Available online at www.sciencedirect.

com

Applied Thermal Engineering 28 (2008) 16761686 www.elsevier.com/locate/apthermeng

Natural convection heat transfer coecients in phase change material (PCM) modules with external vertical ns
Albert Castell a, Cristian Sole a, Marc Medrano a, Joan Roca a, Luisa F. Cabeza a,*, Daniel Garca b
a

` Departament dInformatica i Enginyeria Industrial, Universitat de Lleida, Pere de Cabrera s/n, 25001 Lleida, Spain b ` Departament Projectes dEnginyeria, Universitat Politecnica de Catalunya, Colom 11, 08222 Terrassa, Spain Received 9 May 2006; accepted 4 November 2007 Available online 17 November 2007

Abstract To determine the heat transfer coecient by natural convection for specic geometries, experimental correlations are used. No correlations were found in the literature for the geometries studied in this work. These geometries consisted of a cylindrical module of 88 mm of diameter and 315 mm height with external vertical ns of 310 mm height and 20 and 40 mm length. To determine the heat transfer coefcient by natural convection, experimental work was done. This module, containing PCM (sodium acetate trihydrate), was situated in the middle upper part of a cylindrical water tank of 440 mm of diameter and 450 mm height. The calculated heat transfer coecient changed by using external ns, as the heat transfer surface was increased. The temperature variation of the PCM and the water are presented as a function of time, and the heat transfer coecient for dierent ns is presented as a function of the temperature dierence. Experimental correlations were obtained, presenting the Nusselt number as a function of dierent dimensionless numbers. Dierent correlations were analysed to nd which one t better to the experimental data. 2007 Elsevier Ltd. All rights reserved.
Keywords: Nusselt number; Rayleigh number; Experimental correlation; Heat transfer enhancement; Phase change material (PCM)

1. Introduction Thermal energy storage systems are an important requirement for many applications due to the non-coincidence of heat demand and supply or availability. One of the typical examples of such mismatch is solar energy. Among the thermal energy storage concepts, latent heat thermal storage using PCMs is regarded as a promising technology. Their use in domestic hot water (DHW) tanks would keep hot water for a longer time. In such a system, a lot of energy can be stored as latent heat, but it should be able to be transferred from the PCM to the water when

Corresponding author. Tel.: +34 973003576; fax: +34 973003575. E-mail address: lcabeza@diei.udl.es (L.F. Cabeza).

needed, therefore heat transfer within the PCM and to the water is of high interest [1,2]. There are several methods to enhance the heat transfer in a latent heat thermal store. The use of ns inside the PCM has been extendedly studied. These ns can be axial or radial and are usually attached to the tubes. In this case the most important part is the formulation of phase change problems. Several theoretical techniques have been developed, such as the enthalpy method by Ismail [35], the Landau transform method associated with the nite volumes method, the interface immobilization method together with the nite volumes approach, or the integral energy method [5]. Ismail [3,4] presented a comprehensive review of literature on the subject as well as the results of many experimental and numerical studies on phase change heat transfer into and around simple and complex geometries.

1359-4311/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.applthermaleng.2007.11.004

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

1677

Nomenclature Acond inner conduction heat transfer area (m2) Atransfer convection heat transfer area from the modules to the water (m2) Cp specic heat (kJ/(kg K)) dT distance between thermocouples position in the experimentation work (m) DPCM PCM module diameter (m) Dt tank diameter (m) g gravity acceleration (m/s2) gbT surface T close L3 PCM Gr Grashof number, v2 hPCM convection heat transfer coecient (W/(m2 K)) kPCM PCM thermal conductivity (W/(m K)) LPCM PCM module height (m) Lt tank height (m) Nn number of ns Nu Nusselt number, hL k lC Pr Prandtl number, k p q heat transfer rate from the PCM to the water in the experimental work (W) Ra Rayleigh number, =Gr Pr W fin Rae eective Rayleigh number, Gr Pr LPCM rPCM PCM module radius (m) rPCM,1/4 radius of the quarter of the module (m) s distance between two adjacent ns (m) T temperature (C) Tf lm temperature (C) TPCM,1/2 temperature in the middle of the PCM module (C) TPCM,1/4 temperature in a quarter of the PCM module (C) Tsurface temperature of the surface of the PCM module (C) Tclose temperature of the water close to the PCM module (C) Wn n width (m) Greek symbols b volumetric coecient of thermal expansion (C1) DT temperature dierence, =Tsurface Tclose (C) T T close U dimensionless number of T, T surface T close l dynamic viscosity (kg/(m s)) t kinematic viscosity (m2/s) p Pi number q density (kg/m3)

Mehling et al. [68] and Py et al. [9] proposed a graphitecompound-material, where the PCM is embedded inside a graphite matrix. When using these PCM-graphite composites inside metal modules, the heat barrier is the heat transfer from the metal container to the water [8]. By using ns the heat transfer area is extended and the coecient of heat transfer by natural convection changes, improving heat transfer form the container to the water. Fins geometry is an important parameter when considering the addition of ns in a PCM module. In a vertical module two dierent ns geometries can be considered: horizontal and vertical ns. There is much more literature for horizontal ns [1022], but this geometry interferes with the natural convection in the PCM module. On the other hand there is no literature for vertical ns around circular vertical tubes, but this geometry would improve natural convection in the water side of the PCM module. Because of this, in this work external vertical ns were used to increase the heat transfer from the PCM to the water. There were no literature references for vertical and cylindrical modules with external and vertical ns. Heat transfer coecient for natural convection is typically determined using experimental correlations. Lots of specic geometries have been studied and correlations are available in the literature [23,24]. In this work an experimental set-up was used to evaluate the natural convection heat transfer coefcient for two specic geometries. Correlations of Nusselt number as a function of Rayleigh and eective Rayleigh number were obtained.

2. Experimental work 2.1. Experiments done To determine the eect of adding vertical ns to the external part of the module, some experiments were conducted. The PCM-graphite composite used was sodium acetate trihydrate with graphite (90:10 vol.%). This product has a melting point of 58 C, melting enthalpy between 180 and 200 kJ/kg, density between 1.350 and 1.400 kg/m3, heat capacity of 2.5 kJ/kg K, and thermal conductivity between 2 and 5 W/m K. Three dierent modules were used: one without ns, another one with small ns (20 mm length) and the last one with big ns (40 mm length). Each module had eight external and vertical ns of 2 mm of thickness, providing a 28.45% and 44.28% increase of the heat transfer surface, respectively, compared to the reference module without ns. Fig. 1 shows the considered conguration. The PCM module was 88 mm of diameter and 310 mm of height with a thickness of 2 mm. The experimental work reproduced the behaviour of the PCM modules in a DHW tank. The dimensions of the tank used were 440 mm of diameter and 450 mm of height. The tank was insulated using an elastomeric insulating material of 10 mm of thickness. Five K-type thermocouples and a data-logger (STEP DL01-CPU) instrumented the experimental set-up. Two thermocouples were located inside the PCM module (one in the centre and the other at half

1678

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

Fig. 3. Instrumentation of the experiments.

2.2. Natural convection heat transfer coecient Once the experimental work was done, the heat transfer coecient by natural convection for this specic geometry could be calculated. To determine this coecient a onedimensional study was done in order to simplify the calculations. A three-dimensional behaviour could be expected because of the stratication of the water, but since the experiments were very short it would be negligible. In future works, the stratication in the water tank when using encapsulated PCM inside (considering dierent distributions and geometries) will be studied. The procedure used to determine the heat transfer coefcient by natural convection was the following. 2.2.1. Calculation of the heat transfer rate The PCM (sodium acetate trihydrate) was mixed with graphite in a composite. Therefore, during all the experimen-

Fig. 1. Module with external vertical ns.

distance between the rst one and the metal container). The other ones were situated inside the water, outside the PCM module, one of them in contact with the external surface of it. The distance between the inside water thermocouples was 50 mm, and all thermocouples were at 135 mm distance from the top of the tank. Figs. 2 and 3 illustrate the instrumentation of the experiments. The experimental work consisted of introducing the module containing melted PCM at 70 C into the cold water tank to evaluate the heat transfer phenomenon. The experiment was stopped when PCM and water temperatures were the same.

(1) (2)(3) (4)

(5)

(1) TPCM,1/2

(2) TPCM,1/4

(3) Tsurface

(4) Tclose

(5) Tfar

Fig. 2. Diagram of the experimental set-up.

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

1679
W fin LPCM

tal processes the mixture was not melted inside the PCM module, remaining in solid phase; only the sodium acetate trihydrate went through the melting/solidifying process, but not aecting the solid structure of the composite. Therefore, the heat transfer rate inside the PCM module could be determined using conduction equations (Fourier law). In this work, the transient eect was considered discretizing the data over time, therefore, the heat ux could be calculated using the following equation: q k PCM Acond T PCM;1=2 T surface 1:a where Acond 2 p LPCM rPCM Ln 0:003 1:b

and LW fin 0:063; for 20 mm ns, PCM ns. 2.4. Rayleigh correlations

0:127; for 40 mm

Dierent correlations for Nusselt number were studied and compared with each other to achieve the best t with the experimental results. Three dimensionless numbers (Rayleigh, Prandtl, Grashof) and several combinations of them were used. The most signicant and used dimensionless numbers in natural convection systems and their combination were studied. The uid properties are a function of the temperature and were evaluated at the registered temperature. The following equations were used to calculate them: Gr g b T surface T close L3 PCM m2 1 b Tf T surface T close Tf 2 lH2 O C p PrH2 O k H2 O Ra Gr Pr Raeff Gr Pr W fin LPCM 3:a 3:b 3:c 3:d 3:e 3:f

Note that in the equations above the thermal conductivity of the material is considered constant with time and temperature. Although this assumption might give an error to the calculation, especially during phase change, when the PCM is mixed with graphite the driving force in the conduction is the graphite, therefore the error is low enough. 2.2.2. Calculation of the natural convection heat transfer coecient Once the heat transfer rate was calculated, the natural convection heat transfer coecient could be determined using Newton law q hPCM 2:a Atransfer T surface T close where Atransfer 2 p rPCM LPCM for the module without ns, and Atransfer 2 p rPCM LPCM 2 N fin W fin LPCM for the modules with vertical ns. 2.3. Natural convection heat transfer coecient correlations From the experimental data, experimental correlations were determined for each specic geometry. These equations provided the heat transfer coecient for natural convection as a function of the temperature dierence of the system. The dening dimensionless numbers of the problem, considering its geometry and boundary conditions, were DPCM s Dt 0:284; 0:11; 1:397; LPCM LPCM LPCM Lt x y 1:429; X ; Y LPCM LPCM LPCM T T close U T surface T close g b T surface T close L3 l Cp PCM Gr ; Pr m2 h 2:c 2:b

In Eq. (3f), Rayleigh number is modied by a ratio of n width by n height. These modications introduced in the dimensionless number a measure of using dierent ns. 3. Results The experimental results showed an increase in the heat transfer rate when using PCM modules with vertical ns. This eect can be measured by the time needed by the modules to heat the water. Fig. 4a shows the time needed to heat the water due to the PCM phase change. To cool down the PCM from 60 to 45 C (including the solidication of the material) using a PCM module without ns the time needed was about 17 min. When using 20 mm ns, to achieve the same temperature decrease as in the experiments without ns, the time necessary was about 13 min (a reduction of 23.53%). Fig. 4b shows the heating, and melting and cooling down processes of the water and the PCM, respectively. Finally, PCM modules with 40 mm ns reduced the cooling down time to 7 min for the same temperature decrease (a reduction of 58.82%). Fig. 4c shows the heating and cooling down process for these experiments. To determine the heat transfer coecient a one-dimensional study was done. The stratication of the water was neglected in this work. In future works, the stratication of the water tank when using PCM inside will be studied. The heat transfer coecient for natural convection for each PCM module geometry is compared in Fig. 5 as a func-

1680
70 65 60 55

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

WATER CLOSE SURFACE MODULE PCM 1/4 PCM 1/2 WATER FAR

Temperature (C)

50 45 40 35 30 25 20 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600

Time (s)

Fig. 4a. Water and PCM temperature over time of experimental work using a PCM module without ns.
70 65 60 55

WATER CLOSE SURFACE MODULE PCM 1/4 PCM 1/2 WATER FAR

Temperature (C)

50 45 40 35 30 25 20 0 200 400 600 800 1000 1200 1400 1600 1800

Time (s)

Fig. 4b. Water and PCM temperature over time of experimental work using a PCM module with vertical external ns of 20 mm length.

tion of temperature dierence DT = Tsurface Tclose. When using 20 mm ns, the temperature dierence DT necessary to achieve the maximum heat transfer coecient (h = 179 W/m2 K with no ns; h = 177 W/m2 K with 20 mm ns) experienced a threefold decrease (DT = 3.5 C with no ns; DT = 1.1 C with 20 mm ns). On the other hand, using 40 mm ns the temperature dierence had a fourfold decrease (DT = 0.8 C), but the maximum heat transfer coecient decreased too (h = 159 W/m2 K). Figs. 6a and 6b represent the natural convection heat transfer coecient as a function of the temperature dierence DT, for 20 and 40 mm ns, respectively. The experimental correlations were divided in three parts to achieve the best t with the experimental data. The valid range of temperature dierences was from 0.7 to 9.2 C for 20 mm ns geometry, and from 0.2 to 13.7 C for 40 mm ns.

For 20 mm ns, the experimental correlation is 8 > hPCM 490:6 DT 2 1249 DT 619 0:7 6 DT 6 1:3 < h 35:7 DT 226 1:3 < DT 6 3:7 > PCM : hPCM 1:35 DT 2 24 DT 163:7 3:7 6 DT 6 9:2 4:a For 40 mm ns, the experimental correlation is 8 > hPCM 232:6 DT 2 376 DT 16:5 0:2 6 DT 6 0:8 < h 7:8 DT 2 64:3 DT 185 0:8 < DT 6 3 > PCM : hPCM 0:33 DT 2 9:2 DT 83 3 < DT 6 13:7 4:b In the rst region the heat transfer coecient increases with DT. The maximum value for the coecient is achieved at

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

70 65 60 55

1681

WATER CLOSE SURFACE MODULE PCM 1/4 PCM 1/2 WATER FAR

Temperature (C)

50 45 40 35 30 25 20 0 200 400 600 800 1000 1200 1400 1600 1800 2000

Time (s)

Fig. 4c. Water and PCM temperature over time of experimental work using a PCM module with vertical external ns of 40 mm length.

200

Natural convection heat transfer coefficient (W/m2K)

180

No Fins
160 140 120 100 80 60 40 20 0 0 2 4 6 8 10 12 14 16

Small Fins Big Fins

Temperature difference (C)

Fig. 5. Comparison of the heat transfer coecient for natural convection for dierent PCM modules.

the intersection point between rst and second region. For higher values of temperature dierences the heat transfer coecient decreased to the half of the maximum value. Finally, in the third region the coecient described a parabolic behaviour below values the half of the maximum. For both cases, using 20 or 40 mm ns, the rst and third region were approached with a polynomial regression. In the second region, when using 20 mm ns a lineal regression is good enough, while for 40 mm ns a polynomial regression is necessary. Several experimental correlations representing Nusselt number as a function of dierent dimensionless numbers were determined. As showed before, there were three dierent regions in the heat transfer coecient behaviour, so

Nusselt number should be evaluated in each region individually. First analysis consisted in plotting Nusselt number as a function of Rayleigh number for each geometry, as showed in Figs. 7a and 7b. The obtained correlations were, for 20 mm ns: Nu 4 1022 Ra2 8 1010 Ra 308 5 1011 6 Ra 6 1 1012 Nu 3 1024 Ra2 4 1011 Ra 127:6 1 1012 < Ra 6 3:6 1012 Nu 4 1025 Ra2 7 1012 Ra 61:4 3:6 1012 6 Ra 6 6:5 1012

5:a

1682
200

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

Natural convection heat transfer coefficient (W/m2K)

180 160 140 120 100 80 60 40 20 0 0 1

y = -490,58x2 + 1248,8x - 618,9 R2 = 0,8409 y = -35,668x + 225,99 R2 = 0,9772

First region (20 mm fin) Second region (20 mm fin) Third region (20 mm fin)

y = 1,3527x2 - 24,174x + 163,74 R2 = 0,9815

10

Temperature difference (C)

Fig. 6a. Experimental correlation for natural convection coecient for a PCM module with 20 mm ns.

180

Natural convection heat transfer coefficient (W/m2K)

160

y = -232,63x2 + 375,92x - 16,52 R2 = 0,6416

First region (40 mm fin) Second region (40 mm fin) Third region (40 mm fin)

140

120
y = 7,7697x - 64,279x + 184,88 R2 = 0,8804
2

100

80
y = 0,3303x2 - 9,1883x + 83,13 R2 = 0,8269

60

40

20

0 0 2 4 6 8 10 12 14

Temperature difference (C)

Fig. 6b. Experimental correlation for natural convection coecient for a PCM module with 40 mm ns.

For 40 mm ns: Nu 2 1022 Ra2 2 1010 Ra 4:7 7:1 1010 6 Ra 6 6 1011 Nu 6 10

11 24

Nusselt number as a function of an eective Rayleigh number (Rae) was also studied, as showed in Figs. 8a and 8b. The experimental correlations were, for 20 mm ns: Nu 1 1011 Raeff 98
2

8 1012 6 Raeff < 1:45 1013

Ra 4 10
12

11

Nu 8 1027 Raeff 2 1012 Raeff 122 Ra 91:5 1:45 1013 6 Raeff 6 6 1013 Nu 1 1027 Raeff 2 5 1013 Raeff 61:4 6 1013 6 Raeff 6 1 1014 5:b 6:a

6 10 < Ra < 3 10

Nu 4 1012 Ra 35

3 1012 6 Ra 6 7:5 1012

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

100 90 80 70 60

1683

y = -4E-22x2 + 8E-10x - 307,99 2 R = 0,792

First region (20 mm fin) Second region (20 mm fin) Third region (20 mm fin)

y = 3E-24x2 - 4E-11x + 127,65 R2 = 0,9831

Nusselt

50 40 30 20 10 0 0,00E+00

y = 4E-25x2 - 7E-12x + 61,468 R2 = 0,9908

1,00E+12

2,00E+12

3,00E+12

4,00E+12

5,00E+12

6,00E+12

7,00E+12

Rayleigh

Fig. 7a. Experimental correlation of Nusselt number in function of Rayleigh number for a module with 20 mm ns.

90

80

70

y = -2E-22x + 2E-10x - 4,6824 2 R = 0,6395

First region (40 mm fin) Second region (40 mm fin) Third region (40 mm fin)

60

Nusselt

50

y = 6E-24x - 4E-11x + 91,521 2 R = 0,8871

40

30

y = -4E-12x + 35,128 2 R = 0,7192

20

10

0 0,00E+00

1,00E+12

2,00E+12

3,00E+12

4,00E+12

5,00E+12

6,00E+12

7,00E+12

8,00E+12

Rayleigh

Fig. 7b. Experimental correlation of Nusselt number in function of Rayleigh number for a module with 40 mm ns.

For 40 mm ns: Nu 3 10 Raeff 3 10 5 1011 6 Raeff 6 5 1012

2 24 2 11

Raeff 4:7

Nu 1 1025 Raeff 5 1012 Raeff 93 5 1012 < Raeff < 2 1013 Nu 3 1028 Raeff 5 1013 Raeff 35 2 1013 6 Raeff 6 6 1013
2

6:b

of Rayleigh and eective Rayleigh numbers the Nusselt number decreased to values lower than half of the maximum. Finally, in the third region the Nusselt number described a parabolic behaviour obtaining values lower than a third part of the maximum. For both cases, using 20 or 40 mm ns, the obtained correlations were very similar. Also comparing Rayleigh and eective Rayleigh the results were similar. A polynomial regression was used in most cases. 4. Discussion The increase of the heat transfer rate was a result of the increase of the heat transfer area and the lower temperature

Both correlations showed a similar behaviour. In the rst region Nusselt number increased with Rayleigh and eective Rayleigh numbers, respectively. The maximum value for the Nusselt number was achieved at the intersection point between rst and second region. For bigger values

1684
120

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686

y = 1E-11x - 98,325 2 R = 0,7628

100

First region (20 mm fin) Second region (20 mm fin) Third region (20 mm fin)
2

80

Nusselt

y = 8E-27x - 2E-12x + 121,93 2 R = 0,9767

60

40

y = 1E-27x2 - 5E-13x + 61,468 2 R = 0,9908

20

0 0,00E+00

2,00E+13

4,00E+13

6,00E+13

8,00E+13

1,00E+14

1,20E+14

Effective Rayleigh

Fig. 8a. Experimental correlation of Nusselt number in function of Rayleigh eective number for a module with 20 mm ns.

90 80 70 60

y = -3E-24x2 + 3E-11x - 4,6824 2 R = 0,6395

First region (40 mm fin) Second region (40 mm fin) Third region (40 mm fin)

Nusselt

50 40 30 20 10 0 0,00E+00

y = 1E-25x - 5E-12x + 93,06 2 R = 0,8763

y = 3E-28x - 5E-13x + 35,244 2 R = 0,7332

1,00E+13

2,00E+13

3,00E+13

4,00E+13

5,00E+13

6,00E+13

7,00E+13

Effective Rayleigh

Fig. 8b. Experimental correlation of Nusselt number in function of Rayleigh eective number for a module with 40 mm ns.

dierence necessary to achieve the same heat transfer coefcient for natural convection. The heat transfer coecient was not increased by using vertical ns. When using small ns, a lower temperature dierence was necessary to achieve the same heat transfer coecient as with no ns. Therefore, the needed time to solidify the PCM decreased. The increase of the heat transfer area resulted in an increase of heat transfer rate. When using big ns, the heat transfer coecient was lower. The increase of the ns width may have interfered the natural convection. Nevertheless, the needed time to solidify the PCM was also reduced because of the increase of the heat transfer area.

The increase of the heat transfer rate obtained by using vertical ns could be very useful for applications of PCM modules inside water tanks. These PCM modules are used to store energy in a reduced volume. Using modules with vertical ns could solve the problem of slow heat transfer rate from the PCM to the water and increase the availability of the energy. The storage system would be more exible to match the energy demand. The behaviour of the system with vertical ns was well dened using both correlations, Rayleigh and eective Rayleigh. There were no signicant dierences in the quadratic mean dierence or the complexity of the function describing the system, as showed in Table 1. The accuracy

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686 Table 1 Correlations for modules with vertical ns Experiment Region Correlation (y vs. x) na a b c r2b Valid range Lowc 20 mm n First region hPCM vs. DT Nu vs. Ra Nu vs. Rae hPCM vs. DT Nu vs. Ra Nu vs. Rae hPCM vs. DT Nu vs. Ra Nu vs. Rae hPCM vs. DT Nu vs. Ra Nu vs. Rae hPCM vs. DT Nu vs. Ra Nu vs. Rae hPCM vs. DT Nu vs. Ra Nu vs. Rae 14 490.6 4 1022 1249 8 1010 1 1011 35.7 4 1011 2 1012 24 7 1012 5 1013 376 2 1010 3 1011 64.3 4 1011 5 1012 9.2 4 1012 5 1013 619 308 98 226 127.6 122 163.7 61.4 61.4 16.5 4.7 4.7 185 91.5 93 83 35 35 0.84 0.79 0.76 0.98 0.98 0.98 0.98 0.99 0.99 0.64 0.64 0.64 0.88 0.89 0.88 0.83 0.72 0.73 P0.7 P5 1011 P 8 1012 >1.3 >1 1012 P1.45 1013 P3.7 P3.6 1012 P6 1013 P0.2 P7.1 1010 P5 1011 >0.8 >6 1011 >5 1012 >3 P3 1012 >2 1013 Highd

1685

Second region

15 3 1024 8 1027 1.35 4 1025 1 1027 232.6 2 1022 3 1024 7.8 6 1024 1 1025 0.33 3 1028

Third region

13

61.3 61 1012 <1.45 1013 63.7 63.6 1012 66 1013 69.2 66,5 1012 61 1014 60.8 66 1011 65 1012 63 <3 1012 62 1013 613.7 67.5 1012 66 1013

40 mm n

First region

73

Second region

54

Third region

30

Y = ax2 + bx + c. a n ? Number of experimental data. b 2 r ? Quadratic mean dierence. c Low ? Lower value of the rang of application. d High ? Higher value of the rang of application.

of the correlations was not aected by the n width. Additional experimental work using ns with dierent width should be done in order to determine a unique correlation that represented all the studied cases. In this correlation, the modifying factor LW fin should be important. PCM To simplify the calculations, the authors recommend using the Nusselt over Rayleigh correlation. The accuracy was the same as when using Nusselt over eective Rayleigh correlation, but it was easier to determine the Rayleigh number. In conclusion, the use of external ns in PCM modules reduced the time necessary for the heat transfer to the surrounding water. The temperature dierence necessary to achieve a certain value of the heat transfer coecient by natural convection was also reduced. The bigger the ns were, the faster was the heat transfer process, but the heat transfer coecient was reduced. Finally, the authors recommend the use the Nusselt over Rayleigh correlation in order to simplify the calculations achieving the same precision in the results. References
[1] L.F. Cabeza, M. Ibanez, C. Sole, J. Roca, M. Nogues, Experimentation with a water tank including a PCM module, Solar Energy Materials and Solar Cells 90 (2006) 12731282. [2] M. Ibanez, L.F. Cabeza, C. Sole, J. Roca, M. Nogues, Modelization of a water tank including a PCM module, Applied Thermal Engineering 26 (2006) 13281333. [3] K.A.R. Ismail, Modeling of Thermal Processes; Fusion and Solidication, ISBN 85-900609-3-4, The State University of Campinas, Brasil, 1999 (in Portuguese).

[4] K.A.R. Ismail, Cold Storage Systems; Fundamentals and Modelling, ISBN 85-900-609-2-6, The State University of Campinas, Brasil, 1998 (in Portuguese). [5] K.A.R. Ismail, C.L.F. Alves, M.S. Modesto, Numerical and experimental study on the solidication of PCM around a vertical axially nned isothermal cylinder, Applied Thermal Engineering 21 (2001) 5377. [6] H. Mehling, S. Hiebler, F. Ziegler, Latent heat storage using a PCMgraphite composite material: advantages and potential applications, in: Proceedings of the 4th Workshop of IEA ECES IA Annex 10, Bendiktbeuern, Germany, 1999. [7] H. Mehling, S. Hiebler, F. Ziegler, Latent heat storage using a PCMgraphite composite material, in: Proceedings of Terrastock 2000 8th International Conference on Thermal Energy Storage, Stuttgart, Germany, 2000, pp. 375380. [8] L.F. Cabeza, H. Mehling, S. Hiebler, F. Ziegler, Heat transfer enhancement in water when used as PCM in thermal energy storage, Applied Thermal Engineering 22 (2002) 11411151. [9] X. Py, R. Olives, S. Mauran, Paran/porous-graphite-matrix composite as a high and constant power thermal storage material, International Journal of Heat and Mass Transfer 44 (2001) 2727 2737. [10] Esmail M.A. Mokheimer, Performance of annular ns with dierent proles subject to variable heat transfer coecient, International Journal of Heat and Mass Transfer 45 (2002) 36313642. [11] D.R. Harper, W.B. Brown, Mathematical Equations for Heat Conduction in the Fins of Air-Cooled Engines, National Advisory Committee for Aeronautics, Report No. 158, 1922. [12] A. Ullmann, H. Kalman, Eciency and optimized dimensions of annular ns of dierent cross-section shapes, International Journal of Heat and Mass Transfer 32 (6) (1989) 11051110. [13] R. Karaback, The eect of n parameter on the radiation and free convection from a nned horizontal cylindrical heater, Energy Conversion and Management 33 (11) (1992) 9971005. [14] A. Campo, L. Harrison, Prediction of safe tip temperature in uniform annular ns for the design of thermal exchange equipment via

1686

A. Castell et al. / Applied Thermal Engineering 28 (2008) 16761686 dependence, Journal of Heat Transfer-Transactions of the ASME 108 (1986) 457459. A. Campo, R.E. Stue, Symbolic mathematics for calculation of thermal eciencies and tip temperatures in annular ns of uniform thickness, International Journal of Heat and Mass Transfer 40 (2) (1996) 490492. Lien-Tsaiyu, Chao Kuang Chen, Application of the Taylor transformation to the transient temperature response of an annular n, Heat Transfer Engineering 20 (1) (1999) 7887. J.R. Turnpenny, D.W. Etheridge, D.A. Reay, Novel ventilation cooling system for reducing air conditioning in buildings. Part I: testing and theoretical modelling, Applied Thermal Engineering 20 (2000) 10191037. E. Hahne, D. Zhu, Natural convection heat transfer on nned tubes in air, International Journal of Heat and Mass Transfer 37 (Suppl. 1) (1994) 5963. F. Farhadi, N. Davani, P. Ardalan, New correlation for natural convection of nned tube A-type air cooler, Applied Thermal Engineering 25 (2005) 30533066.

[15]

[16]

[17]

[18]

[19]

symbolic mathematics, International Communications in Heat and Mass Transfer 21 (4) (1994) 531538. P. Razelos, K. Imre, The optimum dimensions of circular ns with variable thermal parameters, Journal of Heat Transfer-Transactions of the ASME 102 (1980) 420425. K. Laor, H. Kalman, Performance and optimum dimensions of dierent cooling ns with a temperature dependent heat transfer coecient, International Journal of Heat and Mass Transfer 39 (9) (1996) 19932003. S.M. Zubair, A.Z. Al-garni, J.S. Nizami, The optimal dimensions of circular ns with variable prole and temperature-dependent thermal conductivity, International Journal of Heat and Mass Transfer 39 (16) (1996) 34313439. H.C. Unal, Determination of the temperature distribution in an extended surface with a non-uniform heat transfer coecient, International Journal of Heat and Mass Transfer 28 (12) (1985) 22702284. A.K. Sen, S. Trinh, An exact solution for the rate of heat transfer from rectangular n governed by a power law-type temperature

[20]

[21]

[22]

[23]

[24]