Renewable Energy 174 (2021) 262e277

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Laboratory tests for the evaluation of the degradation of a photovoltaic

plant of 2.85 MWp with different classes of PV modules*


J.A. Clavijo-Blanco *, G. Alvarez-Tey
-Gonza
, N. Saborido-Barba , J.L. Barbera
lez ,
C. García-Lo
pez , R. Jime

nez-Castan
~ eda

diz Campus Universitario de Puerto Real, 11519, Puerto Real, Ca
Department of Electric Engineering, Superior College of Engineering, University of Ca
diz,
Spain

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

Article history: This paper studies and analyzes the results of performing several laboratory tests of a sample from a PV-
Received 28 December 2020 plant with 2.85 MW of nominal power that has been in operation for 11 years and whose PV-modules are
Received in revised form from different manufacturer and classes. The main purpose is to develop a proper quality inspection that
31 March 2021
allows knowing the degradation in a PV-plant with these special characteristics. The total sample has
Accepted 5 April 2021
Available online 14 April 2021
been taken in proportion to the size of each class through a proportionate allocation stratified sampling
strategy. The tests performed on each sample were detailed visual inspection, power rating, electrolu-
minescence (EL), and electrical insulation. The PV modules tested have been analyzed and evaluated
Keywords:
Electroluminescence (EL)
according to an acceptance/rejection criteria established by the laboratory which has been based on
IeV curves International Electrotechnical Commission standards, bibliographic contributions and the warranty
Visual inspection conditions of PV-modules. Considering the age of the plant and the criteria established in the power
Quality inspection rating, the 80.01% of the sample modules were in good conditions, the 10.27% degraded and the 9.72%
Annual degradation rate rejected. The results also show that the total annual average degradation rate of the PV-plant after 11
years of operation is 0.94%/year.
© 2021 Elsevier Ltd. All rights reserved.

1. Introduction For current and future large-scale PV plants, stakeholders are

interested in improving its reliability optimizing operational and
The reduction of carbon and greenhouse gas emissions to limit maintenance (O&M) cost to secure the invested capital and profit
climate change impact are global issues which have led to a change [3]. In this context, to secure the profitability and to provide
in electric energy markets worldwide. Among all renewables effective follow-up along the lifetime of a PV plant, quality in-
technologies, large-scale solar photovoltaic (PV) power plants have spections, which could be scheduled in preventive maintenance
been broadly installed around the world in recent years. According plans, gain a particular interest [4].
to Solar Power Europe in Ref. [1], the total world installed PV power The study of the degradation, defects and failures in PV plant has
capacity reached 633.7 GW by the end of 2019, resulting in a 23% taken certain significance through times due to the importance of
year-on-year grown compared to 2018 (509.3 GW). Furthermore, getting an accurate prediction of the performance of these systems
based on IRENA future projection, which is reported in Ref. [2], the [5]. Environmental conditions may have an important influence on
cumulative solar PV capacity is expected to grow sixfold by 2030, the degradation rate. Therefore, the degradation rate studies from
with a compound annual growth rate of nearly 9% up to 2050. diverse geographical locations are of great interest [6], since ageing,
performance and degradation are a phenomenon that affects the
power system output and consequently financial figures of merit
*
[7]. In Ref. [7] has been also observed that higher degradation rates
“This work was supported in part by INOMA Renovables S.L.“.

diz, Avenida de la Universidad, nº 10,
are reported when harsher climatic regions are studied. The value
* Corresponding author. Universidad de Ca
11519 Campus Universitario de Puerto Real, C
adiz, Spain. reported in Ref. [8] resulted greater than the 0.8%/year in only 3
E-mail addresses: joseantonio.clavijo@uca.es (J.A. Clavijo-Blanco), german. years of outdoor exposure in Morocco.


alvarez@uca.es (G. Alvarez-Tey), nieves.saborido@uca.es (N. Saborido-Barba), To determine the annual degradation rates of PV modules, three
joseluis.barberagonzalez@alum.uca.es (J.L. Barbera
-Gonza

lez), carmen.garcia@uca.
parameters are usually taken into account: energy efficiency,

pez), rafael.castaneda@uca.es (R. Jime
es (C. García-Lo
nez-Castan~ eda).

https://doi.org/10.1016/j.renene.2021.04.024
0960-1481/© 2021 Elsevier Ltd. All rights reserved.


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

output performance ratio and electrical power [9]. These parame- by PV modules according to their rated power in STC. No string
ters could be studied through the performance of different tests reconfiguration was done, no photocurrent testing or annual
that are usually based on visual inspection, infrared thermography degradation was taken into account. This procedure may aggravate
[10e12] and electrical monitorization, which can sense the overall the degradation of the PV-arrays due to electrical mismatches by
behaviour of the PV plant [13,14] or performing diagnoses at the the use of different classes or manufacturer [27].
panel level [15]. However, each monitorization technique used for In this work, only brands 1, 2 and 3, which represent the 98.29%,
the diagnosis, the measuring of the equipment and the quality of has been taken into consideration to the study of the degradation of
the data may lead to different results [16]. An example could be the PV plant during its years of operation. A sampling method has
observed in Ref. [17], where some authors have measured the been used to select PV panels, which has been described in the
output power of a PV plant to study the presence of dust on the methodology. A total of 360 sample modules has been analyzed.
panel surface. It has been also applied alternative methods which Since sample modules had to be disconnected by an experi-
have been performed in small plants to measure both the IV curves enced technician, each sample was previously labelled during an
of each module [18] and the electrical monitorization in different on-site visit. The electrical coding available on the layout did not
technologies [19]. allow easy identification of the modules in the plant, so, a new
PV panel laboratory tests have also been carried out to study codification focused on the physical location instead of an electrical
degradation, power losses and efficiency. In this case, test mainly code was established to label all sample modules [28].
used for failure detections, which are defined in Ref. [20], are visual To minimize the impact of disconnecting and connecting PV
inspection [21], IR thermography [22,23], indoor and outdoor po- modules in the plant, PV modules were disconnected and sent to
wer measurements using IeV curve tracers [24,25] and electrolu- our labs in 3 different packs. Until a pack of PV did not return to the
minescence (EL) [9]. To get a different evaluation of the possible PV plant the second was not sent. The transport and reloading of PV
defects found in a PV-plant, as may be the power induced degra- modules cause cell cracks [21], so, to minimize it, they were
dation (PID), results of PV modules tested in laboratories can be conveniently packed up.
contrasted with those tested outside, which may lead to improve The on-site PV visit has been performed in May 2019. Apart from
the reliability of the outcome [8]. Furthermore, several tests may be the labelled of the sample modules, a quality inspection was also
performed over several sample modules of a PV plant as has been performed. It consisted of three different tests: visual inspection, IR
done in Ref. [26], where 48 sample modules have been submitted to thermography and electrical monitoring. Tests have been based on
visual inspection, electroluminescence and IeV curves measure- the standard and technical specification published by the Interna-
ment resulting in several defects founded and a global average tional Electrotechnical Commission (IEC) 61215-1-1:2016 [29],
degradation of 9.5% during 15 years of outdoor exposure. 62446e3:2015 [30] and 61724e1:2017 [31] respectively. Results
The paper presents a quality inspection methodology through were published in Ref. [28], and according to the IR thermography
laboratory test with an acceptance and rejection criteria estab- inspection performed, less than 1% of PV modules installed in the
lished by the laboratory itself that get practical results about the power plant shown thermal defect, while, according to the elec-
degradation of a PV-plant operating with different classes of PV trical monitoring, it was noticeable that those inverters whose PV
modules. Results allowed to estimate the annual average degra- arrays contained different classes of PV modules injected slightly
dation rate of the PV-plant according to each manufacturer. less energy.
Although they all were operating in the same emplacement, the All sample modules were subjected to four tests; detailed visual
results obtained were different to one from another due to the inspection, IeV curve tracing, electroluminescence and electrical
quality of the PV modules presented by each manufacturer. isolation. Before carrying out the tests, all of them were cleaning
This article is structured as follows: the second section shows with neutral detergent and distilled water to remove surface dirt
the experimental data of the PV plant, the third section shows the deposits from the glass. The methodology employed in each test is
methodology established by the laboratory, the fourth section described below.
shows the acceptance and rejection criteria, in the fifth section is
where the results are shown and discussed, the sixth section pre- 3. Methodology
sents the annual degradation rate of the PV plant estimated by a
Gauss-Newton algorithm, and finally, in the seventh section, the 3.1. Sampling method
conclusions are bulleted.
A stratified random sampling with proportional allocation has
2. Experimental data been applied to obtain the size of the sample getting results with a
95% of confidence level. The population has been divided into
The PV plant has a rated power of 2.85 MW and was commis- several homogenous groups. The size of each stratum has been
sioned in 2008. It is located in the south of Spain and consists of a taken randomly and in proportion to the size of each group. Thus,
total of 29 inverters, of which 28 have a rated power of 100 kW and the size of the sample represents the total number of PV modules
1 has a rated power of 50 kW. The PV plant is nowadays composed tested while each stratum represents each PV module brand found
of a total of 18 classes of different PV modules of polycrystalline in the PV plant. The sampling procedures applied in this paper has
technology from 5 different manufacturers. Table 1 shows the been based on the standard published by the International Stan-
electrical features of the PV-modules which were operating in the dard Organization 2859e1:1999 [32].
PV power plant. The population considered in the probability sampling method
A PV modules inventory database, which has been provided by proposed was the 98.29% of the PV modules installed in the PV
the electric company responsible for the operation of the plant, has plant. This percentage included PV modules from brands 1, 2 and 3
been studied. According to the database provided, PV modules which have been operational since the commissioning of the PV
brands 1, 2 and 3 were initially installed in the solar plant, which plant. PV modules installed to replace defective PV modules has not
currently represent the 27.12%, 64.50%, and 6.67% of the total PV been included, which represent the 1.71% of the PV plant. As a
modules installed. PV modules brand 4 and brand 5, which repre- result, a total of 360 sample modules were tested. The number of
sent 1.42% and 0.28% respectively, were used to replace defectives sample modules needed for each brand is shown in Table 1.
PV modules of brands 1 and 2. PV modules were replaced directly Once the sample modules were chosen and labelled in the PV-
263


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Table 1
PV-modules electrical features.

PV module Brand PV module class (Wp) Vmp (V) Imp (A) Voc (V) Isc (A) Maximum Power (MPP) (W) MPP Tolerance rating (%) Number of samples

ET SOLAR 1 270 36.40 7.42 43.63 8.10 270.00 ±3 98

TYNSOLAR 220P6 2 195 24.53 7.96 29.52 8.46 195.00 ±5 3

200 24.96 8.03 29.76 8.52 200.00 ±5 10
205 29.58 6.93 35.94 7.47 209.90 ±5 33
210 29.64 7.09 35.94 7.60 214.90 ±5 30
215 29.94 7.18 36.00 7.83 219.90 ±5 30
220 30.12 7.30 36.06 7.95 224.90 ±5 72
225 30.36 7.41 36.42 8.10 229.90 ±5 49
230 30.48 7.55 36.60 8.17 234.90 ±5 8

CANADIAN SOLAR 3 210 46.60 4.51 57.90 4.94 210.00 ±5 5

220 46.90 4.69 58.40 5.10 220.00 ±5 7
230 47.50 4.84 58.80 5.25 230.00 ±5 10
235 29.80 7.90 36.90 8.46 235.00 ±5 3
240 48.10 4.99 59.30 5.40 240.00 ±5 2

ATERSA 4 210 28.99 7.38 36.02 8.03 210.00 ±5 e

215 e e e e e ±5 e
220 29.47 7.59 36.14 8.38 220.00 ±5 e

SOLARDAY PX72 5 260 34.10 7.60 44.40 8.17 260.00 ±5 e

plant, they were disconnected and carried to our labs, where were 3.4. Electroluminescence
subjected to different tests.
To analyze cell cracks in PV modules, all the sample modules
were subjected to electroluminescence (EL). During the EL test,
3.2. Visual inspection several images with high spatial resolution were taken. EL images
provided with high spatial resolution allow detecting even the
The detailed visual inspection has been performed on all the most insignificant PV cell cracks. This test was performed in an
sample modules to find observables defects. It has been based on indoor laboratory by the EL CAM XENICS SWIR camera Bobcat 320.
the standard IEC 61215-1-1:2016 [29], whose images have been EL images were taken based on the technical specification pub-
also studied according to Ref. [21]. Visual inspection was performed lished by the IEC 60904e13:2018 [35], where, each sample has been
exteriorly and always with an illuminance greater than 1000 lux. A polarized in a dark environment through the injection of direct
picture of each PV module was taken while a checklist was used to current (DC). Polarized PV cells emit radiation that is captured by an
write down all visible failures detected. electroluminescence camera [21].
The control of the operation of the EL camera was made by a
specific software which was included in the CAM equipment. This
3.3. Electrical inspection: power rating software allowed selecting the optimal filter, as well as the modi-
fication of settings, among other functions. A filter is used to reduce
To be able to evaluate the total power loss and the yearly the interference light from other sources.
degradation of the sample modules, the current-voltage (IeV) For each sample, it has been obtained three EL images. The first
curve of each one were measured with the IeV tracer PVPM 6020C EL image was taken without polarizing the sample. The second EL
under field conditions. It was measured based on the proceeding image was taken polarizing the sample with a current similar to
published by IEC 61215e2:2016 [33], where, to get good results, the 10% of the PV module short-circuit current in STC. And finally, the
minimum required irradiance level for the measure was set to third EL image was obtained polarizing the sample with a current
700 W/m2 over PV modules surface. Measurements were con- similar to 100% of the PV module short-circuit current in STC. This
ducted between 12:00 and 2 p.m. local time. procedure allows getting the size of the cracks in different oper-
Once taken, IeV curves were then extrapolated to standard test ating condition. Fig. 1 shows an example of different PV modules
conditions (STC) based on procedure 1 described in the IEC subjected to both currents. It can be seen that results can vary
60891:2009 [34]. The translation parameters a and b were supplied significantly from one to another, being the size of the cracks more
by the PV modules datasheets while the incident irradiance and the marked when the sample is polarized with a current similar to
operating cell temperature was measured using a calibrated PV cell 100% Isc.
included in the IeV tracer.
The PV panel total degradation (RT) was obtained using DPmax as
shown below: 3.5. Electrical insulation

DPmax Pmax;0  Pmax;STC;i To evaluate the electrical insulation of the sample modules, the
RT ð%Þ ¼ ¼ (1)
Pmax;0 Pmax;0 insulation resistance has been measured through the FLUKE 1550B.
The test has been made based on the standard IEC 61215e2:2016
Where. [33], which mainly consist of two stages:

 Pmax;0 ; PV panel maximum peak power according to the data-  Stage 1: The PV module is subjected to a ramp voltage of 500 V/s
sheet in STC conditions. up to 3000 V, maintaining such voltage for 1 min.
 Pmax;STC;i ; PV panel maximum peak power tested and extrapo-  Stage 2: The PV module is subjected to a ramp voltage of 500 V/s
lated to STC in the year i. up to 1000 V, maintaining such voltage for 2 min.
264


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 1. Electroluminescence (EL). a) 100% Isc, b) 100% Isc (Zoom), c) 10% Isc.

The final result is taken from stage 2. warranty (WADRPmax) is determined as:

4. Acceptance/rejection criteria WPloss

WADRPmax ¼ (2)
wy
The PV modules tested have been analyzed and evaluated ac-
cording to an acceptance/rejection criteria established by the lab- Where WPloss is the power loss in % and wy the years of modules
oratory which has been based on the level of degradation presented warranty. The manufacturers guarantee a maximum PV module
by each module. Considering the age of the plant, the location and power loss during a certain time, being usually the 20% of the rated
technology used, the criteria establish fit those modules with less power during 20e25 years [20]. Considering the age of the PV plant
than 15% of degradation, degraded with a degradation between 15% when tests were performed, the WADRPmax ¼ 1%.
and 20% and unfit those modules with a degradation greater than Once IeV tracer test was performed and its results extrapolated
20%. to STC, the RT was obtained in each sample. The acceptance/rejec-
The remaining tests that were carried out intend to expand in- tion criteria were established by the laboratory according to the
formation on the origin and nature of the defects/failures, consid- exposure time of the modules, the location of the PV plant, the
ering reference values to evaluate them. warranty condition, the manufacturer features, the references cited
and its own experience in the field. Thus, the results were then
4.1. Visual inspection clustered in three levels of degradation:

According to Ref. [21], defects may be the reason for some PV  Fit modules for those that presented a power degradation lower
modules no performing as well as possible while module failures than 15%. This level of degradation has been established taking
are no-reversed defects that produce power loss or safety issues. So, into consideration the WADRPmax , the age of the PV plant when
as rejection criteria, it has been considered as non-acceptable those tests were performed and the most unfavorable manufacturer
failures who produce a considerable power loss or safety issues. tolerance shown in Table 1.
Although this has been discussed in results and summarized in  Degraded modules for those that presented a degradation be-
Table 2, those sample modules with the front glass broken or tween 15% and 20%. This level of degradation tries to cluster
affected by burned marks in PV cells have been rejected. It is those sample modules that had a considerable degradation but
because the sample modules which suffer from these failures it did not get over the threshold of 20% of power loss.
shown greater power loss, safety issues or both at the same time.  Unfit modules for those that presented a degradation greater
For example, as it has been also reported in Ref. [36], PV modules than 20%.
with broken front glass counted as a severe/strong damage which
has a direct impact on the yield of the PV-plant apart from being a Finally, as acceptance/rejection criteria, those sample modules
hazard safety issue. with a power loss greater than 20% were labelled as non-acceptable
since it fell below the minimum acceptable level established by the
4.2. Electrical inspection: power rating laboratory. So, those sample modules with a power loss lower than
20% were labelled as acceptable, evaluating as degraded those
According to Ref. [6], where nearly 2000 degradation rates have whose power loss were closer to the warranty threshold.
been assembled, the majority of the data reported a degradation Once all the sample modules were evaluated according to the
rate of <1%/year, whose mean and median values are 0.8%/year and acceptance/rejection criteria established, there was a need for more
0.5%/year respectively. However, the degradation also needs to be deep analysis and try to obtain the annual degradation of the PV
interpreted in the context of the useful life of a solar panel and the panels installed in the PV power plant. Therefore, the annual
manufacturer warranty, being both commonly established in 25 degradation of each sample has been obtained.
years in Refs. [6,37]. During its useful life, PV modules are expected In this paper, the module degradation has been assumed as
to not get a power loss greater than the 20% [7]. So, with exception linear during the exposure time. Ref. [6] reported that the degra-
of early exposure and end-of-life stages, if the module degradation dation appeared to be linear for 22 years of exposure, however, it
is considered linear [7,38], the average annual degradation rate of appeared to increase after 30 years of exposure. A linear degrada-
the module’s maximum power according to the manufacturer tion through time has been already considered in similar studies,
265


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Table 2
Analysis of visual defects.

Defects founds Power Power loss over time Safety Occurrence Interpretation of results Final
loss Ref. [21] Ref. [21] decision
range

Broken front glass 48.36 Loss of mechanical integrity. The operation of the module Fire, electrical shock 1.94% Great power loss and Non-
e80.66% would be impaired. Power loss below the detection limit. and physical danger. several safety problems. acceptable
Back-sheet 9.77 Power loss degradation saturates over time. Electrical shock. 1.39% Wide range of power loss. Acceptable
bubbles e17.33% Degradation in steps over time.
Delamination 3.13 Power loss degradation saturates over time. Electrical shock. 1.39% Wide range of power loss. Acceptable
e48.36% Degradation in steps over time.
Anti-reflective 0- Lower short-circuit current, likely caused by loss of This defect has no 12.78% Wide range of power loss. Acceptable
layer 43,34% transparency. effect on safety.
discoloration Linear-shaped power loss degradation over time
Degradation in steps over time.
Yellowing 0 Lower short-circuit current, likely caused by loss of This defect has no 22.22% Wide range of power loss. Acceptable
e21.08% transparency. effect on safety.
Linear-shaped power loss degradation over time
Degradation in steps over time.
Slightly damaged 3,13 Linear-shaped power loss degradation over time. This defect has no 1.67% Wide range of power loss. Acceptable
front glass e12,43% Degradation in steps over time. effect on safety.
Browning 1,13% Linear-shaped power loss degradation over time. This defect has no 31.11% Wide range of power loss. Acceptable
e55.31% effect on safety.
Back-sheet 3.68 e It may affect the 13.06% Wide range of power loss. Acceptable
scratches e43,82% insulation properties.
Burn marks 13.06 Power loss degradation saturates over time. Fire, electrical shock 0.56% Considerable power loss Non-
e22.15% Degradation in steps over time. and physical danger. and several safety acceptable
problems.
PV cell browned 8.74 Power loss degradation saturates over time. Fire. 1.11% Wide range of power loss. Acceptable
more than e30.65%
others
Visible cracked 10.17 Power loss degradation saturates over time. Defect has no effect 5.00% Wide range of power loss. Acceptable
cells e80.66% Degradation in steps over time. on safety.
Snail tracks 6.97 Linear-shaped power loss degradation over time. Fire. 2,22% Wide range of power loss. Acceptable
e17.92%

such as [7,38]. So, the annual degradation rate of each sample module which can lead to a significant power loss. Each cell crack
(ADRPmax) was determined as: type is shown in Fig. 2.
Depending on the pattern of the cracks observed in an EL image,
WPloss the PV cells can be affected by inactive areas, which will no longer
ADRPmax ¼ (3)
y contribute to the power output of the PV module. In this context
and according to Ref. [39], when an inactive part of a PV cell is
Where WPloss is the power loss of the PV modules in the maximum greater than 8% of the total cell area, it results in a significant power
power point in STC and y the exposure time. loss which increases linearly with the inactive cell area. Since cell
cracks class A and B are not involved with a considerable power
4.3. Electroluminescence loss, as acceptance and rejection criteria, those sample modules
whose inactive area was greater than 8% due to cracks class C were
To evaluate the EL images, three types of cracks were established considered as non-acceptable.
based on the technical specification IEC 60904e13 [35], where
three cell crack types (A, B and C) are distinguished according to the 4.4. Electrical insulation
power loss causes on the sample. A-cracks do not cause significant
cell power loss, B-cracks do not imply a significant power loss and The acceptance/rejection criteria of this test have been based on
C-cracks delimit electrically disconnected areas from the PV IEC 61215e2:2016 [33], where those sample modules with an area

Fig. 2. Types of cracks in several sample modules. a) A-cracks, b) B-cracks, c) C-cracks.

266


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

greater than 0.1 m2, the product of the insulation resistance i), and had always a considerable power loss associated. Addition-
measured in stage 2 of the test procedure and the total area of the ally, all the sample modules affected by these two defects may also
PV module must not be lower than 40 MU/m2. cause important safety problems according to Ref. [21]. The analysis
of each visual defect found is summarized in Table 2. For this
5. Results reason, broken front glass and burn marks in PV cells were cate-
gorized as a failure and considered as non-acceptable under the
The results obtained through the different test done in labs are visual inspection test.
developed in this section. The results are shown in an orderly It has been assumed that the impact of each defect on the power
manner according to the followed methodology; detailed visual loss is independent. However, it may be possible that some defect
inspection, power rating, electroluminescence and electrical combined had a stronger impact on the power loss. Fig. 5 shows
insulation. defectives PV modules and their associated power loss. In this
figure, it can be seen that the maximum number of defects found in
5.1. Detailed visual inspection a single PV module is 4.
PV modules have been grouped according to the number of total
Several defects have been found in the detailed visual inspec- defects found in each one. Fig. 6 shows the groups created and its
tion. Fig. 3 shows the different defects found for each PV module power loss, where each point represents a single PV module. It can
class, where it can be seen that the browning is the most recurrent be seen that power loss can vary widely without having any cor-
defect, with a 29.95% of the total, followed by yellowing, back sheet relation. However, there is a light growing trend in the case of PV
scratches and anti-reflective layer, with the 21.39%, 12.57% and modules affected by 4 defects.
12.03% respectively. PV modules with 100% of power loss were due to internal
Once the defects were quantified, they were clustered in two disconnection in the junction box or connectors internal failure,
groups according to the power loss associated. The first group which were not visible during this test.
contain defects which do not have a strong relationship with the
power loss and do not suppose any safety issues. In this group, 5.2. Electrical inspection: power rating
there is not any way to relate the visual defect with a severe power
loss, since it can vary from little to great value. The second group Results obtained to all the sample modules have been grouped
contain failures which had a strong relationship with a consider- according to the manufacturer and presented in Figs. 7e9. In such
able power loss or safety issues and may be detected by a simple figures, the measured peak power of each sample extrapolated to
glance. STC is shown. To be able to evaluate the results obtained, it has been
In Fig. 4 several graphics are presented where each one repre- added in these graphs the rated peak power in STC of each sample
sents the defects found. For each graphic, it is shown the rela- and several thresholds according to the degradation levels estab-
tionship between the presence of the defect and the power loss in lished in the acceptance/rejection criteria. Thresholds are defined
each sample, where it can be seen that the majority of the defects as the value of peak power proportional to the limits established by
do not have a strong relationship between its occurrence and its the laboratory in the year 11 (<15% of RT), the peak power when the
power loss associated. For example, in Fig. 4 d) there exist lots of sample will be considered as degraded (15e20% of RT) and the peak
sample modules which suffer from yellowing and had a different power when the sample is considered as rejected (>20% of RT).
power loss associated, which can vary from 0% to 100%. The same Fig. 7 shows the peak power measured in STC of sample mod-
occurs with the rest of defects except for broken front glass and ules brand 1, where the 87.76% as resulted as fit, the 5.10% as
burn marks in PV cells, which graph are shown in Fig. 4 a) and Fig. 4. degraded and the 7.14% as unfit.

Fig. 3. Detailed visual inspection results by PV module class.

267


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 4. Relationship between visual defects and power losses in the sample modules.

268


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 5. Total number of visual defects in PV modules and its power loss associated.

Fig. 6. Grouping of defectives PV modules and its power loss associated.

Fig. 8 shows the peak power measured in STC of sample mod- However, although the correlation between both parameters is
ules brand 2 from 200Wp to 230Wp classes, where the 74.47% as better than what has been obtained in Ref. [40], it has been
resulted as fit, the 13.62% as degraded and the 11.91% as unfit. considered that it remains weak (R2 ¼ 0.5151). Others author in
Fig. 9 shows the peak power measured in STC of PV modules Ref. [21] considers that this relationship between the number of cell
sample modules brand 3 from 210Wp to 240Wp, where the 100% as cracks in a PV module and its power loss is very noisy.
resulted as fit. Based on the results obtained, this PV module brand The red line drawn in Fig. 10 is the limit established by the
has the best quality compared with the other brands installed in the laboratory for the evaluation of EL test. Thus, all sample modules
PV plant. Due to the positive tolerance given by the manufacturer that got over the red line had an area greater than the 8% affected by
which is shown in Table 1, in some cases, the peak power of the cracks type C and were all considers as non-acceptable. So, a total of
sample modules obtained got over the rated peak power obtained 19 sample modules were rejected, of which 5 were modules brand
from the utility database. 1 and 14 were modules brand 2.
Prior to the laboratory tests, an IR inspection of the complete PV
5.3. Electroluminescence plant was carried out in order to evaluate the behaviour and ther-
mal impacts under real operating conditions. The thermographic
All the sample modules have been analyzed according to the inspection was carried out from the rear of the modules in accor-
level of cracking presented in the acceptance/rejection criteria. dance with the IEC 62446e3 [30] standard and considering pre-
Results are shown in Fig. 17 c). Additionally, other studies have been defined criteria for the location and configuration of the
done to evaluate the relationship between PV cell cracks type C and thermographic equipment [41]. Fig. 11 shows for each brand the
power loss. These results are shown in Fig. 10. percentage of PV modules with thermal incidents and the per-
In this figure, it can be observed that there exits some kind of centage of PV modules with hot spots or hot cells. In addition, the
relationship between the PV cell cracks type C and the power loss. value of the maximum temperature measured in each case is

269


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 7. Power rating results of PV modules brand 1.

Fig. 8. Power rating results of PV modules brand 2.

indicated. In global terms, 0.79% of PV modules in the plant have the laboratory. The sample chosen was the PV module brand 1 of
some kind of thermal incidence, while 0.73% of PV modules have 270Wp, which showed a 49.49% of power loss during the IeV tracer
hot cells or hot spots. test and some visible failures, such as broken front-glass and visual
According to Ref. [12], cracks in cells, which temperature de- cracked cell, were found in the detailed visual inspection. Both
pends on the fracture characteristic and the ambient condition, defects are shown in Fig. 12.
may strongly impact the module performance. IR and EL test have Fig. 13 shows both images, the IR image when the PV module
not commonly shown similar results when both images are was working under operating condition, and the EL image from the
compared. Since both tests have been performed in different con- sample chosen. In this case, the EL test has been performed in a
ditions, it may be caused because the contact resistance between laboratory according to the methodology presented in this paper.
the two sides of the micro-crack varies with module temperature The EL image shows several cracks in solar cells. The affected PV cell
and could be much larger during the day (when hot-spots are area of C-cracks was 40.28% of the PV module respectively, which
observed) than during the night or in a dark environment when EL meant that 59.72% of the area had non-significant power loss. The
images are obtained [40]. IR image, which was taken at the rear of the PV module, shows that
However, it has been observed that some results obtained from a total of 7 PV cells were affected by thermal abnormalities.
EL and IR images show a certain relationship between the inactive Both images were compared and it could be observed that all
area and the hot-spot. Those modules affected by both cracks and the thermal abnormalities found in the IR image were also reflected
hot cells have a greater impact on the module performance [42]. An in the EL image, which have been highlighted in a green box. Ac-
example has been presented in this paper with the result obtained cording to Ref. [43], this correlation occurs when the temperature
in the IR inspection made in-situ [28] and the EL test performed in of the cell increases 10e20  C, having the unproductive cracked

270


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 9. Power rating results of PV modules brand 3.

Fig. 10. Relationship between the PV module area damage by C-cracks and the power loss of the sample modules tested.

Fig. 11. Summary of IR inspection result according to PV module brand.

271


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 12. Visual Inspection a) Cracked cells b) Broken front glass.

Fig. 13. a) IR thermography (at the rear of the module), b) EL front-image at 100% of Isc.

Fig. 14. Electrical insulation results in sample module brands 1. Fig. 15. Electrical insulation results in PV module sample module brands 2.

272


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

only one module has resulted as non-acceptable. The PV module

insulation resistance measured in such rejected sample was 24.06
M U,m2 .

5.5. Overall result following the acceptance/rejection criteria

established

The overall results of the 4 test performed following the eval-

uation of each one and the acceptance/rejection criteria established
in section 4 are shown in Fig. 17. Results for each test are shown by
PV module class.
Fig. 17 a) shows the results of the detailed visual inspection
where the accepted ones have been colored in green while rejected
ones have been colored in red. The PV module class with the
greatest percentage of failure has been the brand 2 of 225Wp with
the 8.16% followed by brand 2 of 215Wp and 210Wp, both with the
Fig. 16. Electrical insulation results in PV module sample module brands 3.
3.33%. Clustering by manufacturer, the percentage of failures found
in brand 1, brand 2 and brand 3 would be the 4.08%, 3.83% and
area a similar behaviour of a shaded area. 0.00% respectively. Globally, only the 3.61% were rejected in the
detailed visual inspection.
Fig. 17 b) shows the results of the power rating test and the
5.4. Electrical insulation acceptance/rejection criteria, where accepted, degraded and
rejected sample modules were colored in green, orange and red
Figs. 14e16 show the results of the electrical insulation test for respectively. In this case, the PV module brand 2 of 225Wp, 215Wp
sample modules brand 1, brand 2 and brand 3 respectively. The and 220Wp, with the 20.41%, 16.67% and 11.11% respectively, were
orange line drawn in such figures is the rejected criteria reference the classes with the greatest percentage of non-acceptance. In
value established by the IEC 61215e2:2016 [33]. terms of degradation, the PV module brand 2 of 200Wp, 215Wp
From all the sample modules tested and as it is shown in Fig. 14, and 210Wp, with the 30.00%, 20.00% and 16.67% respectively, were

Fig. 17. Overall result by each test performed.

273


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

the most affected by. Clustering by manufacturer, the percentage of the sample modules are studied together. In this case, the PDF
rejected of brand 1, brand 2 and brand 3 were 7.14%, 11.91% and obtained also resulted in fitting well with a normal distribution,
0.00% respectively. If all the sample modules are grouped, the which has been obtained applying a non-linear regression fitting
80.00% of the PV modules tested were found to be in good condi- the least square using the Gauss-Newton algorithm. The average
tion, the 10.27% degraded and the 9.72% rejected. annual degradation rate resulted in 0.94%/year while its lower and
Fig. 17 c) shows the results of electroluminescence test clustered upper bounds with a c.i. of 95% according to a normal distribution
by PV module class, where the percentage of accepted and rejected were 0.89%/year and 0.99%/year respectively.
PV modules classes are colored in green and red respectively. The The variation of Pmax losses with losses in Isc, Voc and fill factor
PV module class with the greatest percentage of non-acceptance (FF) has been also studied, where Isc and FF losses were dominant
has been the brand 2 of 225Wp, with the 10.20%, followed by the while Voc had weakness correlation. According to Ref. [7], the
PV module brand 2 of 215Wp and 200Wp, both with the 10%. impact of the Isc on the power loss may be caused by the influence
Clustering by manufacturer, the percentage of rejected PV modules of some observable defects that lead to reduce the light trans-
in brand 1, brand 2 and brand 3 was 8.16%, 7.66% and 0.00% mission to the solar cells.
respectively. As a global result, the 7.22% of the sample modules
were rejected. 7. Conclusion
And finally, Fig. 17 d) shows the results of the electrical insu-
lation test, where only one sample was rejected which corre-  The probability method selected was a stratified random sam-
sponded to the brand 1 of 270Wp. pling with proportional allocation. A total of 360 PV modules
were subjected to the following tests: detailed visual inspection,
6. Annual degradation rates power rating, electroluminescence, and electrical insulation.
 An annual degradation of 0.681%, 1.113% and 0.279% were ob-
Once all the sample modules were evaluated according with the tained to brand 1, 2 and 3 respectively. Which meant that
acceptance/rejection criteria established, the ADRPmax , which eq. is despite being all of them working under the same conditions of
showed in (3), was obtained for each one. The ADRPmax of each location, operation and maintenance, some PV module brands
sample were clustered by the manufacturer to evaluate and showed worst levels of degradation after its years of operation.
compare the degradation between them. Then, the histogram and  When all brands were grouped together, a global average annual
its corresponding probability density function (PDF) has been ob- degradation of 0.9480% was obtained, being the losses in the
tained. All the PDFs obtained resulted in fitting well with a normal short-circuit current and fill factor the dominant factors.
distribution. These PDFs have been obtained applying a non-linear  The browning, yellowing, back sheet scratches and anti-
regression fitting the least square using the Gauss-Newton reflective layer were the most observed defect over the sam-
algorithm. ple modules.
Figs. 18e20 show the histogram and its corresponding PDF of PV  In global terms, the 0.79% of PV modules in the PV plant had
module brand 1, 2 and 3 respectively, which are all compared in some kind of thermal incidence and the 0.73% of PV modules
Fig. 21. Table 3 shows, the annual average degradation rate for each had hot cells or hot spots.
manufacturer and the lower and upper bounds with a confidence  According to the power rating test, which authors considered as
interval (c.i.) of 95% according to a normal distribution. the most suitable as a global result, and its acceptance/rejection
Although all the sample modules presented in this study came criteria established in the methodology, the 80.01% of the PV
from the same areas which shared the same climatic conditions, a modules tested were found to be in good condition, the 10.27%
different average annual degradation was obtained in each one. degraded and the 9.72% rejected.
Fig. 21 shows the comparison of the resulted PDF for each manu-
facturer, where it can be seen that the greater annual degradation This evaluation could be applied to other photovoltaic plants
comes from PV sample brand 2. By contrast, the PV module that with different conditions to analyze its performance.
showed the lower annual degradation was brand 3. If both average
results are compared, there is a difference between them of 0.83%/ CRediT authorship contribution statement
year, which highlight the importance of using one brand or other.
Fig. 22 shows the histogram and its corresponding PDF when all J.A. Clavijo-Blanco: Conceptualization, Formal analysis,

Fig. 18. Histogram and approximation of PDF (normal distribution, x ¼ 0:00683; sd 0:002844) of the PV module sample brand 1.

274


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 19. Histogram and approximation of PDF (normal distribution, x ¼ 0:011134; sd ¼ 0:003908) of the PV module sample brand 2.

Fig. 20. Histogram and approximation of PDF (normal distribution, x ¼ 0:002792; sd ¼ 0:004471) of the PV module sample brand 3.

Fig. 21. Comparison of PV-module brand 1, 2 and 3 PDFs.

Table 3

Investigation, Data curation, Writing e original draft. G. Alvarez-
annual degradation rate for each manufacturer PV-module. Tey: Methodology, Investigation, Resources, Data curation, Writing
lower c.i. mean upper c.i. e review & editing. N. Saborido-Barba: Investigation, Data cura-
tion, Writing e review & editing. J.L. Barbera
-Gonza
lez: Investi-
PV module sample brand 1 0.624992 0.6813 0.737608 %/year
PV module sample brand 2 1.063434 1.1134 1.163366 %/year gation, Data curation, Writing e review & editing. C. García-Lo
pez:
PV module sample brand 3 0.110553 0.2792 0.447847 %/year Investigation, Resources, Writing e review & editing. R. Jime
nez-
Castan~ eda: Validation, Resources, Supervision.

275


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

Fig. 22. Histogram and approximation of PDF (normal distribution, x ¼ 0:009480; sd ¼ 0:004658) to all the sample modules.

Declaration of competing interest Mater. Sol. Cells 107 (2012) 154e164, https://doi.org/10.1016/
j.solmat.2012.07.011.
[13] S. Blanc, P. Yuste, A. Lorente, J.J. Serrano, An integral and flexible wireless
The authors declare that they have no known competing power monitoring system, Renew. Energy Power Qual. J. 1 (9) (2011)
financial interests or personal relationships that could have 1290e1293, https://doi.org/10.24084/repqj09.625.
appeared to influence the work reported in this paper. [14] D. Thevenard, L. Dignard-Bailey, S. Martel, D. Turcotte, Performance moni-
toring of a northern 3.2 KWp grid-connected photovoltaic system, Conf. Rec.
IEEE Photovolt. Spec. Conf. 2000 (2000) 1711e1714, https://doi.org/10.1109/
PVSC.2000.916233. Janua.
Acknowledgements
[15] B. Ando, S. Baglio, A. Pistorio, G.M. Tina, C. Ventura, “ Sentinella, Smart
monitoring of photovoltaic systems at panel level, IEEE Trans. Instrum. Meas.
One of the authors would like to acknowledge the support of Mr. 64 (8) (2015) 2188e2199, https://doi.org/10.1109/TIM.2014.2386931.
Luís Clavijo García for being a source of inspiration. [16] A. Livera, M. Theristis, G. Makrides, G.E. Georghiou, Recent advances in failure
diagnosis techniques based on performance data analysis for grid-connected
photovoltaic systems, Renew. Energy 133 (2019) 126e143, https://doi.org/
References 10.1016/j.renene.2018.09.101.
[17] A. Sayyah, M.N. Horenstein, M.K. Mazumder, Energy yield loss caused by dust
deposition on photovoltaic panels, Sol. Energy 107 (2014) 576e604, https://
[1] Solar Power Europe, “Global Market Outlook for Solar Power: 2020 - 2024, doi.org/10.1016/j.solener.2014.05.030.
2020, https://doi.org/10.4018/978-1-7998-4607-9.ch006. [18] P. Sa
nchez-Friera, M. Piliougine, J. Pel
aez, J. Carretero, M.S. De Cardona,
[2] I. International Renewable Energy Agency, “Future of Solar Photovoltaic. Analysis of degradation mechanisms of crystalline silicon PV modules after 12
Deployment, Investment, Technology, Grid Integration and Socio-Economic years of operation in Southern Europe, Prog. Photovoltaics Res. Appl. 19 (6)
Aspects,” (A Global Energy Transformation: Paper), International Renewable (Sep. 2011) 658e666, https://doi.org/10.1002/pip.1083.
Energy Agency (IRENA), Abu Dhabi, 2019, p. 88 [Online]. Available, www. [19] A. Carullo, A. Vallan, Outdoor experimental laboratory for long-term estima-
irena.org/publications. tion of photovoltaic-plant performance, IEEE Trans. Instrum. Meas. 61 (5)
[3] L. Peters, R. Madlener, Economic evaluation of maintenance strategies for (May 2012) 1307e1314, https://doi.org/10.1109/TIM.2011.2180972.
ground-mounted solar photovoltaic plants, Appl. Energy 199 (2017) 264e280, [20] M.A. Munoz, M.C. Alonso-García, N. Vela, F. Chenlo, Early degradation of sil-
https://doi.org/10.1016/j.apenergy.2017.04.060. icon PV modules and guaranty conditions, Sol. Energy 85 (9) (2011)
[4] Solar Power Europe, Operation & Maintenance Best Practice Guidelines/Version 2264e2274, https://doi.org/10.1016/j.solener.2011.06.011.
4.0, 2019. [21] IEA-PVPS Task 13, Review of Failures of Photovoltaic Modules, 2014 doi: 978-
[5] M. Kumar, A. Kumar, Performance assessment and degradation analysis of 3-906042-16-9.
solar photovoltaic technologies: a review, Renew. Sustain. Energy Rev. 78 [22] Y. Hu, W. Cao, J. Ma, S.J. Finney, D. Li, Identifying PV module mismatch faults
(November 2016) (2017) 554e587, https://doi.org/10.1016/j.rser.2017.04.083. by a thermography-based temperature distribution analysis, IEEE Trans. De-
[6] D.C. Jordan, S.R. Kurtz, Photovoltaic degradation rates - an analytical review, vice Mater. Reliab. 14 (4) (2014) 951e960, https://doi.org/10.1109/
Prog. Photovoltaics Res. Appl. 21 (1) (2013) 12e29, https://doi.org/10.1002/ TDMR.2014.2348195.
pip.1182. [23] S.A. Rahaman, T. Urmee, D.A. Parlevliet, PV system defects identification using
[7] D.A. Quansah, M.S. Adaramola, Ageing and degradation in solar photovoltaic Remotely Piloted Aircraft (RPA) based infrared (IR) imaging: a review, Sol.
modules installed in northern Ghana, Sol. Energy 173 (August) (2018) Energy 206 (April) (2020) 579e595, https://doi.org/10.1016/
834e847, https://doi.org/10.1016/j.solener.2018.08.021. j.solener.2020.06.014.
[8] M.A. Islam, M. Hasanuzzaman, N.A. Rahim, A comparative investigation on in- [24] E. Caaman ~ o, E. Lorenzo, R. Zilles, Quality control of wide collections of PV
situ and laboratory standard test of the potential induced degradation of modules: lessons learned from the IES experience, Prog. Photovoltaics Res.
crystalline silicon photovoltaic modules, Renew. Energy 127 (2018) 102e113, Appl. 7 (2) (1999) 137e149, https://doi.org/10.1002/(sici)1099-159x(199903/
https://doi.org/10.1016/j.renene.2018.04.051. 04)7:2<137::aid-pip249>3.3.co;2-3.
[9] T. Ishii, A. Masuda, Annual degradation rates of recent crystalline silicon [25] E.S. Kopp, V.P. Lonij, A.E. Brooks, P.L. Hidalgo-Gonzalez, A.D. Cronin, I-V curves
photovoltaic modules, Prog. Photovoltaics Res. Appl. 25 (2017) 953e967, and visual inspection of 250 PV modules deployed over 2 years in tucson,
https://doi.org/10.1002/pip.2903. Conf. Rec. IEEE Photovolt. Spec. Conf. (2012) 3166e3171, https://doi.org/
[10] C. Schuss, K. Leppanen, J. Saarela, T. Fabritius, B. Eichberger, T. Rahkonen, 10.1109/PVSC.2012.6318251.
Detecting defects in photovoltaic modules with the help of experimental [26] J.E.F. da Fonseca, F.S. de Oliveira, C.W. Massen Prieb, A. Krenzinger, Degra-
verification and synchronized thermography, Conf. Rec. - IEEE Instrum. Meas. dation analysis of a photovoltaic generator after operating for 15 years in
Technol. Conf. 2015 (2015) 97e102, https://doi.org/10.1109/ southern Brazil, Sol. Energy 196 (November 2019) (2020) 196e206, https://
I2MTC.2015.7151247. July. doi.org/10.1016/j.solener.2019.11.086.
[11] J.A. Tsanakas, L. Ha, C. Buerhop, Faults and infrared thermographic diagnosis [27] Alessandro Massi Pavan, Adel Mellit, D. De Pieri, V. Lughi, A study on the
in operating c-Si photovoltaic modules: a review of research and future mismatch effect due to the use of different photovoltaic modules classes in
challenges, Renew. Sustain. Energy Rev. 62 (2016) 695e709, https://doi.org/ large-scale solar parks, Prog. Photovoltaics Res. Appl. 22 (2014) 332e345, doi:
10.1016/j.rser.2016.04.079. 10.1002/pip.
[12] C. Buerhop, D. Schlegel, M. Niess, C. Vodermayer, R. Weißmann, C.J. Brabec, [28]

J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba, J.L. Barber
a-Gonz
alez,
Reliability of IR-imaging of PV-plants under operating conditions, Sol. Energy

276


J.A. Clavijo-Blanco, G. Alvarez-Tey, N. Saborido-Barba et al. Renewable Energy 174 (2021) 262e277

C. García-Lo
pez, R. Jime

nez-Castan~ eda, Quality inspection of a 2.85 MW PV doi.org/10.1016/j.renene.2017.11.010.
power plant under mismatch loss due to different classes of pv module [37] M.S. Chowdhury, et al., An overview of solar photovoltaic panels’ end-of-life
installed, Renew. Energy Power Qual. J. 18 (18) (2020) 276e281, https:// material recycling, Energy Strategy Reviews (2020), https://doi.org/10.1016/
doi.org/10.24084/repqj18.298. j.esr.2019.100431.
[29] International Electrotechnical Commission, IEC 61215-1-1:2016. Terrestrial [38] C.R. Osterwald, J. Adelstein, J.A. Del Cueto, B. Kroposki, D. Trudell, T. Moriarty,
Photovoltaic (PV) Modules. Design Qualification and Type Approval. Part 1-1: Comparison of degradation rates of individual modules held at maximum
Special Requirements for Testing of Crystalline Silicon Photovoltaic (PV) power, in: Conf. Rec. 2006 IEEE 4th World Conf. Photovolt. Energy Conversion,
Modules, 2016. WCPEC-4, vol. 2, 2006, pp. 2085e2088, https://doi.org/10.1109/
[30] International Electrotechnical Commission, IEC TS 62446-3. PV Systems- WCPEC.2006.279914.
Requirements for Testing, Documentation and Maintenance. Part 3: Photo- [39] I. Kunze, S. Kajari-schr, X. Breitenmoser, B. Bjørneklett, The risk of power loss
voltaic Modules and Plants- Outdoor Infrared Thermography, vol. 7, 2015 no. in crystalline silicon based photovoltaic modules due to micro-cracks, Sol.
4. Energy Mater. Sol. Cells 95 (2011) 1131e1137, https://doi.org/10.1016/
[31] International Electrotechnical Commission, IEC 61724-1:2017. Photovoltaic j.solmat.2010.10.034.
System Performance - Part 1: Monitoring, 2018. [40] R. Moreto
n, E. Lorenzo, L. Narvarte, Experimental observations on hot-spots
[32] International Organization for Standardization, ISO 2859-1:1999 Sampling and derived acceptance/rejection criteria, Sol. Energy 118 (2015) 28e40,
Procedures for Inspection by Attributes. Part 1: Sampling Schemes Indexed by https://doi.org/10.1016/j.solener.2015.05.009.
Acceptance Quality Limit (AQL) for Lot-By-Lot Inspection, 1999, p. 94, https://

[41] G. Alvarez-Tey,
nez-Castan
R. Jime ~ eda, J. Carpio, Analysis of the configuration
doi.org/10.5594/J03363. Iso, vol. 2nd editio. and the location of thermographic equipment for the inspection in photo-
[33] International Electrotechnical Commission, IEC-61215-2. Terrestrial Photo- voltaic systems, Infrared Phys. Technol. 87 (2017) 40e46, https://doi.org/
voltaic (PV) Modules. Design Qualification and Type an Approval. Part 2: Test 10.1016/j.infrared.2017.09.022.
Procedures, 2016. [42] S. Chattopadhyay, et al., Correlating infrared thermography with electrical
[34] International Electrotechnical Commission, IEC 60891:2009. Photovoltaic degradation of PV modules inspected in all-India survey of photovoltaic
Devices - Procedures for Temperature and Irradiance Corrections to Measured module reliability 2016, IEEE J. Photovoltaics 8 (6) (2018) 1800e1808, https://
I-V Characteristics, 2009. doi.org/10.1109/JPHOTOV.2018.2859780.
[35] International Electrotechnical Commission, IEC TS 60904-13 Photovoltaic [43] M. Garcia, L. Marroyo, E. Lorenzo, J. Marcos, M. Pe
rez, “Observed degradation
Devices-Part 13: Electroluminescence of Photovoltaic Modules, 2018. in photovoltaic plants affected by hot-spots, Prog. Photovoltaics Res. Appl. 22
[36] W. Muehleisen, et al., Outdoor detection and visualization of hailstorm (2013) 1292e1301, doi: 10.1002/pip.
damages of photovoltaic plants, Renew. Energy 118 (2018) 138e145, https://

277