International Journal of Thermal Sciences 60 (2012) 153e160

Contents lists available at SciVerse ScienceDirect

International Journal of Thermal Sciences

journal homepage: www.elsevier.com/locate/ijts

Numerical investigation of packed bed storage unit filled with PCM encapsulated
spherical containers e A comparison between various mathematical models
S. Karthikeyan, R. Velraj*
Institute for Energy Studies, Anna University, Sardar Patel Street, Chennai 600025, India

a r t i c l e i n f o a b s t r a c t

Article history: There are several methods adopted by the scientists in modeling the heat transfer phenomenon during
Received 11 February 2012 solideliquid phase change. However, the validity of the various models developed and the complexity
Received in revised form required for a given problem is not well reported in the literature. This paper presents a comparative
13 April 2012
study of three different mathematical models for the packed bed latent heat storage system, comprised
Accepted 24 May 2012
of a cylindrical storage tank filled with paraffin encapsulated spherical containers. The enthalpy
Available online 29 June 2012
formulation technique is used in the models to accommodate the phase change behavior of the paraffin
over a range of temperatures. The results obtained by solving the models using the fully explicit finite
Keywords:
Phase change materials
difference method are initially validated with experimental results. Further numerical analysis is per-
Latent heat storage formed using the models for two different heat transfer fluids of air and water at different mass flow
Packed bed storage rates and ball sizes and the validity of the models are compared and reported. It is found that one model
Air heating is sufficient when air is the heat transfer fluid and another model is recommended when water is the
Enthalpy modeling heat transfer fluid.
Numerical modeling Ó 2012 Elsevier Masson SAS. All rights reserved.

1. Introduction grid size and fixed time step can be employed. Also, in this method
a single equation for the PCM is sufficient to find the temperature
Thermal Energy Storage [TES] technologies are very important variation of the PCM which melts/solidifies at a particular range of
in various fields of engineering applications. The most common TES temperature.
technologies are sensible heat storage and latent heat storage. Many researchers have investigated experimentally and theo-
Latent heat thermal storage units have received greater attention retically the transient behavior of the PCM encapsulated in
by the scientists and engineers working in the areas of solar heat- different geometries and PCM based storage systems in different
ing/cooling, energy management in buildings and waste heat configurations particularly shell and tube, packed bed with capsule
recovery applications due to their high energy storage capacity and of different geometries. Regin et al. [1] presented a review on the
isothermal behavior. heat transfer characteristics of the PCM based TES. Verma et al. [2]
Mathematical modeling of phase change phenomenon involves reviewed the mathematical modeling of the latent heat storage
greater complexity as the solid and liquid domains changes system using PCM, by differentiating the models based on the I and
continuously with respect to time. There are several methods being II law of thermodynamics. The various studies that focus on the
adopted to model and solve for the temperature variation in the numerical analysis of packed bed storage unit are summarized in
domain of interest. The numerical methods for such problems are this section.
reported in two broad categories, such as Temperature based model Saitoh and Hirose [3] have reported the results of theoretical
and Enthalpy model. In the Temperature based model, the phase and experimental investigation on the transient thermal perfor-
change interface is either captured on a grid at each fixed time step, mance of a TES unit based on the capsule diameter, the flow rate of
and therefore, non uniform grid spacing is constructed, or captured HTF, the type of PCMs, the capsule material and the difference
on a uniform grid, and therefore, a non uniform time step is used. In between the inlet HTF temperature and melting temperature of the
the enthalpy model, the interface fitting is totally eliminated by PCM. Chen and Yue [4] developed a one dimensional porous
formulating the energy equation in terms of enthalpy where a fixed medium model to determine the thermal characteristics of a cool
thermal storage system using water/ice as the PCM. The result was
characterized based on five independent dimensionless parame-
* Corresponding author. Tel.: þ91 44 22358051; fax: þ91 44 22351991. ters, such as effective water-to-coolant heat capacity ratio, effective
E-mail addresses: [email protected], [email protected] (R. Velraj). ice-to-coolant heat capacity ratio, Stanton number, Stefan number

1290-0729/$ e see front matter Ó 2012 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ijthermalsci.2012.05.010
154 S. Karthikeyan, R. Velraj / International Journal of Thermal Sciences 60 (2012) 153e160

and Peclet number. Comparisons of the numerical results show Though there are several numerical methods used by the
good agreement with the experiments, against their previous work scientists for the modeling of packed bed latent heat storage
on a one dimensional lump model [5]. Watanabe et al. [6] devel- system, the validity of the model developed and the complexity
oped and studied a heat storage system consisted of horizontal required for the given problem is not reported in the literature.
cylindrical capsules filled with three types of PCM with different Hence, in the present work the three different modeling
melting temperature. A simple numerical model was used to study approaches for the packed bed latent heat storage system is
the effect of flow rate, initial temperature of HTF and the dimen- adopted and after validating the model with experiment results,
sions of the heat storage module on the thermal performance of the the suitability of these models are studied for the commonly used
latent heat storage module. Results proved that the charging and heat transfer fluid, air and water at various mass flow rates.
discharging rate are improved by the use of multiple PCM. Cho and
Choi [7] experimentally investigated the thermal characteristics of 2. Modeling of the packed bed and numerical solution
paraffin in a spherical capsule packed inside a storage tank at
different values of the Reynolds number and inlet temperatures. Fig. 1 shows the physical arrangement of the packed bed
They concluded that the phase change period for the capsule at the considered for the numerical modeling. It consists of a well insu-
edge of the storage tank was shorter than that at the center of the lated cylindrical tank of ‘L/D’ ratio 2. PCM encapsulated in plastic
storage tank, because the porosity at the center was smaller than at spherical containers are packed randomly inside the tank. This
the edge of the storage tank. Also, the influence on the average heat provides a porous structure in the tank with a void fraction of ‘ε’.
transfer coefficients due to the fluid inlet temperature and Rey- The void fraction acts as a space and passage for the flow of the HTF.
nolds number is more during the melting process, than the freezing Initially, the PCM and HTF in the tank are assumed to be at the same
process due to the presence of the natural convection effect of the temperature. The PCM is sub cooled below its melting temperature
melting process. Benmansour et al. [8] carried out a two dimen- and therefore it is completely in solid condition. At time, t > 0 the
sional numerical and experimental analysis to understand the HTF at constant temperature and value greater than the phase
transient axial and radial thermal dispersion on a cylindrical change temperature of the PCM is allowed to enter through the
packed bed storage filled with spherical PCM (paraffin) capsules. bottom of the tank. During the above processes, the PCM undergoes
Air is used as the Heat Transfer Fluid (HTF). The comparisons were three different regimes, such as, sensible heating below phase
made for different Reynolds numbers of 560 and 1120. The energy change temperature, phase change process, and sensible heating
equation for the fluid was resolved by finite difference approxi- above phase change temperature. The maximum velocity used in
mation, and solved by the alternating direction implicit (ADI) the model is ‘vmax’ which is calculated from the HTF flow rate across
scheme, whereas the energy equation for PCM was solved by the the porous cross section of the packed bed. The most commonly
finite difference fully explicit scheme. Regin et al. [9] numerically used HTFs are air, water and thermic fluids. Therefore, air and water
investigated the effect of the phase change temperature range, the are considered as the HTF in the present analysis. Commercial
size of the PCM capsule, inlet heat transfer fluid temperature and grade paraffin wax is considered as the PCM. In order to determine
fluid flow rate on the performance of the packed bed latent heat the phase change behavior of the paraffin used for the validation of
storage system, consisting of spherical capsules for solar water the numerical simulation, a differential scanning calorimetry (DSC)
heating applications. The solidesolid phase change that occurs analysis is carried out. The results show that there are two peaks
before the onset of the actual melting phase change was also existing, the solidesolid phase change, and the solideliquid phase
considered in the numerical solution. Cheralathan et al. [10] per- change. Fig. 2 illustrates the DSC curve during melting of PCM. Fig. 3
formed a numerical analysis and parametric studies on a PCM represents the various regimes of the PCM transformation during
encapsulated cool TES unit integrated with a refrigeration system. the melting process. Table 1 summarizes the thermo physical
They have used the porous medium model with three different properties of the PCM.
time domains as suggested by Chen and Yue [4]. They have repor- Three different mathematical models for a packed bed latent
ted the effect of porosity and various non dimensional parameters heat storage system filled with spherical PCM balls are considered
and conclude that Ste (0.2e0.4), porosity (0.4e0.49) and St for the analysis. The first model is a continuous solid phase model
(0.7e1.0) are to be chosen for higher and faster energy storage. in which all the PCMs at a particular height of the tank are
Nallusamy et al. [11] have also investigated the performance of the
similar storage unit for solar water heating applications using the
same mathematical model. However, their model also accommo-
At x=L; ;
dates the effect of varying inlet temperature to the storage tank to
X=L
simulate the hot water coming from the solar collector. Wu et al.
n=5 ∂
[12] studied thermal performance of a packed bed cool TES system
using n-tetradecane as the PCM. The charging and discharging
processes were numerically simulated by one dimensional gov- n=4

erning equation for HTF and PCM using implicit finite difference
method. The effect of inlet fluid temperature, mass flow rate of HTF, n=3
diameter of the capsule and porosity of the packed bed on the r=0
solidification and temperature distribution of the packed bed were n=2
analyzed. Antony and Velraj [13] performed the transient analysis
on the PCM based heat exchanger module by developing a mathe- x n=1
matical model using apparent heat capacity formulation, in which At r=r 0 ;
the portion of the DSC curve (of the PCM) during the phase change − −
period is interpreted as a triangular profile (299  2 K, wide phase X=0
change temperature range) and a rectangular profile (299  1, 0.4
and 0.25 K, narrow phase change temperature range), and
concluded that the DSC results of 0.5 K/min and rectangular profile
assumption is suitable for slow cooling rate experiments. Fig. 1. Physical arrangement of the packed bed model.
 S. Karthikeyan, R. Velraj / International Journal of Thermal Sciences 60 (2012) 153e160 155

Table 1
Thermo physical properties of the paraffin wax.

Phase change temperature

Solidesolid phase change 46.5e55.5  C
Solideliquid phase change 55.5e66.5  C
Latent heat of fusion
Solidesolid phase change 38 kJ/kg
Solid-liquid phase change 142.7 kJ/kg
Specific heat
Solid 2.4 kJ/kg K
Liquid 1.8 kJ/kg K
Density
Solid 670 kg/m3
Liquid 640 kg/m3
Thermal conductivity
Fig. 2. DSC curves during melting of PCM. Solid 0.0004 kW/m K
Liquid 0.0002 kW/m K

considered to be at the same temperature at a particular time and

the conduction along the axial direction is neglected. In the second
model the conduction effect along the axial direction of the HTF and 8) Enthalpy based formulation is employed for the energy equa-
PCM is also included. The third model is a conduction dominated tion of PCM to accommodate the phase change over a range of
enthalpy based model which considers thermal gradients inside temperature.
the PCM capsules.
The following general assumptions are made in the formulation The governing equation for the HTF and the PCM for the three
of all the mathematical models: models are given in Table 2.

1) The HTF and PCM are two different domains and hence, sepa- 2.1. Initial conditions
rate equations for each region are considered.
2) The thermo physical properties of the PCM are different for the
solid and liquid phases and the variation of properties with At time t ¼ 0;
respect to temperature during the phase change is considered Tf ¼ Tf ini for 0  x  L
linear. Tp ¼ Tp ini for 0  x  L
3) The temperature of the HTF at the inlet to the storage tank is Tp ¼ Tp ini for 0  r  r0 and 0  x  Lðin case of thermal
considered to be constant during the entire charging process. gradient modelÞ
4) Initially the temperatures of the HTF and PCM in the storage
tank are considered to be uniform. 2.2. Boundary conditions
5) The resistance offered by the thin wall of the spherical capsule
is neglected.
6) Radiant heat transfer is ignored. At time t>0
7) The tank is perfectly insulated, and hence, there is no heat loss Tf ¼ Tf inlet at x ¼ 0
from the tank surface to the surroundings. vTf
¼ 0 at x ¼ L
vx
At the PCM wall e HTF interface (i.e. at r ¼ r0 and all values of ‘x’, in
case of thermal gradient model).

vTp

kp ¼ hs Tpjr¼r0  Tf
vr
The convective heat transfer coefficient between the HTF and PCM
capsules is determined by using the correlation given in Eq. (1),
(Perry and Green) [14].

Table 2
Governing equation for the HTF and the PCM for the three models.

Model 1  
vTf vTf
HTF : εrf Ac Lcf þ vmax ¼ hs ap ðTp  Tf Þ
vt  
vx
vHP
PCM : ð1  εÞrP Ac LcP ¼ hs ap ðTf  Tp Þ
vt
Model 2  
vTf vTf v2 Tf
HTF : εrf Ac Lcf þ vmax ¼ kf 2 þ hs ap ðTp  Tf Þ
vt  vx
 vx
vHP v2 T
PCM : ð1  εÞrP Ac LcP ¼ kP 2P þ hs ap ðTf  Tp Þ
vt vx
Model 3  
vTf vTf
HTF : εrf Ac Lcf þ vmax ¼ hs ap ðTpjr¼ro  Tf Þ
vt 2 vx
vHp v Tp 2 vTp
PCM : rp ¼ kp 2 þ $kp
vt vr r vr
Fig. 3. Transition of the PCM during the heating process.
156 S. Karthikeyan, R. Velraj / International Journal of Thermal Sciences 60 (2012) 153e160

!0:365 steel with 0.7 m height and 0.35 m diameter, and is well insulated
h dp dp G with glass wool on the outer surface. The PCM encapsulated in
Nu ¼ ¼ 3:6 (1)
kf mf ε plastic balls of an outer diameter of 70 mm and a wall thickness of
0.5 mm with a total number of 260 balls are packed inside the tank.
A centrifugal air blower, flow adjusting valves and an orifice meter
2.3. Enthalpy e temperature relationship

Hp ¼ cp *Tp   for Tp  Tm1  a1

Tp  ðTm1  a1Þ
Hp ¼ cp *Tp þ la1 for Tm1  a1 < Tp  Tm1
 a1 
Tp  Tm1
Hp ¼ cp *Tp þ la1 þ la2 for Tm1 < Tp  Tm1 þ a2
a2 
Tp  ðTm2  e1Þ
Hp ¼ cp *Tp þ la1 þ la2 þ le1 for Tm2  e1 < Tp  Tm2
e1 
Tp  Tm2
Hp ¼ cp *Tp þ la1 þ la2 þ le1 þ le2 for Tm2 < Tp  Tm2 þ e2
e2
Hp ¼ cp *Tp þ la1 þ la2 þ le1 þ le2 for Tp >Tm2 þ e2

are provided to supply air at varying rates. The air is heated by

2.4. Method of solution finned electric heaters fitted inside an insulated enclosure. The
RTDs (resistance temperature detectors, Type Pt 100) are provided
The equations for the HTF and the PCM are discretized by the at the inlet to the tank, and at five equally spaced locations along
explicit finite difference scheme. The forward difference approxi- the axial direction of the bed, to measure the HTF temperature
mation for the time derivative, and the central difference approx- variations. Similarly, another five RTDs are inserted in the PCM and
imation for the space derivative are used. In particular, the they are located at the center of the capsules which are placed at
convective term of the HTF equation is resolved by the upwind (or five equally spaced locations along the axis of the storage tank, to
backward difference) scheme, as the flow in the axial direction is measure the PCM temperature variations during the experiment.
convection dominated. The discretized equation of HTF and PCM The charging experiments are carried out at two different HTF flow
are solved simultaneously for the new values of temperature of HTF rates of 0.015 kg/s and 0.05 kg/s. The HTF at the inlet is maintained
and enthalpy of PCM at increasing time step. The temperature of at a constant temperature of 70  C during the charging process. The
the PCM is calculated from the enthalpy temperature relationship temperature variations of the HTF and PCM at various heights of the
at every time step. The above calculation is repeated by considering storage tank are recorded and these measurements are used to
this new set of temperature values as the present values for finding evaluate the complete charging time for any given height.
the temperature in the next time step. The simulation is continued
until the temperature of the HTF and the PCM at all the nodes of the
3.2. Error analysis
storage tank reaches the inlet temperature of the HTF.
The error associated with primary experimental measurements
3. Experimental investigation
such as temperature and mass flow rate and the performance
parameters evaluated such as instantaneous and cumulative heat
3.1. Experimental setup
transfer using the root sum square method are listed in Table 3.
An experimental study has been performed with the experi-
mental setup shown in Fig. 4. It consists of a storage tank made of 4. Results and discussion

AIR OUTLET Fig. 5a and b show the charging time of the PCM at various
heights of the bed evaluated experimentally, and also using the
three different models for two different mass flow rates of 0.05 kg/s
and 0.015 kg/s respectively, in order to validate and assess the
capabilities of the models. The charging time refers to the time
required for the PCM at a given height, to attain the temperature of
PCM PACKED the inlet HTF. It is seen from the figure that the results of model 3
BED
are in good agreement with the results of the experiments in both
of the cases. There is absolutely no difference between the results of
AIR BLOWER CONTROL
VALVE
Table 3
Estimated values of uncertainties.

Parameters Error (%)

Temperature 1.05
ORIFICE METER Mass flow rate 1.00
AIR HEATER Instantaneous heat transferred 1.45
Cumulative heat transferred 1.48
Fig. 4. Schematic diagram of the experimental setup.
 S. Karthikeyan, R. Velraj / International Journal of Thermal Sciences 60 (2012) 153e160 157

Air flow rate – 0.05 kg/s Air flow rate – 0.015 kg/s
a 300
b 700
d = 70 mm d = 70 mm
250 600

Charging Time, min

Charging Time, min

200 500
400
150
300
100
expt 200 expt
50 model 1 , 2
100 model 1 , 2
model 3
0 model 3
0
0.2 0.4 0.6 0.8 1
0.2 0.4 0.6 0.8 1
Bed height, x/L
Bed height, x/L

Fig. 5. Charging time of PCM at various height of the bed for air flow rate of a) 0.05 kg/s and b) 0.015 kg/s.

models 1 and 2; however, they show a considerable deviation from It is seen from Fig. 6a and b that model 1 under-predicts the
the experimental results. The difference between models 1 and 2 is charging time, and the difference is higher when water is the heat
the consideration of the additional effect of the axial heat transfer fluid. Further, it is seen that, for a given heat transfer fluid,
conduction of the HTF and PCM in model 2. Since there is no the difference between the results of model 1 and 3 decreases as
difference in the results between models 1 and 2, it is concluded the ball size decreases. It is also observed that for a given ball size, in
that the effect of the axial conduction is negligible compared to the both the cases of air and water as the HTF, the deviation of the
heat transfer by convection of the flowing HTF. Hence, it is results between the models is less at a lower mass flow rate. In
construed that while modeling the packed bed storage system with addition, the deviation in the results is marginally higher at a lower
spherical encapsulated containers using the porous medium bed height. In order to present the difference between the results of
approach, the effect of conduction along the flow direction of the models 1 and 3 quantitatively, the results are presented as
HTF need not be considered in the model. The additional internal percentage deviation in Table 5.
conductive resistance of the PCM considered in the model 3, results The reasons for the above deviations between the results of
in closer agreement with the experimental results. models 1 and 3 are summarized below.
Additional parameters, like the size of the ball, and the There are two resistances, namely, surface convective resistance
surrounding heat transfer coefficient which depends on the type of and internal conductive resistance, offered by the PCM in the
the HTF and its flow velocity, may also influence the results of the spherical capsules, that influence the heat transfer between the HTF
various models. Hence, to assess the performance of the models, and the PCM, which in turn, is reflected in the charging time. Among
a numerical analysis is performed for the commonly used HTF (air the various parameters considered in the present study, the physical
and water) at two different mass flow rates (0.05 kg/s and 0.015 kg/ properties of the HTF, the mass flow rate of the HTF, and the size of
s) and three different ball sizes (100 mm, 70 mm and 50 mm). the ball vary the surface convective heat transfer coefficient. The
The surface heat transfer coefficients for the given mass flow surface heat transfer coefficient is higher, when water is the heat
rate and the HTF during the flow over the spherical capsules were transfer fluid compared to air, and it increases when the mass flow
evaluated, using the correlation given in Eq. (1), (Perry and Green); rate increases, and the diameter of the ball decreases in both the
they are given in Table 4 and these values are used in the numerical cases of air and water, as seen from Table 4. The internal conductive
analysis. resistance, which varies as the melt front moves, is dependent on the
Fig. 6a and b show the charging time of the PCM evaluated at size of the balls. In both cases of the HTF, the difference between the
various heights of the bed at two different mass flow rates of results of the models 1 and 3 decreases as the ball size decreases,
0.05 kg/s and 0.015 kg/s using two different models 1 and 3 for which is due to the decrease in the internal conductive resistance.
various sizes of the balls (50, 70 and 100 mm) drawn for the heat Hence, the results of model 1 approach those of model 3, when the
transfer fluid of air and water respectively. Model 2 is not consid- ball size decreases, that decreases the effective internal conductive
ered, as the results of models 1 and 2 are similar, and hence, the resistance. Both the models have given the same results when air is
simple model is considered for comparison with model 3. Since the the HTF, with the smallest ball size of 50 mm, and at a lower mass
results of model 3 are in good agreement with the experimental flow rate of 0.015 kg/s. However, when water is the HTF, a consid-
results, the percentage deviation of the results of the simplified erable deviation between the results of models 1 and 3 is observed,
model is compared with the results of model 3. even under similar conditions. The deviation is still higher when the
size of the ball increases. This is due to the increase in internal
Table 4 conductive resistance in the PCM as the melting proceeds, and it
Surface convective heat transfer coefficient at different conditions. predominates the surface convective resistance, which is compara-
HTF Ball size, in mm Heat transfer coefficient, W/m2 K tively lesser, when air is the HTF. Hence, when the internal
m ¼ 0.05 kg/s m ¼ 0.015 kg/s
conductive resistance is not considered in model 1, this model under-
predicts the charging time which is much higher when compared to
Air 100 32.9 20.9
70 42.8 27.2 model 3. When air is the HTF, the surface heat transfer coefficient is
50 53.7 34.1 lower, and hence, the variation in the internal conduction resistance
Water 100 224.6 142.7 has only a marginal effect on the overall heat transfer coefficient.
70 291.9 185.5 Hence, the variation of the results of models 1 and 3 is smaller, when
50 366.1 232.6
air is the HTF compared to water as the HTF.
158 S. Karthikeyan, R. Velraj / International Journal of Thermal Sciences 60 (2012) 153e160

HTF : Air HTF : Water

a 900 b 350
800 d = 100 mm 300 d = 100 mm

Charging Time, min

Charging Time, min

700
250
600
500 200
400 150
300
100
200
100 50
0 0
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1
Bed height, x/L Bed height, x/L

700 300
d= 70 mm d = 70 mm
600 250

Charging Time, min

Charging Time, min

500 200
400
150
300
100
200
50
100
0 0
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1
Bed height, x/L
Bed height, x/L

600 250
d = 50 mm d= 50 mm
500
Charging Time, min

Charging Time, min

200
400
150
300
100
200
100 50

0 0
0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1
Bed height, x/L Bed height, x/L

model 1 ; m = 0.05 kg/s model 3 ; m=0.05 kg/s

model 1 ; m = 0.015 kg/s model 3 ; m = 0.015 kg/s

Fig. 6. Charging time of PCM evaluated using two different models at various height of the bed for various sizes of the balls, 50, 70 and 100 mm at the flow rates of 0.05 kg/s and
0.015 kg/s for a) HTF: air and b) HTF: water.

It is construed from the above discussion, that the internal modeling the heat transfer phenomena as in model 1. However,
conductive resistance need not be included when the surface heat when the surface heat transfer coefficient is higher (when water is
transfer coefficient is very low (when air is used as the HTF and also used as the HTF with higher mass flow rates), and when the
at low mass flow rates), and the internal conductive resistance is internal conductive resistance plays a major role as in the case of
comparatively lower as in the case of small sized balls, while large sized balls, the internal conductive resistance needs to be
considered while modeling the heat transfer phenomena as in
Table 5
model 3.
Percentage deviation in charging time between results of model 1 and 3 at different
conditions. Fig. 7 shows the variation in the instantaneous heat transfer,
evaluated using two different models, for the heat transfer fluids of
HTF Ball size, in Percentage deviation in charging
air and water, for three different ball sizes of 100 mm, 70 mm and
mm time between model 1 and 3 at
the bed height (x/L ¼ 0.2) 50 mm, at the mass flow rates of 0.05 kg/s and 0.015 kg/s respec-
tively. It is observed from Fig. 6, that the heat transfer fluid has
m ¼ 0.05 m ¼ 0.015
kg/s kg/s
more effect on the charging time, and hence, in order to compare
the effect of water and air as the HTF, the effect of those two fluids
Air 100 40 17.7
70 35.3 20.6 on the instantaneous heat transfer is shown, and compared in all
50 20.5 0.4 the figures. It is seen from all the figures, that the variation in the
Water 100 81.1 61.4 instantaneous heat transfer is marginal when air is the heat
70 74.2 46.4 transfer fluid, and it is significant when water is the heat transfer
50 58.3 22.2
fluid.
 S. Karthikeyan, R. Velraj / International Journal of Thermal Sciences 60 (2012) 153e160 159

9 2.5

Instantaneous heat transfer, kW

Instantaneous heat transfer, kW

8 d = 100 mm d = 100 mm
7 m = 0.05 kg/s m = 0.015 kg/s
2
6
5 1.5
4
3 1

2
0.5
1
0 0

Time, min Time, min

9 2.5

Instantaneous heat transfer, kW

Instantaneous heat transfer, kW

8 d = 70mm d = 70 mm
2 m = 0.015 kg/s
7 m = 0.05 kg/s
6
1.5
5
4
1
3
2 0.5
1
0 0

Time, min Time, min

9 2.5
Instantaneous heat transfer, kW
Instantaneous heat transfer, kW

8 d = 50 mm
d = 50 mm 2
7 m = 0.015 kg/s
m = 0.05 kg/s
6
1.5
5
4 1
3
2 0.5
1
0 0

Time, min Time, min

model 1 - Air model 3 - Air model 1 - Water model 3 - Water

Fig. 7. Instantaneous heat transfer evaluated using two different models the heat transfer fluids air and water for various sizes of the balls, 50, 70 and 100 mm.

Though an appreciable deviation is seen from Fig. 6 for the

variation in the mass flow rates and size of the balls on the charging
time, similar effects are not visible in the instantaneous heat
Table 6
transfer. This is due to the very low heat transfer involved in the last Processor time for simulation using model 1 and model 3.
stage of the phase change process, with a very small temperature
HTF Ball size, Mass flow Processor time, in s
difference between the HTF and PCM, which increases the differ-
in mm rates of HTF,
ence in the charging time between models 1 and 3, as observed in in kg/s
Model 1 Model 3
Fig. 6. Since, the instantaneous heat transfer is very important in Air 100 0.015 46 145
designing the PCM based storage system, it is concluded from the 0.05 14 72
results that model 1 is sufficient when air is the heat transfer fluid, 70 0.015 31 124
and the effect of the internal conductive resistance considered in 0.05 12 53
50 0.015 27 98
model 3 is important, when water is the heat transfer fluid.
0.05 09 41
In order to assess the variation in the computational time, the Water 100 0.015 11 69
simulations of models 1 and 3 with different operating parameters 0.05 04 44
were performed in a DELL inspiron computer with Intel core i3 CPU 70 0.015 12 56
at 2.27 GHz processing speed. The grid size and time increments for 0.05 03 26
50 0.015 10 41
all the simulation runs are kept the same, and the processor time 0.05 03 19
for each was observed and given in Table 6. It is seen from the
160 S. Karthikeyan, R. Velraj / International Journal of Thermal Sciences 60 (2012) 153e160

results that the processor time for model 3 is almost 3e4 times le latent heat of the PCM during the ‘solideliquid’ transition,
more than that of model 1, and therefore, the simple model is kJ/kg
preferable for conditions, where the simulation results of both the la1 latent heat of the PCM from ‘onset to peak’ during the
models are in close agreement. ‘solidesolid’ transition, kJ/kg ¼ la(a1/(a1 þ a2))
la2 Latent heat of the PCM from ‘peak to end’ during the
5. Conclusions ‘solidesolid’ transition, kJ/kg ¼ la(a2/(a1 þ a2))
le1 latent heat of the PCM from ‘onset to peak’ during the
In the present work three different methods of modeling the ‘solideliquid’ transition, kJ/kg ¼ le(e1/(e1 þ e2))
PCM encapsulated packed bed storage system is adopted and le2 latent heat of the PCM from ‘peak to end’ during the
initially the results obtained from these models are compared with ‘solideliquid’ transition, kJ/kg ¼ le(e2/(e1 þ e2))
the experimental results. There is no variation between the results m dynamic viscosity, kg/m s
of model 1 and 2 and hence it is concluded that the effect of axial r density, kg/m3
conduction is negligible compared to the heat transfer by convec-
tion by the flowing HTF. However, the results of models 1 and 2
show a considerable deviation from the experimental results Subscripts
whereas the results of model 3 are in close agreement with the f HTF
experimental results which is due to the consideration of additional p PCM
internal conductive resistance of the PCM. Further, it is concluded ini initial
from the results of the instantaneous heat transfer, that model 1 is s solid PCM
sufficient in order to reduce the computational time when air is the l liquid PCM
HTF and model 3 is recommended when water is the heat transfer
fluid.
Abbreviation
Acknowledgment PCM Phase Change Material
HTF Heat Transfer Fluid
The authors would like to sincerely thank Professor C. Balaji, TES Thermal Energy Storage
Department of Mechanical Engineering, Indian Institute of LHS Latent Heat Storage
Technology-Madras, Chennai, India for providing valuable SHS Sensible Heat Storage
comments during the research.

Nomenclature References

[1] A. Felix Regin, S.C. Solanki, J.S. Saini, Heat transfer characteristics of thermal
a1 temperature range from ‘onset to peak’ during the energy storage system using PCM capsules: a review, Renewable & Sustain-
‘solidesolid’ transition,  C able Energy Reviews 12 (2008) 2438e2458.
a2 temperature range from ‘peak to end’ during the [2] Prashant Verma, Varun, S.K. Singhal, Review of mathematical modeling on
latent heat thermal energy storage systems using phase-change material,
‘solidesolid’ transition,  C Renewable & Sustainable Energy Reviews 12 (2008) 999e1031.
ap surface area of the PCM sphere, m2 [3] T. Saitoh, K. Hirose, High-performance phase-change thermal energy storage
Ac cross sectional area of the storage tank, m2 using spherical capsules, Chemical Engineering Communication 41 (1986)
39e58.
C specific heat, kJ/kg K [4] S.L. Chen, J.S. Yue, Water thermal storage with solidification, Heat Recovery
dp diameter of the PCM spherical ball, m Systems & CHP 11 (1991) 79e90.
e1 temperature range from ‘onset to peak’ during the [5] S.L. Chen, J.S. Yue, Thermal performance of cool storage in packed capsules for
‘solideliquid’ transition,  C air conditioning, Heat Recovery Systems & CHP 11 (1991) 551e561.
[6] Takayuki Watanabe, Hisashi Kikuchi, Atsushi Kanzawa, Enhancement of
e2 temperature range from ‘peak to end’ during the charging and discharging rates in a latent heat storage system by use of PCM
‘solideliquid’ transition,  C with different melting temperatures, Heat Recovery Systems & CHP 13 (1993)
G mass velocity, kg/m2 s 57e66.
[7] S. Keumnam Cho, H. Choi, Thermal characteristics of paraffin in a spherical
hs surface heat transfer coefficient between HTF and PCM, capsule during freezing and melting processes, International Journal of Heat
kW/m2  C and Mass Transfer 43 (2000) 3183e3196.
H enthalpy, kJ/kg [8] A. Benmansour, M.A. Hamdan, A. Bengeuddach, Experimental and numerical
investigation of solid particles thermal energy storage unit, Applied Thermal
k thermal conductivity, kW/m  C Engineering 26 (2006) 513e518.
L height of the storage tank, m [9] A. Felix Regin, S.C. Solanki, J.S. Saini, An analysis of a packed bed latent heat
ri radius of the PCM at any inner node, m thermal energy storage system using PCM capsules: numerical investigation,
Renewable Energy 34 (2009) 1765e1773.
ro outer radius of the PCM sphere, m [10] M. Cheralathan, R. Velraj, S. Renganarayanan, Heat transfer and parametric
t time, s studies of an encapsulated phase change material based cool thermal energy
T temperature,  C storage system, Journal of Zhejiang University Science 7 (2006) 1886e1895.
[11] N. Nallusamy, R. Velraj, Numerical and experimental investigation on
Tm1 peak temperature of the PCM during the solidesolid
a combined sensible and latent heat storage unit integrated with solar water
transition,  C heating system, Journal of Solar Energy Engineering 131 (2009) 1e8.
Tm2 peak temperature of the PCM during the solideliquid [12] Shuangmao Wu, Guiyin Fang, Xu Liu, Thermal performance simulations of
transition,  C a packed bed cool thermal energy storage system using n-tetradecane as
phase change material, International Journal of Thermal Sciences 49 (2010)
vmax HTF velocity along the bed void fraction, m/s 1752e1762.
x axial direction of the storage tank [13] Aroul Raj Antony, R. Velraj, Heat transfer and pressure drop studies on a PCM-
ε void fraction heat exchanger module for free cooling applications, International Journal of
Thermal Sciences 50 (2011) 1573e1582.
la latent heat of the PCM during the ‘solidesolid’ transition, [14] Robert H. Perry, Don W. Green, Perry’s Chemical Engineers Handbook,
kJ/kg seventh ed. McGraw Hill, New York, 1997.