Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 115 (2015) 264–276
www.elsevier.com/locate/solener

Increased photovoltaic performance through temperature regulation

by phase change materials: Materials comparison in different climates
A. Hasan a,⇑, S.J. McCormack b, M.J. Huang c, J. Sarwar d, B. Norton e
a
Department of Architectural Engineering, College of Engineering, United Arab Emirates University, P.O. Box 15551, Al Ain, United Arab Emirates
b
Department of Civil, Structure and Environmental Engineering, University of Dublin, Trinity College, Dublin 2, Ireland
c
Centre for Sustainable Technologies, University of Ulster, Newtownabbey, N. Ireland BT370QB, UK
d
Sustainable Energy Research Laboratory (SERL), Mechanical Engineering Program, Texas A & M University at Qatar, PO Box 23874, Doha, Qatar
e
Dublin Energy Lab., Focas Institute, School of Physics, Dublin Institute of Technology, Kevin St, Dublin 8, Ireland

Received 29 October 2013; received in revised form 23 December 2014; accepted 6 February 2015

Communicated by: Associate Editor Elias Stefanakos

Abstract

A photovoltaic–phase change material (PV–PCM) system has been developed to reduce photovoltaic (PV) temperature dependent
power loss. The system has been evaluated outdoors with two phase change materials (PCMs); a salt hydrate, CaCl26H2O and a eutectic
mixture of fatty acids, capric acid–palmitic acid in two different climates of Dublin, Ireland (53.33N, 6.25W) and Vehari, Pakistan
(30.03N, 72.25E). Both the integrated PCMs maintained lower PV panel temperature than the reference PV panel. Salt hydrate
CaCl26H2O maintained lower PV temperature than capric–palmitic acid at both the tested sites. The lower PV temperatures effected
by the use of the PCMs prevented the associated PV power loss and increased PV conversion efficiencies. Both the PCMs attained higher
temperature drop in warm and stable weather conditions of Vehari than the cooler and variant weather conditions of Dublin.
Ó 2015 Elsevier Ltd. All rights reserved.

Keywords: Phase change materials; Temperature regulation; Photovoltaics; Performance increase

1. Introduction and associated power drop largely depends on the climate

of the site. In Germany 50% of the solar radiation reaching
Silicon photovoltaics (PV) show a power drop above on a PV panel is above 600 W m2 while in Sudan this
25 °C with a temperature coefficient of up to 0.65% K1 value reaches 80% resulting different operating
(Radziemska and Klugmann, 2002; Radziemska, 2003) temperatures and associated power drop (Bücher, 1997;
depending on the type of the PV cell and the manufactur- Amy de la Breteque, 2009). A maximum PV operating tem-
ing technology (Makrides et al., 2009). Various mathemati- perature of 125 °C has been reported in southern Libya
cal correlations have been developed to describe the (27.6N and 14.2E) resulting in a 69% reduction in the
dependence of PV operating temperature on the climatic nominal power (Nassar and Salem, 2007). The advisable
conditions and PV materials (Skoplaki and Palyvos, operating temperature limit for PV ranges from 40 °C
2009). The operating temperature reached by PV panels to 85 °C (Suntechics, 2008) however in hot and arid cli-
mates, PV temperature frequently rises above this
temperature range (Nassar and Salem, 2007), which results
⇑ Corresponding author. Tel.: +971 555454069.
in temperature induced power failure as well as PV cell
E-mail address: ahmed.hassan@uaeu.ac.ae (A. Hasan).

http://dx.doi.org/10.1016/j.solener.2015.02.003
0038-092X/Ó 2015 Elsevier Ltd. All rights reserved.
 A. Hasan et al. / Solar Energy 115 (2015) 264–276 265

Nomenclature

Symbols Tcv temperature coefficient of voltage (% K1)

C specific heat capacity of the PV–PCM system V0 oc improvement in open circuit voltage (V)
(J kg1 K1)
DT temperature drop (°C) Abbreviations
hca combined convective and radiative heat loss A area of the PV–PCM system in thermal contact
coefficient (W m1 K1) with the ambient environment (m2)
vw wind speed (m s1) D diameter of the PV cooling duct (m)
Ps power savings (W) FF fill factor
Pe electrical power (W) L length of the PV cooling duct (m)
Voc open circuit voltage (V) LHSC latent heat storage capacity (kJ/kg)
Isc short circuit current (A) TES thermal energy storage (kJ/kg)

delamination and rapid degradation (Saly et al., 2001) China at solar radiation intensities of 405 W m2 and
urging a strong need for PV temperature regulation to 432 W m2 (Ji et al., 2006).
maximise both panel lifetime and power output. Passive cooling of BIPV with solid–liquid PCMs was
Different passive and active heat removal techniques experimentally and numerically evaluated using a paraffin
have been used to maintain PV at lower temperatures. wax as the PCM and an a rectangular aluminium container
Passive heat removal in free standing PV relies on the with internal dimension of (300 mm  132 mm  40 mm)
buoyancy driven air flow in a duct behind the PV having selectively coated front surface to mimic the PV cell
(Brinkworth, 2000a). Heat removal depends on ratio of (Huang et al., 2004). Temperature distribution on the front
length to internal diameter (L/D) of the duct surface and inside the PCM was measured experimentally
(Brinkworth, 2000b) with the maximum heat removal and predicted numerically with 2D and 3D finite volume
obtainable at an L/D of 20 (Brinkworth and Sandberg, heat transfer models which showed good agreement
2006). Passive heat removal in building integrated photo- between experimental and numerical results (Huang
voltaics (BIPV) relies on buoyant circulation of air in an et al., 2006b,a). Building on this work, Hasan et al.
opening or air channel, instead of a duct, behind the PV (2010), fabricated and characterised 4 different cell size
(Gan and Riffat, 2004). A theoretical analysis of buoyancy PV–PCM systems to investigate performance of 5 different
driven air flow in such an opening behind a façade inte- types of PCMs to find out the optimum PCM and the PV–
grated PV showed a maximum 5 °C temperature reduction PCM system for BIPV cooling application. Two PCMs, a
in averaged monthly temperature resulting a net 2.5% eutectic mixture of capric acid–palmitic acid, PCM1 and
increase in yearly electrical output of the (Yun et al., a salt CaCl26H2O, PCM2 were found promising in an alu-
2007). Though the temperature reduction and the associ- minium based PV–PCM system (Hasan et al., 2010). A
ated prevention of power drop is very low in such PV sys- temperature drop of 18 °C was recorded for 30 min and
tems, improvements can be made by boosting heat transfer 10 °C for 5 h at 1000 W m2 solar radiation intensity and
through suspending metal sheets and inserting fins in the 23 °C ambient temperature. Huang et al. (2011) evaluated
air channel and optimising distance between the walls the insertion of metallic fins in the rectangular PV–PCM
(Tonui and Tripanagnostopoulos, 2007; Fossa et al., 2008). containers previously investigated (Huang et al., 2006a,b;
Active cooling of PV mostly relies on air or water flow Hasan et al., 2010) PV–PCM and reported an improvement
on the front or back of the PV surface. Effect of air flow in temperature regulation due primarily increased natural
at different inlet velocities and air gaps on front side and convection which affects melting and solidification of
back side of PV temperature was modelled and a maximum PCM in the rectangular containments. Ho et al. (2012) con-
34.2 °C temperature decrease was predicted at air inlet ducted a simulation study of the BIPV performance inte-
velocity of 1 m s1 and front and back air gap of 20 mm grated with microencapsulated PCM attached at PV back
(Mallick et al., 2007). Water flow on the front surface of to cool and enhance performance of BIPV. The simulation
a free standing PV has a decreased cell temperature of up results showed very low temperature drop from 49 °C to
to 22 °C along with decreasing reflection losses from PV 47 °C of in summer and from 35 °C to 30.5 °C in winter
surface yielding an 8–9% increase in electrical power out- condition using a PCM with melting point 26 °C which
put (Krauter, 2004). Water flow on the back of a façade shows microencapsulated PCM show least effectiveness in
integrated PV has theoretically shown optimum electrical BIPV temperature drop due to lower thermal conductivity
and thermal performance at a water flow rate of 0.05 kg s1 of encapsulation materials and the lower mass ratio of
for a particular system in the weather conditions of Hefei, PCM contained in microencapsulation. Biwole et al.
266 A. Hasan et al. / Solar Energy 115 (2015) 264–276

(2013) conducted a mathematical and numerical modelling sufficient heat during day time for PCM melting and tem-
of heat and mass transfer adding PCM at the back of PV perature drop and lower enough temperatures at night time
and reported that PV with PCM took 80 min to reach to allow the PCM solidification and regeneration.
40 °C while PV without PCM took only 5 min to reach
the same temperature. 2. Experiments
The current work builds on Hasan et al. (2010) and
Huang et al. (2011), using the two PCM identified on cell 2.1. Experimental setup
scale inclusion scaled up to the panels size and inserting
metallic aluminium fins to increase natural convection Three 65 W polycrystalline EVA encapsulated PV pan-
and enhance PCM melting and solidification. The work is els with 771 mm  665 mm dimensions were used in the
extended to outdoor deployment in two different weather experiments. The PV panels were characterised outdoors
conditions, a cooler conditions in Dublin, Ireland and war- in Dublin for three days with stable solar radiation inten-
mer conditions in Vehari, Pakistan. In Dublin, Ireland the sity ranging from 400 to 950 W m2 over the duration of
system is subject to lower heat load due to partly cloudy experiment from 09:00 to 18:00 prior to integrating PCM
conditions and lower ambient temperatures during day into the PV panels. The open circuit voltage and short cir-
time which may result lower temperatures on the reference cuit current were measured from all three panels to assure
panels and consequently lower temperature drop. The consistence in the panels which was confirmed with a
advantage is the system may benefit from lower night time measured deviation of below 1% describable by the mea-
ambient temperatures resulting in higher PCM solidi- surement error and instruments accuracy. One PV panel
fication rate to regenerate PCM fully for next day heat without PCM was taken as reference and the other two
absorption. In Vehari, Pakistan, stable radiation and PV panels were each attached at the back with rectangular
higher day time ambient temperatures are expected to offer metallic containers having internal fins to contain PCM.
the advantage of larger heat loads and consequently larger The metallic containers were fabricated from a 5 mm thick
heat extraction by PCM with more temperature drop. The sheet of aluminium alloy (1050A) having internal dimen-
adverse is the higher night time ambient temperatures sions 600 mm  700 mm  40 mm. The containers were
reduce the PCM solidification rate which may result in internally fitted with straight vertical back to back fins of
incomplete PCM regeneration for the next day heat the same alloy with 75 mm horizontal spacing between fins
absorption. In order to understand these conflicting factors as shown in Fig. 1. Thereafter the metallic rectangular
the experiments were conducted in Dublin, Ireland and PCM containers were attached to back of the PV panels
Vehari, Pakistan in suitable weather condition which offer with epoxy resin glue and were held under constant

Fin width=5cm
Inter-fin spacing = 7.5 cm

Internal Fins
PCM container height =77.1 cm

PCM container width =66.5 cm

PCM Container depth=5cm

Fig. 1. Schematics of a PCM containers with vertically installed back to back fins.
 A. Hasan et al. / Solar Energy 115 (2015) 264–276 267

pressure for 48 h to allow settling of the glue to achieve a under halogen lamp with approximate solar spectrum at
strong bond between the PV and the PCM container. 1000 W m2 which showed a ±1% variation within them-
Eutectic mixture of capric–palmitic acid (PCM1 from selves. The whole experimental setup is shown in Fig. 2.
now onwards), was prepared by mixing 75.2% by weight
of 98% pure capric acid with 24.8% by weight of 98% pure
palmitic acid in melted form. The melted mixture was kept 2.2. Experimental procedure
constantly heated at 70 °C and stirred for 12 h to get a uni-
form mixture of the fatty acids. Salt hydrate CaCl26H2O, The reference PV and the two PV–PCM systems were
PCM2 from now onwards, was also melted, raised to 70 °C deployed outdoors facing south at the latitude angles in
and kept stirred for the same duration as PCM1 to get a Dublin, Ireland (53.33N, 6.24W) and Vehari, Pakistan
uniform solution. The PCM were characterised with dif- (30.03N, 72.25E). The experiments were conducted from
ferential scanning calorimetry (DSC) and Temperature 27th August to 13th September in Dublin and from 30th
History Method (THM) to verify their thermo-physical October to 13th November in Vehari from 09:00 AM to
properties (Hasan et al., 2014a). The melted PCM were 07:00 AM daily. The temperatures were measure at the
filled in their respective PV–PCM systems keeping front surface, inside the PCM and at the back surfaces.
100 mm free space on top to allow for expansion. Both At front surface temperature was measured at 9 positions
the PV–PCM were kept at 16 °C for 48 h until all PCM to divide the front surface into equally spaced matrix from
had fully solidified and were ready to be deployed outdoors top to bottom and left to right to account for temperature
for the experiment. The PCMs and their thermo physical distribution over the surface. At the back surface, the tem-
properties reported in literature are given in Table 1. perature was measured at three locations for the reference
Upon complete PCM solidification, the calibrated t-type PV and PV–PCM. The temperature inside the PCM was
copper–constantan thermocouples with a measurement difficult to measure as it caused leakage when inserted from
error of ±0.2 °C were attached on all systems at locations back side due to higher mass of the PCM and it was diffi-
shown in Fig. 1. All the thermocouples were calibrated in cult to hold thermocouples in place when inserted from top
ice- bath and a temperature variation of ±0.2 °C was side, therefore it was decided not to measure inside PCM
observed within the thermocouples. The three panels were temperature. The thermocouples were fixed at the front
connected to an analogue data logger to record tempera- and back surface with strong white tape, were shielded
tures voltages and currents respectively which had an from direct irradiation and were monitored continuously
accuracy of ±0.07%. A pyranometer was installed at the to assure the fixation. The open circuit voltage and short
latitude angle of both locations to measure the solar radia- circuit current were measured for all the three panels at
tion intensity and a weather station was installed to measure the same time to maintain the consistence in the measure-
ambient temperature and wind speed. The pyranometer was ments. The Weather parameters of solar radiation inten-
calibrated under stable outdoor irradiation and also indoors sity, ambient temperature and wind speed were measured

Table 1
Thermo physical properties of PCMs selected for evaluation in the novel PV–PCM systems.
Thermophysical properties Eutectic of capric–palmitic acid (PCM1) Calcium chloride hexahydrate CaCl26H2O (PCM2)
Melting point (°C) 22.5b 29.8a
Heat of fusion (kJ/kg) 173b 191a
Thermal conductivity (W m1 °C1) Solid 0.14c 1.08e
Liquid 0.14c 0.56e
Density (kg d m3) Solid 0.87b 1.71a
Liquid 0.79b 1.56e
Specific heat capacity (kJ kg1 K1) Solid 2b 1.4a
Liquid 2.3b 2.1a
Kinematic viscosity (m2 s1  103) 0.0023b 1.84d
Thermal expansion coefficient (k1) 0.00078b 0.0005e
Thermal cyclic stability Yesh Yesa, Nog
Corrosion to metals Yesf Yesf
Chemical classification Fatty acid Salt hydrate
Material source Sigma Aldrich Sigma Aldrich
a
Tyagi and Buddhi (2008).
b
Sari and Karaipekli (2008).
c
Sharma and Sagara (2005).
d
Huang et al. (2006a).
e
Marı́n et al. (2003).
f
Cabeza et al. (2002).
g
El-Sebaii et al. (2009).
h
Sedat et al. (2005).
268 A. Hasan et al. / Solar Energy 115 (2015) 264–276

PV-PCM System Pyronometer

Data logger
Host computer

Anaemometer

Fig. 2. The experimental set up consists of PV deployed outdoors at latitude angle of the selected sites, a pyranometer, an anemometer, a data logger and a
host computer to record and store the required experimental data.

for the both sites. All the measurements were performed The comparisons on the PV panel only, PV–PCM1 and
every 5 min to allow for a reasonable time step. PV–PCM2 under the same 22 h operations are carried
out for experimental and simulated results discussed in
the Section 3.
2.3. Numerical simulation

The numerical simulation model for the PV/PCM 3. Results and discussions
system has been developed and experimentally validated
by the co-authors (Huang et al., 2006a). The non-linear Although all data was recorded several days, the data is
transient model uses Boussinesq’s approximation and presented for the most stable weather conditions of solar
allows convection and diffusion to be simulated. The radiation intensities and ambient temperatures recorded
developed model can be used to predict the transient tem- on 12th September for Dublin, and 30th October for
perature distribution and fluid flow field within a two and Vehari. On other days, the PCM could not fully melt dur-
three dimensional region in the PV/PCM system for differ- ing day time and achieve the optimum PV cooling. For that
ent solar radiation intensities, ambient temperatures, con- reason, the data obtained for one day at each tested site is
vective and radiative heat transfer boundary conditions presented and discussed.
(Huang et al., 2006b). A series ambient conditions under
steady state and dynamic realistic conditions in difference 3.1. Comparison of PCM performance in Dublin
seasons have been carried out. Furthermore, the numerical
simulation model for PV–PCM system has been modified For Dublin, the climatic parameters of interest of solar
for the study of integrating with two PCMs with different radiation intensity, ambient temperature and wind speed
phase transient temperatures for improving the heat reg- are shown in Fig. 4. It is observed solar radiation generally
ulation on the BIPV. The model has been modified to pre- remained higher on September 12th 2009 in Dublin starting
dict the thermal performance of the multi-PCMs in a at 380 W m2 at 9:00, reaching 950 W m2 at 13:20 and
triangular cell in the PV–PCM system (Huang et al., 2011). dropping to 200 W m2 at 18:00. Although, it was gener-
The current experimental results with full sized PV panel ally a clear and sunny day compared to traditional over
integrated with PCMs have been compared with the cast Dublin weather, there were few moving cloud patches
numerical simulation results using the developed numerical which caused few instances of drop in solar radiation inten-
simulation model. The heat transfer and boundary condi- sity around peak radiation hours. The ambient tempera-
tions are set as shown in Fig. 3 while convection effect of ture was comparatively higher compared to colder
PCM in the PV–PCM system is ignored on the simulation. climate of Dublin starting from 16 °C at 9 AM and
 A. Hasan et al. / Solar Energy 115 (2015) 264–276 269

Insolation incident on the PV-PCM system

Reflection losses from glass cover

Convection and radiation heat loss/gain

Insolation converted to
electricity by PV Heat absorbed
by PV-PCM
1
assembly
2
3
4
5
6
6 5
Convection and
Convection and radiation heat
radiation heat loss/gain
loss/gain 5
7

Heat absorbed by PCM

Convection and radiation heat loss/gain

Fig. 3. The schematics of the energy flow in the PV–PCM system under investigation with labels 1-glass cover, 2-PV cell, 3-PV back sheet, 4-layer of epoxy
glue as interface between PV and PCM container, 5-PCM container wall, 6-PV frame and 7-PCM layer.

30 1000
Amb Temp Insolation

25 4 Wind Speed 800

Wind speed (m/sec)

Solar Radiation (W/m2)

3
Tempeerature (°C)

2
20 600
1

0
9 12 15 18 21 24 27 30
15 Time (hr) 400

10 200

5 0
9 12 15 18 21 24 27 30
Time (hr)

Fig. 4. Weather data of solar radiation intensity, ambient temperature and wind speed measured on 12th September for Dublin, Ireland (53.33N, 6.25W).

reaching to 24 °C at 2 PM. The temperature kept varying and showed a general trend of drop at night towards morn-
during the experiment because of the moving cloud ing. The weather conditions of higher solar radiation,
patches. After 2 PM, the temperature showed a larger drop higher ambient temperature and lower wind speed with
and kept generally dropping with little towards the end of variation during day time caused to PV panel to rise in
the day. At night time, the temperature continued dropping temperature with the similar trend of a general rise with
and 5 °C till towards morning. The wind speed also showed continuous variations.
a very unstable profile starting at 1 m s1 in the morning Fig. 5 shows the experimental and simulated reference
being relatively lower till 2 PM. It started increasing after PV, PV–PCM1 and PV–PCM2 temperatures. As shown in
noon reaching just above 3 m s1 where it kept varying Fig. 5 the reference measured PV temperature (Ref
270 A. Hasan et al. / Solar Energy 115 (2015) 264–276

Ref PV-exper Ref PV-simu PV-PCM1-exper

55 PV-PCM1 -simu PV-PCM2-exper PV-PCM2-simu

45

Temperature ( C)
35

25

15

5
9 12 15 18 21 24 27 30
Time (hr)

Fig. 5. Average measured and simulated temperature at front surface of the reference PV, the PV–PCM1 and the PV–PCM2 for12th September for
Dublin, Ireland (53.33N, 6.25W).

PV-exper) showed jagged temperature rise similar to the they differed in two aspects from the temperature curve for
variation in weather data from start of the experiments reference PV. Firstly, both the PV–PCM systems main-
until it peaked at 49 °C at 13:20 corresponding to peak tained lower temperatures than the reference PV through-
solar radiation intensity of 970 W m2, ambient tempera- out the experiment which due to their latent heat
ture of 24 °C and a lower wind speed of 0.25 m s1. The absorption during melting 2) the variation in temperature
temperature remained around the peak for an hour with followed the trend of variation in weather conditions how-
some variations (between 13:00 and 14:00) followed by a ever the variation was damped compared to the reference
rapid drop to 37 °C (a decrease of 12 °C) in one 1 h PV because of higher thermal mass of both the PV–PCM
30 min caused by the increase in wind speed from system compared to reference PV. Comparing the two
0.25 m s1 to 3 m s1 and decrease in ambient temperature PCMs, PCM2 maintained lower temperature than PCM1
to 18 °C although the solar radiation intensity reduced only during the whole heat absorption cycle which can be attrib-
slightly during the same period. It indicates the impact of uted to difference in their thermo-physical properties. The
increased wind speed on the PV temperature reduction PCM1 had a lower melting point (22.5 °C) compared to
due mainly to convective heat loss which increased from PCM2 (29.8 °C) and was expected to achieve lower tem-
6.65 W m2 K1 at wind speed of 0.25 m s1 to perature on PV, however the higher heat of fusion, thermal
17 W m2 K1 at wind speed of 3 m s1 calculated from conductivity and density of PCM2 compared to PCM1 as
Eq. (1) (Tiwari and Sodha, 2006) 1. shown in Table 1 impacted the PV temperature more than
the melting point. The PCM2 achieved a peak temperature
hca ¼ 5:7 þ 3:8vw ð1Þ
reduction of 10 °C compared to 7 °C of PV–PCM1 against
where hca is the convective heat transfer and vw is the wind reference PV temperature of 49 °C which is much higher
speed. At night, the ambient temperature kept dropping than temperature drop of 2 °C reported by Ho et al.
and so did the reference PV temperature, until the ambient (2012) against the reference PV temperature of 49 °C using
reached 8 °C and the reference PV reached 5 °C at 7 AM. It microencapsulated PCM. It can be attributed primarily to
can be observed that night time minimum temperature at higher effective mass, higher thermal conductivity and
PV panel is lower than the ambient temperature which better heat of fusion of bulk PCM compared to microen-
can be contributed to the radiative cooling effect of PV capsulated PCM. Comparing the experimental and sim-
panels at night. Comparing experimental and simulation ulation results, the simulated and experimental
curves for reference PV temperature in Fig. 5, the simu- temperature profiles of PV–PCM1 shown in Fig. 5 as
lated temperature (Ref PV-simu) also showed the similar “PV–PCM1-simu” and “PV–PCM1-exper” generally shows
trend as of measured reference temperature however with similar trend of temperature rise and drop however the
a slightly higher temperature which may be caused by experimental temperature remain slightly higher during
under-predicting the heat loss coefficients from the PV- day time and slightly lower during night time. The simu-
panel in simulations at day time and radiative cooling effect lated and experimental curves for the PV–PCM2 shown
of PV at night time. The temperature rise in PV–PCM1 and as “PV–PCM2-exper” and “PV–PCM2-simu” respectively
PV–PCM2 shown in Fig. 5 as “PV–PCM1-exper” and “PV– also generally agree in temperature rise however the experi-
PCM2-exper” respectively followed the similar general rise mental temperature remains lower than the simulated at
in temperature till noon and then drop afternoon however start of the experiment and rises above the simulated
 A. Hasan et al. / Solar Energy 115 (2015) 264–276 271

temperature for rest of the day while at night again it falls 3.2. Comparison of PCM performance in Pakistan
below the simulated temperature at start of the cooling
cycle and then rises above the simulated temperature for Fig. 7 shows the solar radiation intensity, wind speed
rest of the night until early morning. In general in both and ambient temperature measured on 30th October in
cases the experimental temperature is closer to simulated Vehari, Pakistan (30.03N, 72.25E). Figure shows that the
temperature which validates the results. solar radiation intensity showed a stable profile during
Since the inclusion of PCM maintained a lower tempera- the day starting from 390 W m2 at 9:00, reached a peak
ture than the reference, they were expected influence the at 950 W m2 at 13:40 and dropped to below 200 W m2
open circuit (Voc) voltage and short circuit current (Isc) of till 18:00. The wind speed remained lower all the day rang-
the PV. Since from the product catalogue, the temperature ing between 0 m s1 and 1 m s1. The ambient temperature
coefficient of voltage is much higher than the temperature also showed a steady increase profile during the day start-
coefficient of current, the later is not presented as it was ing at 18 °C at 9:00 AM, peaked at 34 °C at 14:00 and
too small to actually observe the differences during experi- slowly dropped to 29 °C at 18:00. Fig. 8 shows the
ments. Fig. 6 shows open circuit voltage (Voc) of the refer- measured reference PV (Ref-PV-exper), PV–PCM1 (PV–
ence PV, PV–PCM1 and PV–PCM2 as “Ref-PV–Voc”, PCM1-exper) and PV–PCM2 followed similar trend of tem-
“PV–PCM1–Voc” and “PV–PCM2–Voc” respectively on perature rise with the two PV–PCMs maintaining with
primary vertical axis. It can be observed that the PV– lower temperature than the reference until 6 PM where
PCM1 and PV–PCM2 maintained higher Voc than refer- the reference PV temperature dropped below the two
ence PV because they maintained a lower temperature PV–PCM systems. The two PV–PCM systems had smaller
panel temperature. The increase in open circuit voltage temperature difference at start of the experiment for 25 min
(V0 oc) corresponding to drop in temperature by inclusion however they started deviating from reference afterwards
of the PCMs is also plotted on secondary vertical axis. due to heat absorption by melting PCM at the back of
The higher V0 oc was observed at 13:40 corresponding to the PV. The temperature at the front surface of the refer-
highest solar radiation intensity and ambient temperature ence PV increased from 18 °C to 50 °C in 2 h reaching a
similar to what was observed for temperature difference. peak of 63 °C at 13:45. The temperature remained at or
The peak V0 oc was 0.5 V for PV–PCM1 and 0.8 V for above 50 °C for 6 h until 17:00 when it dropped to 38 °C
PV–PCM2 obtained at 13:40. In order to validate the in the last hour of the experiment till 18:00. After 13:45,
experimental results, the temperature coefficient of voltage the temperature started to decrease in the reference PV
drop (Tcv) is calculated from the measured V0 oc and with a stable gradient (higher than that of PV–PCM1 and
temperature drop (DT) using Eq. (2). PV–PCM2) indicating smooth cooling with a combined
T cv ¼ V 0oc =DT ð2Þ heat loss of 9.1 W m2 K1 calculated from Eq. (1) using
wind speed of 1 m s1.
The calculated Tcv for PV–PCM1 and PV–PCM2 The temperature decrease in the PV–PCM systems was
are calculated to be 71.4 m V K1 and 80 m V K1 very slow which shows the heat retention of the PV–
which agree with the product catalogue value of PCM system due to latent heat and thermal mass of the
77 ± 10 m V K1. PCM. Higher heat retention is required if the heat stored

22 Ref-PV-Voc PV-PCM1-Voc PV-PCM2-Voc PV-PCM1-Voc-Gain PV-PCM2 Voc-Gain

1.9

21
1.4

20
Voltage Gain (V)
Voltage (V)

0.9
19

0.4
18

-0.1
17

16 -0.6
9 12 15 18
Time (hr)

Fig. 6. The measured open circuit voltage (Voc) on primary vertical axis and the open circuit voltage gain (V0oc ) on secondary vertical axis for the reference
PV, the PV–PCM1 and the PV–PCM2 systems measured on 12th September for Dublin (53.33N, 6.25W).
272 A. Hasan et al. / Solar Energy 115 (2015) 264–276

40 1000
Amb Temp Insolation
3
35
Wind Speed 800

Wind speed (m/sec)

2

Solar Radiation (W/m2)

30
Tempeerature (°C)
1
600

25 0
9 12 15 18 21 24 27 30
Time (hr) 400
20

200
15

10 0
9 12 15 18 21 24 27 30
Time (hr)

Fig. 7. Weather data of ambient temperature, solar radiation intensity and wind speed measured on 30th October for Vehari, Pakistan (30.03N, 72.25E).

70
Ref PV-experi PV-PCM1-experi PV-PCM2-experi
Ref PV-simu PV-PCM1-simu PV-PCM2-simu
60
Temperature (ºC)

50

40

30

20

10
9 12 15 18 21 24 27 30
Time (hr)

Fig. 8. Average measured and simulated temperature at front surface of the reference PV, the PV–PCM1 and the PV–PCM2 measured on 30th October for
Vehari, Pakistan (30.03N, 72.25E).

is to be exploited later for domestic water heating. High PV–PCM1 and a peak of 21.5 °C at 13:55 for PV–PCM2
heat retention however discourages nocturnal solidification against the PV temperature of 63 °C which is in close agree-
of the PCMs which may affect its readiness and thermal ment with previously reported temperature drop of 16 °C
drop performance for the next day. In such cases coolant against a reference PV temperature of 78 °C (Maiti et al.,
flow into the PCM to maximise heat extraction may be 2011) employing 6 cm thick bed of paraffin wax Since a
required. A further study needs to be conducted in this previous study (Hasan et al., 2010) found paraffin wax
direction to make such systems energy efficient in hot cli- (used by Maiti et al.) being less effective for temperature
mates by finding mechanisms to extract stored energy in drop compared to capric–palmitic acid and CaCl26H2O
PCM. Fig. 8 shows the measured PV temperature was kept used in current work, the peak temperature drop of
below 45 °C by PV–PCM1 and below 41 °C by PV–PCM2 21.5 °C is quite in agreement with previous research. It is
at the solar radiation intensity of 950 W m2 is in good observed that the PV–PCM2 maintained a higher tempera-
agreement with (Biwole et al., 2013) who reported the PV ture drop than PV–PCM1 for the duration of the experi-
temperature stabilization below 40 °C at 1000 W m2 ment. The PV–PCM2 maintained a temperature difference
through modelling using phase change materials. Fig. 8 from the reference of over 20 °C for two hours between
shows that at the start of the experiment, the temperature 12:20 and 14:20 which illustrates its potential to achieve
drop (temperature difference between the reference and high temperature reduction for longer durations whereas the
the PV–PCM system) was lower. The temperature drop PV–PCM1 maintained a temperature difference of 15 °C
increased (as the incoming solar radiation and the ambient for 2 h 20 min from 10:40 to 13:00. The temperature differ-
temperature increased) to a peak of 16 °C at 12:20 for ence started decreasing after the peak with similar gradient
 A. Hasan et al. / Solar Energy 115 (2015) 264–276 273

for both PCMs however temperature in PV–PCM1 equal- much better than PCM1. Peak measured V0 oc was 1.35 V
led reference PV temperature i.e. zero temperature differ- for PV–PCM1 and 1.75 V for PV–PCM2. Tcv is calculated
ence at 16:30 while temperature in PV–PCM2 equalled from the V0 oc and the DT using Eq. (2) resulting in a value
the reference temperature at 17:50. Afterwards all the night of 84 m V K1 and 83 m V K1 for PV–PCM1 and PV–
the reference PV temperature remained below PV–PCM PCM2 respectively which is in good agreement with manu-
systems temperature which indicates the higher heat retention facturer’s product catalogue Tcv value of 75 ± 10 m V K1
and thermal mass of the PV–PCM systems. The PV–PCM1 (Suntech, 2006).
temperature dropped below the PCM1 melting point of
22.5 °C at 22:30 while the PV–PCM2 temperature dropped 3.3. Comparison of measured power savings
below the PCM2 melting point of 29.8 °C at 21:00 and
remained below this temperature until 7 AM in the morn- Since the use of PCM prevented voltage drop by V0 oc it
ing to allow enough time for PCM to solidify which were consequently saved power loss from the PV. The power
visually inspected to be solid on the next day before start saved, Ps thus was computed by subtracting the power
of the next experiment.
16 Maximum voltage improvement Maximum power improvement
Comparing the experimental and simulation tempera-
ture rise given in Fig. 8, the simulated temperatures of 14
the reference PV (Ref-PV-simu), the PV–PCM1 (PV–
PCM1-simu) and the PV–PCM2 (PV–PCM2-simu) 12

Maximum improvement (%)

generally show a close agreement with the corresponding
experimental temperatures with negligible deviations which 10
shows the validation of the experimental results.
Fig. 9 shows the Voc for the reference PV, PV–PCM1 8
and PV–PCM2 on primary vertical axis while the V0 oc on
the secondary vertical axis caused by reduced PV–PCM 6
temperature compared to the reference PV temperature
for Vehari (30.03N, 72.25E), Pakistan measured on 30th 4
October. PV–PCM2 achieved a higher V0 oc compared to
PV–PCM1 for the duration of the experiment. V0 oc showed 2
a similar trend as the temperature drop in Fig. 8. At the
start of the experiment the V0 oc were similar for both 0
PV–PCM systems when the reference PV temperature Ireland Pakistan

was lower. As the temperature increased, V0 oc for PV– Fig. 10. The maximum power and voltage improvement resulted from
PCM2 became remarkably higher than that of PV–PCM1. temperature regulation of PV by the two PV–PCM systems at both the
This indicates that at higher temperatures PCM2 performs tested cites.

21 4
Ref-PV-Voc PV-PCM1-Voc PV-PCM2-Voc PV-PCM1-Voc-Gain PV-PCM2 Voc-Gain

20
3

19

2
Voltage Gain (V)
Voltage (V)

18

17

0
16

15 -1
9 12 15 18
Time (hr)

Fig. 9. The measured open circuit voltage (Voc) on primary vertical axis and the open circuit voltage gain (V0oc ) on secondary vertical axis for the reference
PV, the PV–PCM1 and the PV–PCM2 systems on 30th October for Vehari, Pakistan (30.03N, 72.25E).
274 A. Hasan et al. / Solar Energy 115 (2015) 264–276

produced, Pe by the reference PV from the Pe by the PV– at the system boundaries of front surface (the PV panel)
PCM1 and PV–PCM2 as shown in Fig. 10. The Pe for each and back surface (the PCM container) of the two PV–
case was calculated by using the measured Voc, Isc and PV PCM systems were measured and an energy balance was
manufacturer’s catalogue values of fill factors (FF). The fill conducted. The temperature at the back surface of PV–
factors, peak and daily averaged Voc and Isc, peak and PCM systems at both the tested sites is presented in
daily averaged Pe and peak and daily Ps are summarised Fig. 11. It shows that the back surface temperatures
in Table 2. It shows that the peak Ps of 4% by PCM1 presented in Fig. 11 also show similar trend as of the front
and 5.1% by PCM2 while daily average Ps of 1% by surface temperatures presented in Fig. 5 for Dublin and
PCM1 and 1.8% by PCM2 was obtained in Dublin. Fig. 8 for Vehari however the variation in temperature in
Similarly the peak Ps of 11.3% by PCM1 and 13% by Dublin was almost unobserved compared to jagged varia-
PCM2 while daily averaged Ps of 4.4% by PCM1 and tion in front surface temperature. The variation does not
7.7% by PCM2 was achieved in Vehari which proves that show up mainly for the back side being shaded and not
more power savings are achievable in hot climate of subject to varying solar radiation intensity which affects
Vehari than the cooler climate of Dublin. strongly the front surface temperature also due to higher
thermal mass of PCM which dampens the variation.
3.4. Comparison of thermal performance Considering front surface and back surface temperatures
and performing the energy balance of the system, the ther-
In order to compare the thermal performance of the mal energy stored (TES) in the PCM1 and PCM2 was cal-
PV–PCM systems in both the climates, the temperatures culated. It was found to be 251 W h (906 kJ) and 382 W h

Table 2
Summary of the temperature regulation and power saving obtained during outdoor PV exposure in Dublin (53.33N, 6.25W) and Vehari (30.03N, 72.25E).
Reference PV PV–PCM1 PV–PCM2
Dublin Vehari Dublin Vehari Dublin Vehari
Insolation (W m2) At peak 990 950 990 950 990 950
Average 674 660 674 660 674 660
Temperature (°C) At peak 49 63 43 46 39 42
Temperature regulation (°C) At peak – – 7 17 10 21
Fill factor (%) Average 72.22 69.64 72.82 71.26 73.22 72.24
Voc (V) At peak 20.1 18.32 20.81 19.71 20.95 20.15
Average 20.41 18.72 20.52 19.42 20.81 19.92
Isc (A) At peak 3.74 3.42 3.70 3.35 3.68 3.33
Average 2.82 2.45 2.77 2.41 2.78 2.39
Pe (W) At peak 53.3 42.52 55.46 47.34 55.99 48.04
Average 41.57 31.94 41.71 33.35 42.3 34.39
Ps (%) At peak – – 4 11.3 5.1 13
Average – – 1 4.4 1.8 7.7

Back-PV-PCM1-Vehari Back-PV-PCM2-Veha ri Back-PV-PCM1-Dublin Back-PV-PCM2-Dublin

50

40
Temperature ( C)

30

20

10
9 12 15 18 21 24 27 30
Time (hr)

Fig. 11. Average temperature at back surface of the PV–PCM1 and the PV–PCM2 measured on 12th September for Dublin (53.33N, 6.25W) and on 30th
October for Vehari, Pakistan (30.03N, 72.25E).
 A. Hasan et al. / Solar Energy 115 (2015) 264–276 275

(1374 kJ) in Dublin and 576 W h (2079.4 kJ) and 875.7 W h phase change materials for increased energy efficiency in
(3161 kJ) in Vehari (Hasan et al., 2014b) for PCM1 and renewable energy systems in buildings for providing an
PCM2 respectively. Since mass of the PCM1 and PCM2 invaluable platform to discuss and develop this work.
contained in PV–PCM system were 13 kg and 20 kg and Thanks also to Professor Javed L. Piracha and Professor
heat of fusions of 173 kJ/kg and 191 kJ/kg respectively, Dr. Nadeem A. Mufti (University of Engineering and
the latent heat storage capacities (LHSC) for PCM1 and Technology Lahore, Pakistan) for their inputs and
PCM2 are 2145 kJ and 3820 kJ. A term utilisation factor cooperation for the outdoor experiments in Pakistan.
(UF) is defined to be the ratio of TES in PCM to the Thanks to Kevin O’ Farrell, Nari Shaqiri, Juanka Carlos
LHSC of the PCM. and Daniel Martinez for their inputs in the fabrication,
UF ¼ TES=LHSC installation and measurement procedure in Ireland.

The UF are found to be 0.40 and 0.36 respectively. This

shows that although PCM1 achieves less cooling effect, it References
has slightly higher utilisation factor than PCM2 primarily
Amy de la Breteque, Emmanuel, 2009. Thermal aspects of c-Si
because of less amount of PCM being used. It also shows photovoltaic module energy rating. Sol. Energy 83 (9), 1425–1433.
that both PCMs are under utilised in cooler climate of Biwole, P.H., Eclache, P., Kuznik, F., 2013. Phase-change materials to
Dublin. The UF for Vehari are found to be 0.924 and improve solar panel’s performance. Energy Build. 62, 59–67.
0.827, much higher than that in Dublin which proves that Brinkworth, B.J., 2000a. Estimation of flow and heat transfer for the
these systems are properly utilised in the warmer climates. design of PV cooling ducts. Sol. Energy 69 (5), 413–420.
Brinkworth, B.J., 2000b. A procedure for the routine calculation of
In Vehari, although PCM1 achieved lower temperature laminar free and mixed convection in inclined ducts. Int. J. Heat Fluid
drop than PCM2, it achieves substantially higher UF Flow 21 (4), 456–462.
primarily due to less amount of material being used. The Brinkworth, B.J., Sandberg, M., 2006. Design procedure for cooling ducts
UF calculated ignore sensible heat capacity effect in order to minimise efficiency loss due to temperature rise in PV arrays. Sol.
to compare the exact PCM technology. Energy 80 (1), 89–103.
Bücher, K., 1997. Site dependence of the energy collection of PV modules.
Sol. Energy Mater. Sol. Cells 47 (1–4), 85–94.
4. Summary and conclusions Cabeza, L.F., Roca, J., Nogués, M., Mehling, H., Hiebler, S., 2002.
Immersion corrosion tests on metal-salt hydrate pairs used for latent
The experimental and simulation results for temperature heat storage in the 48 to 58 °C temperature range. Mater. Corros. 53
rise of reference PV, PV–PCM1 and PV–PCM2 are com- (12), 902–907.
El-Sebaii, A.A., Al-Amir, S., Al-Marzouki, F.M., Faidah, A.S., Al-
pared for the two climatic conditions. The deviations Ghamdi, A.A., Al-Heniti, S., 2009. Fast thermal cycling of acetanilide
between the experimental and simulation results are under and magnesium chloride hexahydrate for indoor solar cooking. Energy
6.3% for all of the three cases in the 22 h operation which Convers. Manage. 50 (12), 3104–3111.
shows that the numerical simulation results have good Fossa, M., Ménézo, C., Leonardi, E., 2008. Experimental natural
agreement with the experimental testing under full sized convection on vertical surfaces for building integrated photovoltaic
(BIPV) applications. Exp. Thermal Fluid Sci. 32 (4), 980–990.
PV–PCM systems in dynamic realistic conditions. Gan, Guohui, Riffat, Saffa B., 2004. CFD modelling of air flow and
Comparing the two PCMs, CaCl26H2O achieved a higher thermal performance of an atrium integrated with photovoltaics.
temperature drop and power savings compared to capric– Build. Environ. 39 (7), 735–748.
palmitic acid at both sites. In the best case CaCl26H2O Hasan, A., McCormack, S.J., Huang, M.J., Norton, B., 2010. Evaluation
achieved 3–4 °C higher PV temperature drop and resulted of phase change materials for thermal regulation enhancement of
building integrated photovoltaics. Sol. Energy 84, 1601–1612.
in 3% more power saving than capric–palmitic acid in Hasan, A., McCormack, S.J., Huang, M.J., Norton, B., 2014a.
Vehari, Pakistan. Comparing the two climates, it was Characterization of phase change materials for thermal control of
observed that both the PCMs achieved higher PV tempera- photovoltaics using Differential Scanning Calorimetry and
ture drop in Vehari than in Dublin, reaching up to 21 °C Temperature History Method. Energy Convers. Manage. 81, 322–329.
and resulting associated power saving of 13% in Vehari. Hasan, A., McCormack, S.J., Huang, M.J., Norton, B., 2014b. Energy
and cost saving of a photovoltaic-phase change materials (PV–PCM)
Since the climate of Vehari is hot with very stable radiation system through temperature regulation and performance enhancement
around the years compared and overcast climate of Dublin, of photovoltaics. Energies 7, 1318–1331.
it is concluded that such systems are only effective in hot Ho, C.J., Tanuwijava, A.O., Lai, C.M., 2012. Thermal and electrical
and stable climatic conditions. performance of a BIPV integrated with a microencapsulated phase
change material layer. Energy Build. 50, 331–338.
Huang, M.J., Eames, P.C., Norton, B., 2004. Thermal regulation of
Acknowledgements building-integrated photovoltaics using phase change materials. Int. J.
Heat Mass Transf. 47 (12–13), 2715–2733.
The authors would like to acknowledge the Higher Huang, M.J., Eames, P.C., Norton, B., 2006a. Comparison of a small-
Education Authority through Strand 3 funding, Science scale 3D PCM thermal control model with a validated 2D PCM
thermal control model”. Sol. Energy Mater. Sol. Cells 90 (13), 1961–
Foundation Ireland through their Research Frontiers
1972.
Program and the Research Support Unit at Dublin Huang, M.J., Eames, P.C., Norton, B., 2006b. Phase change materials for
Institute of Technology. They would also like to acknowl- limiting temperature rise in building integrated photovoltaics. Sol.
edge COST Action TU0802: Next generation cost effective Energy 80 (9), 1121–1130.
276 A. Hasan et al. / Solar Energy 115 (2015) 264–276

Huang, M.J., Eames, P.C., Norton, B., Hewitt, N.J., 2011. Natural Radziemska, E., Klugmann, E., 2002. Thermally affected parameters of
convection in an internally finned phase change material heat sink for the current–voltage characteristics of silicon photocell. Energy
the thermal management of photovoltaics. Sol. Energy Mater. Sol. Convers. Manage. 43 (14), 1889–1900.
Cells 95 (7), 1121–1130. Saly, V., Ruzinsky, M., Redi, P., 2001. Indoor study and ageing tests of
Ji, J., Han, J., Chow, T., Yi, H., Lu, J., He, W., Sun, W., 2006. Effect of solar cells and encapsulations of experimental modules. In: 24th
fluid flow and packing factor on energy performance of a wall- International Spring Seminar on Electronics Technology: Concurrent
mounted hybrid photovoltaic/water-heating collector system. Energy Engineering in Electronic Packaging.
Build. 38 (12), 1380–1387. Sari, A., Karaipekli, A., 2008. Preparation and thermal properties of
Krauter, S., 2004. Increased electrical yield via water flow over the capric acid/palmitic acid eutectic mixture as a phase change energy
front of photovoltaic panels. Sol. Energy Mater. Sol. Cells 82 (1–2), storage material. Mater. Lett. 62 (6–7), 903–906.
131–137. Sedat, K., Kamil, K., Ahmet, S., 2005. Lauric and myristic acids eutectic
Maiti, S., Banerjee, S., Vyas, K., Patel, P., Ghosh, P.K., 2011. Self mixture as phase change material for low-temperature heating
regulation of photovoltaic module temperature in V-trough using a applications. Int. J. Energy Res. 29 (9), 857–870.
metal–wax composite phase change matrix. Sol. Energy 85, 1805– Sharma, S.D., Sagara, Kazunobu., 2005. Latent heat storage materials
1816. and systems: a review. Int. J. Green Energy 2 (1), 1–56.
Makrides, G., Zinsser, B., Georghiou, G.E., Schubert, M., Werner, J.H., Skoplaki, E., Palyvos, J.A., 2009. Operating temperature of photovoltaic
2009. Temperature behaviour of different photovoltaic systems modules: a survey of pertinent correlations. Renewable Energy 34 (1),
installed in Cyprus and Germany. Sol. Energy Mater. Sol. Cells 93 23–29.
(6–7), 1095–1099. Suntechics, 2008. Product Manual, Suntechnich STP065-12/Sb PV Panel.
Mallick, T.K., Eames, P.C., Norton, B., 2007. Using air flow to alleviate Tiwari, A., Sodha, M.S., 2006. Performance evaluation of solar PV/T
temperature elevation in solar cells within asymmetric compound system: an experimental validation. Sol. Energy 80 (7), 751–759.
parabolic concentrators. Sol. Energy 81 (2), 173–184. Tonui, J.K., Tripanagnostopoulos, Y., 2007. Air-cooled PV/T solar
Marı́n, J.M., Zalba, B., Cabeza, L.F., Mehling, H., 2003. Determination collectors with low cost performance improvements. Sol. Energy 81
of enthalpy–temperature curves of phase change materials with the (4), 498–511.
temperature-history method: improvement to temperature dependent Tyagi, V.V., Buddhi, D., 2008. Thermal cycle testing of calcium chloride
properties. Meas. Sci. Technol. 14 (2), 184–189. hexahydrate as a possible PCM for latent heat storage. Sol. Energy
Nassar, Y.F., Salem, A.A., 2007. The reliability of the photovoltaic Mater. Sol. Cells 92 (8), 891–899.
utilization in southern cities of Libya. Desalination 209 (1–3), 86–90. Yun, G.Y., McEvoy, M., Steemers, K., 2007. Design and overall energy
Radziemska, E., 2003. The effect of temperature on the power drop in performance of a ventilated photovoltaic façade. Sol. Energy 81 (3),
crystalline silicon solar cells. Renewable Energy 28 (1), 1–12. 383–394.