Observed degradation in photovoltaic plants affected
by hot-spots
Miguel Garcia , Luis Marroyo , Eduardo Lorenzo , Javier Marcos and Miguel Pérez
ABSTRACT
A number of findings have shown that the test procedures currently available to determine the reliability and durability of
photovoltaic (PV) modules are insufficient to detect certain problems. To improve these procedures, ongoing research into
the actual performance of the modules in the field is required. However, scientific literature contains but few references to
field studies of defective modules. This article studies two different localized heating phenomena affecting the PV modules
of two large-scale PV plants in Spain. The first problem relates to weak solder joints whilst the second is due to micro-
cracks on the module cells. For both cases, the cause is identified, and consideration is given with regard to the effect
on performance, the potential deterioration over time, and a way to detect the problems identified. The findings contained
in this paper will prove to be of considerable interest to maintenance personnel at large-scale PV plants and also to those
responsible for setting module quality standards and specifications, and even the PV module manufacturers themselves.
KEYWORDS
not-spots; hot cells; localized heating; reliability; quality control; commercial PV plants
1.
INTRODUCTION
The term "hot-spot" is widely used to refer to any localized
heating in the photovoltaic (PV) modules. This is the rea-
son that lead the authors to use this term in the title. Nev-
ertheless, it should be clarified that hot-spot normally
describes an overheating caused by cell defects (junction
breakdown). Therefore, is more correct to refer the prob-
lems described in this paper to as localized heating
phenomena.
Localized heatings are the common symptom of many
solar cell defects, such as shading, cracking, weak solder-
ing, polarization, etc. Given the fact that thermographic
cameras are now relatively inexpensive, thermal imaging
has become a very common practice when analyzing the
condition of PV arrays. Although theoretical analyses of
localized heating phenomena are to be found in the litera-
ture from early PV times onwards
[1-6],
there is still a lack
of widely accepted procedures for dealing with those
problems in commercial applications [7]. Significantly,
EEC 61215 [8], the most widely applied standard on PV
module reliability, only considers hot-spot phenomena
caused by partial shading, and whilst IEC 61646: 2009
[9] and some authors [10] mention the usefulness of ther-
mal imaging analyses, no rules are provided for accepting
or rejecting the modules affected. Not surprisingly, this
failure to regulate this phenomenon can lead to controver-
sies between buyers and sellers when localized heatings
are detected in PV arrays. To a certain extent, this is due
to the lack of a clear, empirically supported relationship
between overheating magnitudes, such as temperature or
number and their impact on PV generator performance in
terms of efficiency and reliability.
This paper aims to help on that by providing consistent
experimental observations on large-scale commercial
PV plants affected by two different kinds of localized
heating phenomena caused either by soldering defects
or micro-cracks.
2.
LOCALIZED HEATING CAUSED
BY SOLDERING DEFECTS
The PV plant under study has a maximum PV power output
of 1.8 MWp and comprises 300 6kWp units, each mounted
on a single axis azimuth tracker and coupled to the grid
through a
5
kW inverter. Hereinafter, the group of modules
associated with the same inverter shall be referred to as a
"PV generator" or simply a "generator". Any anomaly af-
fecting the energy production from these generators is easily
detected simply by comparing the 300 values for monthly
energy production given in the billing.
2.1.
Problem detection
This PV plant has been in routine operation since 2006.
During the first year of operation, no significant perfor-
mance anomalies were detected. However, during the sec-
ond year (2008), a slight drop in energy production was
observed for a number of PV generators as a consequence
of a decrease in their standard test condition (STC) power.
The histograms for the generator STC power given in
Figure 1 clearly show a drop from 2007 to 2010. To
calculate the PV generator STC power, the current-voltage
(I-V) curve of some of the generators was measured using
an I-V tracer (as described in [11]). These measurements
served to calculate the STC power of the remaining gener-
ators,
based on the methodology described in [12].
During the first year of operation, all the PV generators
had an STC power of more than 90% of the nominal value.
However, from 2008 onwards, a number of generators had
power outputs of less than 90% of the nominal value. In
2009,
the drop in production (i.e., in the STC power out-
put) for the aforementioned generators became more pro-
nounced and other cases also appeared. At the same time,
a significant number of blackened points in the TEDLAR
were observed on these PV generators, clearly due to a lo-
calized heating phenomenon. The presence of those black-
ened points appeared to be directly related to this lower
power output (Figure 2).
Figure 1. PV generator STC power histograms for the plant
from 2007 to 2010. The STC power is expressed as a deviation
from its nominal value.
Figure 2. Thermal image of the defective generator, April 2009.
Contrary to former findings [6], the blackened points in
this case were located over the two busbar tabs inside the
cells (Figure 3). In some of these points, the TEDLAR
backsheets were punctured, representing a risk of insula-
tion failure (Figure 3.b).
Given the fact that the blackened points were randomly
distributed on the affected generators, there is reason to think
that there is no causal relation between the appearance of the
localized heating phenomena and the PV generator shading.
During an initial visual inspection of the modules, it
was only possible to detect those localized heatings that
had actually damaged the TEDLAR backsheet. However,
a subsequent inspection of the plant using a thermal imag-
ining camera made it possible to establish that there were
some localized heatings that had reached temperatures
close to 150 °C and also that there were a large number
of localized heatings that had not yet damaged the
TEDLAR. Figure 4 shows one of those localized heating
not visible to the naked eye.
It was also observed how some of the blackened points
in the TEDLAR, were no longer hot, as shown in Figure 5.
In such cases, the corresponding bypass diode was fre-
quently conducting.
In 2010, the problem grew to become more serious,
with more than 17% of the generators with STC power out-
puts of below 90% of their nominal value. However, the
module manufacturer, aware of the problem, replaced the
affected modules. This explains the increase in the number
of generators with an STC power output of over
95%
of the
nominal value in 2010, as shown in Figure 1.
Table I shows the mean value for the STC power devi-
ations from the nominal value for the PV generators and
the standard deviation for each year.
The aforementioned table shows how the mean plant
STC power output remained approximately 6% below its
nominal value during the first year (2007) and how this
dif-
ference slowly increased with time. The generator power
dispersion (i.e., the standard deviation) also showed a
marked increase, particularly for 2009 and 2010.
2.2.
Cause of the phenomenon (lab tests)
To determine the causes of the decreased power output of
those generators with localized heatings, a series of tests
Figure 3. Localized heating effects on the modules of the PV plant. The overheating were produced on the two busbar tabs inside
each
cell:
(a) Blackened points in the TEDLAR and (b) Punctured TEDLAR.
Figure 4. (a) Image of a module with a localized heating; (b) Thermal image of the module showing how it has a temperature over
85 °C. The presence of the overheating is not yet visible to the naked eye.
Figure 5. (a) Image of the blackened point in theTEDLAR of one of the modules caused by
a
localized heating. The heat had punctured
the TEDLAR. (b) Thermal image of the module in which the overheating does not show up.
Table I. Mean value and standard deviation of the STC power
difference from its nominal value from 2007 to 2010.
2007 2008 2009 2010
Mean (%)
STD deviation (%)
6
1.25
6.2
1.61
6.9
2.22
7.6
3.1
were run on the affected modules. Firstly, the I-V curve of
some of the modules with blackened points was measured.
The measurements were taken with capacitive load, whilst
also measuring the radiation and cell temperature with a
calibrated module [11]. Figure 6 shows the I-V curves
for 4 of the 36 modules of a generator (Figure 6a) and
power-voltage curve (Figure 6b) under standard test
Figure 6 illustrates how the presence of those localized
heatings has the same effect on the I-V curve as a resistor
connected in series with the module. Furthermore, the im-
pact of the localized heating is not only dependent on
quantity but also on the state of degradation and on
whether the localized heatings are on the same block or
on different blocks. For instance, although module 18
shows fewer blackend points in the TEDLAR than module
8, it actually has a greater effect on the pressure-volume
curve.
As a simple test to check the condition of the soldering
in those cells affected by localized heatings, the busbar
tabs were pulled out (once the TEDLAR and EVA layers
had been removed). Figure 7 shows the analysis of the im-
print left by the tabs: the only electrical contact between
the tabs and the metal surface covering the back of the cells
occurred at the solder
joints.
In these modules, there were
four solder joints per tab. For the remaining points, the
space between the tab and the conducting surface were
filled by an encapsulant (electrical insulator). The solder
joints for one of the two tabs pulled off (the upper tab)
had a very small surface area, indicating a weak weld.
Interestingly, the overheating did not appear on the
weakest solder joints but on those tabs that were better sol-
dered (lower tab on Figure 7). To explain this fact, it
should be borne in mind that the two collecting tabs are
the only path for the current generated by the cell. If both
tab solderings have the same electrical resistance, then
the current will be divided equally between them. How-
ever, if one of these tabs has an abnormally high electrical
resistance, then the current would tend to flow through the
path with the lowest electrical resistance. The resistance of
defective solder joints may be high enough to force almost
all the current to pass through the strongest solder joints.
This concentration of the current in small areas is responsi-
ble for the temperature increase in these areas and is there-
fore the cause of the hot-spots observed. This cause is
different from the one found in other plants also experienc-
ing soldering problems [6].
Figure 7. Imprints left by the busbar tab on the back of the cells after being pulled off by hand.
20 30
Voltage (V)
• module 17 (0 blackened points) -
module 18 (4 blackened points)
• module 16
(1
blackened point)
module 8 (8 blackened points)
(a)
180
160
140
120
100
80
60
40
20
0
J¿
""X \
*» \
* \
\ \ 1
\ * \
y ^
J\ 1 j _.._.._._
._ZJ
i i '
U
\\
0 10 20 30
Voltage (V)
40
50
• module 17{0 blackened points)
module 18 (4 blackened points)
• module 16
(1
blackened point)
module 8 (8 blackened points)
(b)
Figure 6. (a) I-V curve and (b) P-V curves, in STC for four
mod-
ules of a generator severely affected by localized heatings.
conditions (STC). There was no sign of localized
heatings in one of the modules, whereas it was possible
to observe 1, 4 and 8 blackened points in the TEDLAR
respectively, in the other three.
M.
García
et
al.
Observed
degradation
in
photovoltaic plants
Weak solder joints were observed
not
only
on
cells
af-
fected
by
localized heating phenomena
but
also
on
cells
that were apparently sound
and
even
on
cells
of
modules
that
had
never been exposed
to
solar radiation.
All the
modules affected
by
localized heating phenomena were
seen
to
have been manufactured between August
and
September 2004, whilst
the
least affected modules
had
been manufactured
in
some other month. This
all
suggests
that the root cause was
a
manufacturing problem basically
occurring during those 2 months.
2.3.
Effects
on the
power output
At the
end of
2009,
a
thorough inspection was made
of all
the
PV
generators
in the
most affected areas
of
the plant,
quantifying
the
number
of
modules with blackened points
in
the
TEDLAR. Figure
8
shows
the STC
power output
for those generators
in
relation
to the
number
of
modules
with blackened points
in the
TEDLAR.
The linear fitting
of
the points gave the following rela-
tion:
P*
/
PNOMINAL
=
0.94-0.00047
•
NH-S,
where
N
H
_
S
is
the number
of
modules with blackened points
per
genera-
tor.
The
slope
of the
line indicates that
the
average
de-
crease
of the
generator power output
is 0.47% for
each
module with blackened points
in the
TEDLAR. Each
PV
generator comprises 216 blocks, where
a
“block”
is a
mod-
ule sub-string
of
cells associated with
a
single bypass
di-
ode.
Therefore,
one of
these blocks accounts
for 0.46%
of
the
entire generator, which approximately corresponds
to
the
decrease
in the PV
generator mean power output
per module with blackened points.
It
could therefore
be
concluded that,
in
the majority
of
cases,
the presence
of
lo-
calized heatings tends
to
force the bypass diode to conduct,
due
to
the total loss
of
electrical contact
in
the affected cell
(s),
as
verified
in the
thermographic inspection.
2.4.
Degradation over time
As mentioned
at
the start
of
this paper, the STC power out-
put decreased
as the
number
of
affected generators mark-
edly increased each year.
The
decrease
in the
generator
power output over time could be explained
by
the fact that
Figure
8.
STC power output
of the
generators from
the
most
severely
affected areas
of the
plant,
in
relation
to the
number
of
modules with blackened points
in the
TEDLAR.
Prog.
Photovolt:
Res.
Appl.
(2013)
©
2013
John
Wiley
&
Sons,
Ltd.
DOI:
10.1002/pip
laboratory inspections revealed weak solder joints
in ap-
parently sound modules
and
even
in
modules that
had
never been exposed
to
solar radiation.
The
probability
of
such modules developing localized heatings
is
fairly high,
and this appears
to
have been
the
case
for
those modules
installed
at the
plant
and
which initially showed
no
signs
of localized heating problems.
In
fact,
the
presence
of lo-
calized heatings not visible
to
the naked eye
in
the thermal
images taken
in
mid 2010 (4 years after the plant commis-
sioning) suggest that the deterioration process had still
not
stabilized
at
that time. Up till then, the number
of
modules
affected amounted to 1545, accounting
for
13%
of
the total
number
of
modules installed. Indeed,
the
following year,
new blackened points appeared
and the
power output
of
the generators affected continued
to
fall. However,
the
mass replacement
of the
defective modules
by the
manu-
facturer from that year onwards made
it
impossible
to
con-
tinue the degradation study.
3. LOCALIZED HEATING CAUSED
BY MICRO-CRACKS
Solar cell micro-cracks
are a
well-known problem
in the
PV sector [13–18]. Today,
it is not
easy
to
quantify their
influence
on the
efficiency
of PV
modules
and
generators
over their service life. Although
a
number
of
laboratory
studies have been made [16–18],
the
available literature
contains no consistent experimental data relating
to
the oc-
currence
of
this type
of
problem
in
commercial plants.
Al-
though
it is
possible
to
clearly identify micro-cracks
through electroluminescence,
the use of
thermal imaging
is,
in
many cases,
a
simpler
and
cheaper method,
and, as
discussed
in
this paper,
it
could
be
effective
in
detecting
problems
of
this type when the cracks leave part
of
the cell
completely isolated. As shown
in
earlier studies
[18],
these
particular cracks
are
precisely those that cause significant
power losses
in PV
modules.
In
fact, over
the
last
few
years,
the systematic thermal imaging
of
the PV generators
has become common practice, and
at
some plants, this has
revealed cells with
a
significantly higher temperature than
the other cells. These hot cells might be
a
sign
of a
number
of problems, including micro-cracks.
We shall now go on to discuss the specific case
of a
PV
plant with
a
peak power
of
more than 3.8 MW, comprising
more than 550, 6.8 KWp generators mounted
on a
single
azimuth axis. Each generator delivers power
to the
grid
through
a
5 kW inverter.
As in the
case
of the
aforemen-
tioned PV plant, any anomaly affecting energy production
is easily identified simply
by
comparing
the
more than
550 values
of
monthly energy production
in
the billing.
3.1.
Identification
of
the problem
The plant was commissioned in
2007.
A thermal imaging in-
spection carried out
in
2010 showed
a
large number
of
cells
with temperatures that were considerably higher than the rest