You are on page 1of 18

Solar Energy 211 (2020) 1283–1300

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Investigation of the operational performance and efficiency of an


alternative PV + PCM concept
Nikolaos Savvakis, Evangelia Dialyna, Theocharis Tsoutsos *
Renewable and Sustainable Energy Lab, School of Environmental Engineering, Technical University of Crete, GR 73100 Chania, Greece

A R T I C L E I N F O A B S T R A C T

Keywords: The significance of the temperature effect on the performance of photovoltaics (PVs) has implied the necessity to
Photovoltaic develop PV cooling methods in recent years. In this research, the inclusion of phase change materials (PCMs)
Monocrystalline silicon cells through an alternative type of enclosure in tubular shape was proposed and investigated as an option of miti­
Phase change materials
gating the PV operating temperature to enhance their efficiency and lifetime. Two PVs incorporating different
Operating temperature
Energy efficiency
PCMs (PV + PCM systems) and a conventional PV module (reference case) were experimentally tested to assess
PV+PCM system their energy performance under the Mediterranean conditions in Chania, Crete. As PCMs employed were selected
Paraffins RT 27 and RT 31. The results indicated that a peak temperature decrement of 6.4 ◦ C and 7.5 ◦ C could be
observed by using 260 g of PCM27 and PCM31, respectively. Hence, PV + PCM27 and PV + PCM31 systems
exhibited increased energy generation by 4.19% and 4.24%, respectively, while the increment in the PV con­
version efficiency by PCM integration ranged from 2.86 to 4.19%. The proposed configuration of PCM enclosures
took advantage of the synergistic effect of wind, as demonstrated by the recorded daily temperature profiles of
the PV + PCM27 and PV + PCM31 systems, even after the time of complete PCMs’ melting.

Savvakis and Tsoutsos (2015) investigated the effect of high tem­


peratures in the Mediterranean on the energy behaviour of a thin-film
1. Introduction PV system. They reported that the main electrical parameters’ values
(i.e. DC-voltage, DC-power, efficiency) were lessening linearly as the PV
The performance of photovoltaic (PV) modules is dependent on cli­ operating temperature was raising. Taking into consideration that PV
matic variables, primarily the incoming sunlight and secondarily the modules usually operate at temperature values over 50 ◦ C during the
operating temperature (Lobera and Valkealahti, 2013). In practical summer period, a significant deviation from the efficiency appraisals
terms, the PV module operating temperature has a significant effect on received under standard conditions appears (Mavromatakis et al.,
the solar-to-electricity conversion process, as the PV module voltage 2014). Accordingly, the regulation of PV operating temperature as low
generation and, by extension, its power output diminishes with the as reasonably achievable could result in a substantial improvement in
rising temperature. This temperature is controlled by the physical PV efficiency (Siddiqui et al., 2012). Recently, this literature has been
characteristics of the module materials and the ambient conditions enlarged. In these techniques, various fluids as means of heat storage (i.
(Skoplaki et al., 2008; Tsitoura et al., 2016). e. air, water, PCMs, etc.) are exploited for the absorption of the excessive
For instance, outdoor experiments with mc-Si PV modules demon­ heat from the PV module surface. PV cooling via PCM is considered
strated that the PV power generated (P), fill factor (FF) and efficiency amongst the most effective options since the achievement even up to
(nELEC) were dropping by 6.5%, 2.0% and 0.8% respectively, per 10 ◦ C 20% in PV modules efficiency is possible (Ali, 2020).
increase of PV module operating temperature above 25 ◦ C (Radziemska, In the present study, an alternative configuration for the PCM in­
2003). A few years later, Chander et al. (2015) examined the impact of clusion in a typical PV module was experimentally tested. More specif­
operating temperature on the mc-Si PV modules’ energy efficiency. ically, two PCM containers were designed and fabricated in the form of a
Several experiments via solar simulator as concerns the light intensity thermal collector for the optimization of the correlation of the PCM
and the PV cell temperature shown a downward trend in the range from quantity and the energy performance of PV + PCM systems. The moti­
− 0.13%/oC till − 0.25%/oC in temperature coefficients of VOC, FF and vation of exploiting a tubular configuration was to improve the free
PMAX, with the rise of PV cells temperature.

* Corresponding author. Tel.: +30 2821037825.


E-mail address: Theocharis.Tsoutsos@enveng.tuc.gr (T. Tsoutsos).

https://doi.org/10.1016/j.solener.2020.10.053
Received 17 May 2020; Received in revised form 16 October 2020; Accepted 17 October 2020
Available online 1 November 2020
0038-092X/© 2020 International Solar Energy Society. Published by Elsevier Ltd. All rights reserved.
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Nomenclature t Measurement interval


T Reference period
Tmb PV back-side operating temperature (◦ C) xi ith value of the independent variable
Tmf PV front-side temperature (◦ C) yi ith value of the dependent variable
TPCM,melt Melting temperature of the PCM (oC) bj regression coefficients for j = 0, 1, 2, 3…
Tα Ambient temperature (◦ C) nELEC electrical efficiency (%)
GT Solar irradiance (W m− 2 m2) R correlation coefficient
A Effective PV module area (m2) R2 coefficient of determination
Vw Wind velocity (km h− 1) R2adj adjusted coefficient of determination
IMAX Current at the maximum power point (A) KT Clearness index
ISC Short Circuit current (A)
I Current (A) Abbreviations
VMAX Voltage at the maximum power point (V) PV photovoltaic
VOC Open-Circuit voltage (V) PCM phase change material
V Voltage (V) PV + PCM27 Photovoltaic module coupled with PCM RT 27
P PV module power output (W) PV + PCM31 Photovoltaic module coupled with PCM RT 31
PMAX PV module maximum power output (W) PV reference Photovoltaic module used as a reference case
FF fill factor PV/T Photovoltaic/thermal
EDC,d Daily total DC energy output (kWh) PV/T + PCM Photovoltaic/thermal module coupled with phase
N Number of time series change material

convective thermal-exchange from the back of the PV module, which case and two in conjunction with PCMs. To this goal, two rectangular
affects the PCM melting process in the rectangular containers as enclosures were produced of 0.5 cm aluminum alloy and vertically
mentioned by many researchers (Kamkari and Groulx, 2018; Groulx placed fins of 5 cm in width. The experiments took place on selected
et al., 2020). The comparative analysis of the energy behaviour of two dates in Ireland and Pakistan. The PCMA used was a compound of cap­
PVs incorporating different PCMs (PV + PCM systems) and a conven­ ric/palmitic acid, and the PCMB was salt hydrate CaCl2⋅6H2O. The main
tional PV module (reference case) was carried out to measure the ben­ outcomes demonstrated a temperature decline of 7–10 ◦ C in Ireland,
efits from the proposed modification. Paraffins RT 27 and RT 31 were whereas the analogous measurements in Pakistan were in the range of
employed to upgrade the thermal response and energy generation in the 10–21 ◦ C. Correspondingly, PV electricity generation enhanced over
PV + PCM systems. 10% in the warm environment of Pakistan, while 4% more electricity
produced under the weather conditions of Ireland.
2. State-of-the-art To estimate the efficiency of a PV + PCM system in a hot climate
region, Hasan et al. (2017) accomplished a year-long monitoring pro­
2.1. PV + PCM typical shapes cess. Except for the commercial PV modules, a PCM enclosure fabricated
from 4 mm aluminum plates was exploited for the experiments. They
Since the first systematic research of a PV + PCM system performed reported that the existence of PCM (paraffin RT42) declined the mean
by Huang et al. (2004), several researchers have tried to answer to value of the PV temperature by 10.5 ◦ C, achieving a rise of 5.9% in
multiple questions (e.g. the optimal design for the PCM enclosure, the electric power generation on an annual basis.
selection of the proper PCM for a PV + PCM system) aiming to form In parallel Nada et al. (2018) studied experimentally the response of
design principles of such an application. employing pure PCM and PCM mixed with Al2O3 nanoparticles to the
In a later study, the same researchers executed a series of trials to overall performance of a conventional PV module. The experimental
evaluate the efficiency of different PV + PCM systems. For their analysis, arrangement consisted of three PV modules of the identical type; two
three PCM enclosure schemes and two sorts of PCM (RT25 and GR40) connected with the chosen PCMs and the reference case. According to
examined to identify the importance of systems’ configuring on their their results, both PCM enclosures with and without nanoparticles could
operating temperature evolution. Their findings revealed that the PV + mitigate the PV temperature by 10.6 ◦ C and 8.1 ◦ C and raising the PV
PCM operating temperature did not exceed over 30 ◦ C for about 2.2 h by modules efficiency by 13.2% and 5.7%, respectively.
using RT25 and depth of 4 cm for the PCM enclosure. Finally, they re­ Khanna et al. (2018, 2019) have examined two PV-PCM system
ported that the finned PCM enclosures regulated more effectively the configurations to optimize the PCM enclosure design. Their conclusions
temperature rising of PVs, as compared with the un-finned enclosures are summarized as follows: (i) the ideal depth of PCM storage enclosures
(Huang et al., 2006). without and with fins was calculated to equal 390 mm and 460 mm,
The ability of PCMs for thermal regulation in BIPV systems was respectively, (ii) the integration of fins arranged in a spacing of 25 cm
experimentally explored by Hasan et al. (2010) with diverse PCMs and a width of 0.2 cm was proposed to enhance the PV + PCM system
(RT20, mixtures of capric/palmitic acid and capric/lauric acid, SP22 efficiency, (iii) finned PCM enclosure was stated as capable of the PV
and CaCl2⋅6H2O) under certain irradiance levels. Four enclosure types module’s thermal maintenance on clear-sky days, mostly for the warm
from perspex and aluminium in a rectangular scheme were exploited to climates as the improvement in electricity generation attained 12.1%.
encapsulate the selected PCMs. A clear benefit of 10 ◦ C in the prevention Wongwuttanasatian et al. (2020) examined the possibility of
of the temperature rise for 300 min was achieved by CaCl2⋅6H2O with applying palm wax as an operating medium for PV thermal regulation.
TPCM,melt = 29 ◦ C, while the highest temperature decrement of 18 ◦ C for In their analysis, a finned PCM enclosure employed to achieve the
0.5 h was seen by employing CaCl2⋅6H2O. Besides, they reported a sig­ maximum cooling effect on the examined PV module. They highlighted
nificant interaction between the thermal parameters (i.e. thermal con­ that the adopted design approach achieved a temperature decrease of
ductivity and melting point) and the PV module operating temperature. 6.1 ◦ C on the PV front surface, increasing its efficiency by 5.3%.
Five years later, Hasan et al. (2015) implemented an experimental Mostly in the literature, a rectangular enclosure formation (with or
procedure to check the operational parameters of three pc-Si PVs; a base without inner fins) with the selected PCM is fitted at the backside of the

1284
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

PV module. Often these enclosures fabricated by aluminium are fixed the energy output characteristics of three PV units (i.e. a commercial PV
behind the PV frame with thermal paste. Obviously, the selection of the unit, a PV/T + PCM and PV/T + PCM + Metal matrix unit). They re­
fabrication material of the PCM enclosure is directly correlated to the ported that the enhancement of 8.6% in PV energy conversion efficiency
need for enhanced heat transfer from the PV module to PCM. Further­ appeared due to PCM + Metal matrix integration (Shastry and Aru­
more, the quantity of PCM to be employed is a critical parameter, as nachala, 2020).
demonstrated by the option of depth ≥ 4 cm in the most above cases. Additional investigations have been conducted by some scholars to
balance the aggregated effect of a PCM and thermal collector incorpo­
2.2. PV/Τ + PCM applications ration for PV cooling and to estimate the overall energy generation of
such a system. In these researches, various syntheses for the optimisa­
PV/T applications can provide both electric and thermal energy, tion of the thermal conductivity and water-flow rate, as well as several
while the aggregated efficiency of such applications can be further categories of PCM, have been examined. Many of these papers reported
improved by incorporation of PCM. In such a system, the excessive heat that the PV/T + PCM concept presents high solar energy conversion
of the PV module is firstly engrossed by PCM as latent heat, resulting in efficiency because of its potentiality to generate low-temperature ther­
PV temperature decline and electrical efficiency increment. Subse­ mal energy that can be further exploited in pre-heating water systems.
quently, the working fluid in the thermal collector immersed in the PCM However, a PV/T + PCM unit could only substitute solar applications in
soaks up the energy stored by PCM resulting in higher thermal efficiency cases that the utilisation of hot water produced is meaningful (e.g. urban
(Ali, 2020). environment, industry, tourist accommodation).
The notion of a modified PV/T + PCM application was introduced
first-ever and theoretically studied by Malvi et al. (2011). This type of 2.3. Alternative PCM-centred solar systems
system was considered as a PV plate back coated by a copper layer with
fixed tubes, a PCM layer and an insulation plate. Based on this approach, Recently, a more targeted analysis of PV + PCM systems is taking
they developed a one-dimensional model for the parametric analysis of place to achieve their optimal adaptation. Towards this direction,
significant variables (e.g. water flow rate, inlet temperature, PCM depth) several researchers have introduced alternative shapes of PCM enclo­
aiming to determine an optimal energy balance between the thermal sures to enhance the PV heat removal rate, the cost-effectiveness, and
and electrical efficiency of the proposed application. The researchers the sustainability of these solar systems.
concluded that PV energy output enhancement of 9% could be realised Huang (2011) developed an alternative method for thermal regula­
under certain circumstances (i.e. water flow rate of 2 L/h and 5 cm PCM tion of PVs by introducing the simultaneous utilisation of different PCM
thickness). Finally, they pointed out that the melting point is related to couples in triangular cells fitted in the PCM enclosure. To check the
the local ambient. efficiency of this modified PV + PCM approach, a suitable adjustment of
An innovative PV/T + PCM application was evolved and tested in a numerical model evolved for the single PCM exploitation took place.
Dublin by Browne et al. (2016). The experimental set up constituted of a Subsequently, several numerical simulations of multiple PCM combi­
PV module with a thermal collector nested in a filled PCM enclosure, a nations run for the summer conditions in the UK. Two types of metal
typical PV/T configuration, as well as two conventional PV modules cells shapes (i.e. triangular and half-circular) were considered for the
with and without any enclosure attached at the rear-side, respectively. encapsulation of the two separate PCMs as concerns to their melting
Their results via the comparative analysis of the individual subsystems points. In all combinations, PV + PCM unit temperature was maintained
indicated that PV/T + PCM system presented the highest efficiency, as it under 30 ◦ C leading to a rise in PV-efficiency; the RT27–RT21 combi­
achieved to rise by 5.5 ◦ C the temperature of the water and to enhance nation handled most efficiently the PV temperature under 22 ◦ C during
the energy generated by the PV. Also, they referred that such a system the selected period. However, RT27-RT27 combination was accepted as
could be used as a preheating water system. a more capable option for long-term regulating of the PV temperature
Preet et al. (2017) accomplished a day-long experimental procedure rise. Finally, the proposed alternative prevailed against the typical
in the outdoor environment of Gurdaspur (India) to examine the approach of using a single PCM, as regards to the thermal regulation
aggregated performance of a common PV module, a PV/T and a PV/T + capability and temperature uniformity on the PV surfaces.
PCM systems. Their experiment was undertaken at flow velocities from In the same pattern, Atkin and Farid (2015) examined both experi­
0.013 − 0.031 kg/s, while the PCM utilised in the PV/T + PCM system mentally and numerically the enhancement of the thermal management
was paraffin wax RT30. The main conclusions of this study were sum­ potential for PVs by the infusion of graphite into PCM, as well as the
marised as follows: a) both PV/T systems presented a significant tem­ configuration of a PCM enclosure with external fins. To this end, three
perature decline above 45% as compared to the reference case; b)the test cases for the controlling of PV temperature were developed; a PV +
peak energy efficiency of 12.6% was recorded for the PV/T + PCM PCM unit filled in graphite-infused PCM, a PV module coupled with a
system. finned plate, and a PV + PCM unit that employed an externally-finned
Based on similar experimental set-up, Li et al. (2019) evaluated container to embed the graphite-mixed PCM. They found that the
under real-field conditions, the efficiency of a PV + PCM/T and PV + enclosure which incorporates graphite soaked PCM and fins is the option
PCM module in comparison with a single PV unit. Before the trial pro­ of peak performance, as the PV efficiency could be increased by 12.97%.
cess, two different PCM pots were manufactured by polypropylene with Japs et al. (2016) tested the capability of pristine and mixed PCMs for
and without the addition of a thermal collector to integrate the paraffin PV thermal mitigation. Both PCMs macro-encapsulated in bags had the
wax with melting range 35–42 ◦ C. They found that the PCM exploitation same melting point, while the mixed PCM exhibited lesser thermal
leads to a significant PV temperature reduction raising PV energy output storage capacity and better thermal conductivity because of the graphite
by 5.18%. Besides that, they referred to a noticeable enhancement of the inclusion.
total energy (i.e. energetic sum) of the PV + PCM/T system due to its As stated by the findings, the energy-economic performance of the
additional capability for thermal energy generation. However, it should tested PV + PCM systems was recorded unfavourable as against the
be highlighted that the electric power generated by the PV + PCM/T reference case, even though the positive trend has appeared during the
application was slightly lower than the corresponding value for the PV 1st half of the tested days. Finally, they concluded that a mixed PCM
+ PCM system. would supply more effectively PV cooling as compared with a pristine
Shastry and Arunachala (2020) investigated another type of PV/T + PCM.
PCM concept, which incorporated an additional metal matrix in the Waqas et al. (2017) offered an inspired approach of applying PCM for
form of a honeycomb in the PCM container. An outdoor experiment the thermal regulation of PVs. In their approach, the paraffin RT24 was
carried out to test the overall performance of their concept comparing filled in copper pipes, which were settled at the rear layer of the PV

1285
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 1. Schematic outline of the selected PV module and PV + PCM system prototype.

module with suitable thermal paste. To measure the effects of the pro­
Table 1
posed PV + PCM system, the researchers conducted a 2-day experi­
Characteristic parameters of the investigated type of PV module.
mental procedure in Hefei, China. The main findings indicated that a
temperature drop by 8.5 ◦ C in the PV front layer, as well as amelioration Parameter Value

of 3% in energy efficiency, is possible. As concluded, the pipe based PCM Type of module Monocrystalline solar module
enclosure was far more cost-effective than a typical aluminium rectan­ Manufacturer LUXOR
Rated power 10 Wp
gular enclosure as a significantly lower quantity of PCM was utilized. In
Number of cells 4×9
a repetition study, Waqas et al. (2019) found that the PCM (with melting Rated voltage 17.39 V
range 30–32 ◦ C) encapsulation in aluminium tubes achieved the miti­ Rated current 0.58 A
gation of PV temperature by 2.8 ◦ C (on average) during daylight hours, Open circuit voltage 21.60 V
enhancing its conversion efficiency up to 2%. Short circuit current 0.64 A
Cell size 62.5 × 31.25 mm
In 2017, Hachem et al. presented an experimental examination of the
Weight 1.5 kg
impact of utilizing diverse types of PCM (i.e. pure and mixed) on the Power tolerance 0/+5%
thermal response and energy output of a PV module. For achieving this Temperature coefficient of power − 0.49%/◦ C
goal, a testing trial consisting of three exemplar cases was developed. Temperature coefficient of current − 0.05%/◦ C
Temperature coefficient of voltage − 0.35%/◦ C
The exemplar A was a commercial PV module as the baseline, while
exemplar cases B and C were formed with PCM enclosures that included
the pure PCM and mixed PCM (in a mass of 70% PCM, 20% copper and In details, three mc-Si PV modules manufactured by LUXOR Solar
10% graphite). Their results demonstrated a mean enhancement of 3% GmbH (Solo Line LX-10M) with a nominal power of 10Wp and di­
and 5.8% in the PV efficiency due to the integration of pure PCM and mensions of 353 mm × 293 mm × 35 mm served as testing cases in the
mixed PCM, respectively (Hachem et al., 2017). experimental procedure. One PV module without any modification was
The evidence presented in this section suggests that the optimisation considered as reference, and the other twain PV modules were coupled
of the correlation between the cost of the PCM enclosure and its effi­ with the copper-based PCM enclosures to encapsulate the selected sorts
ciency is critical for further development of the PV + PCM technology. In of paraffin RT27 and RT31, respectively. Each of the PCM enclosures
that sense, an optimal system design should be based on the recognition was made of copper pipes and fittings with an inner diameter of 20 mm,
of the real needs and potential obstacles of PV + PCM application, as arranged properly in a 2 mm thick copper plate to mimic a heat sink. In
well as the gaining of the highest PCM enclosure’s performance from the such a configuration, the copper pipes with PCM were expected to play
minimum required quantity of PCM. the role of external fins. The dimensions of these tubular formations
were in good accordance with the length and the width of the PV
3. Method and materials modules. In the present study, PCMs used were paraffin waxes RT27 and
RT31 which are characterised by heat storage capacities of 179 kJ/kg
3.1. Experimental set up and 165 kJ/kg and melting temperatures of 27 ◦ C and 31 ◦ C, respec­
tively. As the PCMs are positioned at the PV modules back, any potential
Before the analysis, two similar PV + PCM systems were devised and malfunction linked to the PCM volume expansion and its density during
fabricated to enable the quantification of the gains arising from diverse the phase change process should be avoided. Therefore, the PCM
PCMs inclusion at the rear surface of conventional PV modules. These enclosure fabrication was suitably adapted considering these specific
PV + PCM prototypes were exploited to undertake real-field experi­ parameters. Hence, the peak operative volume of each these enclosures
ments monitoring their efficiency. Fig. 1 depicts an indicative outline of was 340 mL, while the total mass of the inserted PCM was approximately
the adopted PV + PCM concept. 260 g. A short overview of the technical details for the PV modules and

1286
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Table 2 data were gathered from the meteorological station installed at the
Thermal and physical properties of the selected PCMs. campus of Technical University of Crete and is part of the developed
Parameter units PCM RT 27 PCM RT 31 network of the National Observatory of Athens (Lagouvardos et al.,
values 2017).
Melting area ◦
C 25–28 27–33 The electrical measurements acquisition system comprised of a
Main melting point ◦
C 27 31 validated I-V tracer and a hosting computer. In detail, the I-V tracer was
Congealing area ◦
C 28–25 33–27 wired by means of a switch to the PV module and the PV + PCM systems
Heat storage capacity [±7,5%] kJ/kg 179 165 to enable their output current and voltage signals recording. By using
Specific heat capacity kJ/kg⋅K 2 2
Density solid kg/L 0.88 0.88
the proper commercial software (RealTerm version 2.0.0.70), the elec­
Density liquid kg/L 0.76 0.76 trical parameters (i.e. ISC, VOC, IMAX, VMAX) were measured for the three
Heat conductivity W/m⋅K 0.2 0.2 PV modules with a 10 s delay considered being simultaneous. The entire
Volume expansion % 12.5 12.5 experimental set-up is depicted in Fig. 2 and detailed info about the
Flash point C 146 157
experimental sampling devices in Table 3.

the main characteristics of the selected PCMs are reported in Tables 1


and 2, respectively.
Throughout the experimental process, the three modules were Table 3
mounted on the same base fixed at an inclination of 30◦ from the hori­ Summary description of the data sampling devices and sensors.
zontal and an azimuth angle of 0◦ (south-facing) to be examined under Measurement Device Model Measurement Measurement
the same conditions. Apart from the systems to be tested (i.e. PV refer­ parameter range error

ence, PV + PCM27 system, PV + PCM31 system) and the components for Solar in-plane Kipp & Zonen CMP6 0–2000 Wm− 2
<4%
their outdoor installation, the experimental setup consisted of an irradiance pyranometer
Current I-V tracer NA 0.5–40 A
appropriate data acquisition system including sampling devices and ±2%
Voltage 10–250 V
parts utilised to monitor their thermal and the electrical performance. Back surface Omega digital HH309A − 200 to ±0.2% (rdg +
The operating temperature of PVs characterising their thermal per­ temperature data logger 1370 ◦ C 1 ◦ C)
formance was monitored utilising three K-type thermocouples, which thermometer
were centrally settled at the backside of the PVs and between of the PV Front surface Powerfix − 50 to 380 ◦ C ±0.5 ◦ C
temperature infrared
+ PCM systems’ elements. To assure enhanced conduction and precise
thermometer
temperature observations, an appropriate thermal paste was applied. Thermal image Irisys thermal 4010 NA NA
These measurements were collected by a digital data logger thermom­ camera
eter and delivered to a hosting computer through proper software. Ambient Davis Wireless − 40 to 65 ◦ C ±0.3 ◦ C (rdg
temperature1 Weather Vantage 16–38 ◦ C)
Moreover, an infrared temperature probe having an accuracy of ±0.5 ◦ C
Wind speed1 station Pro2 3–241 km/h ±5%
was used to record manually the PV modules temperature at the front Wind 0–360◦ ±3◦
surface. Supplementary, the thermal behaviour of the PV + PCM systems direction1
and PV reference could be recognised as concerns the temperature dis­ Solar 0 to 1800 W/ ±5%
tribution by employing a thermal camera (IRYSIS 4010). irradiance on m2
a horizontal
А pyranometer CMP6 (KippZonen) fixed on the top of the experi­
surface1
mental set-up connected with a digital multimeter were used to compute
1
the total incident radiation reaching the surface of the PVs. The weather Source: Lagouvardos et al. (2017).

Fig. 2. Schematic depicture of experimental set-up.

1287
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 3. Overall presentation of the experimental procedure adopted.

3.2. Experimental procedure data from 22nd of February, 14th of March, 10th of April, 3rd and 13th
of May, 20th and 25th of June 2019 are obtained as representative for
The experiments were executed in the ReSEL facilities (latitude: the goal-oriented analysis.
35o31′ 0N, altitude: 24o04′ 0E) under the real-field conditions of Medi­ Before the beginning of the experiments, the three PV modules
terranean coastal areas, such as Chania, Greece. Generally, the major without any modification (e.g. PCM enclosures integration) were
part of metering data was logged on 10-minute intervals, from 8:00 to checked for two days as concerns to their energy performance consis­
18:00. An indicative flowchart of the experimental procedure is given in tency. According to the measurements received, they operated similarly
Fig. 3. The experimental measurements indicating the energy perfor­ presenting a slight difference of 2%, which is associated with the I-V
mance of the three PV systems were gathered along the five-month tracer accuracy.
monitoring period (12 February 2019–12 July 2019). In this research,

1288
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 4. Variation of the operating temperature of the PV-ref, PV + PCM27 and PV + PCM31 systems, ambient temperature, wind speed, solar irradiance during the
selected experimental days: (a) 22/2/2019, (b) 14/3/2019, (c) 10/4/2019, (d) 3/5/2019, (e) 13/5/2019, (f) 20/6/2019 and (g) 25/6/2019).

3.3. PV and PV + PCM systems performance analysis performance requires a deep insight into the heat transfer mechanisms
involved (e.g. conduction, convection, radiation exchange). In such an
To assess the PV and PV + PCM systems’ performance a set of in­ approximation, the calculation of Tmb (PV+PCM) should consider the
dicators were calculated using data logged during the experimental complexities of these mechanisms, as well as the specific features of PCM
procedure (i.e. solar irradiance, current, voltage, open-circuit voltage (Huang, 2011). On the other hand, there is a significant number of
and short circuit current, temperature): explicit equations expressing the PV module temperature, in response to
the relevant weather parameters (i.e. Tα, VW, GΤ) in the literature
• Instantaneous maximum PV power generation (PMAX) (Skoplaki et al., 2008). These equations are simplest for use, and their
• Daily PV power output (EDC,d) prior evaluation is suggested for better estimation of Tmb in different
• PV module’s electrical efficiency (nELEC) climate zones.
• Fill factor (FF). Bearing in mind the shaping details of the proposed alternative PV +
PCM system (e.g. PCM quantity used, PCM container design, etc.), an
More specifically, the incident sunlight power on the front surface of empirical model for the quick estimation of Tmb (PV+PCM) can be
a PV module is calculated by (Kalogirou, 2009): formulated using multiple linear regression. More specifically, a linear
equation for the dependent variable (i.e. Tmb (PV+PCM)) using three in­
PINC = GT ⋅A (1)
dependent variables X1,X2 and X3 representing GT, Tα and VW, respec­
The instantaneous maximum power generation of a PV module is tively can be given in a general form, as follows:
computed by the formula (Kalogirou, 2009):
yi = b0 + b1 ⋅xi,1 + b2 ⋅xi,2 + ⋯ + bk− 1 ⋅xi,k− 1 + ei (6)
PMAX = IMAX ⋅VMAX (2)
where yi is the predicted Tmb (PV+PCM) and assumes ith independent error
The electrical efficiency of a PV module with and without PCM is ei ~ N(0,σ2) (Savvakis and Tsoutsos, 2015; Charles Lawrence Kamuyu
possible to be defined by the following equation (Kalogirou, 2009): et al., 2018).
PMAX IMAX ⋅VMAX Correspondingly, the proposed model can be expressed as:
nELEC = = (3)
PINC GT ⋅A Tmb(PV+PCM) = b0 + b1 ⋅GT + b2 ⋅Ta + b3 ⋅VW (7)
The fill factor is estimated by the formula (Kalogirou, 2009):
where b0, b1, b2, b3 are the regression coefficients for the tested system.
IMAX VMAX
FF = (4)
ISC VOC 4. Results and discussion
The daily energy output over a daily period [t1, tN] is computed via
Simpsons’ rule by the simplified equation (Hansen et al.,2012): 4.1. PV thermal behaviour under outdoor weather conditions

t ∑N
In the current section, the experimental data of the PV modules
EDC,d = ⋅ PMAX (ti ) (5)
T i=1 operation with and without the selected PCMs are examined. To meet
this objective, the weather records, and the temperature curves of the
conventional PV module and PV + PCM systems along seven represen­
3.4. PV and PV + PCM systems operating temperature prediction
tative days were studied. Fig. 4 illustrate the evolution of the operating
temperatures of the PV reference (Tmb (reference case)/Tmf (reference case)), the
The accurate estimation of a PV + PCM system’s thermal

1289
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 5. Short-Circuit Current (ISC) and Open-Circuit Voltage (VOC) for the PV-ref, PV + PCM27 and PV + PCM31 systems during the selected experimental days: a)
22/2/2019, b)14/3/2019, c)10/4/2019, d)3/5/2019, e)13/5/2019, f)20/6/2019 and g)25/6/2019.

1290
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

PV + PCM27 system (Tmb (PV+PCM27)/Tmf (PV+PCM27)) and the PV + 50 ◦ C for about 5 h (between 11:40 and 16:20). The peak temperature of
PCM31 system (Tmb (PV+PCM31)/Tmf (PV+PCM31)), as well as the weather 55.7 ◦ C was found at 14:30 when the GT, Tα and VW were 906 W m− 2,
parameters including in-plane solar irradiance (GT), ambient tempera­ 28.7 ◦ C and 8 km h− 1, respectively. The related values of the Tmf (reference
ture (Tα) and wind velocity (VW) during the selected experimental days. case) were presented marginally lowered, while an identical evolution is
The curves related to the working conditions of the PV reference and pointed out. As concerns to the thermal behavior of the PV + PCM27 and
the PV + PCM systems during a sunny winter day of weak wind PV + PCM31 systems, the Tmb (PV+PCM27) and Tmb (PV+PCM31) recorded
(February 22nd, 2019) are displayed at first in Fig. 4. The Tmb (reference following the same trend of Tmb (reference case) curve; however, both the PV
case) reached its maximum value of 51.9 C at 13:00, when the GT, Tα and + PCM systems maintained lower temperature than the case of reference

VW were 1006 W m , 13.4 C and 11.3 km h− 1, respectively. At the


− 2 ◦
until 16.00 when the Tmb (reference case) equalled for 1.5 h to their corre­
same time, the Tmb (PV+PCM27) and Tmb (PV+PCM31) observed lightly lower sponding values (Fig. 4(f)). Comparing the impact of the PCMs
by 2.4 ◦ C and 2.6 ◦ C, respectively. At the start of the experimental day’s employed, PCM27 kept a slightly lower temperature than PCM31 until
cycle, the operating temperature of the PV + PCM27 system has noon times which can be recognised due to the differentiation in their
increased steadily from 10.2 ◦ C at 8:00 to 40.8 ◦ C at 11:10 (in about 3.2 specific properties (i.e. thermal capacity). The maximum temperature
h) continuing under 50 ◦ C for the rest of the day. The positive impact of mitigation of 3.9 ◦ C was achieved by PCM31 at 9:00, compared to the
the PCM27 embedding observed mainly in the period from 9:40 to 3.8 ◦ C of PV + PCM27 at 8:40, against the corresponding values of the
11:00, when the ΔΤ (Tmb (reference case)/Tmb (PV+PCM27)) presented a mean PV reference. Also, the PCM inclusion assisted in the declining of the
value of 3.1 ◦ C. The Tmb (PV+PCM31) varied from 10.2 ◦ C at 8:00 to 40.3 ◦ C average PV operating temperature of about 2 ◦ C. Taking into consider­
at 11:10, while the maximum value of ΔΤ (Tmb (reference case)/Tmb ation that a PCM quantity of 0.26 kg was used in the tested PV + PCM
(PV+PCM31)) equalled to 4.3 C at 11:30. Additionally, the temperature systems, the peak temperature difference cannot be considered as

deviation was presented negative, varying between − 0.2 ◦ C at 15:10 and negligible, since the index of effective PCM mass coefficient of 0.67 kg/
− 4.7 ◦ C at 17:00 for the PV-PCM27 system, indicating a slower time- m2⋅oC is also significantly lower than the reported values in similar
response of the PCM27-based system to the rapid decrease of GT. A studies.
similar trend has been also recorded for the PV + PCM31 system at the The variation of the operating data of the PV reference and the PV-
same time. In general, the effect of GT variation on Tmb (PV+PCM27) and PCM systems during a summer day with strong wind is presented in
Tmb (PV+PCM31) can also be recognized in the temperature curves logged Fig. 4(g). The Tmb (reference case) reached its maximum value of 55.5 ◦ C at
on the dates of 14/3/2019 and 13/5/2019, when a smoother variation 14:00, which is about 30 ◦ C higher than the Tα. The temperature profile
of temperature and lag of changes as against the Tmb (reference case) noted as of the PV reference mimics the variation trend of the GT and Tα, while
the result of PCMs’ use. the effect of VW can also be considered as significant. On the other hand,
Under the influence of weather on 10th of April 2019, the recorded the Tmb (PV+PCM27) and Tmb (PV+PCM31) are noted mildly lower in the value
values of Tmb (reference case), Tmb (PV+PCM27) and Tmb (PV+PCM31) lied within area 27.7–53.6 ◦ C, because of the mitigation influence of the embedded
the ranges of 13.7–43.1 ◦ C, 13.3–39.2 ◦ C and 13.4 ◦ C − 39.1 ◦ C, PCMs. At the beginning of the test, the operating temperature of the PV
respectively. In more detail, Tmb (reference case) rose almost evenly during + PCM27 system rose progressively from 27.7 ◦ C at 8:00 to 45.8 ◦ C at
the first daytime hours from 13.7 ◦ C at 8:00 to 30.1 ◦ C at 10:40 and 10:00 (in 2 h) remaining under 50 ◦ C for a total period of about 3.6 h.
retained between 31.9 ◦ C and 43.1 ◦ C for a period> 6 h. The maximum Similarly, Tmb (PV+PCM31) ranged from 28.1 ◦ C at 8:00 to 45.4 ◦ C at
Tmb (reference case) of 43.1 ◦ C was found at 13:50, which is 25.5 ◦ C greater 10:00, while a temperature-rise rate of 3.5 ◦ C h− 1 observed after that
than Tα. The thermal profiles of the PV PCM27 and PV PCM31 systems time for 2 h. However, the effect of PCMs’ minimised after 11:00 to
(i.e. Tmb (PV+PCM27) and Tmb (PV+PCM31)curves) followed the Tmb (reference 14:00 when the mean temperature deviation among the conventional
case) curve variation. However, both PCM-based systems operated in PV module and the PV + PCM systems was about 1.7 ◦ C. This limitation
lower temperatures than the Tmb (reference case) values, due to the observed appeared simultaneously with the decrease of wind velocity from 41.8
synergistic effect of the PCMs and VW. That effect presented until 17:00 km h− 1 at 11:00 to 27.4 km h− 1at 14:00, demonstrating its significant
when their values observed higher than the reference at the end of the impact on the developed PCM enclosure as the main determinant of the
experimental day’s cycle. The comparative assessment of the tested free-convective heat-exchange process. Moreover, the temperature
PCMs effectiveness pointed out that PCM31 held a marginally lower alteration presented its maximum value of 3.8 ◦ C at 16:30 for the PV-
temperature than PCM27 until midday which stemmed from their PCM27 and a maximum of 3.6 ◦ C at 16:10 against the Tmb (reference case)
different thermal properties (i.e. melting temperature range). The peak of 52.9 ◦ C. This phenomenon can be attributed to the cumulative in­
temperature reduction of 7.5 ◦ C obtained by the PCM31-based system at fluence of (i) the decreasing trend of GT, (ii) the lower values of Tα and
12:30, compared to the 6.4 ◦ C by the PV + PCM27 system recorded at (iii) the high values of VW during the afternoon time. During the
the same time, against the matching values of the reference case. Be­ experimental process, the VW ranged between 24.1 and 49.9 km h− 1, the
sides, the embedding of PCM contributed to the diminishing of the mean Tα varied from 23.4 ◦ C to 26.6 ◦ C and GT was in the range from 98 W
PV operating temperature by about 3 ◦ C. Considering that a PCM m− 2 at 8:00 to 920 W m− 2 at 13:40. Consequently, these results indicate
quantity of 0.26 kg was exploited in the tested PV + PCM systems, the that VW, as well as the PCM inclusion, achieved a synergistic effect in PV
achieved temperature difference at the peak is far from negligible. This cooling along the testing day.
finding explained by the calculated effective PCM mass coefficient of
0.34 kg/m2⋅oC, which is substantially lower than the reported values in 4.2. PV electrical characteristics
the literature (Waqas et al., 2018).
As shown in Fig. 4(f), GT increased steadily from 98 W m− 2at 8:00 to The temperature decrease achieved by PCM inclusion influenced the
906 W m− 2 at 13:40, while it remained above 750 W m− 2for about 4.5 h. voltage (V) and current (I) output of the tested PV modules, as defined
At the end of the experiment, the corresponding value was equal to 463 from the relative temperature dependence coefficients. Fig. 5(a–g) show
W m− 2. The VW ranged from 1.6 km h− 1/h to 11.3 km h− 1with a mean of the variation of the short-circuit current (ISC) and open-circuit voltage
6.8 km h− 1, indicating a day of weak wind. The Tα fluctuated from (VOC) for the PV reference, PV + PCM27 and PV + PCM31 systems for
22.4 ◦ C at 8:00 to its peak value of 29.1 ◦ C at 13:20 with a mean value of the selected days.
26.9 ◦ C. Under the given weather conditions on 20th of June 2019, the Indicatively, Fig. 5(b) indicates the comparing of VOC and ISC output
observed values for Tmb (reference case), Tmb (PV+PCM27) and Tmb (PV+PCM31) among the investigated systems. Along the main part of the experi­
were in the ranges of 28.8–55.7 ◦ C, 27.0–54.2 ◦ C and 27.4− 54.1 ◦ C, mental day’s cycle, the cells of the conventional PV module generate
respectively. More specifically, Tmb (reference case) grew smoothly in the lower VOC (in the range:19.75–22.81 V) while the PV cells in the PV +
morning hours from 28.8 ◦ C at 8:00 to 50.8 ◦ C at 11:40 and retained over PCM27 and PV + PCM31 systems observed with higher values (within

1291
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 6. Temperature difference at the back surface between the PV-ref, PV + PCM27 and PV + PCM31 systems, the variation of their power output and energy
conversion efficiency during specific experimental days of spring and summer: (a) 14/3/2019, (b) 10/4/2019, (c) 20/6/2019, (d) 25/6/2019.

the ranges: 20.24–22.86 V and 20.22–22.85 V, respectively) and limited similar variation trend of the tested PV modules’ electrical parameters in
fluctuation. The deviations of the VOC (PV+PCM27) and VOC (PV+PCM31) unstable weather conditions was pointed out in Fig. 5(e).
profiles against the related VOC (PV reference) curve are resulted mainly by As shown in Fig. 5(c), the ISC (PV reference) elevated steadily from 0.06 A
the ability of the PCMs’ employed to ensure lower operating tempera­ at 8:00 to 0.68 A at 13:10, and thereafter declined to 0.28 A at 16:50
tures and limit their sharp changes. As concerns the response of ISC (PV under the influence of GT variation. Proportionally, the recorded ISC
reference), ISC (PV+PCM27) and ISC (PV+PCM31), the related values were mainly (PV+PCM27) and ISC (PV+PCM31) were 0.06 A and 0.07A at 8:00; 0.67 A and
affected by the large variation of GT while the impact of the Tmb values 0.66 A at 13:10; and 0.29 A and 0.28 A at 18:00. This minor alteration of
received by the three tested systems was practically insignificant. A ISC (PV reference) may be caused by the higher working temperatures of the

1292
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 6. (continued).

reference case in comparison with the PCM-supported systems. In decreased to 0.37 A at 18:00, following the GT curve with time. Corre­
contrast, VOC (PV+PCM27) and VOC (PV+PCM31) values found greater than spondingly, the values for the PV + PCM27 and PV + PCM31 systems
VOC (PV reference) along the experimental cycle while their mean values were 0.11 A and 0.10A at 8:00; 0.71 A and 0.72 A at 13:40; and 0.34 A
were of 21.17 V, 21.18 V and 20.95 V, respectively. In reliance on the and 0.36 A at 18:00. The marginal increase of ISC (PV reference) attributes to
fluctuation lines of VOC, a negative impact underlined for the tested PV the higher operating temperatures of the reference case as compared to
modules’ output due to the rising of the PV temperature. However, this the PCM-based systems. Besides, VOC (PV+PCM27) and VOC (PV+PCM31) were
effect on the PV + PCM systems recognized as smoother, while a clear retained higher than VOC (PV reference) during the experiment with peak
VOC difference between the PV reference against the PCM-embedded values of 21.76 V, 21.64 V and 20.85 V, respectively. Based on the
systems occurred between 9:00–17:00. variation curves of VOC, it can be highlighted that the relative values for
According to the results shown in Fig. 5(f), the ISC (PV reference) was the PV PCM systems raised in compliance with the solar irradiance
0.11 A at 8:00 that raised gradually 0.72 A at 13:40, and subsequently evolution until 9:20 when a negative trend presented as the result of the

1293
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 7. Daily electric energy output of the PV-ref, PV + PCM27 and PV + PCM31 systems for the selected experimental days.

rising of the PV temperature. This temperature effect lasted for about ref, respectively. These values steadily elevated to 9.58 W and 9.34 W at
6.5 h, while the maximum VOC difference between the PV reference and 13:10 when the passive cooling effect was limited. During the period
the PV + PCM systems appeared in two distinct periods (between from 14:00 to 18:00, a downward trend appeared, while the PV modules
8:00–9:50 and 13:10–14:30). In correspondence, the data for the elec­ were performing under negligible alteration. For example, at 14:50, the
trical characteristics (i.e. ISC-VOC curves) of the experimentally evalu­ power by the reference PV module, the PV + PCM27 system and PV +
ated PVs recorded on May 3rd, 2019, presented a similar response to the PCM31 system were 8.85 W, 8.94 W and 8.76 W, respectively. Overall,
fluctuation of the weather parameters (Fig. 5(d)). The high values of the energy output per day by the PV + PCM27 system and the PV +
wind velocity during a summer day (June 25th, 2019) affected reason­ PCM31 system was 69.82 Wh/day and 69.27 Wh/day, while the PV
ably the VOC output of the PV + PCM systems, considering that a con­ reference generated 67.59 Wh/day.
stant temperature difference against the PV reference occurred. As In summary, these results revealed that both PCMs selected
recognised in Fig. 5(g), the measured maximum and mean VOC of the PV contributed to PV power increment and indicated that a greater PCM
reference, PV + PCM27 and PV + PCM31 systems were 20.2 V and 18.1 quantity could reach better results and time-consistency with the period
V, 21.4 V and 18.7 V, 21.3 V and 18.6 V, respectively; also, a gradual of high solar irradiance.
decline in the VOC values took place from 9:30 to 18:00 due to the PV Following the power output profiles of the tested cases, a significant
modules’ temperature rise. As concerns to the measurements of ISC for improvement, ranged from 5% to 9%, in the efficiency of the PV + PCM
the three modules, a similar fluctuation with no significant deviations systems was monitored between 8:00 and 9:50 (Fig. 6(c)). After that
among the PV reference and the PV + PCM systems was recorded during time, the PV modules’ efficiencies were lessening reasonably even
the whole experimental procedure. Furthermore, the profiles of ISC (PV though the PCM-based systems operated in marginally lower tempera­
reference), ISC (PV+PCM27) and ISC (PV+PCM31) were not affected by the tem­ tures as compared to the reference case. The mean and the peak energy
perature rise, while the corresponding values were proportional to solar conversion efficiencies were 10.22% and 10.92%, 10.17% and 10.88%,
irradiance. 9.87% and 10.57% for the PV + PCM27 system, the PV + PCM31 system
and PV the PV reference, respectively.
4.3. Power output and conversion efficiency From Fig. 6(d), it is apparent that the PV modules cooled by the
tubular PCM enclosure presented improved efficiencies and higher
The current section of this study gets ahead in analysing the essential electricity production than the conventional module in the length of
issues regarding the electrical efficiency of the experimentally investi­ time during a summer day with increased wind velocity. More specif­
gated cases. The recorded measurements of V and I for the PV module ically, the ranges of power generation were from 1.17 W to 9.44 W for
and the PV + PCM systems were fully tapped to compute the power the PV + PCM27, from 1.15 W to 9.39 W for the PV + PCM31 and from
output and solar energy conversion efficiency during the selected days. 1.06 W to 9.18 W for the PV reference. The mean value of the power
Hereafter, the maximal and daily mean values of the energy earnings enhancement achieved by the PCM27 and PCM31 inclusion was 3.56%
were estimated from the variation of the power output curve along the and 2.59%, respectively. As shown in Fig. 6(d), the power output pro­
experiment periods. files of the tested systems follow the GT curve, while the wind effect is
According to the obtained results, the electricity generated by the PV notably significant. For instance, at 12:50, the power generation of the
+ PCM27 and PV + PCM31 systems over the major part of a typical reference PV module, PV + PCM27 system and PV + PCM31 system
summer day with low wind velocity was slightly higher than the pro­ were observed 9.07 W, 9.25 W and 9.26 W, respectively, when the GT,
duced by the PV reference because of the temperature difference among Tα and VW were 892 W m− 2, 26.2 ◦ C and 35.4 km h− 1, respectively.
them, as can be identified by Fig. 6(c). In particular, the power outputs Comparing the corresponding values of 8.96 W, 9.17 W and 9.15 W, at
of the PV + PCM27 and PV + PCM31 systems were 1.02 W and 0.98 W at 14:00, when the VW decreased to 26.4 km h− 1 while the GT and Tα
8:00, which corresponds to a rise of 6.2% and 2% compared to the PV remained constant, the synergistic effect of wind by increasing the free-

1294
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Table 4
Multiple linear regression analysis for the PV + PCM31 system under diverse weather conditions.
Tmb (PV + Testing conditions
PCM31) model
ΚT > ΚT > ΚT > ΚT > ΚT > ΚT > ΚT > 0.5, ΚT > 0.5, ΚT > 0.5, ΚT > 0.5, ΚT > ΚT > ΚT >
0.5, 0.5, 0.5, 0.5, 0.5, 0.5, VW ≥ 19 VW ≥ 19 VW ≥ 19 VW ≥ 19 0.5, 0.5, 0.5,
VW ≥ 19 VW ≥ VW ≥ VW ≥ VW ≥ 19 VW ≥ km h− 1 km h− 1 km h− 1 km h− 1 VW ≥ 19 VW ≥ VW ≥
km h− 1 19 km 19 km 19 km km h− 1 19 km km h− 1 19 19
h− 1 h− 1 h− 1 h− 1 km km
h− 1 h− 1

ANOVA

df SS MS F Significance F

Regression 3 3 49,350 26,218 16,450 8739 2502 1028 <0.0001 <0.0001


Residual 425 204 2794 1735 7 9
Total 428 207 52,144 27,953

Coefficients Standard Error t-Stat P- value Lower 95% Upper 95%

Constant 4.414 4.815 0.5414 1.186 8.154 4.058 <0.0001 <0.0001 3.350 2.476 5.479 7.154
GT 0.025 0.029 0.0005 0.001 48.338 34.611 <0.0001 <0.0001 0.024 0.027 0.026 0.030
VW − 0.079 0.101 0.0388 0.025 − 2.038 4.098 0.0422 <0.0001 − 0.155 0.053 − 0.003 0.150
Tα 0.938 0.685 0.0208 0.027 45.071 25.066 <0.0001 <0.0001 0.898 0.631 0.979 0.739

Regression Statistics

Multiple R 0.97 0.97


R2 0.95 0.94
R2 adjusted 0.95 0.94
Standard Error 2.56 2.92
Observations 429 208

convection heat transfer from the rear side of the PV + PCM systems, can average increase of 3.4% and 2.5%, respectively, when compared with
be recognised. the PV reference output (67.59 and 68.09 Wh/day). Contrariwise, PV +
Along the same lines, the results of the PV module efficiency (Fig. 6 PCM31 system presented slightly higher energy outputs (i.e. 54.79,
(d)) demonstrated a clear enhancement in the energy conversion effi­ 48.26, 69.42, 57.72 Wh/day) in the cases that weather instability
ciency of the PV modules with PCM as their variation ranged from appeared (e.g.14/3/2019, 13/5/2019, etc.), demonstrating that except
9.45% to 11.81% in case of using PCM27 and from 9.31% to 11.82% in for the melting area, the heat storage capacity plays a key role in the
case of using PCM31. The corresponding values for the PV module thermal response of a PV + PCM system under varying operating
without PCM was from 9.19% to 11.21%. Thus, the average increment weather conditions.
of the efficiency of the PV + PCM27 and the PV + PCM31 systems
compared to the reference case was 3.74% and 2.59%, respectively. 4.4. Empirical models to estimate the operating temperature of the
Also, it should be underlined that the observed efficiency improvement investigated systems
presented a positive trend, from 8:00 to 10:10 and 15:50 to 17:10, due to
the aggregated influence of the used PCMs and the wind velocity. Exploiting further the available experimental data, three similar
From Fig. 6(b), it becomes clear that the PV + PCM systems operated empirical approaches considered to estimate the operating temperature
in better efficiencies and enlarged electricity output in comparison with of the proposed alternative PV + PCM systems and the PV reference.
the PV reference during a spring day with moderate wind velocity. The These modelling approximations correlating Tmb, GT, Tα and VW at
monitored power generation of the PV + PCM27 system ranged from different conditions were established using multiple linear regression.
0.94 W to 9.87 W, while the matching values for the PV + PCM31 system The sample of data (i.e. 10% of the total) utilized for the analysis was
and the PV reference were from 0.94 W to 9.93 W and from 0.96 W to indicative of the prevailing conditions along the experimental period
9.41 W, respectively. Hence, the highest power enhancement achieved while in parallel the restrictions (ΚT ≥ 0.5) have been taken into
by the PCM27 and PCM31 embedding was 4.19% and 4.24%, respec­ consideration. Then, the received results examined statistically to
tively. it should be noted analysing further the recorded energy data that evaluate the accuracy of the identified approximations.
the power output curves of the researched systems meet the GT curve
trend, while the wind effect on them is observed. Indicatively, at 13:30, If ΚT ≥ 0.5 and VW < 19 km h− 1:
the PV, PV + PCM27 and PV + PCM31 systems’ power output were
equal to 9.41 W, 9.84 W and 9.93 W, respectively, under the conditions: Tmb(PV+PCM31) = 4.414 + 0.025⋅GT + 0.938⋅Ta − 0.079⋅VW [8a]
GT of 1033 W m− 2, Tα of 16.9 ◦ C and VW of 19.3 km h− 1. As illustrated in An outline of the statistical analysis results for the PV + PCM31
Fig. 6(d), the PV efficiency improved by the PCM27 inclusion ranging system is given in Table 4. GT, Tα and VW examined as independent
from 8.99% to 10.21%, while the corresponding values using PCM31 variables of the gained relation under specific restrictions (i.e. ΚT ≥ 0.5
were in the range 9.01% to 10.22%. The efficiency range (i.e. 8.47% to and VW < 19 km h− 1). The obtained values of correlation coefficient (R
10.18%) observed for the case of the PV module without PCM reflected = 0.97) and coefficient of determination (R2 = 0.95) highlight that the
its poor performance. Consequently, the average efficiency increment proposed linear equation (8a) can be applied to estimate Tmb (PV+PCM31).
achieved by the PV + PCM27 and the PV + PCM31 systems as against Further, the adjusted coefficient of determination (R2adj) of 0.95 in­
the reference case was 4.14% and 4.19%, respectively. dicates that the model can be broadened for the total of the population
In Fig. 7, the energy production of the PV reference, PV + PCM27 data. As stated by the analysis of variance (ANOVA), the dependent
system and PV + PCM31 system is illustrated. In the selected experi­ relation among the proposed equation’s variables validated since its
mental days in summer, the PV + PCM27 system produced more energy results recognized as statistically significant considering that the
(69.72 and 70.45 Wh/day) than the generated by the PV + PCM31 observed F-ratio is lower than the value of 0.001. To be interpreted the
system (69.27 and 69.77 Wh/day). These results correspond to a slight main parameters of the model, some remarks explained in more details:

1295
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Table 5
Multiple linear regression analysis for the PV + PCM27 system under diverse weather conditions.
Tmb (PV + Testing conditions
PCM27) model
ΚT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT > ΚTT >
0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5,
VW < 19 VW ≥ VW ≥ VW ≥ VW ≥ 19 VW ≥ VW ≥ 19 VW ≥ 19 VW ≥ 19 VW ≥ 19 VW ≥ 19 VW ≥ VW ≥
km h− 1 19 km 19 km 19 km km h− 1 19 km km h− 1 km h− 1 km h− 1 km h− 1 km h− 1 19 19
h− 1 h− 1 h− 1 h− 1 km km
h− 1 h− 1

ANOVA

df SS MS F Significance F

Regression 3 3 52,243 25,741 17,414 8580 2479 1195 <0.0001 <0.0001


Residual 424 193 2978 1386 7 7
Total 427 196 55,221 27,127

Coefficients Standard Error t-Stat P- value Lower 95% Upper 95%

Constant 3.205 6.983 0.557 1.132 5.752 6.167 <0.0001 <0.0001 2.110 4.750 4.301 9.217
GT 0.024 0.032 0.001 0.001 45.308 40.098 <0.0001 <0.0001 0.023 0.030 0.026 0.033
VW − 0.087 0.095 0.039 0.023 − 2.226 4.048 0.027 <0.0001 − 0.164 0.049 − 0.010 0.141
Tα 1.037 0.564 0.022 0.025 47.820 22.137 <0.0001 <0.0001 0.995 0.513 1.080 0.614

Regression Statistics

Multiple R 0.97 0.97


R2 0.95 0.95
R2 adjusted 0.95 0.95
Standard 2.65 2.68
Error
Observations 428 197

• Tmb (PV+PCM31) raises with rising GT and Tα, as a consequence of the • By virtue of t-values, the predictor variables are arranged in the
corresponding coefficients positive sign. following manner: GT, Tα and VW
• Conversely, Tmb (PV+PCM31) diminishes with increasing VW, as the If ΚT ≥ 0.5 and VW ≥ 19 km h− 1:
corresponding coefficient displayed a negative sign.
Tmb(PV+PCM31) = 4.815 + 0.029⋅GT + 0.685⋅Ta + 0.101⋅VW [8b]
• The executed t-tests for the regression coefficients linked with the Tα
and GT (independent variables), yielded p < 0.001, so these variables By the results indicating the PV + PCM31 system’s thermal behav­
deemed as statistically significant. Contrariwise, the p-value of iour (Fig. 4.), in cases, the lower temperatures observed due to the
0.0422 (≫0.0001) for the coefficient related to VW constituted an higher VW. Thus, the determination of a correlation under the specific
indication of its minor statistical significance. constraints (i.e. KT ≥ 0.5 and VW ≥ 19 Km h− 1) characterized as a
reasonable option. Correspondingly, multiple linear regression

Table 6
Multiple linear regression analysis for the PV reference under diverse weather conditions.
Tmb (PV Testing conditions
reference) model
ΚT > ΚT > ΚT > ΚT > ΚT > ΚT > ΚT > 0.5, ΚT > 0.5, ΚT > 0.5, ΚT > 0.5, ΚT > ΚT > ΚT >
0.5, 0.5, 0.5, 0.5, 0.5, 0.5, VW ≥ 19 VW ≥ 19 VW ≥ 19 VW ≥ 19 0.5, 0.5, 0.5,
VW < 19 VW ≥ VW ≥ VW ≥ VW ≥ 19 VW ≥ km h− 1 km h− 1 km h− 1 km h− 1 VW ≥ 19 VW ≥ VW ≥
km h− 1 19 km 19 km 19 km km h− 1 19 km km h− 1 19 19
h− 1 h− 1 h− 1 h− 1 km km
h− 1 h− 1

ANOVA

df SS MS F Significance F

Regression 3 3 68,744 26,120 22,915 8707 4742 1408 <0.0001 <0.0001


Residual 628 188 3035 1162 5 6
Total 631 191 71,779 27,282

Coefficients Standard Error t-Stat P- value Lower 95% Upper 95%

Constant 8.214 8.347 0.386 1.057 21.255 7.895 <0.0001 <0.0001 7.455 6.261 8.972 10.433
GT 0.029 0.032 0.000 0.001 85.241 41.503 <0.0001 <0.0001 0.028 0.030 0.029 0.033
VW − 0.091 0.088 0.022 0.022 − 4.123 4.086 <0.0001 <0.0001 − 0.135 0.045 − 0.048 0.130
Tα 0.810 0.594 0.013 0.025 64.536 23.750 <0.0001 <0.0001 0.785 0.545 0.834 0.644

Regression Statistics

Multiple R 0.98 0.98


R2 0.96 0.96
R2 adjusted 0.96 0.96
Standard 2.20 2.49
Error
Observations 632 192

1296
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 8. Simulation results and experimental data for the PV-ref, PV + PCM27 and PV + PCM31 systems during two typical summer days (July 5th and July 12th,
2019) of weak wind (0–19 km h− 1).

employed to develop a linear equation among Tmb (PV+PCM31), GT, Tα and Tmb (PV+PCM31) and VW is converse than expected since a synergistic
VW. The developed equation (8b) presented a satisfactory ability to effect of PCM31 and VW has been recorded in temperature variation
predict the Tmb (PV+PCM31) as depicted by the statistical analysis results curves (Fig. 4). However, the small number of observations and
(Table 4). In detail, the R, R2 and R2adj found equal to 0.97, 0.94 and seasonality seemed to play a key role to form this result.
0.94, respectively, determining a high correlation between the regres­ • The explanatory variables identified as significant, under the values
sion variables. Moreover, the significance of the proposed model, as of the corresponding coefficients (p < 0.001) by the undertaken t-
concerns the estimation of the dependent variable (i.e. Tmb (PV+PCM31)), tests.
is illustrated by the F-ratio value (<0.0001). To be explained the vari­ • In reliance on t-values, the regression variables are ranked as follows:
ables of the linear regression model, the main observations can be out­ GT, Tα and VW.
lined as follows:
In a similar line with the analysis carried out for Tmb (PV+PCM31)
• Tmb (PV+PCM31) raises with rising GT, Tα and VW, because of the estimation under the diverse weather condition, the details related to
regression coefficients positive sign. The observed relation among the statical investigations for the Tmb (PV+PCM27) and Tmb (PV reference)

1297
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Fig. 9. Simulation results and experimental data for the PV-ref, PV + PCM27 and PV + PCM31 systems during a summer day (June 25th, 2019) of high wind (>19
km h− 1).

estimation models summarized in Table 5 and Table 6, respectively. with the corresponding real-field results observed for the experimental
setup concurrently. As shown in Fig. 8(a–b), the simulation outcomes for
4.4.1. Empirical models validation the PV reference, PV + PCM27 and PV + PCM31 systems during two
Following the statistical analysis section, the developed empirical typical summer days of weak wind (i.e. 0–19 km h− 1) exhibited satis­
equations validated by applying experimental data as illustrated in factory match with the experimental results. In the same lines, the
Figs. 8 and 9. More precisely, the available data of three sunny days in measured and modelled results for the PV reference, PV + PCM27 and
Chania utilized to simulate the operating temperature curves of the PV + PCM31 systems during a summer day of high wind (>19 km h− 1)
investigated systems (i.e. PV + PCM27, PV + PCM31 and PV reference). presented limited deviation as can be seen in Fig. 9(a–b). Mainly, the
The prevailing weather conditions (i.e. GT, Tα and VW) were compliant observed errors appeared in the morning and afternoon times when the
to the set constraints. In the sequel, the simulation results compared effect of the irradiance rapid changes took place. The mean typical

1298
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

errors linked to the estimation of Tmb (reference case), Tmb (PV+PCM27) and Tmb References
(PV+PCM31 are calculated as 0.4 C, − 0.3 C and 0.5 C, respectively,
◦ ◦ ◦

while the corresponding maximum errors equalled to 5.5 ◦ C, 5.1 ◦ C and Ali, H.M., 2020. Recent advancements in PV cooling and efficiency enhancement
integrating phase change materials-based systems – a comprehensive review. Sol.
5.2 ◦ C recorded at afternoon times. Energy 197, 163–198.
Atkin, P., Farid, M.M., 2015. Improving the efficiency of photovoltaic cells using PCM
5. Conclusions and recommendations infused graphite and aluminium fins. Sol. Energy 114, 217–228.
Browne, M.C., Norton, B., McCormack, S.J., 2016. Heat retention of a photovoltaic/
thermal collector with PCM. Sol. Energy 133, 533–548.
In this research, the capabilities of PCM inclusion to alleviate the Chander, S., Purohit, A., Sharma, A., Nehra, A. S.P., M.S. Dhaka, M.S., 2015. A study on
impact of high temperatures on the PV module’s efficiency were photovoltaic parameters of mono-crystalline silicon solar cell with cell temperature,
Energy Reports, vol. 1, pp. 104–109.
examined. To this aim, two alternative PCM enclosures were devised Charles Lawrence Kamuyu, W., Lim, J.R., Won, C.S., Ahn, H.K., 2018. Prediction model
and fabricated to encapsulate the selected sorts of paraffin RT27 and of photovoltaic module temperature for power performance of floating PVs. Energies
RT31. The formed PV + PCM systems, as well as a conventional PV 11 (2), 447.
Groulx, D., Biwole, P.H., Bhouri, M., 2020. Phase change heat transfer in a rectangular
module as a reference, were utilised to perform real-field experiments in
enclosure as a function of inclination and fin placement. Int. J. Therm. Sci. 151,
southern Greece. For the performance assessment of the testing PV and 106260.
PV + PCM systems, a set of relevant indicators was calculated using data Hachem, F., Abdulhay, B., Ramadan, M., El Hage, H., El Rab, M.G., Khaled, M., 2017.
logged during the experimental procedure (February 2019 - July 2019). Improving the performance of photovoltaic cells using pure and combined phase
change materials – experiments and transient energy balance. Renew. Energy 107,
Based on the undertaken analysis of the results from the testing systems, 567–575.
the main conclusions derived are summarised as follows: Hansen, C.W., Stein, J.S. and Riley, D., 2012. Effect of time scale on analysis of PV system
performance. SANDIA Report.
Hasan, A., McCormack, S.J., Huang, M.J., Norton, B., 2010. Evaluation of phase change
• The peak temperature reduction against the reference that achieved materials for thermal regulation enhancement of building integrated photovoltaics.
by the inclusion of the PCM27 and PCM31 was 6.4 ◦ C and 7.5 ◦ C Sol. Energy 84 (9), 1601–1612.
(during a spring day), respectively. Although these values considered Hasan, A., McCormack, S.J., Huang, M.J., Sarwar, J., Norton, B., 2015. Increased
photovoltaic performance through temperature regulation by phase change
as moderate, the effective PCM mass coefficient was equal to 0.34 materials: materials comparison in different climates. Sol. Energy 115, 264–276.
kg/m2 oC. Hasan, A., Sarwar, J., Alnoman, H., Abdelbaqi, S., 2017. Yearly energy performance of a
• The design of PCM enclosures in tubular form gained enhanced free photovoltaic-phase change material (PV-PCM) system in hot climate. Sol. Energy
146, 417–429.
convection coefficient, especially in times of high wind velocities, as Huang, M.J., Eames, P.C., Norton, B., 2004. Thermal regulation of building-integrated
demonstrated by the observed temperature profiles of the PV + photovoltaics using phase change materials. Int. J. Heat Mass Transf. 47 (12–13),
PCM27 and PV + PCM31 in these periods. 2715–2733.
Huang, M.J., Eames, P.C., Norton, B., 2006. Phase change materials for limiting
• PV + PCM 27 and PV + PCM 31 systems exhibited an improved daily
temperature rise in building integrated photovoltaics. Sol. Energy 80 (9),
energy production by 4.19% and 4.24%, respectively, while the 1121–1130.
increment in the PV conversion efficiency by PCM incorporation was Huang, M.J., 2011. The effect of using two PCMs on the thermal regulation performance
in the range of 2.86–4.19%. of BIPV systems. Sol. Energy Mater. Sol. Cells 95 (3), 957–963.
Japs, E., Sonnenrein, G., Krauter, S., Vrabec, J., 2016. Experimental study of phase
• Taking into consideration that 0.34 kg/m2 of PCM mass required to change materials for photovoltaic modules: energy performance and economic yield
decrease 1 ◦ C of PV temperature, further research adjusted on the for the EPEX spot market. Sol. Energy 140, 51–59.
developed PV + PCM concept will take place, by optimizing the Kalogirou, S.A., 2009. Solar Energy Engineering: Processes and Systems. Elsevier/
Academic Press.
quantity the used PCMs, varying the type of PCMs and life cost Kamkari, B., Groulx, D., 2018. Experimental investigation of melting behaviour of phase
analysis implementation. change material in finned rectangular enclosures under different inclination angles.
• The development of separate empirical models to estimate the Tmb Exp. Therm Fluid Sci. 97, 94–108.
Khanna, S., Reddy, K.S., Mallick, T.K., 2018. Optimization of finned solar photovoltaic
(reference case), Tmb (PV+PCM27) and Tmb (PV+PCM31) values under outdoor phase change material (finned pv pcm) system. Int. J. Therm. Sci. 130, 313–322.
conditions was accomplished. According to the obtained results, Khanna, S., Newar, S., Sharma, V., Reddy, K.S., Mallick, T.K., 2019. Optimization of fins
more detailed investigation to correlate Tmb and VW should take fitted phase change material equipped solar photovoltaic under various working
circumstances. Energy Convers. Manage. 180, 1185–1195.
place, so that the synergistic effect of high-wind and PCM inclusion Lagouvardos, K., Kotroni, V., Bezes, A., Koletsis, I., Kopania, T., Lykoudis, S.,
on PV cooling to be clarified. Mazarakis, N., Papagiannaki, K., Vougioukas, S., 2017. The automatic weather
stations NOANN network of the National Observatory of Athens: operation and
database. Geosci. Data J. 4 (1), 4–16.
Declaration of Competing Interest
Li, Z., Ma, T., Zhao, J., Song, A., Cheng, Y., 2019. Experimental study and performance
analysis on solar photovoltaic panel integrated with phase change material. Energy
The authors declare that they have no known competing financial 178, 471–486.
interests or personal relationships that could have appeared to influence Lobera, Diego Torres, Valkealahti, Seppo, 2013. Dynamic thermal model of solar PV
systems under varying climatic conditions. Sol. Energy 93, 183–194.
the work reported in this paper. Malvi, C.S., Dixon-Hardy, D.W., Crook, R., 2011. Energy balance model of combined
photovoltaic solar-thermal system incorporating phase change material. Sol. Energy
Acknowledgements 85 (7), 1440–1446.
Mavromatakis, F., Kavoussanaki, E., Vignola, F., Franghiadakis, Y., 2014. Measuring and
estimating the temperature of photovoltaic modules. Sol. Energy 110, 656–666.
The authors would express their appreciation to Dr Kostas Lagou­ Nada, S.A., El-Nagar, D.H., Hussein, H.M.S., 2018. Improving the thermal regulation and
vardos (National Observatory of Athens) for the provision of site mete­ efficiency enhancement of PCM-Integrated PV modules using nano particles. Energy
Convers. Manage. 166, 735–743.
orological data, Prof. Fotis Mavromatakis (Hellenic Mediterranean Preet, S., Bhushan, B., Mahajan, T., 2017. Experimental investigation of water based
University) for his support in the I-V tracer development and Rubitherm photovoltaic/thermal (PV/T) system with and without phase change material
GmbH for their offer of PCM RT31. (PCM). Sol. Energy 155, 1104–1120.
Radziemska, E., 2003. The effect of temperature on the power drop in crystalline silicon
solar cells. Renew. Energy 28 (1), 1–12.
Appendix A. Supplementary material Savvakis, N., Tsoutsos, T., 2015. Perfomance assessment of a thin film photovoltaic
system under actual Mediterranean climate conditions in the island of Crete. Energy
90 (Part 2), 1435–1455.
Supplementary data to this article can be found online at https://doi.
Shastry, D.M.C., Arunachala, U.C., 2020. Thermal management of photovoltaic module
org/10.1016/j.solener.2020.10.053. with metal matrix embedded PCM. J. Storage Mater. 28, 101312.
Siddiqui, M.U., Arif, A.F., Kelley, L., Dubowsky, S., 2012. Three-dimensional thermal
modeling of a photovoltaic module under varying conditions. Sol. Energy 86 (9),
2620–2631.

1299
N. Savvakis et al. Solar Energy 211 (2020) 1283–1300

Skoplaki, E., Boudouvis, A.G., Palyvos, J.A., 2008. A simple correlation for the operating Waqas, A., Ji, J., Bahadar, A., Xu, L., Zeshan, Modjinou, M., 2019. Thermal management
temperature of photovoltaic modules of arbitrary mounting. Sol. Energy Mater. Sol. of conventional photovoltaic module using phase change materials—an
Cells 92 (11), 1393–1402. experimental investigation. Energy Explorat. Exploit. 37 (5), 1516–1540.
Tsitoura, M., Michailidou, M., Tsoutsos, T., 2016. Achieving sustainability through the Waqas, A., Ji, J., Xu, L., Ali, M., Alvi, J., 2018. Thermal and electrical management of
management of microclimate parameters in Mediterranean urban environments photovoltaic panels using phase change materials – a review. Renew. Sustain. Energy
during summer. Sustain. Cities Soc. 26, 48–64. Rev. 92, 254–271.
Waqas, A., Jie, J., Xu, L., 2017. Thermal behavior of a PV panel integrated with PCM- Wongwuttanasatian, T., Sarikarin, T., Suksri, A., 2020. Performance enhancement of a
filled metallic tubes: an experimental study. J. Renew. Sustain. Energy 9 (5), photovoltaic module by passive cooling using phase change material in a finned
053504. container heat sink. Sol. Energy 195, 47–53.

1300

You might also like