PhotoVoltaic Cost Reduction, Reliability, Operational performance, Prediction and Simulation
Grant Agreement no.:
Project Acronym:
Project Title:
Instrument:
Thematic Priority:
Deliverable nº.:
Deliverable Title:
Date of preparation:
Author(s):
Deliverable lead partner:
WP Leader:
Partners involved:
Dissemination level:
PROJECT
308468
PVCROPS
Photovoltaic Cost Reduction, Reliability, Operational
performance, Prediction and Simulation
Collaborative Project
FP7-ENERGY.2012.2.1.1
DELIVERABLE
9.4
Description of testing kits
31/05/2015
Nikolay Tyutyundzhiev, Francisco Martínez
CL-SENES
CL-SENES
CL-SENES, UPM, ACCIONA
PUBLIC
1
INDEX
1. INTRODUCTION.
1
2. TESTING KITS.
3
2.1 TESTING KIT 1. Reference PV modules: Global performance of the PV system. 4
2.1.1. Summary.
2.2. TESTING KIT 2. I-V curve tracers: performance of PV modules/arrays.
10
11
2.2.1. Single I-V curve tracer.
12
2.2.2. Twin I-V curve tracer.
23
2.2.3. Summary.
25
2.3. TESTING KIT 3. Climatic box: detailed performance of PV modules.
2.3.1. On-site measurements at STC.
26
26
2.3.2. On-site measurements of temperature coefficients and efficiency parameters
of the PV module.
2.3.3. Summary.
2.4. TESTING KIT 4. PID tester: Potential Induced Degradation detection.
31
34
35
2.4.1. Laboratory degradation and detection tests.
35
2.4.2. Detection tests on the field.
37
2.4.3. Summary.
44
2.5. TESTING KIT 5. Hot-spot tester: thermal inspection.
46
2.5.1. Traditional hot-spot detection.
48
2.5.2. Hot-spot detection with PV drones.
48
2.5.3. PV drone aerial photogrammetry.
53
2.5.4. Summary.
56
2.6. TESTING KIT 6. Full PV system tester: detailed performance analysis.
58
2.6.1. AC characterization.
60
2.6.2. DC characterization.
62
2.6.2.1. DC characterization integrated into SCADA.
65
2.6.3. Inverter characterization.
67
2.6.4. Portable SCADA tool (microSCADA).
69
2.6.5. Summary.
74
3. CONCLUSIONS.
77
4. REFERENCES.
79
1. INTRODUCTION.
In the last years, grid-connected PV installations (BIPV and large commercial PV
plants) have become an interesting financial product, even without subsidized feed-in
tariff. This is the reason why less than 10 years have been enough to achieve a total
installed PV power[1] above 100 GW. As for any financial product, some cares must be
taken into account in order to guarantee, first, the investment recovery and, second, a
good profitability of the PV installation during its whole lifetime. These objectives will
be achieved if the PV system finally constructed fulfils the energy production
expectations established at its initial design.
On the one hand, in order to achieve these objectives, the PV project must be properly
defined through a set of technical specifications that report about how the PV
installation must be implemented and what are the characteristics that each individual
device should have. So, if these technical specifications are met, good practices will
dominate in the system, what will enable to reduce the maintenance costs; and bad
practices that lead to mistakes, failures and a lower energy production will be avoided[2].
This way, profitability of the real PV installation will be equal than the predicted one, or
even above it, with a higher probability.
On the other hand, the fulfilment of these technical specifications must be checked
during the implementation of the PV installation, but also once it has been constructed
and during its lifetime. This can be achieved through a set of testing procedures whose
objective is making sure if the actual devices installed perform like those defined at the
initial design.
Both, the technical specifications defined to achieve the most profitable performance of
the PV system and the quality control procedures designed to analyze the fulfilment of
these technical specifications, have been already reported in the PVCROPS project and
have been validated by several experts of the PV community (Deliverable 2.2,
Deliverable 9.3). These documents try to show to the different actors of the PV industry
that the final quality of a PV system can be improved if a detailed definition of the
characteristics that should present the individual parts of a PV installation is done and if
the responsabilities in case of misoperation (constructor, PV modules manufacturer, PV
1
inverter manufacturer, maintenance staff, etc.) are clearly established before the
construction of the PV system. The source of each particular mistake can be detected
thanks to the quality control procedures specifically designed to check the individual
performance of each PV device or even the general performance of the whole PV
system. Besides, these quality control procedures can be also useful to get a more
accurate characterization of each device, which allows reducing the uncertainty and
achieving more accurate values for the yearly energy production estimations.
Now, this report shows the testing kits that have been designed in the PVCROPS project
to execute the quality control procedures proposed, what are the devices that should be
used and what are the devices that can even be implemented. Also several testing
campaigns which have been done in real PV installations using the proposed devices are
reported. In these testing campaigns, the new devices have been applied in order to
validate its usefulness. The results obtained, together with some recommendations,
demonstrate that these testing kits constitute suitable equipment to properly analyse the
behaviour of grid-connected PV systems and/or the behaviour of their main
components.
2
2. TESTING KITS.
In this chapter the different testing kits proposed by the PVCROPS project to perform
the quality control procedures previously defined are presented.
The first one is the more basic testing kit and allows evaluating the global performance
of the PV system as a black box: it reports if the output energy production injected into
the grid is the proper one, only taking into account the power installed and the
meteorological conditions at which the PV system has operated during the considered
period (effective irradiance and cell temperature).
The next 4 testing kits are intended for the study of PV modules/arrays: measurement of
the module main electrical variables, its behaviour at different operating conditions and
the detection of abnormal operation due to internal anomalies (Potential Induced
Degradation and hot-spots).
The final testing kit has the objective of obtaining a very accurate and detailed analysis
of the PV installation, from the whole PV system performance, reporting clearly about
anomalous situations in its operation, to the individual performance of PV arrays and
inverters, reporting about the fulfilment of their actual characteristics related to the ones
stated by the manufacturers.
Some of these testing kits are based on the combination of general and accessible
commercial equipment and detailed testing procedures expressly designed to properly
evaluate the performance of the main devices of a PV installation. And some other
testing kits have been specifically designed and fully assembled by the PVCROPS team
in order to customize the test related to some PV devices, getting accurate results in a
less expensive way.
3
2.1. TESTING KIT 1. Reference PV modules: Global performance of the PV
system.
Technical performance of a grid-connected PV installation is usually assessed by means
of the Performance Ratio, PR, observed during a given period of time. This index is
defined[3] in IEC 61724 and is calculated as
𝑃𝑅 =
𝐸AC,real
𝐺efΔT
∗
𝑃NOM
𝐺∗
1
where the symbol * refers to Standard Test Conditions (STC), EAC,real is the real AC
energy injected into the grid during the test period T (it can be obtained from the
energy meters as the difference between the reading at the end of the test and its initial
∗
value), 𝑃NOM
is the contractual nominal power of the PV system under study, GefΔt is the
effective global solar irradiation in the plane of the array during the test period T and
G* is the global solar irradiance at STC (G* = 1000 W/m2). So, calculating PR only
requires integrating records of effective irradiance, Gef.
This mere PR is adequate for qualifying global technical quality of a PV installation if
full year periods are considered. This is because for a given PV installation and site, the
PR value tends to be constant along the years, as much as the climatic conditions tend to
repeat. Therefore, PR value is time and site dependent. This is due to its dependence on
the operation temperature which, in turns, depends on the climatic conditions and,
therefore, on the site and on the time of the year. So, the mere PR is generally not
adequate for sub-year periods (days, weeks, months…).
As an example, Figure 1 shows the weekly PR variation (PRW) measured at a PV plant
located in Navarra (Spain) during one year. The figure shows values larger than 1
because the real peak power of the modules is slightly larger than the nominal value and
because winter is typically very sunny but very cold in the location of this PV plant. The
key point here is to observe that PR varies up to 10% along the year and even up 5%
along the same month.
4
Figure 1.
Observed evolution of the weekly PR and PRSTC during a whole year at a PV plant located
in Navarra (Spain).
When sub-year periods are considered, the PR dependence on unavoidable and timedependent losses requires corresponding correction in order to properly qualify the
technical quality of a PV plant. These losses are the ones derived from the efficiency
variation with temperature and irradiance, from intrinsic to PV design phenomena
(shades and inverter saturation) and from possible angular and spectral response
differences between the PV array and the irradiance and cell temperature sensors. A
convenient way of doing such correction is to consider the so-called Performance Ratio
at STC
PRSTC, t 
PRt
u (1  Eu )
2
where E represents energy losses during the considered period and the subscript “u”
extends to all the unavoidable energy losses phenomena. So, the PRSTC calculation
requires records of Gef, but also records of cell temperature, TC, and some modelling.
5
In our case, we are going to work with the PV performance model adopted by SISIFO, a
PV simulation software developed at PVCROPS, free available at www.sisifo.info and
whose main equation is:
∗
𝑃𝐷𝐶 (𝐺𝑒𝑓 , 𝑇𝐶 ) = 𝑃𝑁𝑂𝑀
·
𝐺𝑒𝑓
𝐺𝑒𝑓
𝐺𝑒𝑓
· [1 + 𝛾 · (𝑇C − 𝑇C∗ )][𝑎 + 𝑏 ∗ + 𝑐 · ln ∗ ]
∗
𝐺
𝐺
𝐺
3
This model describes thermal losses by means of gamma, , a value which is always
found at the manufacturer datasheets (TC* is the value of the cell temperature at STC,
TC* = 25ºC). Moreover, the three parameter a, b and c, describing the efficiency
dependence on irradiance are obtained from values corresponding at two other than G*
irradiance values, which must also be found at datasheets, providing they comply[4] with
EN 50380.
For example, thermal losses are typically the most significant at the global energetic
balance of a PV plant. In energy terms, ∆𝐸 𝑇𝐶 ≠𝑇𝐶∗ result from weighting the power
thermal losses by the incident irradiance. That is:
∆𝐸
𝑇𝐶 ≠𝑇𝐶∗
=
∫∆t 𝛾 · (𝑇C − 𝑇C∗ ) · 𝐺𝑒𝑓 · dt
∫∆t 𝐺𝑒𝑓 · dt
4
Similarly, losses derived from efficiency dependence on irradiance are:
∆𝐸𝐺𝑒𝑓 ≠𝐺∗ =
𝐺𝑒𝑓
𝐺𝑒𝑓
∫∆t [(𝑎 − 1) + 𝑏 𝐺 ∗ + 𝑐 · ln 𝐺 ∗ ] · 𝐺𝑒𝑓 · dt
5
∫∆t 𝐺𝑒𝑓 · dt
As can be seen in Figure 1, PRSTC,W is significantly more constant than PRW (in the
graph there are only two anomalous values, which are probably related with bad
weather or with problems at the temperature measurements). In our opinion, the
advantages derived from the use of the PRSTC is large enough to pay the price of, first,
recording operation temperature and, second, modelling just based on PV manufacturer
datasheet information.
6
So calculation of PR requires records of Gef; and calculation of PRSTC requires also
records of TC. For the measurements of such variables we recommend to use reference
PV modules of the same type of that the concerned array installed in the same
supporting structure. This way, correction due to the transposition from horizontal to the
plane of array (solar radiation sensor used to be horizontal pyranometers) and also
correction for angular, spectral and soiling responses are avoided, as well as their
corresponding uncertainties.
In order to measure effective irradiance, Gef, a module of the same type has to be shortcircuited by using a calibrated resistor (shunt), which allow to obtain the short-circuit
current of the PV module. This variable is directly related to irradiance, as is established
in the international standards[5][6][7]:
Gef  G *·
I SC, ref
I SC,refC
*
·
1

1   Isc · TC  TC*

6
In this equation ISC,ref is the short-circuit current measured, ISC,refC* is the value of this
current at STC (given by an accredited laboratory after the calibration process), and αIsc
is the module’s temperature coefficient of short-circuit current.
On similar lines, in order to measure cell temperature, TC, a module of the same type
has to be open-circuited, which allow to obtain the short-circuit voltage of the PV
module. This variable is related to cell temperature, as is established in the international
standards[8].
VOC,ref  VOC,refC  β Voc · N S,ref · TC*
*
TC 

m· k
Gef 
N S,ref ·  β Voc 
· Ln * 

q
G 

7
In this equation VOC,ref is the open-circuit voltage measured, VOC,refC* is the value of this
voltage at STC (given by a laboratory after the calibration process), NS,ref is the number
of PV cells composing the sensor, m is the diode ideality factor (in the one diode model
its value is between 1 and 2, typically 1.3 from crystalline silicon), k is the Boltzmann
7
constant, q is the electron charge and βVoc is the module’s temperature coefficient of
open-circuit voltage.
This is in practice a better indicator of TC than the direct temperature measurements
given by thermocouples, PT100 or PT1000 sensors glued to the back of modules. The
open-circuit voltage avoids possible sensors sticking failures and also the uncertainty
associated to non-homogeneous temperature distributions inside the PV modules. This
is because the sensor is glued to a single point while the open circuit voltage of the PV
module integrates the values corresponding to all the solar cells.
On the other hand, as the reference module of cell temperature is in open circuit and the
PV array is in operation delivering power, hence slightly cooler[9][10][11], a small
correction must be done in the cell temperature. This is because the reference module
has to dissipate all the solar energy reaching it as heat, while the PV array dissipates a
part of the whole solar energy by delivering power. It is easy to see that the following
equation applies:
TC,A  TC,MR 
NOTC  20
800
 G  ηGef , TC 
8
where TC,A is the array cell temperature, TC,RM is the reference module cell temperature,
NOCT is the nominal operation cell temperature and  (Gef, TC) is the module efficiency
(𝐺𝑒𝑓 , 𝑇𝐶 ) =
𝐺𝑒𝑓
𝐺𝑒𝑓
𝑃∗
· [1 + 𝛾 · (𝑇C − 𝑇C∗ )] [𝑎 + 𝑏 ∗ + 𝑐 · ln ∗ ]
∗
𝐴𝑟𝑒𝑎 · 𝐺
𝐺
𝐺
𝐺
𝐺
𝑒𝑓
𝑒𝑓
= ∗ · [1 + 𝛾 · (𝑇C − 𝑇C∗ )] [𝑎 + 𝑏 ∗ + 𝑐 · ln ∗ ]
𝐺
𝐺
9
Both reference PV modules to measure Gef and TC are not commercially available and
they have to be specifically stabilized and calibrated in recognized laboratories. The
stabilization requires a minimum light soaking of 60 kWh/m2, as established at
international standards[12][13].
Sometimes is difficult to find a free place in the structure to install two reference PV
modules to measure operation conditions Gef and TC. Then, an option is to use only one
8
reference module as the unique sensor of both Gef and TC. This can be achieved by
taking advantage of the module internal connections in the junction box. Thus, a part of
the module can be short-circuited with a shunt resistor for measuring Gef while the other
part is in open-circuit for measuring TC (Figure 2).
(a)
Figure 2.
(b)
(a) Single reference module that has been modified to measure simultaneously boh Gef and
TC. (b) Connection box added to the module with the shunt resistor and the wiring to get
both signals simultaneously (in the module’s junction box can be noticed the internal
modification).
The signals from the reference PV modules to measure Gef and TC should be recorded
with a datalogger that must be able to store data with a good accuracy at least each
minute. If so high frequency is not possible, the datalogger has to be able to store mean
data that are representative of the whole interval. Otherwise, mistakes on the daily,
weekly, monthly or yearly irradiation can drastically affect to the final result of PR or
PRSTC. In fact, we have analysed the impact of storing instantaneous or mean data each
15 minutes during a whole year from reference modules in a PV installation.
Differences even above 10% in daily irradiation and close to 3% in monthly irradiation
are obtained. These differences could affect drastically to the result of performance
values.
9
2.1.1. Summary.
In order to obtain the global performance of a PV system PR (for yearly
evaluation) or PRSTC (for sub-year periods) records of effective irradiance, Gef, and cell
temperature, TC, are needed. These values should be stored each minute or higher
frequency with a datalogger of good accuracy. In case of a lower data acquisition
frequency, the stored values must be mean values of the whole interval in order to be
representative. For the measurements of Gef and TC we recommend to use reference
modules of the same type of that the used in the PV array, previously stabilized and
calibrated in a recognized laboratory. This ensures that both PV array modules and
reference modules will have the same spectral, angular and thermal responses and a
similar degree of soiling, thus minimizing the uncertainty of the measurements of these
parameters. These reference PV modules (or one single module modified to act
simultaneously as both Gef and TC sensor) should be installed some time before the
beginning of test (so, they will be affected by similar soiling) and they must be placed in
a place of the PV installation which provides representatives values of Gef and TC for the
full PV array.
The use of reference modules is a better option than the current state of the art
based on pyranometers and thermocouples. On the one hand, pyranometers are
broadband sensors that use to be installed to measure horizontal radiation. So, as their
angular, spectral and soiling responses are different than the ones corresponding to a PV
module, some corrections are needed in order to obtain the effective irradiance on the
PV module. When a reference module is short-circuited to act as an irradiance sensor all
this corrections are avoided and, consistently, the uncertainty of the final value is
minimized because the effective irradiance is measured directly. On the other hand,
thermocouples or similar devices (PT100, PT1000) can have sticking failures and they
are only glued to a single point. So, mistakes and higher uncertainty due to nonhomogeneous temperature distributions inside the modules are more probable. Again,
these problems are avoided when using a reference module in open-circuit because its
voltage value is directly related to the internal cell temperature and, besides, it integrates
the individual values corresponding to all the solar cells.
10
2.2. TESTING KIT 2. I-V curve tracers: performance of PV modules/arrays
The key device in a PV installation is the PV module because it is responsible of
converting solar energy into electricity. Therefore, the better the PV module
performance is, the more electrical energy will be delivered by the PV system.
Before reaching the market, the production line of a model of PV modules have to get
the design qualification established at IEC 61215 and IEC 61646. The objectives of the
tests in these standards are ensuring the quality and the reliability of the modules that
are going to be manufactured as is defined in the production line. Although the modules
belong to a production line certificated by the previous standards, it does not assure
totally that the final module’s power delivered at the market is the one reported. In fact,
there are studies that estimate the deviation of PV real power between 2% and 8%
below their actual nominal power announced by the manufacturer[14][15]. Besides, the
buy-sell contracts of PV modules use to be related to the total power of the modules
delivered. For this reason, PV modules should be tested in order to know if their
electrical characteristics are coherent to those reported by the manufacturer. Otherwise,
the final performance of the PV system could be lower than the expected one and,
therefore, the final economical incoming could not be enough to monetize the purchase
of the modules.
PV modules electrical characteristics use to be presented by its I-V curve at STC; that is,
their working curve: the set of currents that the PV module can deliver when it is
polarized at an established voltage and when their cells are at 25ºC and the irradiance
reaching the module is 1000W/m2. Thanks to this characteristic, malfunction of PV
modules or PV strings/arrays inside an installation can be discovered: actual power
below manufacturer information, light induced degradation –LID–, potential induced
degradation –PID–, mismatching losses higher than the expected ones, hot-spots,
unforeseen shadows, etc. All these undesired phenomena result on a lower power value
of the final system which adversely affects the energy production of the installation and,
therefore, its economic income.
These are several commercial I-V tracers available in the market[16], but in the
PVCROPS project we have decided to develop our own I-V curve tracer for on-site
11
testing. This device can be sized to be adapted to the system which is going to be tested,
from single modules –some hundreds of watts–, through single strings –some kilowatts–
and reaching entire PV arrays –up to 2 megawatts. As this device is scalable, currents
from several amps to a thousand of amps can be measured. Nowadays, commercial
devices are limited to 100 A, so they are able to measure PV systems lower than 100
kW power. Besides, the devices implemented allow measuring every kind of PV
modules because they are based on the charge of capacitors through insulated gate
bipolar transistors (IGBTs), not like other commercial alternatives based on transistors
which are not able to measure some kind of special modules. Finally, the proposed I-V
tracer allows the PV system characterization on-site without the need to uninstall it and
send individual modules back to the laboratory.
In fact, the on-site characterization of PV modules is not yet a common practice within
the framework of quality control procedures, despite of the fact that published results of
outdoor rating procedures are very positive in terms of both accuracy and
repeatability[17][18].
2.2.1. Single I-V curve tracer.
Figure 3 shows the power circuit of the capacitive load developed:
Figure 3.
Power circuit of the developed capacitive load based on Isolated Gate Bipolar Transistors
(IGBT).
12

Insulated gate bipolar transistor 1 (IGBT1) is switched on-off sequentially
for charging and discharging, respectively the capacitor C. The size of this
capacitor is selected according to the required charging time, whose
recommended value[19] is between 50 and 200 ms in order to avoid the
capacitance effects of the PV module (faster charging times) and the
variation of operating conditions during the tests (slower charging times).
The charging time, tC , is slightly higher than the value given by:
tC 
VOC
C
I SC
10
Where VOC and I SC are, respectively, the open-circuit voltage and the shortcircuit current of the PV module/string/array. The size of the capacitor
should be adapted to the characteristics of each PV technology (between
other issues, it has to have a nominal value higher than the maximum
expected value of VOC ). A good practice is to calculate the needed
*
capacitance to get the desired charging time at STC, that is with VOC
and
*
. Besides, these short times reduce the overheating and the size of
I SC
components, such as IGBTs and capacitors.

Diode D1 protects the load against the reverse polarity connection of the PV
module/string/array.

The negative pre-charging circuit (sub-circuit composed of push button P1,
fuse F2, resistor RP and voltage source VB), when the button P1 is pushed,
applies a negative voltage (-9V) to the capacitor, which allows the voltage
drop across the load to be compensated. This ensures that the I-V
characteristic starts in the second quadrant (V<0, I>0) and crosses the shortcircuit point (V=0, I=ISC). The Fuse F2 protects this sub-circuit against the
possibility of pushing the button P1 without performing a previous discharge
of the capacitors.
13

The fuse F1 protects the IGBT2 against the direct connection of the PV
module, which may occur in the case of an eventual failure of the IGBT1.

The resistor RD allows the capacitor C to be discharged when the IGBT2 is
switched ON, which should be carried out before tracing a new I-V curve.
During discharging, the capacitor voltage decreases exponentially from VOC
to zero with a time constant RDC. The selection of RD is a trade-off between
a fast discharging with high power dissipation (low RD) or a slower
discharging with lower power dissipation (high RD).

Finally, diode D2 is used to avoid the discharge of the capacitor C through
RD and the diode of the IGBT2 when the negative pre-charging voltage has
been applied.
The current (I) of the PV module is measured with an external calibrated resistor
(0.5% accuracy), which is not displayed in Figure 3, and the voltage (V) is directly
measured at the output terminal of the PV module in order to carry out a “four-wire”
measurement. These signals must be registered using a differential four-channel
oscilloscope (1% voltage accuracy).
Each IGBT is switched from the OFF state to the ON state, and vice versa, by a
drive circuit, whose schematic is shown in Figure 4. The switch placed on the left-hand
side of the circuit is used to select the charge (IGBT1) or the discharge (IGBT2) of the
capacitor, and the push button switches ON the selected IGBT (1 or 2) when it is
pressed.
Finally, each control circuit is powered by the power supply displayed in Figure
5. The circuit is made up by three DC-DC converters connected to a rechargeable 12V
battery, which provide three outputs: one with +5V for the supply of the debounce
circuits, and two with +15V for the optocouplers. This supply circuit is designed to
provide a safe isolation between control and power circuits.
14
Figure 4.
Drive circuit to control the IGBTs.
Figure 5.
Supply circuit for the drive circuit.
The I-V tracer works this way: with the capacitor initially discharged, once the
IGBT1 is ON, the PV device under test (PV module, string or array) delivers its short-
15
circuit current ISC and the capacitor start to charge up. From that moment, the capacitor
voltage increases until it reaches the open circuit voltage VOC. Then, the capacitor
charge finishes and the current goes down to zero (Figure 6). Finally, the capacitor has
to be discharged in order to get a new I-V curve. So, when IGBT2 is switched ON the
resistor RD allows the capacitor C to be discharged.
This process is registered by a differential four-channel oscilloscope. Four
channels are needed because I-V curve (two channels) and effective irradiance, Gef, and
cell temperature, TC, (other two channels) from reference modules (see 2.1) have to be
registered simultaneously. So, I-V curve under real operating conditions is obtained,
which can be extrapolated to STC using well-known procedures[5] to compare with the
expected values (datasheet information).
Figure 6.
Current and voltage waveforms during the charging process.
Figure 7 shows the first I-V curve tracer implemented by the IES-UPM. In order
to avoid damages and to improve isolation during field tests, it has been implemented in
two separated sections: the left-hand one has the control circuitry, while the right-hand
one presents the high voltage capacitors. This device allows measuring from individual
PV modules to large PV arrays. In order to measure a PV module or a PV array it is
only needed to select the proper calibrated resistor adapted to the currents the system is
going to deliver and the capacitors to obtain a charging time between 50 and 200 ms,
taking into account the characteristics of the system (it has to have a nominal value
16
higher than the maximum expected value of VOC). Usually, in order to measure PV
arrays it is used to be needed to include additional capacitors.
Figure 7.
External and internal views of the first capacitive I-V curve tracer implemented by the IESUPM. Left box contains the IGBTs and the driver and supply circuits. Right box contains
the capacitors and the negative pre-charging circuit.
This capacitive load is totally manual and allows the user to select its favourite
differential four-channel oscilloscope. Figure 8 shows the I-t, V-t , Gef-t, TC-t and the I-V
curves from a PV module (left) and from an 84 kW PV array (right) obtained with this
equipment.
Later, also based in this first I-V tracer, IES-UPM has implemented another two
devices specifically designed to measure the I-V curve of single modules (Figure 9 (a))
and the I-V curve of PV arrays up to 2 MW (Figure 9 (b) and Figure 9 (c)). The last one
has been used to measure an 800 kW PV array (Figure 10). As far as we know, this is
the unique PV tracer that is able to obtain I-V curves with currents above 1000A (Figure
10 (c)).
17
Again, these capacitive tracers are totally manuals and allow the user to select its
favourite differential four-channel oscilloscope to register the I-V curve and the Gef and
TC signals from reference modules.
(a)
Figure 8.
(b)
(a) I-t, V-t, Gef-t and TC-t signals –up– and I-V curve –down– from a single PV module
measured with the first capacitive tracer assembled by the IES-UPM. (b) I-t, V-t, Gef-t and
TC-t signals –up– and I-V curve –down– from an 84 kW PV array also measured with this
device.
(a)
Figure 9.
(b)
(c)
(a). View of the I-V curve capacitive tracer implemented by the IES-UPM specifically
designed to measure individual PV modules. It includes all the required devices and
circuits in a single box. (b). View of the I-V curve capacitive tracer implemented by the
IES-UPM specifically designed to measure large PV arrays up to 2 MW. This box includes
the IGBTs, the driver circuit, the supply circuit and the negative pre-charging circuit.
(c) View of the one of the boxes with the capacitors for the I-V curve tracer of picture b.
18
(a)
(b)
1400
1095 A
642 V
703 kW
1200
I (A)
1000
1226 A
800
600
400
820 V
200
0
0
200
400
600
800
1000
V (V)
(c)
Figure 10.
(a). View of the I-V curve capacitive tracer implemented by the IES-UPM specifically
designed to measure large PV arrays up to 2 MW. The box with the IGBTs and all the
circuits as well as the two boxes with the capacitors are connected at the input wires of a
real PV array of 800 kW nominal power. (b) View of the class 0.5 shunt resistor used to
measure the DC current. The nominal current of this particular shunt is 1000 A. (c) I-V
curve obtained from the 800 kW PV array with this capacitive tracer. As far as we know,
this is the largest I-V curve measured all around the world.
ACCIONA has also implemented his own I-V tracer based on the previous
scheme. Figure 11 shows the device. In this case, depending on the position of the
selector and the plugs selected (right side or left side) a PV module or a PV array up to
150 kW can be measured. But this time the procedure can be manual or semiautomatic:
the user can push each one of the buttons of the front panel (the left-hand ones)
following the sequence of charge-discharge explained before; or it can push the
“automatic button” (the right-hand one) which execute the whole charge-discharge
process. The user has only to prepare the oscilloscope to be ready for the acquisition of
the signals, its storage and the final process to obtain the I-V curve.
19
(a)
(b)
(c)
Manual
buttons
Automatic
button
º
(d)
Figure 11.
Views of the I-V curve capacitive tracer implemented by ACCIONA: (a) Front part of the
tracer, with the front panel to control the measure process –up– and the switch to select the
device that is going to be tested–down–, a single module or an array. (b) Rear part of the
tracer, with the control circuit and the batteries to feed them –up– and the two sets of
capacitors to measure PV arrays –left hand– or PV modules –right hand–. (c) Side part of
the tracer, with the different plugs that allow measuring PV arrays (on the other side are the
plugs to measure PV modules). (d) Detail of the front panel, with the output signals (up,
BNC connectors) and the buttons to measure manually (left ones) or automatically (right
one).
Figure 12 shows the I-t, V-t , Gef-t, TC-t and the I-V curves from a 170 W PV
module (left) and from a 150 kW PV array (right) obtained with this equipment.
20
(a)
Figure 12.
(b)
(a) I-t, V-t, Gef-t and TC-t signals –up– and I-V curve –down– from a single 170 W PV
module measured with the capacitive tracer assembled by ACCIONA. (b) I-t, V-t, Gef-t and
TC-t signals –up– and I-V curve –down– from a 150 kW PV array also measured with this
device.
Finally, CLSENES has developed more professional and automatic I-V tracers
ready to be introduced into the market: one light-kit for BIPV tests, whose size and
weight are suitable for easy transportation and quick test even on building roofs (Figure
13 (a)) and a second heavy-kit for PV plant tests, with a portable rolling container that
carries all the heavy components which is also useful as a table for the data recording
equipment (a laptop and a USB oscilloscope, Figure 13 (b)).
In order to evaluate quickly the real power of PV arrays and to compare real
performance with the nominal one, CLSENES has also developed an open software
application based on Labview for automatic data acquisition and data processing of I-V
characteristics, with a friendly graphical user interface (GUI) to present curves and table
data of measurements at real outdoor conditions automatically in near real-time. After
additional filtering, the basic parameters from the I-V curve (ISC, VOC, IM. VM, PM and
FF) are extracted for fast diagnosis of failure in the PV array. Figure 14 shows the GUI
implemented by the CLSENES, with the I-t, V-t and I-V curves from a PV string
measured with the BIPV testing kit. This string is affected by a high mismatch between
modules characteristics, what is evidenced by the flat stair that appears in the I-V curve.
21
(a)
(b)
Figure 13.
Views of the I-V curve capacitive tracers implemented by CLSENES. Both work
automatically: (a) Light-kit for test in BIPV and small installations. It is inside a briefcase
which can be easily transported. (b) Hard kit for large PV installations up to 100 kW. It is
included in a portable rolling container which can be used as a table for the laptop and the
oscilloscope.
Figure 14.
Graphical User Interface of the open source application implemented by CLSENES in
Labview. This application executes the automatic data acquisition and data processing of
the I-V curve measured with the capacitive tracers implemented by CLSENES once the
MEASURE button is pressed (upper left corner). The menus showing the I-t and V-t curves
–up– and the associated I-V curve –down– of the module measured are presented.
22
2.2.2. Twin I-V curve tracer.
Previous outdoor rating procedures[17][18] rely on using a reference module and
the measurement of both the test module and the reference module, calibrated by an
accredited laboratory. This module, as has been said previously in 2.1, should be of the
same type as the modules being tested in order to ensure that the spectral, optical and
thermal responses are very similar. In a simple way, these procedures consist of the
following steps. First, the two I-V curves of the sample and the reference are traced
quasi-simultaneously. Second, the operating conditions (irradiance and cell temperature)
are calculated through the measured short-circuit current and the open-circuit voltage of
the reference. Third, the I-V curves of both reference and sample are corrected to
STC[5]. And, finally, the deviations of the calculated parameters under STC of the
reference regarding its calibration are calculated, and these same deviations are applied
to the parameters of the sample in order to obtain its final rating under STC.
In the practice, the two I-V characteristics are measured with a certain delay
because common electronic loads are usually able to trace only one I-V curve at the
same time. The recommended maximum interval between sample and reference module
measurements is 3 minutes[17]. Obviously, the lower this delay the less different the
operating conditions are. But the main advantage of minimizing this delay is the
substantial reduction of the testing time, which is especially important when a high
number of samples must be measured.
This is the reason why IES-UPM has developed a twin capacitive load that is
able to trace two I-V characteristics simultaneously in the framework of the PVCROPS
project. This device may reduce the testing time and the rating uncertainty.
Basically, the twin capacitive load is just the duplication of the single capacitive
load that has been described in section 2.2. Figure 15 (a) shows the first version of the
twin capacitive load and Figure 15 (b) shows the final version, which internally includes
the oscilloscope and the calibrated resistors to measure currents. This way, external
wiring needed to connect PV modules to the I-V tracer is reduced. Figure 16 shows the
V-t and I-t of two PV modules (the reference one and the sample one) which has been
23
registered during the charge of the capacitors using a differential four channel
oscilloscope.
(a)
(b)
Figure 15.
(a). View of the first version of the twin I-V curve capacitive tracer implemented by the
IES-UPM specifically designed to simultaneously two PV modules: the sample one and the
reference one. It includes all the required devices and circuits for measuring two modules in
a single box. (b). Views of the final version of the twin I-V curve capacitive tracer (open
–left– and closed –front part up and rear and side parts down–). This new version includes
inside the box –left– the differential four-channel oscilloscope.
Figure 16.
(a) I-V curves of both a sample module and a reference module measured with the twin
capacitive tracer assembled by IES-UPM.
24
2.2.3. Summary.
In order to know if the electrical characteristics of a PV system are coherent to
those reported by the modules manufacturer, PV modules, strings and/or arrays must be
tested. The electrical characteristics of a PV system can be obtained from its I-V curve.
Thanks to this characteristic, malfunction of PV modules or PV strings/arrays inside an
installation can be discovered: actual power below manufacturer information, light
induced degradation –LID–, potential induced degradation –PID–, mismatching losses
higher than the expected ones, hot-spots, unforeseen shadows, etc. All these undesired
phenomena result on a lower power value of the final system which adversely affects
the energy production of the installation and, therefore, its economic income.
In this section we have shown the design of an I-V curve tracer for on-site testing
and some results. This device can be sized to be adapted to the system which is going to
be tested, from single modules –some hundreds of watts–, through single strings –some
kilowatts– and reaching entire PV arrays –up to 2 megawatts. It is based on the charge
of capacitors through insulated gate bipolar transistors (IGBTs). For the PV system
characterization it is needed reference modules as proposed in section 2.1 to obtain the
effective irradiance and the cell temperature at the same time than the I-V curve is
traced; a differential four-channel oscilloscope, to register both the I-V curve and the
operating conditions; and a laptop to store the data and extrapolate it to STC.
Several I-V curve tracers have been presented: single devices to measure
manually or automatically PV modules and/or PV strings/arrays up to 2 MW; and twin
devices that are able to measure simultaneously two PV modules.
25
2.3. TESTING KIT 3. Climatic box: detailed performance of PV modules.
In the previous section I-V curve tracers have been presented to measure on-site at real
operating conditions that use to be different than the STC. So, the measurements
obtained have to be extrapolated to STC so they can be compared to the expected values
as the module manufacture reports in the datasheet.
Measurements at STC are usually done in laboratories with flash testers, what is the
common practice today for controlling the power delivered by the PV manufacturer.
Obviously, the temperature and the irradiance can be controlled at a laboratory thanks to
the factory’s air conditioning system and with a complex solar simulator. Therefore, the
22 I-V curves required by international standard[6] (IEC 61853-1) that proposes to obtain
the PV module performance at 4 cell temperature (from 15ºC to 75ºC) and at 7
irradiance levels (from 100W/m2 to 1100W/m2) can be measured quickly. This
procedure is clearly defined for the characterization with a solar simulator inside a
laboratory because in the field is hardly possible to obtain simultaneously these specific
values of irradiances and cell temperatures.
2.3.1. On-site measurements at STC.
Nevertheless, the PVCROPS team encourages testing on site because the best
“simulator” of solar radiation all around the world is the Sun. This is the reason the
PVCROPS team has designed and implemented a cheap device to measure outdoors the
module power at STC: a climatic box that can be built anywhere a PV installation has
been constructed to test the modules and characterize their behaviour. So, the modules
can be tested on-site at STC. The only requirement is to have a sunny day, an I-V tracer
(like the ones presented previously) and a datalogger to record the irradiance and the
module temperature at different points. Besides, the PV module sample can be tested in
the PV plant without the need of sending it to a laboratory that can be far away, with the
consequent transport cost.
Figure 17 shows the first version of the climatic box, which was made
completely of isolated material (polystyrene). The PV module inside the climatic box is
26
cooled down to 25ºC with an external climatic device and 5 temperature sensors
(PT1000) attached to the PV module back sheet are used to monitor its temperature
(Figure 17 (a)). As the box is mounted on a tracking structure, it can be manually
positioned (tilt and azimuth) and an external secondary solar cell helps to move it until
incident irradiance is very close to 1000 W/m2 (Figure 17 (b)). Then the box cover is
retired which forces the PV module to operate at Standard Test Conditions, STC (Figure
17 (c)). An I-V curve is then recorded and later adjusted to precise STC, using the
incident irradiance given by a primary solar cell, also placed inside the box (Figure 17
(d)), and the PV module temperature given by the 5 PT1000 sensors attached to the PV
module back sheet. These temperatures must be recorded with a datalogger
simultaneously to the I-V curve acquisition.
Figure 17.
(a)
(b)
(c)
(d)
First version of the climatic box used for testing PV modules outdoors. It is completely
made of polystyrene (a) Climatic box opened, with the PT1000 sensors inside. (b) Before
the test, the PV module is placed inside the thermally insulated box, the inside temperature
is forced to 25ºC by means of an air conditioner and the box is manually positioned to reach
incident irradiance close to 1000 W/m2 thanks to the external reference solar cell (c) The
cover is removed and I-V curves are recorded at 25ºC and also along the natural heating
process. (d) Detail of the external and the internal reference solar cells.
27
Figure 18 shows the improved second version of the climatic box. This time the
box is made of wood, because the previous one was very weak, and it is filled with
white polystyrene to decrease the heating of the whole box once the cover is removed.
Now 9 temperature sensors can measure the module temperature at 9 different points
(Figure 18 (a)). In order to cool down faster the module, the climatic device is located at
a side of the box and a set of four fans, placed at the box corners (Figure 18 (d)), helps
to homogenize the temperature inside the box (Figure 18 (b)). The cover box is done of
polystyrene lined with a thin reflective layer to avoid heating of the inside.
Figure 18.
(a)
(b)
(c)
(d)
Second version of the climatic box used for testing PV modules outdoors. This second
version is made of wood and it is filled with white polystyrene to decrease the heating of
the whole box once the cover is removed (a) Climatic box opened, with the PT1000
sensors, the fans and the internal reference solar cell inside. (b) Before the test, the PV
module is placed inside the thermally insulated box, the inside temperature is forced to
25ºC by means of an air conditioner and the box is manually positioned to reach incident
irradiance close to 1000 W/m2 thanks to the external reference solar cell (c) The cover is
removed and I-V curves are recorded at 25ºC and also along the natural heating process.
(d) Detail of the fans and the internal reference solar cell.
Figure 19 to Figure 22 show the result of measuring along a whole year a PV
module inside the climatic box. The measurements until December were done inside the
28
first box. The remaining measurements were done inside the second improved box. As
can be seen, there is an increment of the currents when changing the box, probably due
to a misalignment of the internal primary reference cell. Nonetheless, repeatability of
the measurements is very good: almost all the values of ISC, IM, VOC, VM and PM are
inside a range of 1% regarding the associated mean values.
Figure 19.
I*SC –blue diamonds– and I*M –red squares– of a PV module measured 21 times with the
climatic box along a whole year. The continuous lines represent the mean values of all the
measurements and the dotted lines represent the deviation of 1% –internal ones, red and
blue– and 2% –external ones, yellow and green– respect to these values. The
measurements until December were made inside the first climatic box. The remaining
measurements were made inside the second version.
Figure 20.
V*OC of a PV module measured 21 times with the climatic box along a whole year.
continuous line represents the mean values of all the measurements and the dotted
represents the deviation of 1% –internal ones– and 2% –external ones– respect to
value. The measurements until December were made inside the first climatic box.
remaining measurements were made inside the second version.
29
The
line
this
The
Figure 21.
V*M of a PV module measured 21 times with the climatic box along a whole year.
continuous line represents the mean values of all the measurements and the dotted
represents the deviation of 1% –internal ones– and 2% –external ones– respect to
value. The measurements until December were made inside the first climatic box.
remaining measurements were made inside the second version.
The
line
this
The
Figure 22.
P*M of a PV module measured 21 times with the climatic box along a whole year.
continuous line represents the mean values of all the measurements and the dotted
represents the deviation of 1% –internal ones– and 2% –external ones– respect to
value. The measurements until December were made inside the first climatic box.
remaining measurements were made inside the second version.
The
line
this
The
30
2.3.2. On-site measurements of temperature coefficients and efficiency
parameters of the PV module.
The box allows measuring electrical characteristics at STC. But if we continue
measuring I-V curves once the cover box is removed, the solar radiation heats the PV
module until the equilibrium temperature is reached (roughly about 30ºC over the
ambience). Heating rate is between 3 to 6ºC per minute, which is slow enough to record
up to 25 I-V curves along the process, allowing the measurement of the PV module
power, current and voltage temperature coefficients. These coefficients must be
measured for the calibration of reference PV modules in order to reduce uncertainty and
avoid deviations in effective irradiance and in cell temperature. Figure 23 shows an
example of the results of such measurements.
Figure 23.
Measured characteristics of a PV module versus operation temperature, ranging from 22ºC
to 52ºC, using the climatic box.
But these coefficients are not only needed for reference modules. They are also
appreciated to achieve more accurate energy estimations of the PV installation, reducing
the uncertainty of such predictions. A good energy estimation relies not only on the
goodness of the module power characterization at standard test conditions (STC), but
also on the goodness of the characterization of the module behaviour related to the
variation of irradiance and temperature. So, it is closer to the reality running a
31
simulation exercise of energy production with the actual values measured than with the
values obtained from datasheet. In section 2.1 (equation 3) we presented the model
adopted by SISIFO to describe the performance of a PV system. It relies on information
from manufacturer: nominal power of PV modules, P*M; power variation of module
with temperature, , and efficiency dependence on irradiance, which is described by
three parameter a, b and c that can be obtained from power corresponding at two other
than G* irradiance values, which must also be found at datasheets, providing they
comply with EN 50380[4].
This climatic box allows measuring also these parameters, a, b and c from a
batch of modules that are going to be installed in the PV system to characterize with
higher accuracy the behaviour of the whole system. Following the procedure presented
in Figure 17 P*M and  are obtained. As the box is mounted on a tracking structure, it
can be oriented to achieve the desired amount of irradiance during the I-V curve
measurements, thanks to the external secondary solar cell. So, if the explained
procedure is repeated to measure power at 25ºC but at lower irradiances, as for example
600 W/m2 and 200 W/m2, values of a, b and c can be calculated.
As an example, this exercise has been done with a batch of 7 modules. Figure 24
shows the result of measuring the temperature coefficients of each one. Individual
values are similar and close to the datasheet value, but they differ slightly one from
another one. So, better than using the manufacturer value of  for the energy estimation
is to use the mean value obtained from real measurements.
Figure 25 shows the result of measuring at 25ºC the PV modules, but at lower
irradiances. These values allow calculating a, b and c values and, therefore, the module
efficiency curve, which is presented in the graph (normalized by the module efficiency
at STC). As can be noticed, all the modules perform better than is reported in the
datasheet; that is, this information in the datasheet is conservative. Therefore, if this
information is used in simulation software the final energy prediction will be less
accurate. It would be better to use the average of the module efficiency curves measured
to perform a more accurate simulation.
32
Figure 24.
Temperature coefficients, γ and β, of seven different modules from the same batch
measured inside the climatic box.
Figure 25.
Efficiency curve of seven different modules from the same batch measured inside the
climatic box. They are obtained by measuring the PV modules at 25ºC and at 1000, 600 and
200 W/m2. The efficiency curve from the information of manufacturer datasheet is also
included.
33
2.3.3. Summary.
A climatic box has been presented in order to characterize the behaviour of a PV
module in the same place where it is installed. This device –that is composed by the
isolated box, a manual tracker in which it is mounted, an external air conditioning
machine, several temperature sensors (PT1000) and the datalogger to record these
temperatures, four fans at the corners, and two reference cells (one secondary external
and another primary one internal)–, in combination with the I-V curve tracers proposed
in the previous section, allows measuring the I-V curve at STC, as well as the
temperature coefficients and the module efficiency at different irradiances.
The temperature coefficients are needed for the calibration of reference PV modules.
And both the temperature coefficients and the parameters defining the efficiency
behaviour at different irradiances are very useful to reduce the uncertainty and to
achieve more accurate predictions of the system energetic production.
34
2.4. TESTING KIT 4. PID tester: Potential Induced Degradation detection.
Potential induced degradation (PID) of modules in large grid-connected PV installations
is an alarming phenomenon[20] from 2006. Modules affected by PID have power losses
due to leakage currents. Initially it was associated to special manufacturing procedures
for modules, as SunPower modules (with high-efficiency back contact cells) and
Evergreen modules (string ribbon cells)[20]-[23]. Nevertheless, five years ago was
reported that crystalline silicon modules are also susceptible to PID[24]. From that
moment, PV manufacturers and researchers are investigating the origin of PID, the
solutions to avoid it in the manufacturing process and the test to detect PID in a PV
module[23]-[29].
PID has become a key issue last years because industry is worried about their effects: it
can cause yield losses[26] up to 80%. The explanation why PID did not appear before is
simple: PID is related to high system voltages (some hundreds of volts), which are
typical of systems installed after 2005. These high voltages joined with humidity and
high temperatures facilitate PID come on the scene[20][23]-[28].
There is not a standard test procedure to detect PID yet, although there is a proposal of
PID standard under discussion (draft stage[30]). It will be integrated into the design
qualifications and type approval European standards for PV modules[12][13]. This lack of
reference is the reason why manufacturers and researchers have proposed different tests
for the detection of PID in laboratories[23][25]-[27]. Another researchers have developed
new and expensive tests for the on-site PID detection in full daylight[29].
2.4.1. Laboratory degradation and detection tests.
One of the most typical accelerated test to degrade modules in the laboratory in
order to verify if they could be affected by PID is the next one[23]: applying nameplate
system voltage (typically ±1000V through a high voltage source) during 7 days between
short-circuited active poles and the PV module frame once its front surface has been
covered with conductive layer (aluminium foil or water). The test is passed if the power
degradation after this period of seven days is lower than 5%.
35
It is also possible to detect its PID degradation through its electroluminescence
(EL) image. This is done in a dark room though a CCD camera (Figure 26) while the
PV module is biased with a current source. As an example, Figure 27 shows the
electroluminescence image before and after executing this degradation test to two PV
modules at the IES-UPM laboratory. The up module is free of PID because all their
cells have almost the same brightness at both pictures –left-hand and right-hand EL–. At
the contrary, the down one is affected by PID because after the degradation period it
presents dark cells (inactive cells) that are shunted due to PID. The electroluminescence
image reveals clearly if the module is affected by PID, but it does not report about how
this degradation is affecting to the PV module efficiency.
(a)
Figure 26.
(b)
(a) CCD camera and (b) current source to bias the PV module in order to obtain its
electroluminescence image.
BEFORE
7 DAYS AT ±1000 V
AFTER
(a)
(b)
(c)
Figure 27.
Accelerated test to degrade modules in the laboratory. (a) First, the EL image of the
module is taken. (b) Then, a voltage of ± 1000V is applied between the short-circuited
active poles and the frame of the module, once its front surface has been covered with an
aluminium foil. (c) Finally, after a period of 7 days the EL image of the module is taken
again. If the module is prone to PID phenomenon some of their cells become inactive
(dark ones, down module).
36
2.4.2. Detection tests on the field.
The electroluminescence imaging can be done also on-site, but it is more
difficult because this test only can be done without light, that is to say, at night. Besides
it is needed to power all the equipment (CCD camera and current source) and CCD
camera has to stand totally stable without movements to acquire a good
electroluminescence.
This test at night has been done at a real PV installation with modules affected
by PID. Figure 28 shows the electroluminescence images of 24 modules connected in
series in a string. The electroluminescence image reveals clearly if the module is
affected by PID, but it does not report about how this degradation is affecting to the
module efficiency.
Figure 28:
Electroluminescence images of 24 modules connected in series in a string affected by PID:
the modules closer to the negative pole (module number 1) have dark cells (evidence of
PID); the modules closer to the positive pole (module number 24) have their cells with
almost the same brightness (free of PID). The module number 1 has not dark cells because
the original one, severely affected by PID, was replaced.
37
The module number 1 is the module connected to the negative pole of the string;
and the module number 24 is the module connected to the positive pole of the string. As
can be noticed, the modules closer to the positive pole have almost the same brightness,
while the modules closer to the negative pole have dark cells: they are affected by PID.
It has to be pointed that the module number 1 has not dark cells even though it is close
to the negative pole because the original module, severely affected by PID, was
replaced.
Nevertheless, the typical on-site test to know if a module is free of PID is to
measure its I-V characteristic with an electronic tracer as the presented in the section
2.2. Only the shape of the I-V curve can report the presence of anomalies in the
characteristic of the module, as can be seen in Figure 29.
1.2
1.0
I / ISC
0.8
0.6
0.4
No anomalies
0.2
PID
0.0
0.0
0.2
0.4
0.6
0.8
1.0
1.2
V / VOC
Figure 29.
Normalized IV curve of a module with no anomalies (circles) and normalized IV curve of
a module with PID (crosses).
But not always the differences on the shape are so evident. Then, it is needed to
compare the power of different modules at STC, as is shown in Figure 30. In this figure
the normalized power of the 24 modules of the string analyzed (the maximum power of
the module extrapolated to STC, Pm, divided by the nominal value from datasheet of
the maximum power, Pm nom) is represented. These results are coherent with the
electroluminescence images: the modules with more dark cells are those that deliver a
lower power.
38
Pm nom
Pm /Pm
Pm/ nom
1.02
1.00
0.98
0.96
0.94
0.92
0.90
0.88
0.86
0.84
0.82
0
Figure 30.
5
10
15
Module
20
25
Normalized maximum power of the 24 modules connected in series in a string affected by
PID
These power measurements have to be done in a sunny day and in a short period
of time to guarantee that all the measurements are done at the same irradiance and cell
temperature conditions to compare the value of the different modules with the lowest
uncertainty, what is not always possible. Another on-site option to detect PID is to
measure the voltage of the module when it is biased with a current source and without
illumination (dark conditions). It can be done with a blanket or a cardboard covering all
the module area, as is shown in Figure 31.
Figure 31.
Cardboard covering the area of a module which is being biased with a current source
(lower left corner).
Figure 32 shows the results of measuring this dark Vbias in the previous 24
modules. The modules were biased with a current of -1.5 amps (about 20% of the
datasheet short-circuit current). In the graph the normalized Vbias (the biased voltage of
39
the module, Vbias, divided by the nominal value from datasheet of the open-circuit
voltage, Voc nom) is represented. As can be noticed, the measurements are coherent
with the tests of electroluminescence imaging and I-V curve.
Vbias
Vbias
/ Voc/ Voc
nomnom
1.02
1.00
0.98
0.96
0.94
0.92
0.90
0.88
0.86
0.84
0.82
0
Figure 32.
5
10
15
Module
20
25
Normalized biased voltage (dark conditions) of the 24 modules connected in series in a
string affected by PID.
These points represent one of the points of the dark I-V characteristic of the
modules, the point associated with Ibias = -1.5 amps. In fact, a value of this current
closer to zero would be better to detect PID (slightly below 10% of I*sc nameplate
value). Figure 33 shows the dark I-V curve from 6 of the 24 modules under analysis. As
can be noticed, the more negative Ibias (higher absolute value, |Ibias|) the less
difference between the Vbias of all the modules; and the less negative Ibias (lower
|Ibias|) the more difference between the Vbias of the healthy modules (modules 1, 21
and 24, with almost the same Vbias value) and the Vbias of the modules affected by
PID (modules 3, 4 and 8, with lower and different Vbias values, in function of their
degradation). So, the dark I-V curve reports about the PID because it shows if the shunt
resistance (related to the slope of the curve for low |Ibias|) is high or low and
consequently the associated low or high leakage current, respectively.
Maybe it would be enough to compare the Vbias value when the module is
biased with a current between 5% and 10% of the datasheet short circuit current (in this
40
case, between -0.5A and -1A). As can be seen in the graph, in this range, differences are
more evident. Besides, this option is faster than measuring the whole dark I-V curve.
Vbias (V)
0
5
10
15
20
25
30
35
40
0
-0.5
-1
Ibias (A)
-1.5
-2
Module 1
-2.5
Module 3
-3
Module 4
-3.5
Module 8
Module 21
-4
Module 24
-4.5
Figure 33.
Dark I-V curves from 6 of the 24 modules of the array under analysis. The modules
affected by PID have lower Vbias values when biased at lower |Ibias|.
In any case, it must be highlighted that it is needed a “reference” module free of
PID to compare its dark I-V curve with the dark I-V curve of the suspect module. This is
because the dark I-V curve of modules with the same degree of degradation uses to be
similar. As an example, Figure 34 shows the dark I-V curve of 8 modules of the same
manufacturer and model just after unpacking an with the same temperature. It can be
seen that the curve of all the modules are almost identical.
Vbias (V)
0
5
10
15
20
25
0.0
Ibias(A)
-0.5
-1.0
Module 1
-1.5
Module 2
-2.0
-2.5
-3.0
-3.5
-4.0
Module 3
Module 4
Module 5
Module 6
Module 7
Module 8
-4.5
Figure 34.
Dark IV curves of 8 new modules just after unpacking.
41
30
35
40
The main advantage of this method to detect PID in the PV installation is that
these measurements can be done every day at every hour: there are not time restrictions
because the voltage is measured in dark conditions. These conditions can be reproduced
by covering the module with a blanket or a cardboard, as has been previously shown in
Figure 31. The only restriction is to do these measurements in the healthy module and in
the suspect modules at the same cell temperature conditions to reduce uncertainty.
Obviously, this test can be also performed in the laboratory.
Finally, the fastest and cheapest option (only some voltmeters are needed) is to
measure the voltage of each module while they are operating. Figure 35 shows the
normalized operating voltage (the operating voltage of the module, Vop, divided by the
nominal value from datasheet of the maximum power point voltage, Vm nom) of the
previous 24 modules connected in series in a string. The modules more affected by PID
have an operating voltage lower than the healthy modules. This is because the modules
with PID have higher leakage currents and, consequently, to deliver the same current
than the remaining modules in the string, they have to operate at a lower voltage.
Taking
into
account
the
results
of
these
measurements
and
the
electroluminescence images, a 5% difference between the operating voltage of a module
with PID and another one free of PID is indicative of the presence of this phenomenon.
Besides, this decrease of operating voltage is proportional to the power losses related to
PID.
Again, as in previous tests, the measurement Vop has to be done in a sunny day
and in a short period of time to guarantee that all the measurements are done at the same
irradiance and cell temperature conditions to compare the value of the different modules
with the lowest uncertainty.
The easier method to measure Vop is to open the junction box to have access to
the positive and negative poles of the module. Nevertheless, it is more and more usual
that PV modules have a sealing label in the junction box or even a insulation paste that
fills it that avoids to access to the active poles (Figure 36 (a)). If this label is broken or
the internal insulation paste is damaged, the module warranty could be lost. So, another
option is to prepare a testing kit composed of “T” connections in the poles of the
42
suspect modules and in the poles of at least a healthy module in order to compare their
Vop values (Figure 36 (b)). A voltmeter is the only additional equipment to complete
the kit.
Vm nom
VopVop
/ Vm/ nom
1.02
1.00
0.98
0.96
0.94
0.92
0.90
0.88
0.86
0.84
0.82
0
Figure 35.
5
10
15
Module
25
Normalized operating voltage of the 24 modules connected in series in a string affected by
PID.
(a)
Figure 36.
20
(b)
Sealing label of “warranty” in the junction box of a PV module. Insulation paste that fills
the junction box. “T” connections in the pole of two modules to measure its operating
voltage without open the junction box.
43
2.4.3. Summary.
Several manufacturers and researchers developed some tests in order to reveal if
a PV module is going to be affected by PID once it is installed in a PV system. The
typical procedure is to test the PV modules is applying ±1000V through a high voltage
source during 7 days between the short-circuited active poles and the frame of the
module. If the power of the specimen, measured with I-V curve tracers like the
presented in the section 2.2, has decreased above 5%, this kind of modules is prone to
this
degradation
phenomenon.
Another
option
is
to
obtain
the
module’s
electroluminescence image through a CCD camera when it is placed in a dark room and
it is biased with a current source. Then, the presence of dark areas reveals that the
module is affected by PID. Nevertheless, this second test does not quantify the power
losses derived from this effect. Both tests are designed basically to be done with
expensive devices (high voltage sources and CCD cameras) and in laboratories,
although the electroluminescence imaging can be also performed on field at night hours.
In order to detect PID at the location of the PV system, the PVCROPS project
proposes new and cheaper tests. The more typical test is to measure its I-V curve with a
PV curve tracer (section 2.2). PV modules affected by PID, apart of presenting a lower
power in comparison with a module free of PID, can have a strange shape in its I-V
curve, but sometimes this difference is not so evident. The problem of this test is that it
only can be performed on-site in sunny days.
A new test that can be performed at any time is to obtain the dark I-V curve of
the module when it is biased through a current source and covered with a blanket or a
cardboard. This test only requires the current source and a voltmeter to measure the
voltage delivered by the module. Moreover, it would be enough to measure the voltage
delivered at a current between 5% and 10% of the datasheet short-circuit current to
detect PID on a module. In any case, it is needed to compare these measurements with
the ones corresponding to a module free of PID at the same temperature.
Finally, the fastest and cheapest option to detect PID is to measure the voltage of
the module while it is operating (only some voltmeters are needed). A testing kit has
been produced to perform this test, consisting of “T” plugs to be connected in the poles
44
of the suspect modules and in one module free of PID to measure their operating
voltage. A difference higher than 5% between the operating voltage of a module with
PID and another one free of PID can be indicative of the presence of this phenomenon.
45
2.5. TESTING KIT 5. Hot-spot tester: thermal inspection.
A hot-spot consists of a localized overheating in a photovoltaic (PV) module due
to any anomaly. It appears when, because of the anomaly, the short circuit current of the
affected cell becomes lower than the operating current of the whole module, giving rise
to reverse biasing and thus dissipating the power generated by other cells as heat. Figure
37 shows two infrared (IR) images of hot-spots. The anomalies that cause hot-spots can
be external to the PV module (shading or dust) or internal (micro-cracks, defective
soldering, short-circuited by-pass diodes, potential induced degradation –PID). In
general, when a hot-spot persists over time, it entails both a risk for the PV module’s
lifetime and a decrease in its operational efficiency.
Figure 37.
(a)
(b)
(c)
(d)
Hot-spots in three modules. (a) General view of a tracker with hot-spots caused by PID. (b)
Hot-spot caused by micro-cracks. The operating temperature of the hot-spot is 87ºC while
the mean temperature of the rest of the module is 53 ºC. (c) Two columns of cells up to 4ºC
hotter that the remaining cells of the module. (d) This is due to a defective by-pass diode
that is short-circuited.
46
Hot-spots are relatively frequent in current PV generators and this situation will
likely persist as the PV technology is evolving to thinner wafers, which are prone to
develop micro-cracks during the manipulation processes (manufacturing, transport,
installation, etc.). Fortunately, they can be easily detected through IR inspection, which
has become a common practice in current PV installations. However, there is a lack of
widely accepted procedures for dealing with hot-spots in practice as well as specific
criteria referring to the acceptance or rejection of affected PV modules in commercial
frameworks. So there is not a clear answer to the next question: which affected PV
modules should be changed under the PV manufacturer’s responsibility.
PVCROPS has investigated the hot-spot impacts on the lifetime and on the
operational efficiency of the affected module (that is, the efficiency of the PV module
when it is integrated in a PV generator and connected to an inverter able to track the
maximum power point). First, hot-spots are characterized by the temperature increase of
the defective cell ∆𝑇𝐻𝑆
in relation to the non-defective ones and normalized to the
STC irradiance, ∆𝑇𝐻𝑆 ∗ :
∆𝑇𝐻𝑆
= max(𝑇𝑀𝑜𝑑𝑢𝑙𝑒 ) − min(𝑇𝑀𝑜𝑑𝑢𝑙𝑒 )
∆𝑇𝐻𝑆 ∗ = ∆𝑇𝐻𝑆
𝐺∗
𝐺
11
12
Then, using observations on 200 affected modules as experimental support, the
following acceptance/rejection criteria are proposed[31]:

If ∆𝑇𝐻𝑆 ∗ < 10°𝐶, to consider the module non-defective, except in the
case that one or more by-pass diodes are defective.

If ∆𝑇𝐻𝑆 ∗ > 20°𝐶, to consider the module defective.

If 10°𝐶 < ∆𝑇𝐻𝑆 ∗ < 20°𝐶, to consider defective all the modules with an
effective power loss (measured as a decrease in the operating voltage in
relation to a non-defective module of the same string) that exceeds the
allowable peak power losses fixed at standard warranties.
47
2.5.1. Traditional hot-spot detection.
The main device used to detect hot-spots in PV modules is an infrared camera. It
provides a thermal image of the object that is under study that is very easy to interpret:
the image shows the different temperatures of each one of the modules parts (see Figure
37), so an area of the module significantly hotter is detected immediately. In the last
years, these devices have decreased its price drastically and even there are new gadgets
that can be included on smartphones[32][33] to convert their internal camera into a
thermographic camera. This situation has allowed that IR cameras are more and more
frequent in the set of tools needed for the supervision of a PV system not only for the
detection of hot spots in modules, but also for the detection of overtemperatures inside
connection boxes and plugs due to defective connections between live wires[2].
The typical way of using these devices is as follows: the operator carries an IR
camera in hand and he inspects the temperature of PV modules when they are at normal
operation (injecting power into the grid) by looking at the IR camera screen. When a
much hotter area of a module is found, the operator takes a thermographic image and a
visual image of the suspect module, as well as he writes down the irradiance at which
the pictures are taken, the module location inside the system and its serial number (all
this information uses to be needed in case of a claim to the manufacturer). The
temperature of a normal area of the affected module and the maximum temperature of
the hotter area can be extracted from the picture with the software of the own IR camera
and, then, ∆𝑇𝐻𝑆 ∗ can be calculated. The value obtained reports about if the module must
be rejected regarding the criteria proposed above.
2.5.2. Hot-spot detection with PV drones.
The previous procedure is very easy to perform, but it is also very timeconsuming. The importance of fast methods has increased as the size of PV installations
has become larger and larger. Large PV plants and Building Integrated PV (BIPV) of
difficult access are doing harder and more time consuming to execute complete on-field
IR test to asses a proper performance of all the modules of the system.
48
Some drones are already used in the current state of the art to save time, but they
are too expensive. This is the reason why PVCROPS project has developed a cheaper
PV drone with open-source software and hardware to quickly detect visual defects and
hot-spots in PV systems through Hi-res cameras and IR imaging (aerial inspection
based on an unmanned aerial vehicle, UAV). Recent small and cheap multirotor
helicopters have enough power to carry video and IR cameras and their associated video
transmitters. With these devices it is possible to remotely control the quadcopter or even
planning automatic missions to detect defects in a PV installation.
CLSENES has taken advantage of the progress in airframe technology,
components costs and technical methods required, as well as of the open-sharing of
hardware designs based on Arduino. This open-source hardware offers to PV
community the option of designing and assembling in a cheap way an electronic
multirotor platform controlled by low-cost programmable microcontroller and free
software. Multirotor aircraft (quad-, hex- and octo-rotors), widely known as “drones”,
are a popular choice for monitoring tasks due to their ability to take off and land in
compact spaces and its increased maneuverability. Specially, the copter platforms,
equipped with GPS (Global Positioning System), IMU (Inertial Measurement Unit),
stabilizing platform, autopilot and digital cameras have excellent application prospects
for its use in visual and thermal analysis in PV installations.
Figure 38 shows the first version of the developed PV drone. This robocopter is
equipped with an analog visible camera and a modified IR camera. The visible camera
resolution is 600 TV lines, what is enough to detect optical imperfections on the
modules at first glance, and its video signal is transferred from the drone to a ground
station where a TV screen displays it to check the PV modules remotely. The IR camera
has 80x60 pixels and takes pictures following a command from the ground through a
radio controlled trigger.
As any UAV, our quadcopter requires the following components to achieve
functional autonomous flight: an airframe (the X-shape structure) and its associated
components (batteries, propellers and motors), onboard sensors for telemetry
(gyroscope, 3-axis accelerometer, barometer, 3-axis magnetometer GPS, altimeter and
49
compass), an autopilot to act upon telemetry data, a radio controlled flying device and
software for autonomous UAV control.
Figure 38.
First version of the developed quadcopter (PV drone). It has an IR camera and a visible
camera for monitoring a PV installation.
In our case, the quadcopter is equipped with an ArduCopter autopilot which
allows for autonomous and manual control of the aerial vehicle. The level of control
exerted by the autopilot is determined through tuning parameters associated with the
autopilot’s internal Proportional-Integral-Derivative (PID) controller. It provides tuning
settings for yaw, pitch, roll and throttle gain and cross-track behaviour.
The radio controlled (RC) flying device that has been assembled operates at 3
frequency ranges. One of these frequency ranges is for manually control the drone from
the ground by 2.4 GHz RC control station using 4 channels for Throttle, Yaw, Pitch and
Roll. Other channels are configured for video switching and still picture triggering. The
frequency ranges, based on 5.8 GHz wireless video-link, is for video transmission from
the air to ground. Finally, the additional 433 MHz wireless connection can upload to
Arduino controller waypoints with GPS coordinates for autonomous flying missions.
These flight plans can be developed and flashed to the ArduCopter hardware using free
and open-source software such as the Ardupilot Mission Planner. This software also
allows analyzing all the telemetry collected during flight of the vehicle to the laptop.
Basically, autonomous UAV needs two persons: an operator and ground crew
station. The operator is responsible to control the drone during launching and landing
operation to avoid any damages on the device. Ground crew station is responsible to
50
desing the flight trajectory at the study area and monitor UAV altitude, battery status,
number of satellite and data link between the drone and the ground computer.
The visual and thermo-graphic inspection with the drone is as follows: first, the
drone does its flying mission and records a video stream by a digital video recorder
(DVR) at the ground station. Simultaneously, the person who is supervising the drone
during the flight can both recognize suspicious PV modules and take thermo-graphic
pictures (the operator triggers the IR camera using RC commands). Then, on the ground
the operator checks the video again and the IR pictures. If a defect is recognized in a PV
module, the operator goes directly to its location to take more detailed pictures. Thus,
the operator avoids having to walk the entire facility looking for defects and should only
review those suspected modules detected by the drone. This method is less timeconsuming especially in large PV plants and in BIPV of difficult access.
Figure 39 shows pictures related to a flying mission for a 150-kW PV plant
located in Bulgaria. The mission planning involves 22 waypoints for zig-zag trajectory
exactly above the PV rows, starting from the NW corner of the PV installation and
flying at 5 m altitude. The useful flight time is about 15 min with fully charged
batteries, which is large enough for inspection all the PV installation. Figure 39 (a)
shows the screen of the laptop with the flying mission that was planned. Figure 39 (b)
shows the quadcopter in flight during the mission.
(a)
Figure 39.
(b)
(a) Quadcopter flying mission planned for a 150 kW PV plant. (b) Quadcopter during the
inspection flying mission above the array PV modules.
51
Figure 40 shows several pictures reporting a defect in a module during the flight
mission. Figure 40 (a) reveals a PV module with its front glass broken (this picture was
taken directly with the drone). Later, when the operator went to the location of this
module he could check that the module had a burnt just in the busbar connecting two
solar cells that was visible even in the back of the module (Figure 40 (b)). As the right
front-side busbar of the solar cell is evaporated all the current generated by the cell
flows through the left busbar and this part of the cell is overheated. As a consequence,
Figure 40 (c) shows the IR picture of the module, reporting about a hot-spot that is
higher than 83 ºC.
(a)
Figure 40.
(b)
(c)
(a) Picture recorded during the drone flight: a module has its front glass broken. (b) Picture
of the back of the module taken on site. The internal layers and contacts are burnt.
(c) Thermographic picture of the back of the module. This module has a hot-spot reaching
more than 83ºC and half of the cell affected is overheated.
Finally, Figure 41 shows other defects detected by the drone during the flight
mission: a cell-crack located on the corner of a PV module (Figure 41 (a)); electrocorrosion of busbar because rainwater has reached the internal layers of the module
(encapsulation problems, Figure 41 (b)); and temperature differences in solar cells of a
PV module above 10ºC, that could be an early indication of accelerated PV module
degradation (Figure 41 (c)). The two first defects was initially detected by the drone
during its panoramic flight and snapshots was taken later in detail with a Hi-Res
camera. The last one is an IR picture directly taken from the drone with the IR camera
installed on it.
52
(a)
Figure 41.
(b)
(c)
(a) Cell cracked detected on the corner of a PV module. (b) Electro-corrosion of the busbar
due to rainwater which has reached the internal layers of the module. (c) IR picture taken
from the drone. It shows two solar cells that are hotter than their neighboring cells (above
10ºC).
2.5.3. PV drone aerial photogrammetry.
The use of UAVs can be also useful to perform a shade analysis of the PV
system before its installation. A set of high resolution pictures taken with the PV drone
can be mosaiked using photogrammetric software1 to create from single orthomosaic up
to 3-D representations of the area of interest thanks to techniques based on ‘structure
from motion’ (SfM) that use to be included in this kind of software. So, mutual shading
between buildings, shading between roof elements, shading from surrounding
vegetation, etc. can be characterized before the PV system is installed. This
characterization allows evaluating what is the optimal arranging of the PV modules in
the PV array to minimize the reduction of energy production due to shading. This is
especially useful in BIPV located in urban areas.
Image processing algorithms need sharp accurate images to combine them in a
composite picture. Unfortunately, wide angle field of view (FOV) cameras like the one
installed on the PV drone do not perform well because of large edge distortion. This
setback is solved using a good compact digital camera which gets accurate results due to
more advances optics. But the original design of the PV drone is not enough powerful to
include a camera of these characteristics.
1
Many new software programs like LPS, Pix4D, ImageModeler, PhotoModeler and 123D catch offer
photogrammetric processing for end users.
53
CLSENES has improved the previous X-airframe design and has implemented a
second PV drone which includes: a new larger fibre frame (900 mm); four new more
powerful brushless motors; a new RC control (FrSky Taranis); two 500mAh LiPo heavy
duty batteries; a 3-axis magnetometer; and an additional compact digital camera
(Canon) with automatic shutter at the bottom of the UAV platform. As can be seen in
the pictures of Figure 42, this second PV drone is larger and more powerful than the
previous one. These new features allow a remote controlling distance up to 300 meters
and a maximal operation slot of 25 minutes.
Several pictures must be shooted to obtain a good 3-D reconstruction of the area
of interest. So, the camera firmware was reflashed through open source software (Canon
Hack Development Kit, CHDK) and specific intervalometer script was used to set a 5second shooting interval. Shooting pictures with this interval at an altitude of 3-5 meters
above the object can document a 25m2 area in about 10 minutes.
Figure 42
(a)
(b)
(c)
(d)
(a) and (b) PV drone configuration enabling 3-D mapping missions. (c) IR camera and
visible camera for monitoring a PV installation and compact digital camera for
photogrammetry analysis. (d) UAVs developed by PVCROPS team
54
A composite orthophoto can be obtained by software stitching of successive
string of single images. Overlapping of photos must be more than 60% to generate
realistic picture. It can easily be adjusted by straight-line flight and constant air speed of
the quadcopter. Photogrammetric processing of raw imagery enables the production of a
3-D mesh of the surface via SfM software. During processing the images are simply
selected and auto-aligned. The program automatically identifies common points
between images based on color and texture. From the multiple perspectives of common
points, stereoscopy (and thus three-dimensionality) is reconstructed, as can be observed
at Figure 43. However, photogrammetric processing requires significant computational
power and RAM capacity. Our first imagery test has taken more than 12 hours
computational time of conventional PC.
(a)
Figure 43.
(b)
3-D orthophotos of a sun tracker generated by the developed equipment.
The most accurate form of topographical data is Light Detection and Ranging
data (LiDAR). LiDAR data can be used to create three-dimensional digital elevation
models (DEMs) that analyze the impacts of shading obstructions, identify roof tilt, and
estimate the amount of roof area that can be used for a particular installation. Where
LiDAR data are not available, the PV community can use high-resolution
orthophotography along with building footprint and parcel data for feature and building
identification.
Another methodology, called Solar Automated Feature Extraction (SAFE™)
assesses the solar potential of buildings through a combination of aerial imagery and
advanced 3D modeling. The method takes into account factors such as roof obstructions
55
(air conditioning units, chimneys, vents), azimuth (the direction of the sun), shadowing
from other buildings, and roof slants. This methodology also calculates total roof area,
usable roof area for solar panels, the amount of electricity the panels can produce, the
electricity cost reduction, and resulting CO2 reduction.
2.5.4. Summary.
Nowadays there is a lack of widely accepted procedures for dealing with hotspots. Due to this lack, PVCROPS has proposed a specific criteria referring to the
acceptance or rejection of affected PV modules by hot-spots in commercial frameworks.
Hot-spots can be easily detected through IR inspection, which has become a
common practice in current PV installations. The main device used to detect hot-spots
in PV modules is an infrared camera. It provides a thermal image of the object that is
under study that is very easy to interpret because any area of the module significantly
hotter is detected immediately
In order to detect hot-spots, typically an operator carries an IR camera in hand
and, while is walking by the PV installation, he is inspecting the temperature of PV
modules when they are at normal operation (injecting power into the grid) by looking at
the IR camera screen. When a much hotter area of a module is found, the operator takes
a thermographic image of the suspect module, as well as the irradiance at which the
picture has been taken in order to calculate ∆𝑇𝐻𝑆 ∗ . The value obtained reports about if
the module must be rejected regarding the criteria proposed above. This procedure is
very easy to perform, but it is also very time consuming.
Some drones are already used in the current state of the art to save time, but they
are too expensive. This is the reason why PVCROPS project has developed a cheaper
PV drone with open-source software and hardware to quickly detect visual defects and
hot-spots. It is a quadcopter with an IR camera and a visible camera for monitoring a
PV installation. It can be manually controlled or even planning automatic missions to
detect defects in a PV installation. Examples of flying mission executed in real PV
installations have been presented.
56
But this device can be also useful to perform a shade analysis of the PV system
before its installation thanks to photogrammetry. So, a second open source PV drone
larger and more powerful equipped with an additional compact digital camera has been
constructed. This improved device allows creating 3-D representations of the area of
interest and characterizing the shade profile to evaluate what is the optimal arranging of
the PV modules in the PV array to minimize the reduction of energy production due to
shading.
57
2.6. TESTING KIT 6. Full PV system tester: detailed performance analysis.
Until now we have presented several testing kits that are useful to know if the
global performance of the installations is as initially expected or to characterize and/or
to detect failures in the main devices of the system: the PV module/array (I-V curve
tracers; climatic box to characterize it in several conditions on site; PID detection in the
laboratory and on-field; and thermal inspection looking for hot-spots). Nevertheless,
none of them allow knowing simultaneously the global performance of the installation
and the particular behaviour of both PV arrays and PV inverters, the main devices of a
PV system.
Testing kit 1 (reference modules and energy meter records, section 2.1) is
enough to obtain the global performance of a PV system, PR and/or PRSTC. But a better
and more accurate option for the characterization of the global behaviour of a PV
installation is to compare the real energy production during the test period (one or a few
weeks) with the simulated energy output. The last one can be calculated from the
corresponding recorded operating conditions (Gef, TC) and taking into account the
characteristic power of the array at STC and also its thermal and low irradiance
behaviour, the inverter efficiency and adopting the same baseline losses scenario
defined at the energy forecast phase: thermal losses, variation of module efficiency with
irradiance, shading, DC cable losses, inverter efficiency, inverter saturation, AC cable
losses, etc.
PVCROPS estimates the power delivered by the PV system with the PV
performance model adopted by SISIFO, a PV simulation software developed in this
project that is available at www.sisifo.info. The main equations are:
∗
𝑃𝐷𝐶 (𝐺𝑒𝑓 , 𝑇𝐶 ) = 𝑃𝑁𝑂𝑀
·
𝐺𝑒𝑓
𝐺𝑒𝑓
𝐺𝑒𝑓
· [1 + 𝛾 · (𝑇C − 𝑇C∗ )][𝑎 + 𝑏 ∗ + 𝑐 · ln ∗ ] 𝑓𝐷𝐶
∗
𝐺
𝐺
𝐺
𝑃𝐴𝐶 = 𝑃𝐷𝐶 𝑝 𝑓𝐴𝐶
58
13
14
𝑝 (𝑝𝑎𝑐 ) =
𝑝𝑎𝑐
2 )
𝑝𝑎𝑐 + (𝑘0 + 𝑘1 𝑝𝑎𝑐 + 𝑘2 𝑝𝑎𝑐
𝑝𝑎𝑐 =
𝑃𝐴𝐶
𝑃𝐼𝑀𝐴𝑋
15
16
𝑡=𝑇
𝐸𝐴𝐶 = ∫ 𝑃𝐴𝐶 𝑑𝑡
17
𝑡=0
In these equations, the symbol
*
refers to STC, 𝑃𝐷𝐶 is the DC power output of
∗
the PV array; 𝑃𝑁𝑂𝑀
is its nameplate DC power; 𝐺𝑒𝑓 is the effective global solar
irradiance in the plane of the array; 𝐺 ∗ is the global solar irradiance at STC (𝐺 ∗ = 1000
W/m2); 𝑇𝐶 is the cell temperature; 𝑇𝐶∗ is the cell temperature at STC (𝑇𝐶∗ = 25ºC);  is the
coefficient of power variation due to cell temperature; 𝑎, 𝑏 and 𝑐 are three parameters
related with the variation of module efficiency with solar irradiance; 𝑓𝐷𝐶 is a coefficient
that lumps together all the additional system losses in DC, e.g., technology-related
issues, wiring, soiling and shading; 𝑃𝐴𝐶 is the AC power output of the PV array from
this DC power at the inverter entry; 𝑝 is the power efficiency of the inverter; 𝑓𝐴𝐶 is a
coefficient that lumps together all the additional system losses in AC, e.g., technologyrelated issues and wiring; 𝑝𝑎𝑐 is the load factor, that is, the normalised AC output
power; 𝑃𝐼𝑀𝐴𝑋 is the nominal power of the inverter; 𝑘0 represents the no-load loss; 𝑘1
represents the losses that depend linearly on the current (as the voltage drop across
diodes); 𝑘2 represents the losses that depends on the square of the current (as the
resistive losses); and, finally, 𝐸𝐴𝐶 is the energy produced during a period of time T (a
year, for example). These equations properly define the performance of a PV system
with high accuracy as demonstrated by other authors[34][35].
It is interesting to note that concerned parameters (𝑃∗ ,  𝑎, 𝑏 and 𝑐) are not only
given at the information datasheet of modules, but also considered as a part of the
design qualification international norms (𝑎, 𝑏 and 𝑐 are obtained from module power
corresponding at three irradiance values, which is usually found at datasheets, providing
they comply with international standards)[4][12][13]. Besides, if a batch of modules
installed in the system has been previously tested as proposed in the previous sections
59
(2.3), all these values are coming from real measurements, what will reduce the
uncertainty of this power estimation.
Regarding the inverter response, 𝑘0 , 𝑘1 and 𝑘2 parameters that define its power
efficiency can be estimated from several values of the inverter efficiency curve, as those
reported in the manufacturer’s datasheet.
So, PVCROPS project has developed a “full PV system tester”: a testing kit that
allows characterizing the whole PV system from a global point of view, by knowing
both its total power performance and its total energetic performance, but also from a
particular point of view, by knowing the power performance of each single component.
It is composed by reference PV modules, a watt-meter together with AC clamps and a
shunt resistor, in order to measure DC and AC powers with high accuracy, and the
energy meter of the system under study. These devices allow recording simultaneously,
with a sampling time lower than one minute, Gef and TC from the reference modules and
PDC and PAC from the inverter entry and output. Thanks to these measurements a
detailed characterization of the whole system can be done.
2.6.1. AC characterization.
A watt-meter is able of measuring AC voltages and AC currents, the last ones
directly through its internal circuitry or through external AC clamps (Figure 44). Then,
it calculates the experimental active AC power (PAC,EXP). If simultaneously the
operating conditions, Gef and TC are measured, the simulated active AC power (PAC,SIM)
is obtained applying the previous equations. So, the AC power response can be
characterized when PAC,EXP are plotted versus PAC,SIM, as it is shown in Figure 45. In the
graph it can be noticed not only the normal operation (linear dotted behaviour), but also
anomalous situations such as: shading due to clouds affecting more to reference PV
modules than to PV array and vice-versa; string switched off; inverter saturation;
inverter stops; etc. In the particular case of this figure, the PV systems delivers to the
grid a power that is a 2.7% lower than the expected one (reported by the slope of the
linear behaviour) in normal operation. This value should be similar to the difference
between the energy injected into the grid during the test period (equation 17 with
60
PAC,EXP) and the one that is expected (equation 17 with PAC,SIM) when there is not
anomalous situations.
(a)
(b)
Figure 44
(a) AC clamps (the yellow ones) to measure AC currents up to 400 A. (b) AC clamps (the
red ones) to measure AC currents up to 2000A. These red clamps have such length that
allows measuring the AC current flowing through several large section cables.
Figure 45.
AC power response of a 110 kW nominal power PV array measured on-site with a wattmeter. The normal operation is represented by the linear dotted behaviour. The other
anomalous situations can be noticed in the graph: shadow over sensors or over PV array
modules due to clouds, strings switched off, inverter saturation and inverter stop.
Besides, the AC characterization with watt-meter has the advantage that the
accuracy of the procedure can be calculated: as the AC power is measured, the energy
injected into the grid during the test period can be calculated, EAC,EXP (equation 17 with
PAC,EXP). And this energy obtained from the integration of the power measurements can
61
be compared with the real energy injected into the grid, which is obtained just as the
difference between the readings of the energy meter at the end of the test and its initial
value, EAC,REAL. Typically, the differences between AC energy from energy meters and
the one from integration of AC power measurements is lower than 0.6%, which is
indicative of the accuracy of the measurements with the watt-meter.
Finally, as EAC,REAL, Gef and TC have been recorded, the global performance of
the PV system (values of PR and/or PRSTC) can be also calculated (section 2.1).
2.6.2. DC characterization.
A watt-meter is also able of measuring DC voltages and DC currents, the last
ones directly through its internal circuitry or through external shunt resistors class 0.5 to
get a good accuracy (Figure 46).
(a)
Figure 46.
(b)
(a) 300A/150mV shunt resistor added in a 100 kW inverter to measure accurately the DC
input current. (b) 1000A/150mV shunt resistor that is going to be added in an 800 kW
inverter to measure accurately the DC input current.
Then, the experimental DC power, PDC,EXP, can be calculated. If the operating
conditions Gef and TC are simultaneously measured, the DC power corrected to 25ºC,
PDC(Gef, 25ºC) can be obtained and plotted versus Gef in order to obtain the STC power
of the PV array, 𝑃∗ .
𝑃𝐷𝐶 (𝐺𝑒𝑓 , 25ºC) =
62
𝑃DC,EXP
1 + 𝛾(𝑇𝐶 − 𝑇𝐶∗ )
18
Figure 47 shows the result of doing this process, once the anomalous situations
have been disregarded and the irradiance range has been restricted to Gef >800W/m2.
So, the STC value of the PV array, 𝑃 ∗ , is given by the best fit to the line
𝑃𝐷𝐶 (𝐺𝑒𝑓 , 25ºC) = 𝑃∗
Figure 47.
𝐺
𝐺∗
19
DC power corrected at 25ºC of a 110 kW nominal power PV array measured on-site with a
watt-meter. The DC power records related to anomalous situations and to low irradiances
have been previously removed to obtain the DC peak power value at STC (the value
indicated by the red-dotted line in the “y” axis.
This peak power characterization can also be done by means of I-V tracers, as
presented in 2.2. Nevertheless, due to the module temperature, the dispersion
uncertainty of a single measurement tends to be large (a single I-V curve provides a
single value of maximum power). As the DC characterization with watt-meter must be
done at least during a whole sunny day, hundreds of points contribute to minimize the
uncertainty derived from module temperature.
63
Figure 48 shows the results of the DC characterization for 11 consecutive days
on a 700 kW nominal power PV array. Each individual point is the result obtained for
each single day. As can be seen, day 2 has not results because it was a very cloudy day
(effective irradiance –blue– and cell temperature –red– profiles are also plotted above).
All the results for every single day are inside a range of 1.3% of the mean power
obtained for all the days (red dotted line). Even in cloudy days with moments of sun we
have obtained acceptable results, but with a higher uncertainty: power from day 9 –good
result, very close to the mean value– and power from day 4 –not so good result, the
farthest from the mean– differs in 1.5%. Nevertheless, the mean of the power is almost
the same once the day 4 is removed (yellow-dotted line): there is a difference lower than
0.2 %. On the other hand, the results from very sunny days (days 3 and 8, for example)
are inside a range of 0.6% of the mean power. So, this DC characterization should be
done in sunny days to reduce the uncertainty as much as possible.
Figure 48.
Results of the P* characterization on a 700 kW nominal power array for 11 consecutive
days. The points represent the DC power characterization for each individual day. The
dotted lines represent the mean value for all the period: the red-one is the mean of all the
days and the yellow-one is the mean once the value from day 4 is removed. The profiles of
effective irradiance and cell temperature of days 2, 3, 4, 8 and 9 are also represented in the
graph.
64
Similar results have been obtained when this characterization has been
performed in different seasons of the year. Figure 49 shows the results of measuring P*
with a watt-meter in different sunny days of different months along a year, from April
2014 to November 2014, on a 5.8 kW PV array grid-connected. Again, each individual
point is the result obtained for a single day and all the points are inside a range of 0.6%
of the mean power (red-dotted line).
Figure 49.
Results of the P* characterization on a 5.8 kW nominal power array from April 2014 to
November 2014. Each point represents the DC power characterization for an individual
sunny day of the month considered. The dotted line represents the mean value for all the
measurements.
2.6.2.1. DC characterization integrated into SCADA.
Indeed, this DC characterization can be integrated into the SCADA of the
PV system. As DC instantaneous voltages and currents have not a difference of
phase as in AC, DC power can be obtained directly as the product of voltage and
current. So, these variables can be measured with a good accuracy through a
datalogger (it is not needed a watt-meter). The only cares that must be taken are,
first, to integrate shunt resistors at the inverter entry to measure the DC currents
65
and, second, to measure simultaneously these variables and the operating
conditions, Gef and TC.
This characterization has been done in a 5.8 kW PV system for a whole
year, measuring DC voltages, DC currents, Gef and TC each 30 seconds (grey
triangles). The selected days to perform the DC characterization have been those
that have more than 1.5 hours with an effective irradiance above 600W/m2. As
can be seen in Figure 50, all the resulting P* of the 238 days evaluated are inside
a range of ±2.1% of the mean power (grey continuous line, 5572 kW; the
difference between the maximum and the minimum values is 3.6%) and their
related uncertainty (calculated as two times the standard deviation) is 1.35% of
the mean. Only 7 days are out of a range of 1.5% of the mean power and this
figure reaches 25 days if the range is of only 1% of the mean.
As the signals in the SCADA of a PV installation use to record data only
each 15 minutes (this is the typical frequency configured in energy meters and/or
inverters to store data), we have repeat the same analysis but with the
instantaneous records each 15 minutes (blue squares). Now the mean power is
5571 kW (blue-dotted line), that differs 0.03% respect to the 30 seconds
analysis, and the related uncertainty is 1.44%). In this occasion, 10 days are out
of a range of 1.5% of the mean power and this figure reaches 27 days if the
range is of only 1% of the mean. That is: the only drawback of measuring with a
lower frequency in order to characterize the DC power is that the uncertainty is
slightly higher, but almost negligible.
66
Figure 50.
Results of the P* characterization on a 5.8 kW nominal power array from January 2014 to
November 2014. Power values and instantaneous values of meteorological conditions –Gef
and TC– are registered through a datalogger each 30 seconds or each 15 minutes. The daily
results related to 30 seconds records are plotted with grey triangles (the mean value is
represented with the continuous-grey line). The daily results related to 15 minutes records
are plotted with blue rectangles (the mean value is represented with the dotted-blue line).
The dotted-lines represent a deviation from the mean value of 1.5% (red one) or 1%
(yellow one).
2.6.3. Inverter characterization.
In the previous sections, it has been said that a watt-meter is able to measure AC
and DC power accurately. As all the signals are measured instantaneously and
simultaneously, it is possible to characterize the inverter efficiency.
Figure 51 shows an example of the relation between a 800 kW inverter
efficiency (PAC,EXP/PDC,EXP) and the load factor (PAC,EXP/PIMAX) –blue points– together
with DC voltage associated–red points–. AC clamps able of measure up to 2000A (see
Figure 44(b)) and a shunt resistor of 1000 A (see Figure 45 (b)) have been installed in
the inverter to measure AC and DC currents. Steps at observed efficiency reflect the
master-slave inverter configuration, and dispersion reflects differences on VDC and also
on internal inverter consumption, which is mainly due to refrigeration fans (sometimes
67
ON and sometimes OFF, especially at the low power range). The yellow triangles
represent the efficiency values derived from the manufacturer’s datasheet. And the blue
squares represent the experimental adjust of the efficiency once the parameters 𝑘0 ,
𝑘1 and 𝑘2 have been calculated (equation 15). Differences at low power between
observed values and those derived from datasheet are usual. This is due to the internal
inverter losses (mainly fans consumption) which use to be just given as additional
information in the datasheet but not considered at the efficiency load curve of the
manufacturer. So, the energy impact due to the fans consumption can be now included
in the simulations. Then, thanks to this improvement, the energy simulations will be
more accurate.
Figure 51.
Inverter efficiency versus load factor observed at an inverter of 800kW (dark-blue points).
They are also plotted: the experimental adjust to the measured values using equation 15
(clear-blue squares); the efficiency as reported by manufacturer (yellow triangles); and the
DC voltage measured that is related to each efficiency point (red points).
Once the power efficiency has been measured, the corresponding “European
efficiency”, EUR, can be calculated from particular power efficiency values (equation
20). It is important to notice that European efficiency from datasheet not necessarily
coincides with the energy efficiency observed during the test, because the corresponding
irradiance distributions could not be the same.
68
ηEUR  0,03 · ηp (0,05)  0,06 · ηp (0,1)  0,13 · ηp (0,2) 
 0,1 · ηp (0,3)  0,48 · ηp (0,5)  0,2 · ηp (1)
20
2.6.4. Portable SCADA tool (microSCADA).
The use of a professional watt-meter for on-site measurement is expensive when
PV systems are small, as it is in the case of BIPV. This is the reason why CLSENES has
implemented an alternative low cost testing kit for a detailed performance analysis of a
small PV installation, that is able to perform simultaneously measurements of effective
irradiance, cell temperature and DC and AC power, as well as calculate power quality
and performance indicators. It is based on low-cost SCADA data acquisition systems,
which offer real-time datalogging and flexible control of sensors.
This portable SCADA tool (hereafter microSCADA) is implemented by a
portable laptop computer and a set of smart sensors. Figure 52 shows the electrical
scheme of the microSCADA. All sensors are connected sequentially in a RS-485
network using MODBUS protocol. All currents sensors are split-core sensor for easy
and safe measurement setup at outdoor conditions.
The first group of sensors is related to the meteorological variables: incident
solar irradiance, solar cell temperature, environment temperature, wind speed and wind
direction, etc. As in small PV systems it is usually hard to find free room for a new PV
module of the same size as the ones installed, CLSENES has also implemented a
portable meteo-tower set with several sensors (Figure 53): a calibrated module that
provides both incident solar irradiance (Sensor 1) and solar cell temperature (Sensor 2),
a sensor of ambient temperature (Sensor 3), an ultraviolet sensor (Sensor 4) and wind
speed sensor (Sensor 5). The signals from all these sensors are measured through an 8
channel analog input module with MODBUS protocol (see Figure 52, blue device).
69
Figure 52.
Low cost portable SCADA scheme for DC and AC measurements, as well as operating
conditions measurements.
Figure 53.
Portable set with several sensors: a calibrated module that provides both incident solar
irradiance (Sensor 1) and solar cell temperature (Sensor 2), environment temperature
(Sensor 3), ultraviolet sensor (Sensor 4) and wind speed (Sensor 5).
The second group of sensors is composed by high-class DC and AC current and
voltage sensors and an AC power analyzer (Figure 54). This group is responsible of DC
and AC power measurement. The DC current and voltage sensors are directly connected
to the RS-485 network via MODBUS, while the AC current and voltage sensors are
connected to the AC power analyzer, which is connected to this network to send the
70
information about the AC power measurements (active power –kW–, phase of voltage
relative to current –cosPhi–, power factor –PF–, total harmonic distortion –THD–,
reactive power –kVAR–, apparent power –kVA–).
(a)
Figure 54
(b)
(c)
(a). DC current and DC voltage sensor. (b) AC current transformer. (c) 3 phases AC power
analyzer.
In order to show the information of the measurements with the microSCADA
system, a Graphical User Interface (GUI) based on Windows 8 style has been
implemented. The user can start this software like any other Windows application.
Figure 55 shows the initial screen, which presents the basic functions. The user can
access to any of the applications (METEO, INVERTER, INDICATORS) by clicking
the associated rectangular button. Then, a set of mathematical formulas executes the
digital to analog conversion (D/A) and sets the calibration values of the different
sensors. The buffer collects data from all the sensors, calculate the values for each one
and present these values in real time in the corresponding screen. Besides, these
readings are recorded in internal files for further processing (the monitoring time
interval can have a high resolution of up to 1 minute). It is worth to mention that the
accuracy of power measurement is not affected by the time of data readings because the
total time for sensors sequential addressing and data reading is negligible in comparison
to temporal change of the atmospheric outdoor conditions.
71
Figure 55.
Initial screen of the microSCADA application. It allows access to the different sub-menus:
METEO, INVERTER or INDICATORS by clicking the associated rectangular button.
All the input signals can be monitored in real time. The METEO submenu
(orange rectangle in the initial screen, Figure 56) and the INVERTER submenu (blue
rectangle in the initial screen, Figure 57) show graphically the values of the different
meteorological and AC and DC variables that are being measured: incident solar
irradiance, solar cell temperature, environment temperature, ultraviolet index, wind
speed, as well as DC input and AC output voltages, currents and power. In the
INVERTER submenu are also shown the efficiency of the inverter (EFF) and the losses
related to the DC/AC conversion (LOSSES).
Finally, the INDICATORS submenu (green rectangle in the initial screen, Figure
58) show graphically the values of performance indicators based on the previous
variables measured. The performance indicators implemented are:

Specific Yield (SY, in kWh/kWp), which is the ratio between the real AC
energy injected into the grid during the test and the total PV array peak
power.
72

Performance Ratio (PR, in %), which is the ratio between the SY and the
solar irradiation that has reached the PV array during the test period
(normalized by the irradiance at STC).

Performance Index (PI, in %), which is the PR but taking away thermal
losses and inverters losses associated to the DC/AC conversion. So, this
indicator is not site-dependent and nor season-dependent, unlike PR.

Capacity Utilization Factor (CUF, in %), ratio between the real AC
energy injected into the grid during the test and the theoretical AC
energy injected into the grid considering that the system is working at
STC all the time (it takes into account even blank periods like nighttime).
Figure 56.
Screen of the METEO submenu of the microSCADA application.
Figure 57.
Screen of the INVERTER submenu of the microSCADA application.
73
Figure 58.
Screen of the PERFORMANCE submenu of the microSCADA application.
In this submenu other performance indicators are also presented, as those related
to the AC signal (active and reactive power and their associated power factor and total
harmonic distortion), as well as the losses due to temperature, the losses due to DC/AC
conversion and the total CO2 savings during the test thanks to the use of the evaluated
PV system.
It must be also mentioned that this software has been developed taking into
account the different kind of users that can execute the application. More in detail, two
profiles of users have been developed. The first one is the “Administrator user profile”.
This kind of users can access databases, control sensors remotely and execute the main
program. The second one is the “Normal user profile”. In this last case, the user has not
enough priority and some functions are restricted. Basically, they can just execute the
main program.
2.6.5. Summary.
In this section the most powerful testing kit has been presented for the
performance analysis of a PV system. A professional watt-meter (together with AC
clamps installed at the output of the PV inverter and a DC shunt resistor installed at the
74
input of the PV inverter in order to measure currents with high-accuracy) in
combination with testing kit number 1 (reference modules, section 2.1) allow knowing
simultaneously and with a high degree of details the global performance of the
installation and the particular behaviour of both PV arrays and PV inverters, the main
devices of a PV system.
Records of Gef, TC and experimental AC power, PAC,EXP, allow to obtain the
power response, that is, the comparison between the real AC power, PAC,EXP, and the
simulated one, PAC,SIM which is calculated by using a very simple model (the same
model used by SISIFO, the PV simulation software developed in the PVCROPS
project) that is based on the information that is available on the manufacturer datasheet
(modules and inverters). The power response reports graphically about the behaviour of
the PV systems. Both the normal performance and the anomalous situations are clearly
represented in the power response and are easily identified and quantified. Besides,
these records are the needed to calculate the global performance of the PV system as it
was presented in section 2.1 (PR, PRSTC)
Records of Gef, TC and experimental DC power, PDC,EXP, allow to obtain the STC
power of the PV array. This characterization must be done in very sunny days and with
a high frequency of records (each minute or faster) to obtain accurate results with the
lower uncertainty. In fact, as the DC voltage and current have not a difference of phase,
DC power can be obtained directly as the product of these variables. So, if DC voltage
and current signals are included in the SCADA system, together with Gef and TC, the
DC characterization can be done daily in the routines of maintenance. The only cares
that must be taken are, first, to integrate shunt resistors at the inverter entry to measure
the DC currents and, second, to measure simultaneously these variables and the
operating conditions, Gef and TC.
Finally, simultaneous records of experimental AC and DC power, PAC,EXP and
PDC,EXP, allow to obtain the power efficiency of the inverter. Differences at low power
between observed values and those derived from datasheet are usual. This is due to the
internal inverter losses (mainly fans consumption) which use to be just given as
additional information in the datasheet but not considered at the efficiency load curve of
the manufacturer. So, the energy impact due to the fans consumption can be now
75
included in the simulations. Then, thanks to this improvement, the energy simulations
will be more accurate.
Several measurement examples from real PV installation tested on-site have
been also reported to show that the performance of the proposed testing kit is good and
that it provides a very detailed information about the behaviour of the system, from both
a global point of view of the whole PV system and a particular point of view of each PV
device (arrays and inverters). Besides, some of the results obtained in these tests have
been also presented and reported about the high accuracy of this testing kit.
An advantage of using a watt-meter is that inverters and PV arrays ranging from
kW to MW can be tested with the same equipment just by selecting the proper shunt
resistor size that must be installed at the inverter entry and the adequate AC clamps that
must be installed at the inverter output.
Nevertheless, the use of a professional watt-meter for on-site measurement can
be very expensive, especially when small PV systems have to be tested. So, a low cost
testing kit (microSCADA) has been developed as an alternative to perform a detailed
analysis of a BIPV. The microSCADA testing kit is implemented by a portable laptop
computer and a set of smart sensors (meteorological sensors and AC and DC sensors).
All these sensors are connected sequentially in a RS-485 network using MODBUS
protocol. In order to present to the user the results of the PV system analysis in an easy
way, a Graphical User Interface (GUI) has been implemented to show in an easy way
the results of the PV system.
76
3. CONCLUSIONS
PVCROPS team has been working from the beginning of 2013 on the definition of
quality testing procedures and the associated testing kits to evaluate PV installation. The
first result of this work was the definition of the technical specifications that a gridconnected PV system should fulfil in order to achieve (and even overcome) the energy
production expectation that was established at the initial design of the PV system
(Deliverable 2.2).
Once the technical characteristics of the PV installation have been clearly defined, it is
needed a set of quality control procedures whose objective is to check if the system that
has been finally installed fulfils these technical characteristics previously established.
The proposed quality control procedures have been already reported (Deliverable 9.3).
Both, technical specifications and quality control procedures have been reviewed and
validated by several experts of the PV community, but also by its successfully
applications in several commercial PV systems that are already working.
Now, this report shows what are the testing kits that have been designed to execute the
proposed on-site quality control procedures, what are the devices that should be used
and what are the devices that can even be implemented. These testing kits allow
evaluating on-site a whole PV system, both from a global point of view (general
performance of the installation) and from a particular point of view (individual
performance of PV modules, PV arrays and/or PV inverters). Some examples of the use
of the different testing kits during the evaluation of real PV systems have been also
reported.
The next table summarizes the testing kits that have been presented in this report, what
are their objectives, what are the main devices needed in each testing kit and what are
their main characteristics.
77
TESTING KIT
1
Reference PV modules
2
I-V curve tracer
3
Climatic box
4
PID tester
5
Hot spot tester
6
Full PV system tester
OBJECTIVES
 Global performance of the PV
system
 Calculation of PR, PRSTC
 Electrical behaviour of PV
modules/arrays
 Detection of malfunction of PV
modules
 Electrical behaviour of PV
modules/arrays
 At STC
 At low G
 Dependence on Tc
 Detection of PV modules
affected by potential induced
degradation (PID)
 Detection of overheating in PV
modules due to any anomaly
DEVICES
Reference modules
Shunt resistor
Datalogger
Single/twin self-made I-V tracer based on
IGBT
 Reference PV modules
 Shunt resistor class 0.5
 Differential 4 channel oscilloscope
 Self-made climatic box
 Single I-V tracer
 Datalogger
 Temperature sensors (PT1000)
 2 irradiance sensors (reference cells)
a) High voltage source + I-V tracer
 Power degradation
b) Current source + CCD camera
 Electroluminiscence image
c) I-V tracer
 I-V shape/power
d) Current source +Blanket/cardboard
 Dark I-V curve
e) Voltmeter + “T” connectors
 Operating voltage
CHARACTERISTICS
 Modules of the same type of the
PV system
 Shunt resistor class 0.5
 Based on IGBT and capacitors
 Scalable for measuring from PV
modules to large PV arrays
 Single or twin
 Manual or automatic
a) Hand thermographic camera
b) Self-made PV drone
 Visual and termographic camera
 High resolution digital camera
b) Based on free-software and free
hardware, fast, automatic flight
missions. If includes a digital
camera for aerial photogrammetry
(shadow analysis)




 Global performance of the PV
a) High precision watt-meter + AC clamps +
system
shunt resistor + reference PV modules
 Calculation of PR, PRSTC
b) Self-made microSCADA
 Detection of anomalous operation
 High-accuracy AC and DC sensors
 Performance of PV arrays.
 Gef and TC sensors
 Performance of PV inverters
78
 “Thermally isolated”
 Fans to homogenize temperatures
 Accurate internal reference cell
a) Expensive, only laboratory
b) Expensive, dark room
c) Sunny day, slow
d) Cheap, any moment
e) Cheap, sunny day, fast
a) Very expensive
b) Cheap, conceived for small PV
systems, cheaper. Based on
MODBUS protocol.
4. REFERENCES
[1]
Global market outlook for photovoltaics 2013-2017. May 2013. European
Photovoltaic IndustryAssociation (EPIA).
Available at www.epia.org/news/publications/
[2]
F. Martínez-Moreno, N. Tyutyundzhiev. Good and bad practices. Manual to
improve the quality and reduce the cost of PV systems. PVCROPS project
(available at http://www.pvcrops.eu/modelling-and-simulation).
[3]
IEC 61724: 1998. Photovoltaic system performance monitoring – Guidelines for
measurement, data exchange and analysis.
[4]
EN 50380: 2003. Datasheet and nameplate information for photovoltaic
modules.
[5]
IEC 60891: 2009. Photovoltaic devices. Procedures for temperature and
irradiance corrections to measured I-V characteristics.
[6]
IEC 61853-1: 2011. Photovoltaic (PV) module performance testing and energy
rating. Part 1: Irradiance and temperature performance measurements and power
rating.
[7]
IEC 60904-2: 2007. Photovoltaic devices. Part 2: Requirements for reference
solar devices.
[8]
IEC 60904-5: 2011. Photovoltaic devices. Part 5: Determination of the
equivalent cell temperature (ECT) of photovoltaic (PV) devices by the opencircuit voltage method.
[9]
R. Auer, U. Jahn, B. Buerhop-Lutz. Infrared analysis of PV modules for
improving quality. 22nd EUPVSEC. 2519-2522, Milan, Italy. 2007.
[10]
M.B. Nieto, J.P. Silva, F. Chenlo. Diferencia de temperature de operación de
módulos FV según tecnología de células y tipo de encapsulado. XIV Congreso
Ibérico y IX Congreso Iberoamericano de Energía Solar. 2: 727-732. Vigo,
Spain. 2008
[11]
G.M. Tina, R. Abate. Experimental verification of thermal behaviour of
photovoltaic modules. 14th IEEE Mediterranean Electrotechnical Conference.
MELE-CON 2008, 579-584.
[12]
IEC 61215: 2005.Crystalline silicon terrestrial photovoltaic (PV) modules Design, qualification and type approval.
79
[13]
IEC 61646: 2008. Thin-film terrestrial photovoltaic (PV) modules - Design,
qualification and type approval.
[14]
J. Leloux, L. Narvarte, D. Trebosc. Review of the performance of residential PV
systems in Belgium. Renewable and sustainable energy reviews 16, 1. 178-184
(2012).
[15]
J. Leloux, L. Narvarte, D. Trebosc. Review of the performance of residential PV
systems in France. Renewable and sustainable energy reviews 16, 2. 1369-1376
(2012).
[16]
J. Neuenstein, C. Podewils. Los módulos y sus curvas. Photon. La revista de
fotovoltaica, 54-71, 2009
[17]
E. Caamaño, E. Lorenzo, R. Zilles Quality control of wide collections of PV
modules: lessons learned from the IES experience. Progress in Photovoltaics:
Research and Applications 1999; 7: 137–149.
[18]
G. Nofuentes, J. Aguilera, R. L. Santiago, J. de la Casa and L. Hontoria. A
Reference-module-based Procedure for Outdoor Estimation of Crystalline
Silicon PV Module Peak Power. Progress in Photovoltaics: Research and
Applications 2006; 14:77–87.
[19]
G. Blaesser, D. Munro. Guidelines for the Assessment of Photovoltaic
Photovoltaic Plants. Document C. Initial and Periodic Tests on PV plants. Joint
Research Centre Ispra. European Commission 1995.
[20]
J. Siemer, C. Haase. Stress relief? High-voltage stress is a widespread cause of
PV system performance degradation that has only recently come to light. Photon
International. January 2011, 136-143
[21]
P. Welter. These cells are simply too good. An explanation of poor yields from
SunPower’s modules. Photon International. May 2006, 68.
[22]
I. Rutschmann. Siempre se aprende algo nuevo. También los módulos de
Evergreen presentan un comportamiento de polarización. Photon España. Enero
2008, 110-111.
[23]
S. Koch, C. Seidel, P. Grunow, S. Krauter, M. Schoppa. Polarization effects and
test for crystalline silicon cells. 26th EUPVSEC September 2011, 1726-1731.
[24]
J. Berghold, O. Frank, H. Hoehne, S. Pingel, B. Richardson, M. Winkler.
Potential Induced Degradation of solar cells and panels. 25th EUPVSEC
September 2010, 3753-3759.
80
[25]
P. Hacke, K. Terwilliger, R. Smith, S. Glick, J. Pankow, M. Kempe, S. Kurtz, I.
Bennett, M. Kloos. System Voltage Potential-Induced Degradation Mechanisms
in PV Modules and Methods for Test. 27th IEEE PVSC June 2011, 814-820.
[26]
I. Rutschmann. Power losses below the surface. Laboratory testing is needed to
diagnose the reasons for negative impact of PID on solar modules. Photon
International. November 2012, 130-137.
[27]
J. Berghold, S. Koch, A. Böttcher, A. Ukar, M. Leers, P. Grunow. Potentialinduced degradation (PID) and its correlation with experience in the field.
Photovoltaics International. 19th edition. 85-92.
[28]
Advanced Energy. White Paper. Understanding Potential Induced Degradation.
[29]
S. K. Chunduri. Out of the dark and into the light. A new device aims to take
module luminescence imaging out of the factory and into the field.
[30]
IEC draft 62804. System voltage durability test for crystalline silicon modules –
Qualification and type approval.
[31]
R. Moretón, E. Lorenzo, J. Leloux, J.M. Carrillo. Dealing in practice with hotspots. 29th EUPVSEC, 2014, 2722-2727.
[32]
www.thermal.com/thermal-cameras/
[33]
www.flir.com/flirone/
[34]
M. Fuentes, G. Nofuentes, J. Aguilera, D.L. Talavera, M. Castro. Application
and validation of algebraic method to predict the behaviour of crystalline silicon
PV modules in Mediterranean climates. Solar Energy 81, pp 1396-1408 (2007).
[35]
M. Jantsch, H. Schmidt, J. Schmidt. Results of the concerted action on power
conditioning and control. 11th EUPVSEC, 1992, 1589-1592.
81