Você está na página 1de 11

Solar Energy Vol. 70, No. 1, pp.

51–61, 2001
 2001 Published by Elsevier Science Ltd
Pergamon PII: S0038 – 092X( 00 )00112 – 2 Printed in Great Britain
0038-092X / 01 / $ - see front matter
www.elsevier.com / locate / solener

AN ANALYSIS OF ISOTHERMAL PHASE CHANGE OF PHASE CHANGE


MATERIAL WITHIN RECTANGULAR AND CYLINDRICAL CONTAINERS
B. ZIVKOVIC and I. FUJII †,1
Meiji University, Department of Mechanical Engineering, 1-1-1 Higashi-Mita, Tama-ku, 214-8571
Kawasaki, Japan

Received 10 January 2000; revised version accepted 7 June 2000

Communicated by ERICH HAHNE

Abstract—In this paper, a simple computational model for isothermal phase change of phase change material
(PCM) encapsulated in a single container is presented. The mathematical model was based on an enthalpy
formulation with equations cast in such a form that the only unknown variable is the PCM’s temperature. The
theoretical model was verified with a test problem and an experiment performed in order to assess the validity
of the assumptions of the mathematical model. With very good agreement between experimental and
computational data, it can be concluded that conduction within the PCM in the direction of heat transfer fluid
flow, thermal resistance of the container’s wall, and the effects of natural convection within the melt can be
ignored for the conditions investigated in this study. The numerical analysis of the melting time for rectangular
and cylindrical containers was then performed using the computational model presented in this paper. Results
show that the rectangular container requires nearly half of the melting time as for the cylindrical container of
the same volume and heat transfer area.  2001 Published by Elsevier Science Ltd.

1. INTRODUCTION Ghoneim, 1989). Using the shell-and-tube model


for the LHES unit, Morrison and Abdel-Khalik
Effective and economic thermal energy storage of
(1978) performed a long-term analysis of air-
a daily surplus of irradiated solar energy is an
based (air as HTF) and liquid-based (water as
unavoidable necessity for the efficient use of solar
HTF) solar heating systems, assuming that the
energy for heating purposes (Duffie and Beck-
thermal conductivity of the PCM in the direction
man, 1991). Among the various methods of
of the HTF flow, as well as the thermal resistance
energy storage, latent heat thermal energy storage
of the PCM in the direction normal to the HTF
is particularly attractive. The motivation for using
flow, can both be ignored. In a later analysis,
phase change materials (PCM) is their high
Ghoneim (1989) took into account both the
energy storage density and their ability to provide
conduction within the PCM in the direction of the
heat at a constant temperature (Abhat, 1983).
HTF flow and the direction normal to the HTF
In order to perform a long-term performance
flow and showed that a substantial error in
analysis of a specified solar heating system, an
estimating the solar fraction (amount of total
adequate model of the heat storage unit is needed
heating load supplied by solar energy) was intro-
(Klein et al., 1976) which, naturally, depends on
duced by neglecting the conduction within the
its design. A survey of the previously published
PCM. However, the common conclusion from
papers dealing with latent heat storage reveals that
both works is that the reduction in storage volume
the most intensively analyzed latent heat energy
by using PCM is not as nearly as pronounced for
storage (LHES) unit is the shell-and-tube LHES
liquid-based systems as it is for air-based systems.
unit with the PCM filling the shell and the heat
It is, therefore, preferable to use a LHES unit
transfer fluid (HTF) flowing through the tubes
coupled with an air-based system. For that reason,
(Lacroix, 1993; Bansal and Buddhi, 1992; Esen et
air is considered as the HTF in the subsequent
al., 1998; Soma and Dutta, 1993; Ismail and
analysis.
Alves, 1986) or, vice versa, PCM filling the tube
The problem of the phase change of PCMs falls
and HTF flowing parallel to it (Esen et al., 1998;
into the category of moving boundary problems.
When the PCM changes state, both liquid and

Author to whom correspondence should be addressed. Tel. / solid phases are present and they are separated by
fax: 181-44-943-7395; e-mail: fujii@isc.meiji.ac.jp the moving interface between them. There have
1
ISES member. been many different numerical methods de-

51
52 B. Zivkovic and I. Fujii

veloped for dealing with the phase change prob- accounted for (Voller et al., 1987; Lacroix, 1993;
lem (Bansal and Buddhi, 1992; Soma and Dutta, Soma and Dutta, 1993). While in most works,
1993; Murray and Landis, 1959) of which the convective heat transfer from the HTF to the
most attractive and commonly used are the so- PCM is calculated through the mean value of the
called enthalpy methods (Voller and Cross, 1981; heat transfer coefficient (aconv ), some authors
Voller, 1985; Voller, 1990). The major reason for solved the problem of phase change coupled with
this is that the method does not require explicit transient convective heat transfer between HTF
treatment of the conditions on the phase change and PCM, i.e. with the complete Oberbeck’s set
boundary (cf. Carslaw and Jaeger, 1959), i.e. there of equations solved for the HTF (Belleci and
is no need for tracking the phase change boundary Conti, 1993; Trp et al., 1999).
throughout the phase change domain. However, In the present analysis, the mathematical model
beside the fact that implicit finite difference for phase change of the encapsulated PCM is
discretization results in a set of nonlinear equa- derived under the following assumptions:
tions, the method has several other drawbacks. (A) Thermal conductivity of the PCM in the
These are the quite cumbersome calculation of direction of the HTF flow is ignored.
liquid fraction updates and the fact that the (B) The effects of natural convection within the
temperature field within the PCM is not calculated melt are negligible and can be ignored.
explicitly but via enthalpy-temperature correla- (C) The PCM behaves ideally, i.e. such phe-
tion. In this paper, a slightly modified enthalpy nomena as property degradation and supercooling
method, which enables decoupling of temperature are not accounted for.
and liquid fraction fields, is presented. Derivation (D) The PCM is assumed to have a definite
of discretization equations is straightforward and melting point (isothermal phase change).
the method itself is very easy to implement. (E) Thermophysical properties of the PCM are
different for the solid and liquid phases but are
independent of temperature.
2. MATHEMATICAL MODEL
(F) The PCM is homogeneous and isotropic.
The subject of the present investigation is a (G) Thermal resistance across the wall of the
single container filled with PCM (Fig. 1). Packing container is neglected.
the PCM in single containers enables modular (H) Lateral sides of the rectangular container
construction of the LHES unit and is also very are well insulated, i.e. heat transfer occurs only on
economic from the viewpoint of mass production. sides x 5 0 and x 5 d (cf. Fig. 1).
Moreover, complete melting of the PCM, which is In order to validate the first assumption (A), the
absolutely necessary for long-term (seasonal) heat influence of the conduction within the PCM in the
storage (Zivkovic and Fujii, 1999), is difficult to direction of HTF flow was numerically investi-
obtain using a shell-and-tube LHES unit where gated in the following way. Retaining all the
large masses of the PCM are involved. above assumptions and neglecting the conduction
There are several approaches to the mathemati- within the PCM in the direction normal to the
cal modeling of LHES units. In some models, HTF flow, the governing equations for heat
conduction within the PCM in both the direction transfer and phase change can be written as:
of the HTF flow and the direction normal to the
HTF flow are taken into account (Lacroix, 1993;
Ghoneim, 1989), while in others the effect of
≠H ≠ k ≠T
]5] ] ]
≠t ≠z r ≠z S D (1)

natural convection within the molten PCM is also ≠T air


mf c pf ]] 5 aconv A htsT 2 T aird (2)
≠z
where all the variables are defined in the nomen-
clature table. Detailed description of the numeri-
cal solution of the above set of equations is
omitted as it is described later in the paper. Here,
only the results obtained are discussed briefly.
Even though conduction within the PCM in the
direction of HTF flow plays an important role for
long LHES units (containers) and relatively high
convective heat transfer coefficients (Fig. 2), its
Fig. 1. Geometry of the rectangular container. influence can be ignored (Fig. 3) for low convec-
An analysis of isothermal phase change of phase change material within rectangular and cylindrical containers 53

Fig. 2. Temperature variation within the PCM in the direction of the HTF flow for l53 m and aconv 5300 W/ m 2 K.

tive heat transfer coefficients (such are those when With the foregoing assumptions, the enthalpy
air is used as the HTF) and relatively short formulation for the conduction-controlled phase
containers (from 200 to 400 mm). Furthermore, change can be written as (Voller, 1990; Voller et
comparison of the solution obtained using the al., 1987):
mathematical model defined with the set of Eqs.
(1) and (2) and the solution obtained solving the
same problem using the lumped mass method
≠H k
S
] 5 div ] grad T
≠t r D (3)
shows that no significant improvement of the
results is obtained if the conduction within the An alternative form of Eq. (3) can be obtained by
PCM is accounted for (Fig. 4). These results are splitting the total enthalpy H into sensible and
in accordance with the results obtained by latent heat components:
Ghoneim (1989), who concluded that there is no H 5 h 1 L ? fl (4)
significant change in the prediction of the solar
fraction with reduction of Dz to less than 200 mm. where

Fig. 3. Temperature variation within the PCM in the direction of the HTF flow for l50.314 m and aconv 515.5 W/ m 2 K.
54 B. Zivkovic and I. Fujii

Fig. 4. Comparison between the lumped mass method and the case when the axial conduction is accounted for.

h5 E c dT (5)
≠h ≠ k ≠T
S D
≠f
] 5 ] ] ] 2 L ]l
≠t ≠x r ≠x ≠t
(8)
Tm

The fully implicit discretization equation for an


and T m is the melting temperature of the PCM.
internal node ‘i’ can be written as (Fig. 5):
For the problem of isothermal phase change, the
local liquid fraction fl is defined as: ≠fl
≠h i k
] 5 ]]2 sT i 21 2 2T i 1 T i 11d 2 L ]i (9)
fl (T ) 5 H
1 if T . T m
0 if T , T m
(6)
≠t r Dx ≠t

where the sensible enthalpy term and liquid


Substituting Eq. (4) into Eq. (3) gives: fraction term are deliberately left in the differen-
tial form for the convenience of subsequent
≠h
≠t
k
r S ≠f
] 5 div ] grad T 2 L ]l
≠t D (7) numerical computation. It should be noted that the
discretization equations for boundary nodes de-
Eq. (7) represents, together with Eqs. (5) and (6) pend on specific boundary conditions and are
and the appropriate initial and boundary con- derived from an energy balance on boundary
ditions, the mathematical model of conduction control volumes.
controlled isothermal phase change. The key feature of the proposed method is the
fact that for an isothermal phase change (which is
the case for most salt hydrates), the temperature
3. NUMERICAL SOLUTION
of the PCM within a given control volume
For the problem of one-dimensional isothermal remains constant and equal to its melting tempera-
phase change of the PCM encapsulated within a ture until the PCM has melted completely. Con-
rectangular container (Fig. 1), the governing sider first the case when control volume ‘‘i‘ is
equation for the PCM follows from Eq. (7): fully solid or fully liquid. In that case, from the

Fig. 5. Discretization domain for a one-dimensional phase change problem.


An analysis of isothermal phase change of phase change material within rectangular and cylindrical containers 55

definition of sensible enthalpy, Eq. (5), and the k Dt


liquid fraction, Eq. (6), it follows that: fl i 5 f old
l i 1 ]] sT 2 2T m 1 T i 11d (18)
r L Dx 2 i 21
≠h ≠T
]i ; c ]i (10) which is the equation for updating the liquid
≠t ≠t fraction field within the control volume that is
and undergoing a phase change. As Eq. (15) shows,
liquid fractions are updated from the temperature
≠fl field and not from the sensible enthalpy field
]i ; 0 (11)
≠t (Voller, 1990; Lacroix, 1993). The temperature
and the liquid fraction field are decoupled, the
where c is the specific heat of the solid or liquid
temperature field within the PCM being calculated
phase, depending on state of the control volume.
independently from Eq. (13) forcing a i 2 1 5
After introducing Eqs. (10) and (11), Eq. (9)
a i 1 1 5 0, a i 5 1 and b i 5 T m for the control
reduces to ordinary heat diffusion equation:
volumes which are undergoing the phase change.
≠T k Furthermore, on the basis of Eq. (18), it can be
]i 5 ]]2 sT i 21 2 2T i 1 T i 11d (12)
≠t r c Dx inferred that for an isothermal phase change, all
the heat supplied to the control volume undergo-
Backward differencing of the left side term gives, ing a phase change is used for changing the
after rearranging, the fully implicit finite differ- amount of latent heat content of that control
ence equation of the form: volume.
At this point it is worthwhile to describe the
a i 21 ? T i 21 1 a i ? T i 1 a i 11 ? T i 11 5 b i (13)
implementation of the computational model,
where coefficients a i 2 1 5 a i 1 1 5 2 Fo, a i 5 1 1 which is as follows:
2 3 Fo and b i 5 T iold are introduced for the sake (a) Coefficients ‘a’ of Eq. (13) are formed. For
of computational simplicity. Superscript ’old’ nodal points where the liquid fraction fli is strictly
refers to the previous time step and Fo is a finite in the interval [0,1], 0 , fli , 1, the coefficients of
difference form Fourier number: Eq. (13) are set to: a i 2 1 5 a i 1 1 5 0, a i 5 1 and
bi 5 Tm .
a ? Dt (b) The set of linear algebraic Eqs. (13) is
Fo 5 ]] (14)
Dx 2 solved using the Gauss-Seidel iterative procedure.
(c) Liquid fractions are updated from the tem-
where a 5 k /r c is the thermal diffusivity of the
perature field using Eq. (18).
PCM. The thermophysical properties in Eq. (14)
(d) A check for the ‘start’ and for the ‘end’ of
depend on the state of the control volume. Those
phase change is performed. Explicitly, if the state
of the solid phase should be inserted if the control
of the liquid fraction field changes within the
volume is in the solid state and those of the liquid
given time step, i.e. a finite volume commences or
phase if the control volume is in the liquid state.
terminates with the phase change, the coefficients
Now, consider the case when melting (or
of Eq. (13) need to be updated and steps (a)
freezing) occurs around a certain node ‘i’. In that
through (d) repeated for the same time step.
case, the liquid fraction fli lies strictly in the
In practice, for most time steps only one
interval [0,1]. Recognizing that for an isothermal
iteration is needed per time step. The only time
phase change:
when two iterations are needed is when the phase
Ti ; Tm (15) change boundary moves from one control volume
to the next one.
and from Eq. (5):
3.1. Checking for start /end of the phase change
≠h
]i ; 0 (16) At the end of each time step, the check for start
≠t
and / or end of the phase transition is performed
Eq. (9) becomes: throughout the entire domain. For the case of
melting, checking for the ‘start’ and ‘end’ of the
≠fl i k
L ] 5 ]]2 sT i 21 2 2T m 1 T i 11d (17) phase change is performed in the following
≠t r Dx fashion:

Backward differencing of the liquid fraction term 3.1.1. START of melting. For a given time step,
gives: if T i $ T m while T iold , T m , it indicates that within
56 B. Zivkovic and I. Fujii

this time step, the finite volume in question begins 3.1.2. END of melting. For a given time step, if
with melting. In that case, the coefficients of Eq. fli $ 1 while f liold , 1, it indicates that within this
(13) are updated as described above and the time step, the control volume in question has
calculation for that step is performed again. It melted completely. In that case coefficients of Eq.
should be noted, however, that in the time step (13) are again set to a i2 1 5 a i 1 1 5 2 Fo, a i 5
old
when a control volume has just begun with 1 1 2 3 Fo and b i 5 T i and the calculation for
melting, Eq. (18) has the form: that time step is performed again. In the time step
in which the phase change boundary moves from
k Dt the control volume in question to the next one,
fl i 5 f old
l i 1 ]] sT 2 2T m 1 T i 11d coefficient b i has the following form:
r L Dx 2 i 21
c L
2 ] sT m 2 T old
i d (19) b i 5 T m 2 ] s1 2 f old
li d (20)
L c
The last term on the right-hand side of Eq. (19) where the last term on the right hand side can be
represents the amount of sensible heat that is described as the amount of heat needed to com-
needed to raise the temperature of the control pletely melt the control volume in question within
volume from the temperature in the previous time the time step and which consequently can not be
step (T old used to raise the temperature of the PCM.
i ) to the melting temperature (T m ).
Consequently, that amount of heat can not be used The flow chart for the computational procedure
for melting the PCM. is given in Fig. 6.

4. VERIFICATION OF THE MATHEMATICAL


MODEL

4.1. Test problem


The performance of the presented method is
first verified with a one-dimensional phase change
test problem explained in Voller (1990). A pure
liquid initially at 28C occupies the semi-infinite
space x $ 0. At time t 5 0 the surface at x 5 0 is
fixed at the temperature of 2108C, which is
below the freezing point of the substance T m 5
08C. As time proceeds, a solid layer builds up on
the surface x 5 0 and moves out into the liquid.
Simply stated, the problem is to determine how
the solid–liquid surface moves with time. The
thermal properties of the material in question are
assumed to be constant and equal for both solid
and liquid phase: k52 [W/ mK], r c52.5310 6
[J / m 3 K] and r L510 8 [J / m 3 ]. In numerical
analysis 50 time steps of Dt54.32310 4 [s] (1 / 2
days) and 20 space increments of Dx50.125 [m]
were used.
The position of the phase front after 25 days
(i.e. 50 time steps) obtained with proposed meth-
od is x50.8405 [m]. The difference between this
result and the result of x50.8415 [m] obtained by
Voller (1990) is merely 0.12%. Therefore, it could
be concluded that the accuracy of the proposed
computational model for conduction controlled
isothermal phase change is satisfactory.

4.2. Experimental verification


As was discussed in the previous paragraph, the
Fig. 6. Flow chart. solution obtained with the proposed method gives
An analysis of isothermal phase change of phase change material within rectangular and cylindrical containers 57

Table 1. Thermophysical properties of CaC1 2 ?6H 2 O sides. A thermocouple was placed in the centre of
Melting point [8C] 29.9 the container in the manner indicated in Fig. 7.
Latent heat [KJ / kg] 187
Density [kg / m 3 ]: Solid 1710 The container with the solid PCM was then
Liquid 1530 placed vertically in the constant temperature bath,
Specific heat [kJ / kgK]: Solid 1.4 where the temperature was set to T ` 5608C.
Liquid 2.2
Thermal conductivity [W/ mK]: Solid 1.09 The computational model was set up to re-
Liquid 0.53 produce experimental conditions within the con-
stant temperature bath. The convection heat trans-
very satisfactory results for the conduction-con- fer coefficient between the air and the container
trolled one-dimensional phase change test prob- wall was determined using the correlation given
lem. However, melting of the PCM in sealed in Incropera and DeWitt (1985), and was calcu-
containers is generally multi-dimensional (both lated to be aconv 516 [W/ m 2 K]. Furthermore, a
radial and axial conduction exist) and also natural time step Dt55 s and space increment Dx52 mm
convection occurs within the melted PCM. There- were used in the calculation. In Fig. 8 the
fore, an experiment was performed in order to variation with time of numerical and experimental
investigate the influence of the assumptions of the values of the temperature at the centre of the test
mathematical model. The PCM used for ex- container is shown.
perimental analysis is calcium chloride hexahy- From the results shown in Fig. 8, it can be
drate (CaCl 2 ?6H 2 O) with the thermophysical concluded that the agreement between numerical
properties as listed in Table 1 (Fujii and Yano, and experimental data is well within experimental
1996). uncertainties (i.e. positioning of the thermocou-
A rectangular container, made of stainless steel, ple’s tip exactly in the centre of the container is
with dimensions of l5b5100 mm and d 520 mm quite difficult and experimental data were read
(Fig. 1), was filled with the calcium chloride from charts, which reduces accuracy). The higher
hexahydrate and well insulated on the lateral slope of the theoretical curve in the liquid region

Fig. 7. Test container.

Fig. 8. Variation with time of PCM’s temperature at the centre of the rectangular container (cf. Fig. 7.).
58 B. Zivkovic and I. Fujii

is due to the fact that the Grashof number was vestigated for the rectangular and cylindrical
calculated for the temperature difference of T ` 2 containers.
T m ¯30 K, while in reality this difference be- The mathematical model for the isothermal
comes smaller as the temperature of the liquid phase change of the PCM filling the cylindrical
PCM increases. Consequently, the convective heat container (Fig. 9) was derived under the same
transfer coefficient between the air and the assumptions as in the case of the rectangular
container wall used in the calculation is higher container. However, the governing equation for
than the ‘real’ one. Furthermore, it can be ob- the two models is different and for the case of the
served from Fig. 8 that in theory PCM reaches its cylindrical container it has the following form:
melting temperature faster than in the experiment.
This is assumed to be due to the basic assumption
made in the mathematical model that the conduc-
≠H 1 ≠ k ≠T
]5] ] ] ]
≠t r ≠r r ≠r S D (21)
tion resistance of the container wall is neglected.
In the above equation, H represents the total
Moreover, from the same figure it can be ob-
enthalpy, which can be split into its sensible and
served that the calculated PCM’s melting time is
latent component (cf. Eq. (4)). The solution
slightly longer than the experimental one, which
methodology and numerical procedure for Eq.
may be due to the fact that the natural convection
(21) is the same as for Eq. (3).
within the liquid PCM is ignored. However, it can
be seen that neglecting both the natural convec-
tion within the liquid PCM and the conduction 5.1. Results and discussion
within the PCM in the direction of the HTF flow The dimensions of the containers were chosen
do not introduce significant error in the prediction in such a manner that the volume as well as the
of the PCM’s temperature variation during melt- convective heat transfer area for both the cylindri-
ing. cal and rectangular containers were equal. The
fixed dimensions were chosen to be the length of
the cylindrical container l c 50.2 m (cf. Fig. 9) and
5. COMPARISON OF THE MELTING TIME
the width of the rectangular container b50.1 m
FOR RECTANGULAR AND CYLINDRICAL
(cf. Fig. 1). The air velocity was assumed to be
CONTAINERS
w55 m / s and its temperature T air 5608C. The
The melting time of the incapsulated PCM is thermophysical properties of the air, as well as the
one of the essential parameters for determining convective heat transfer coefficient between the
the size and the shape of the container, as it must air and the container’s wall aconv , were taken
correspond to the total amount of daily insolation. from Incropera and DeWitt (1985). The PCM
To be specific, the container containing PCM filling the containers was chosen to be calcium
should be designed in such a way that at the end chloride hexahydrate, with the thermophysical
of the day, complete melting of the PCM is properties as listed in Table 1.
achieved. In that way, the maximum efficiency of Fig. 10 shows the variation with time of the
the LHES unit is achieved. Furthermore, complete PCM’s temperature in the centres of both the
melting of the PCM is a necessary condition for rectangular and cylindrical containers for the
long-term (seasonal) thermal energy storage. In different dimensions of the containers. For small
light of that, the influence of the container’s values of d (cf. Fig. 1) and ro (cf. Fig. 9), the
dimensions and its shape were numerically in- difference in melting time between the rectangular

Fig. 9. Geometry of the cylindrical container.


An analysis of isothermal phase change of phase change material within rectangular and cylindrical containers 59

Fig. 10. Comparison of the variation with time of the PCM’s temperature at the centre of the rectangular and cylindrical
containers.

and cylindrical containers is not so pronounced. PCM filling the container on the melting time of
However, on increasing the mass of the PCM PCM. It can be observed that for larger quantities
filling the container, i.e. with increasing d for the of the material filling the container, the difference
rectangular container and ro for the cylindrical in the melting time between the rectangular and
container, the difference in the melting time cylindrical containers is very pronounced, with
increases considerably, with the rectangular the melting time of the cylindrical container being
container showing a much shorter melting time nearly twice that of the rectangular one. It should
than the cylindrical container of the same volume be pointed out that in order to make the com-
and heat transfer area. parison between the two geometries significant,
A series of numerical experiments were per- the numerical analysis was performed under the
formed and their results are summarized in Fig. condition of equal volume and heat transfer area
11 which shows the influence of the amount of for both the rectangular and cylindrical containers.

Fig. 11. Comparison of the melting time for rectangular and cylindrical containers of equal volume and heat transfer area.
60 B. Zivkovic and I. Fujii

It is interesting to note that the correlation be- mf Mass flow rate of air [kg / s]
r Radius of cylindrical container [m]
tween the melting time and the mass of the PCM T Temperature of PCM [8C]
is nearly linear. T ` , T air Air temperature [8C]
Tm Melting point of PCM [8C]
t Time [h]
a Thermal diffusivity of PCM [m 2 / s]
6. CONCLUSION aconv Heat transfer coefficient [W/ m 2 K]
d Thickness of rectangular container [m]
A simple and easily implemented computation- Dx Space increment [m]
al model simulating transient behaviour of the Dt Time step [h]
isothermal phase change is developed. A test r Density of PCM [kg / m 3 ]
problem was solved and results obtained with the
proposed method show very close agreement with
the results obtained from other computational REFERENCES
models (cf. Voller, 1990). An experiment was Duffie J. A. and Beckman W. A. (1991) Solar Engineering of
performed and numerical results verified with Thermal Processes, John Wiley, New York.
experimental data. Abhat A. (1983) Low temperature latent heat thermal energy
storage: heat storage materials. Solar Energy 30(4), 313–
Taking into consideration the basic assumptions 331.
of the mathematical model and recognizing very Klein S. A., Beckman W. A. and Duffie J. A. (1976) A design
good agreement between the numerical and ex- procedure for solar heating systems. Solar Energy 18, 113–
127.
perimental results, the following conclusions can Lacroix M. (1993) Numerical simulation of a shell-and-tube
be made: latent heat thermal energy storage unit. Solar Energy 50(4),
1. Heat conduction resistance of the container’s 357–367.
Bansal N. K. and Buddhi D. (1992) An analytical study of a
wall can be neglected without losing accuracy latent heat storage system in a cylinder. Energy Convers.
in the computational data. Mgmt 33(4), 235–242.
2. Conduction within the PCM in the direction of Esen M., Durmus A. and Durmus A. (1998) Geometric design
of solar-aided latent heat store depending on various param-
the HTF flow can be ignored as its influence eters and phase change materials. Solar Energy 62(1),
was found to be negligible. 12–28.
3. For flat thin containers, the effects of natural Soma D. and Dutta T. K. (1993) Mathematical modeling and
experimental studies on solar energy storage in a phase
convection within the liquid PCM can be change material. Solar Energy 51(5), 305–312.
ignored without introducing a significant error Ismail K. A. R. and Alves C. L. F. (1986) Analysis of
in the prediction of the temperature variation shell-tube PCM storage systems. In Proceedings of the
Eight International Heat Transfer Conference, San Fran-
with time within the PCM. cisco, pp. 1781–1786.
Moreover, comparison between the melting Ghoneim A. A. (1989) Comparison of theoretical models of
time for the rectangular and cylindrical containers phase-change and sensible heat storage for air and water-
based solar heating systems. Solar Energy 42(3), 209–220.
was performed and the results show that the Morrison D. J. and Abdel-Khalik S. I. (1978) Effects of
rectangular container requires half of the melting phase-change energy storage on the performance of air-
time as for the cylindrical container of the same based and liquid-based solar heating system. Solar Energy
20, 57–67.
volume (i.e. equal mass of the PCM filling the Murray W. D. and Landis F. (1959) Numerical and machine
container) and the same heat transfer area between solutions of transient heat-conduction problems involving
HTF and the container’s wall. It is, therefore, melting or freezing. Transactions of ASME. Journal of
Heat Transfer 81, 106–112.
preferable to use rectangular containers for en- Voller V. and Cross M. (1981) Accurate solutions of moving
capsulating the PCM. boundary problems using the enthalpy method. Int. J. Heat
Mass Transfer 24, 545–556.
Voller V. R. (1985) Implicit finite-difference solutions of the
NOMENCLATURE enthalpy formulation of Stefan problems. IMA Journal of
Numerical Analysis 5, 201–214.
Voller V. R. (1990) Fast implicit finite-difference method for
the analysis of phase change problems. Numerical Heat
A ht Heat transfer surface [m 2 ] Transfer, Part B 17, 155–169.
b Width of rectangular container [m] Carslaw H. S. and Jaeger J. C. (1959) Conduction of Heat in
c pf Specific heat of air [J / kgK] Solids, 2nd edn, Oxford University Press, Great Britain, pp.
c Specific heat [J / kgK] 282–287.
fl Liquid fraction Zivkovic B. and Fujii I. (1999) Experimental results on
Fo Fourier number prevention of supercooling of low melting temperature salt
H Total enthalpy [J / kg] hydrates. In Proceedings of JSES /JWEA Joint Conference,
h Sensible enthalpy [J / kg] 25–26 November, Kyoto, Japan, pp. 257–260 (in Japanese).
k Conductivity [W/ mK] Voller V. R., Cross M. and Markatos N. C. (1987) An enthalpy
l Length of container [m] method for convection / diffusion phase change. Int. J.
L Latent heat of fusion [J / kg] Numerical Methods in Engineering 24, 271–284.
An analysis of isothermal phase change of phase change material within rectangular and cylindrical containers 61

Belleci C. and Conti M. (1993) Transient behaviour analysis Fujii I. and Yano N. (1996) A consideration on phase change
of a latent heat thermal storage module. Int. J. Heat Mass behaviour of latent heat storage material. In Proceedings of
Transfer 36(15), 3851–3857. the Society of Heating, Air-conditioning and Sanitary
Trp A., Frankovic B. and Lenic K. (1999) An analysis of Engineers of Japan (in Japanese).
phase change heat transfer in a solar thermal energy store. Incropera F. P. and DeWitt D. P. (1985) Introduction to heat
In Proceedings of ISES World Congress, July, Jerusalem, transfer, John Wiley & Sons, New York, pp. 288–308.
Israel.

Você também pode gostar