the insulation around the test cell be removed; therefore, these ‘observations were made as rapidly and as infrequently a possible to minimize heat losses from the test cell 0 the ambient laboratory environment. Thermocouple EMF and flow meter readings were recorded at the same time, 1In freezing experiments, the procedure was similar except that the material was precooled to a uniform temperature just above the fusion temperature before solidification was started. IT the experiment was to be melting followed by freezing, as many of them were, ere was a period of atleast 15 min between the two parts ofthe experiment when no water was circulated through the heater blocks. This was to allow the constant temperature bath to adjust to the new temperature condition. ANALYSES Neumann analysis Consider first the solidification of a liquid layer Which is intially at a uniform temperature Ty. Sud- deny at time ¢ > O the temperature ofthe boundary at y= Ois lowered toa value T, which s below the fusion temperature 7, ofthe material. The physical properties of the two phases are assumed 10 be different but independent of temperature. The bulk temperature of the liquid at a large distance from the solid-liguid Interface is T,and constant. This implies thatthe liquid thickness is effectively infinite, even when it is quite small. If uid motions due to density differences between the two phases and due to the buoyancy forces in the liquid are neglected [1,4], the tempera- ture distribution in the solid and liquid phases are governed by the equations a 2 =a. o0, 4 rae neem TOyt)+ Ty asy (6) and ao Ana aL May, The temperature distribution in the solid phase is sven by [4,18] @,- ~ enya vavertune ve) and inthe liquid phase by O,= 1 —erfe(y 2 /arieries2%) (9) ‘The constant i is determined from the transcendental equation 1 (2). fis ida) Ste \ 2) Vm erttay/ V4) fs ‘T, — T, \exp(— 32/4a, VEG eage and the solid-liquid interface position is given by ay For the melting of a solid the formulation of the problem is similar, and the temperature distribution in the liquid and solid phatoe ie expressed by equations of the same form as equation 9) and (8), respectively. The solid-liquid interface position is given by sit) = ay vf s= Gif. (2) ‘The constant 2 is determined from the transcendental equation 1 (2). Ste 2) Vm erffayv2 Jal] (% —T\erp(- 2/44) Ba Mew Vite) ay Ee : Introduction ofthe dimensiones tme dtd as ‘els ~ TY 28) H ‘hy permits scaling of the Stefan number for the melting problem, for example. Sine thete is wo characteristic length for the one-dimensional diffusion considered here, the height H ofthe test cell was arbitrarily chosen for the purpose of nondimensionalizing the distance and the Fourier number. pe, Solidification from above with natural convection atthe solid-liguid interface Aslongas the system remains stable the assumption that heat conduction is the only mode of energy transfer and therefore the Stefan analysis is appro- priate. The onset of instability and development of natural convection in the liquid phase as a result of bbuoyaney forces due to temperature dilerences would invalidate the model Here we presenta simple analysis to account for natural convection heat transfer atthe solid -liquid interface during solidification from above Tilly, he Liquid temperature 7 is taken to be above the fusion 7;, Since the solid liquid interface is colder than the bulk of the liquid, natural convection may develop in the liquid. The temperature distri ‘bution in the solid is governed by equation (1), and after the critical Rayleigh umber has heen exceeded [18] and natural convection develops the temperature distribution in the liquid phase will become and remain uuifori al Tj. An energy balance at the solid-liquid interface, equation (7), can then be ex- pressed as as_ | ay pad, + WT; — Te) at y = 0,5) Where fis the convective heat transfer coefficient available, for example, from correlations [19] for natural convection heat transfer in a fluid confined between two horizontal parallel rigid wall mn terms of dimensionless variables the governing equation (1) and the initial and boundary conditions ‘can be written as (16) and “ as) 9) 20) &, The nonlinearity introduced by the boundary con- dition, equation (20), precludes a closed form exact analytical solution. The solution of equation (16) with the initial and boundary conditions, equations (17) through (20), can be obtained using numerical [10,20] perturbation (21, 22] or integral [23] methods. We adopt here an integral method of solution ‘AS un approximation for the temperature distsi= bution in the solid layer the functional form for ©,(é,r5) is postulated in terms of two undetermined Functions s*(r,) and gs) sueh that o-4(S)sa—o(S) ‘The function d(s,)is determined by requiring that (,) be independent of ds*/d, [23]. This can be derived it the following way. By differentiating equation (19) with respect to r, we obtain en (2) IN. W, Hue Jn and R. isnot The elimination of dst/de, between equations (20) and (22) and substitution ofthe resulting equation (0 climinate Cyr, {rom equation (16) yields the nec- essary condition for determining (tr,) fou (°Y (2) 21 He Substitution of equation (21) into equation (23) and solution results in =0.03) 2-de[b-o+ Ve FEA 2H whee bers step Now with @ determined, an energy balance at the [Phase change interface boundary, equation (20), yields the following differential equation for the interface position s*(s,) ast set “ 2s) Integration of equation (25) yields ad " 26 | .t@==ra “ Unfortunately, this integral cannot be evaluated analytically because both ¢ and I” depend on s* ina ‘complicated way For the special case of negligible neat capacity inthe solid (Ste, +0}, equation (26) becomes 3 2 In the absence of convection at the solid liquid interface (7 = 0) evaluation ofthe integral in equation 27) gives Vin (8) iFit ie assumed that s# = 0 at «, = 0 -Molting from below with natural convection in the melt It has been experimentally observed [8, 12] and theoretically predicted [24] that a convective in- stability develops in a horizontal melt layer created by a solid-to-liquid phase change. Ifthe Rayleigh number is sufficiently large, natural convection in the melt could significantly affect heat transfer and the interface ‘motion. Here we present a simple analysis to predict the Solid-tiquid interface position during melting of « solid heated from below A solid of sufficient lateral extent so that the edge ‘effets could be neglected ic intially ata temps T, below the fusion temperature T;, At time ¢ > 0 heating is initiated at the lower bounding surface (= 0)as a step increase in the wall temperature 10 8 value T,. During the early stages ofthe phase-change process when the met layer is stable, the temperature ‘oua-nguio pnase-cnange neat cranser distribution and the interface postion are predicted by the Neumann analysis. After the critical Rayleigh number, Ra., and the corresponding critical time, t., hhave boon exceeded, the interface motion i determined. by the convective heat transfer to the phase-change boundary. The energy balance at the interface is still valid, but the conductive heat flux in the liquid — Tey is replaced by the convective heat flux so that the energy balance at the interface becomes, or, MT, T+ bya Where the convective heat transfer coefficient h is available from correlations [19] for natural con: vextion. The temperature isttibution in he sold is Predicted from the energy equation in the solid, ‘equation (I), with appropriate intial and boundary conditions. For time ¢ > f, the temperature distribution in the solid and the phase-change houndary are predicted from the energy equation 20, _ #0, ste (29) 4 idly 5, Gee OEE TE 60) and the following initial and boundary conditions +0) fortyrte ands, G1) Ont agnor 2) and Mu atg=st 3) Because ol the nonlinearity introduced by the boun- dary motion, equation (33), an exact, closed-form analytical solution is not possible. Therefore, an approximate solution will be obtained using an in- tegral method (22, 25] If the energy equation, equation (30), is integrated cover the dimensionless thermal boundary layer thick- ness A, there results 04) {A second order polynomial for the temperature distri- bution in the solid, eg @,= [14-94 — SF, 6s) hich satisfies the conditions O,=10 agast, 26) +0 asgra en and @/2-+0 asg+A, 08) is assumed as an approximation of the profile. Substitution of equation (35) into equation (34) and use of the boundary condition gives 1 3, d §(2_ Vio) waned Si {(52 Yo) - nave} (39) Combining equation (35) with equation (3) yields oat wa Equations (39) and (40) are solved with the initial conditions Staal and Ost Amst ate =e, (Al) where 4, isthe initial penetration depth at r, = t which canbe obtained by matching equation 10} with cauation (35) at r~ 5 Depending on the Raleigh number range, the natural convection heat transfer between two rigid parale! boundaries maintained at diferent temper tures can be expressed as [19] (2) emarre = ann (ay where 4 is independent of s*. The constant C and the exponents m and n are different for the creeping, laminar, transition and turbulent regions. For the special case ofthe solid initially and held atthe fusion temperature, the salution of equation (40) ie (25a) I'm = 1/2, equation (42) indicates that the melt layer thickness will increase linearly with time while the Neumann analysis predicts that the melt layer will grow as /s,. It should be emphasized, however, that ‘equation (42) hay been determined on the basis of steady-siate natural convection heat transfer in the absence of phase-change, and therefore use of the ‘equation in a situation where one ofthe boundaries is moving may not be appropriate (in) (43) RESULTS AND DISCUSSION Qualitative discussion of experimental results ‘Numerous experiments were performed which yild- ced valuable qualitative and quantitative data about the solid-liquid interface motion and temperature distribution during solid-liquid phase-change heat transfer in the materials. A detailed discussion of the rerult is available [26], and thus only a few selected ‘experiments will be highlighted here Figure 2 shows comparison of the interface motion for melting of seasie wed frou above aad below. The solid was purposely not preheated and its initial temperature was over 40°C below the fusion tempera- ture. The solid-liquid interface velocity is seen to be ‘much larger for melting from below than above. For the later the interface motion nearly comes to a halt at 2a IN. W, HAL JR. and R. ViskANTA times. The differences between the two experiments is material resolidfied as a salt hydrate complex. The attributed to the higher heat transfer due to natural material was heated and upon cooling and stirring & convection which occurs when the material is heated large portion of the material did form as crystals but 2 from below but not when the material is heated from layer of water remained on the top. Howaver, after above. In experiments where the heating surface was remelting there was again a layer of solid material facing upward natural convection set in very rapidly deposited on the bottom of the test cell. And alter the but was never observes when the heating surface faced material had been cooled below the fusion tempera: downward. Natural convection manifesed itself in a _ture,only a very small erystal layer formed. Nucleating, form ofeasily observable cells which srew in sizes the and thickening agents have been recommended [27] 10 ‘melt layer thickness increased. The solid-liquid in- aid nucleation and prevent segregation of a Glauber's terface was irregular and provided evidence of cellular salt. Upon addition of sodium tetraborate (borax) only ‘convection, These observations concur with experi- about 25%, of the salt hydrate formed cryetals, and ‘mental results for meting of ice from below [6]. Dut- these crystals were inhomogeneous mixed with water ing melting from above the interface remained flat, _In the next two subsections the experimental results indicating the absence of convection. Furthermore, obtained for n-octadecane (¥9, pure) will be com- convection cells were not observed for this pared with analytical predictions because the physical arrangement. and transport properties are well established [28] 10 Experiments with sodium phosphate dodecahyd- makesuch a comparison meaningful. The physical and rate (Na;HPO, °12H,0, T; = 36°C)revealed that the transport properties of n-octadecane used in the data ‘material began to break down with the frst melting) reduction were taken from Ref. [29]. For stearic acid freezing cycle and stratified into various layers: an- the transport property data are incomplete. No com hydride," dilferent hydrates (believed 10 be parisons of experimental data obtained for sodium NNa,HPO, 12H,0, Na;HOP,-/H,0, NaHPO,- phosphate dodecanydrate and sodium sulfate de: 5H,O, et. but no chemical analysis was made), cthydrate are possible with analysis because of the and water. In addition, observations indicated that nonhomogeneous structure of the materials asa result water was trapped in’ some of the crystal layers, ofsegregation ofthe hydrates into anhydrous salts and ‘making the layers inhomogeneous. For example, a water, bulk motion ofthe crystals during melting and short ime after the second melting cycle was started unpredictable as well as unrepeatable behavior during the layers were: solid anhydride, liquid hydrate, phase change. Despite the promise that the salt dodecahydrate crystals, heptahydrate crystals and hydrates have for low temperature latent heat thermal ‘water. After the underlying dodecahyarace crystal euergy storage, the experience with these materials layer had melted the heptahydrate crystal layer fell to indicate it will be difficult to predict the system the bottom of the test cell and mixed with the liquid. performance due to their complex behavior during This type of unpredictable behavior made it very eyclic melting and freezing difficult to interpret the results quantitatively. The test celland the experimental procedure were not designed Heat transfor under stable conditions to study bulk motion of the solid hydrate crystal layer When a liquid is cooled from below or a solid is after it had lost support atthe cell walls heated from above, the fui layer is stable, and the In the first experiment with sodium sulfawe Ue- Neumann viel for onediaeusional phase-change ccahydrate (NaSO, 10H,O; T; = 324°C), the hy- heat transfer should be appropriate. Figure 3 shows a rate was completely solid before heating from below was started. However, as the melting proceeded, °° pug avaiis Sieg a anhydride was deposited atthe bottom ofthe test eel SOs ‘When the cell was cooled only a small fraction of the a - 2 ors o Va o . < ow ° oot A Co wot 4 ae ct 2 BERET. ee Care Bees es. Q + ar ee Jaen a 1 G0 Ce ne ae A ® Pa Comparion of measur ant peice ‘mode soid-liquid interface positions during sb Fic. 2. Variation of solid-tigui imerface position with time of n-octadecane fom below (3) Ste, ‘during melting ofsearie acid from above and below: and (b) Ste, = 00725 (f, Sovia-tguta poasechange Hen ear comparison of the solid-liquid interface position predicted from the Neumann analysis with experimen- tal data during solidification, For Ste, = 0.0725 the ‘analysis predicts the phase change boundary which ‘overlaps the curve for Ste, = 0.0685 and separate lines ‘could not be drawn. At early times (r, <0004) the analysis predicts a faster rate of solidification than experimentally observed, but at later times there is ood agreement between predictions and data. A comparison of the predicted temperature variation with experimental data at a few selected locations is. sven in Fig. 4. Again, there i substantial agreement between the two results. An increase in interfacial velocity was obtained for solidification of hexadecane at later time [14]. This Was attributed to the departure from one-dimensional analysis and the neglect of heat sain from the environment as the interface approached a cooled top plate. ‘The predicted and measured solid-liguid interface positions for melting from above are compared i Fig, 5. Separate curves could not be clearly drawn for the {wo Stefan numbers, The agreement between data and analysis is reasonably good for early umes (r,< 2008), “| xp bate ost Noor maces noone { oo al xO of dy fe of aa Fic. 4. Comparison of measured and predicted temy Aistributions during solidification of moctadecane low: Ste, = 00728 (7, = 202°C), yan Fie, 5. Comparison of measured end predicted (Nouann, model) solid lguid interface positions for meling of m- ‘ctadecane from above! (a) Ste,= 0175 (T, = 434°C) and 1b) Ste,= 0180 (T, = 440°C), butat later times the observed melt layer thickness was greater than that predicted. The fluid motion due to the density differences of the two phases could possibly account for part of the diserepaney, but this effect is ‘expected to be relatively small (7, 30}. The primary reason for the discrepancy between data and pre- dictions at later time (c; > 0008) i believed to be due to weak natural convection which developed as 4 result of heat conduction along the vertical est cell walls, Heat transfer under unstable conditions ‘A comparison of the experimental data with pre-

    You might also like