Professional Documents
Culture Documents
www.elsevier.com/locate/apthermeng
Abstract
Application of a novel PCM package for thermal management of portable electronic devices was investigated experimentally for
effects of various parameters e.g. power input, orientation of package, and various melting/freezing times under cyclic steady conditions.
Also, a two-dimensional numerical study was made and compared the experimental results. Results show that increased power inputs
increase the melting rate, while orientation of the package to gravity has negligible effect on the thermal performance of the PCM pack-
age. The thermal resistance of the device and the power level applied to the PCM package are of critical importance for design of a pas-
sive thermal control system. Comparison with numerical results confirms that PCM-based design is an excellent candidate design for
transient electronic cooling applications.
2007 Published by Elsevier Ltd.
Keywords: Phase change materials; Electronics cooling; Thermal management; Cyclic steady state
Nomenclature
A(/) porosity function, A(/) = C(1 /)2/(/3 + e), l dynamic viscosity, kg/m s
e = 0.001 / liquid fraction
A, B constant in Eq. (5) q density, kg/m3
cp specific heat, J/kg K RT
h specific enthalpy, h ¼ href þ T ref cp dT Subscripts
k thermal conductivity, W/m K i coordinate component
L latent heat, J/kg l liquid phase
S momentum source term, S = A(/)u m melting
Sh latent heat source term, S h ¼ oto ðqLÞ n the nth fluid
Dt time step, s ref reference
T temperature, C s solid phase
u velocity, m/s w wall
x Cartesian coordinate, m
a fluid’s volume fraction
Greek symbols
b thermal expansion coefficient, K1
dij Kronecker delta
Gaye [9] also analytically examined one-dimensional solid- quad PCM package [19,20] was fabricated for this purpose.
ification and melting of a slab with uniform volumetric Furthermore, a generalized two-dimensional computa-
energy generation. However, the low thermal conductivity tional model is developed to assess the heat transfer perfor-
of PCMs presents a significant challenge in design of elec- mance of the PCM package.
tronic cooling systems. In order to overcome this draw-
back, researchers have proposed various heat transfer 2. Experimental setup
enhancement techniques, e.g. use of partitions/fins [10],
graphite/metal matrices [11,12], dispersed high-conductiv- The experimental setup was designed to simulate the
ity particles in the PCM [13], micro-encapsulation of applications of typical portable electronic device. The sche-
PCM [14,15], and PCM-based heat sinks [16]. matic of the setup is shown in Fig. 1a and the illustration of
For thermal management of electronic components, the the actual setup is shown in Fig. 1b. The over-sized enclo-
chip and other functional electronic components must be sure of the PCM package is intentionally designed to pro-
kept below their respective allowable maximum tempera- vide good resolution of the internal temperature profile
ture at all times during normal operation. Generally, with the PCM. Considering that this is a transient heat
considering that each chip has a different maximum transfer problem in a 3D domain, spatially spaced thermo-
operating temperature, the global maximum allowable couples were placed inside the PCM slab at fixed spatial
temperatures range from 85 to 120 C to prevent the com- intervals to record the evolution of temperatures with time
ponents from overheating. Furthermore, the temperature at the monitored points. As shown in Fig. 1, thermocouple
tolerated by humans must be considered as well. Henry location P1 was selected for temperature measurement at
Dreyfuss Associates (1993) [17] report that the maximum the mid-center of the PCM, while P2 and P3 were referred
environmental temperature human can tolerate for a one to temperature of heater and plastic case, respectively. The
period is 49 C and that metals and non-metals become dis- experiments were limited to a single PCM (paraffin wax),
ruptive to human tissues at 50 and 62 C, respectively. whose properties are shown in Table 1.
Based on this report, Leoni and Amon [18] assumed that The heat dissipation by the chip module was simulated
humans can comfortably hold plastic objects up to 45 C. by placing a single uniform electrical heater on the sides
Such basic criteria are involved in the thermal control of the PCM package. The rest of the other sides were
design of portable electronic devices. encased in using a 3 mm perspex sheet. As shown in
The objective of this work is to investigate the feasibility Table 1, perspex is chosen because its low thermal conduc-
of using PCMs for application in thermal control of porta- tivity is 0.17 W/m/K, which is the same as that of ABS. and
ble electronic devices which are used intermittently. The it is transparent as well so that usual observation of the
heat transfer characteristics of a PCM-based control unit melting phenomenon could be made. ABS is used in most
are tested experimentally for various power input levels, external casings of portable electronic devices.
orientations of the PCM package, and various melting/ The data acquisition unit used was an Agilent 34970A
freezing times under cyclic steady state conditions. A flat Data Acquisition Unit, linked to a PC. This setup enables
2824 R. Kandasamy et al. / Applied Thermal Engineering 27 (2007) 2822–2832
Table 1
Properties of paraffin wax and perspex
ql (kg/m3) qs (kg/m3) kl (W/m K) ks (W/m K) cp (J/kg K) Tm (C) L (J/kg)
Paraffin wax 750 900 0.12 0.21 2890 46–48 173,400
Perspex – 1190 – 0.7 1500 80 900
the temperature of each thermocouple to be captured at external temperatures etc. was derived for three orienta-
fixed time intervals and plotted in real-time. Thirty one tions of the package (horizontal, vertical and slanted at
T-type thermocouples were attached to the internal PCM 45) at several power levels which were either steady or
cavity as well as its external surfaces. An additional ther- time dependent.
mocouple was fixed in the vicinity of the PCM package
to record the ambient temperature. The material properties
for our experiments are listed in Table 2. 3. Experimental results
For each power level, at least three different runs were
conducted to verify reproducibility of the data. One typical In this experimental study, the effects of the following
test for local temperature (P1 in Fig. 1) evolution with time parameters were examined:
9 W case in the 45 orientation for both melting and freez-
ing phases are shown in Fig. 2. The uncertainties for tran- • Power level. For the vertical orientation, power input
sient temperature recorded and power measurement for a levels were set at 6 W, 9 W, and 12 W giving power den-
power level of 12 W were within ±3.5% and ±0.5%, respec- sity values of at 2.4 kW/m2, 3.6 kW/m2, and 4.8 kW/m2,
tively. Hence the resulting uncertainty for calculated tem- respectively.
perature was within 4%, which also can be seen from Fig. 2. • Orientations with respect to gravity. For fixed power
From the data collected, information such as melting input of 12 W, orientations examined were: horizontal,
time, time to attain steady state, maximum heater and vertical, and inclined at 45 to the vertical.
• Melting/freezing times until cyclic steady state (CSS) is
Table 2 reached. During CSS for the vertical orientation case
Properties of materials with 12 W power input, various melting/freezing times
Material k (W/m K) q (kg/m3) cp (J/kg K) were selected for testing.
Silicon (die) 117.5–0.42 (T 100) 2330 700
Substrate 14.0 3690 880 To analyze the thermal performance of the PCM pack-
Aluminum 180 2700 963 age, values of some important variables must be clarified
PCB 0.3 1200 880
before detailed discussion. During experimental testing,
R. Kandasamy et al. / Applied Thermal Engineering 27 (2007) 2822–2832 2825
Fig. 2. Typical reproducibility of results for the wax temperature at location P1 for 45 orientation (9 W).
the temperature of the heater should be kept under 85 C Fig. 3 shows temperatures of the heater (P2 in Fig. 1) at
since it is typically the highest operating temperature for various power levels. Since the heater is directly attached
most chips to ensure reliability. Also, the performance of to the PCM package, the heater temperature reflects the
the CCD in digital cameras, for example, degrades above status of the PCM in the package. Since the heater reached
85 C. Since the plastic casing is the surface users actually its maximum temperature in Fig. 3 data, it implies that the
hold during use of portable electronic devices, the external PCM had melted fully. It is observed that a lower power
temperature should be held below 45 C for use comfort. level results in longer time for full melting and also a lower
These constraints were considered in this study. maximum temperature. However, it must be noted that the
power input is decided by electronic device itself and not by
3.1. Effect of power levels users. Hence, various power input levels have to be consid-
ered in design. With increase of power from 6 W to 9 W to
First, a similar experimental setup but without the PCM 12 W, the time for full melting decreased from 8850 s to
package was built and tested for direct comparison of its 8050 s to 6400 s, while the maximum temperature of the
thermal performance with the PCM package. Results monitored point varied from 58 C to 73 C to 82 C,
showed that the heater and plastic case temperatures of respectively.
the former device exceeded 100 C within 20 min. This Also, as can be seen from Fig. 3, the time for solidifica-
means the PCM package, as expected, has the capacity to tion of the PCM package during which the heat accumu-
hold back temperature rise of the device at critical points. lated during melting is dissipated is much longer than
Fig. 3. Comparison of the heater temperature (P2) for various power input levels.
2826 R. Kandasamy et al. / Applied Thermal Engineering 27 (2007) 2822–2832
that for the melting phase. Note that the setup was cooled There could be an optimum PCM size for each power
by natural convection cooling rather than by forced con- input that needs to be tested for various input parameters
vection cooling. The PCM at the mid-plane takes a longer and PCM materials. This optimized size can be defined
time to freeze due to the poor thermal conductivity of par- as the smaller size of PCM slab required to maintain the
affin wax. However, the effect was minimized by the small heater below 85 C and the external plastic case under
thickness of the package. Actually this is also the core idea 45 C for the duration of life of the battery life of the
of small capsulated PCMs for increasing the effective ther- device. The latent heat of fusion for the PCM must be uti-
mal conductivity. Hence, for small packages, similar to lized in the 9 W and 12 W cases. The 6 W case shows that
ones discussed here, methods for enhancing the thermal the PCM could be of a smaller size and yet achieve the
conductivity of PCM such as use of metal meshing struc- designed temperature profiles. This reduction in size is crit-
tures or partitions may not be necessary as they increase ical for cost considerations and for keeping the overall size
the cost of such devices. However, for large PCM pack- and weight of the device small.
ages, the improvement of the thermal conductivity of
PCM itself would enhance its thermal performance and 3.2. Effect of orientation
hence may be worth considering.
Fig. 4 shows the temperature of the plastic case (P3 in Fig. 5 shows the heater temperature (P2 in Fig. 1a) for
Fig. 1) at various power levels. As mentioned earlier, the various orientations for 12 W power input: horizontal, ver-
plastic case is held by the user. Hence, its maximum tem- tical, and inclined at 45 to the vertical. Experimental
perature should be controlled to be comfortable for users. results show that the orientation of the simulated package
However, poor heat spreading can result in undesirably affected its steady state temperature and the cooling time of
high temperatures at some parts of the device. At the back PCM because of free convection effects in the PCM melt.
of the plastic case next to the heater, temperature profiles For the vertical orientation case, the time for full melting
recorded were beyond the comfort level within 5 min of is decreased due to the increased effect of natural convec-
operation. However, on the front side of the plastic casing, tion. However, as seen from this figure, the effects of orien-
steady temperatures for 9 W and 12 W were slightly below tations were small enough to be ignored for practical
the melting point of the PCM (T = 42 C). The reason for applications.
this observation is that the PCM near the walls undergoes Fig. 6 demonstrates the measured temperature evolution
phase change at the later times. On the other hand, for the of the plastic case (P3 in Fig. 1a) for three typical orienta-
case of 6 W, the steady state temperature at the same loca- tions. Similar to the temperature distribution over the hea-
tion was much lower than the PCM’s melting temperature ter, the effect of orientation on the plastic case temperature
(T = 34 C) as the PCM near the walls was still in solid is small. The maximum temperature decreases for horizon-
phase. Therefore, if a heat conduction path of low thermal tal to slanted, to the vertical position. During use of porta-
resistance could be provided from the back of the heater to ble electronic devices such as mobile phones or PDAs, the
the PCM package, the temperatures on all sides of the plas- position and orientation of the device is subject to random
tic case could be sustained at a comfortable level. Alterna- change. Considering the negligible effect of orientation on
tively, both sides of the heater can be clad with PCM the thermal performance of PCM-based thermal system,
packages that are half the thickness of the original to these models can be expected to work safely in real
accomplish the same effect. environment.
Fig. 4. Comparison of the plastic case temperature (P3) for various power inputs.
R. Kandasamy et al. / Applied Thermal Engineering 27 (2007) 2822–2832 2827
Fig. 5. Comparison of the heater temperature (P2) for various orientations (12 W).
Fig. 6. Comparison of the plastic case temperature (P3) for various orientations (12 W).
3.3. Effects of melting/freezing times ing characteristics of a small PCM package rather than
improving the internal thermal resistance within the PCM
Fig. 7 shows the evolution of heater temperature with package. For the application of PCMs in the portable elec-
time for different melting/freezing times under cyclic steady tronic industry, the size of the PCM package should not
state (CSS). Comparing the cyclic and constant power exceed about ten percent of the whole device. In the case
experiments for 12 W in vertical orientation, the cyclic of cell phones with camera functions and digital cameras,
one was able to sustain temperatures at all points of the this would be about the size of two strips of chewing
simulated device at very low levels (55 C for heater and gum. Considering the small size required, the effect of ther-
34 C for external casing temperatures) for large on/off mal resistance to the ambient would have greater impact on
cycles. Thus, when designing a PCM package for electronic the performance rather than the improvement of the inter-
cooling, the average power (such as RMS value) over the nal thermal resistance of the PCM.
duration of the battery life could be used as a conservative There is no concrete conclusion that can be drawn from
design criterion. Else, if constant peak power was assumed the infra-red images of the melting patterns of the PCMs as
for design, the resultant PCM package would be over-sized shown in Fig. 8 since they are not the direct reflection of
to meet the cooling needs in operation of the device. The the PCM melting temperature. Initially, these thermal
unnecessary bulk would add to the cost, size and weight images show the external case temperature variations of
of the final device. the enclosure rather than the melting phenomena. Thermal
The thermal resistance between the PCM and the ambi- resistance from the PCM to the ambient has a significant
ent has a significant effect on the equilibrium temperatures effect on the equilibrium temperatures of the device and
of the device and the duration of temperature suppression. the duration of temperature suppression. This aspect is
This aspect is important to improve the melting and freez- more important to improve the melting and freezing
2828 R. Kandasamy et al. / Applied Thermal Engineering 27 (2007) 2822–2832
Fig. 7. Comparison of the heater temperature (P2) for different melting/freezing times at cyclic steady state.
Fig. 8. Experimental IR images for melting at vertical 12 W constant power (external slide profile).
characteristics of a small PCM package rather than neously for the solid fins, solid and liquid PCM, and air,
improving the internal thermal heat resistance within the while allowing for PCM expansion, convection in the
PCM package. fluid media (melted PCM and air), and solid motion in
the liquid. Detailed temperature and phase fields have
4. Numerical model been obtained as function of time, showing evolution of
the heat transfer in the system as the phase change material
A schematic diagram of the two-dimensional physical melts.
model is shown in Fig. 9. Similar to our experimental setup As shown in Fig. 9, the PCM package is 12 mm wide
as shown in Fig. 1, the PCM (paraffin wax) is filled in the and 132 mm high with 2 mm thick aluminum wall, with a
aluminum casing with 10 mm empty space at the top, PCM stored in the cavity 10 mm thick and 120 mm high
which is open to ambient to allow expansion of PCM vol- (top 10 mm is left for ambient air). The left wall is attached
ume during melting. Modeling of phase change processes to a constant heat flux source with 50 mm high to supply
presents a significant challenge due to the complexity and heat for melting of PCM. From above, the PCM is exposed
conjunction of the involved physical phenomena such as to ambient air. Thus, it is assumed that heat is transferred
volumetric expansion due to the phase change, convection between the PCM and the base, the left wall, the right wall
in the liquid phase, and motion of the solid in the melt due and the ambient air. Heat transfer between the fin tips and
to density differences. the ambient is neglected. In all the cases, the PCM used is a
The present work attempts to overcome the limitations commercially available paraffin wax, whose properties
and to solve complete conservation equations simulta- shown in Table 1.
R. Kandasamy et al. / Applied Thermal Engineering 27 (2007) 2822–2832 2829
Heater 2
q"
where A = 4.25 and B = 1790 following Reid et al. [23].
Other properties of PCM, aluminum, and air are summa-
130 mm rized in Tables 1 and 3.
50 mm 0 3 Since the proposed PCM-based heat sink problem is
PCM time dependent, initial and boundary conditions have to
be set appropriately before calculation. For the studied
cases here, initially, the whole system was kept at 25 C.
4
Then, a constant or variable temperature Tw was put at
the bottom wall of the heat sink, with continuous or dis-
crete heaters. The top of the air was open to ambient using
5
10 mm 2 mm constant pressure outlet conditions.
The numerical solution was carried out using the Fluent
6.2 software [21]. The PISO algorithm was used for pres-
sure–velocity coupling. The number of computational grids
Base was around 25,300 for the 2D model after grid-indepen-
dence tests. Furthermore, after a careful examination of
Fig. 9. 2D physical model.
the preliminary calculations, the time step in the simula-
For the phase change region in the PCM, the enthalpy- tions was as small at Dt = 0.01 s. The convergence was also
porosity method [21] was used. Hence, the porosity in each checked at each time step, with convergence criteria of 104
computational cell is set equal to the liquid fraction in that for velocity components and 107 for energy.
cell. Accordingly, the two-dimensional governing equa-
tions used here for the PCM-air system are [21]: 5. Numerical results
Fig. 10. Temperature distribution of the 2D PCM slap for various time (48 W): (a) 50 s; (b) 100 s; (c) 200 s; (d) 300 s.
Fig. 11. Liquid fraction distribution of the 2D PCM slap for various time (48 W): (a) 50 s; (b) 100 s; (c) 200 s; (d) 300 s.
shown in Fig. 8, the evolution of the predicted temperature easily. However, for the 2D numerical model, heat can only
contours is different. First, the experimental setup is three be conducted through the bottom wall and the ambient air.
dimensional, where the heat is conducted through the outer It is noted that the bottom is not a good thermal path,
aluminum walls as well. Also, the air gap in the experimen- which is also demonstrated in experiment. Hence, heat is
tal case is rather small since the compressed can be leaked mainly carried away by the heated air to the right side.
R. Kandasamy et al. / Applied Thermal Engineering 27 (2007) 2822–2832 2831
6. Conclusions
expected operating time of the device should be used [9] L.M. Jiji, S. Gaye, Analysis of solidification and melting of PCM with
instead of the peak power level over the same amount of energy generation, Appl. Therm. Eng. 26 (2006) 568–575.
[10] C. Bauer, R. Wirtz, Thermal characteristics of a compact, passive
time. After adding in a safety factor, a PCM package can thermal energy storage device, in: Proceedings of the 2000 ASME
be designed which can meet not only the cooling needs of IMECE, Orlando, USA, 2000.
the device, but also the cost and size requirements. [11] H. Mehling, S. Hiebler, S. Ziegler, Latent heat storage using a PCM-
Another aspect to consider for a PCM thermal manage- graphite composite material: advantages and potential applications,
ment system is the thermal resistance of the PCM to the In: Proceedings of the 4th Workshop of IEA ECES IA Annex 10,
Bendiktbeuern, Germany, 1999.
ambient. This is important for both the transient cooling [12] M. Kamimoto, Y. Abe, K. Kanari, S. Swata, T. Tani, T. Ozawa, Heat
during the peak operation of the device and for the freezing Transfer in Latent Heat Thermal Storage Units Using Pentarythritol
of the PCM during low power times. Improvement in this Slurry, Thermal Energy Storage, World Congress of Chemical
respect means a smaller PCM package can be used for the Engineering, Tokyo, Japan, 1986.
same cooling needs. [13] A. Benmansour, M. Hamdan, A. Bengeuddach, Experimental and
numerical investigation of solid particles thermal energy storage unit,
Appl. Therm. Eng. 26 (2006) 513–518.
References [14] R. Velraj, R.V. Seeniraj, B. Hafner, C. Faber, K. Schwarzer, Heat
transfer enhancement in a latent heat storage system, Sol. Energy 65
[1] Z.X. Gong, A.S. Mujumdar, A new solar receiver thermal store for (3) (1999) 171–180.
space-based activities using multiple composite phase change mate- [15] B. Zalba, J.M. Marin, L.F. Cabeza, H. Mehlin, Free-cooling of
rials, ASME J. Solar Energy Eng. 117 (1995) 215–220. building with phase change materials, Int. J. Refrig. 27 (2004) 839–
[2] Z.X. Gong, A.S. Mujumdar, Cyclic heat transfer in a novel storage 849.
unit of multiple phase change materials, Appl. Therm. Eng. 16 (10) [16] D. Pal, Y.K. Joshi, Melting in a side heated tall enclosure by a
(1996) 807–815. uniformly dissipating heat source, Int. J. Heat Mass Transfer 44
[3] Z.X. Gong, A.S. Mujumdar, Enhancement of energy charge– (2001) 375–387.
discharge rates in composite slabs of different phase change materials, [17] Henry Dreyfuss Associates, The Measure of Man and Women,
Int. J. Heat Mass Transfer 39 (4) (1996) 725–733. Human Factors in Design, Whitney Library of Design, Watson-
[4] Z.X. Gong, A.S. Mujumdar, Exergetic analysis of thermal storage Guptill Publications, 1993, pp. 82–83.
using multiple phase change materials, ASME, J. Energy Res. [18] N. Leoni, C.H. Amon, Transient thermal design of wearable
Technol. 118 (1996) 242–248. computers with embedded electronics using phase change materials,
[5] Z.X. Gong, A.S. Mujumdar, Thermodynamic optimization of the ASME, Natl. Heat Transfer Conf. 7 (1997) 49–56.
thermal process in energy storage using multiple phase change [19] R.R. Tummala, E. Rymaszewski, Microelectronics Packaging Hand-
materials, Appl. Therm. Eng. 17 (11) (1997) 1067–1083. book, Van Nostrand Reinhold, New York, 1989.
[6] Z.X. Gong, A.S. Mujumdar, A finite element model for convection- [20] B. Joiner, S. Neelakantan, Integrated circuit package types and
dominated melting and solidification problems, Int. J. Numer. thermal characteristics, Electron. Cool. 12 (1) (2006).
Methods Heat Fluid Flow 8 (4) (1998) 393–408. [21] http://www.fluent.com.
[7] Z.X. Gong, S. Devahastin, A.S. Mujumdar, Enhanced heat transfer [22] W.R. Humphries, E.I. Griggs, A design handbook for phase change
in free convection-dominated melting in a rectangular cavity with an thermal control and energy storage devices, NASA Technical Paper
isothermal vertical wall, Appl. Therm. Eng. 19 (1999) 1237–1251. 1074NASA Scientific and Technical Information Office, 1977.
[8] G. Casano, S. Piva, Experimental and numerical investigation of the [23] R.C. Reid, J.M. Prausnitz, B.E. Poling, The Properties of Gases and
steady periodic solid–liquid phase-change heat transfer, Int. J. Heat Liquids, McGraw-Hill, New York, 1987, p. 439–456.
Mass Transfer 45 (20) (2002) 4181–4190.