Loughborough University
Institutional Repository
Performance of photovoltaic
modules in a temperate
maritime climate
This item was submitted to Loughborough University's Institutional Repository
by the/an author.
Citation: WILLIAMS, S.R., GOTTSCHALG, R. and INFIELD, D.G., 2003.
IN: Proceedings of 3rd World Conference on Photovoltaic Energy Conversion
2003, Vol.2, Osaka, Japan, 11th-18th May, pp. 2070-2073.
Additional Information:
•
This
conference
paper
[ c
IEEE]
is
also
available
at:
http://ieeexplore.ieee.org/xpl/mostRecentIssue.jsp?punumber=9136.
Personal
use
of
this
material
is
permitted.
However,
permission
to
reprint/republish this material for advertising or promotional purposes or
for creating new collective works for resale or redistribution to servers or
lists, or to reuse any copyrighted component of this work in other works
must be obtained from the IEEE.
Metadata Record: https://dspace.lboro.ac.uk/2134/8207
Version: Published
Publisher:
c
IEEE
Please cite the published version.
This item was submitted to Loughborough’s Institutional Repository
(https://dspace.lboro.ac.uk/) by the author and is made available under the
following Creative Commons Licence conditions.
For the full text of this licence, please go to:
http://creativecommons.org/licenses/by-nc-nd/2.5/
May 11-18,2003 Osah. Japan
3rd World Conference on Photovoltaic Energy Conversion
6P-D5-19
PERFORMANCE OF PHOTOVOLTAIC MODULES IN A TEMPERATE MARITIME CLIMATE
S.R. Williams, R. Gottschalg, D.G. Infield
CREST, Centre for Renewable Energy Systems Technology, Department of Electronic and Electrical Engineering,
Loughborough University, Loughborough, LE1 1 3TU, UK, Tel: +44-1509-228148,
Fax: +44-1509-61003 1, e-mail: [email protected]
ABSTRACT
The energy production of photovoltaic modules is a
topic of growing importance. Under real operating
conditions two modules with the same name-plate
efficiency may have very different energy productions,
despite being installed at the same site. In this paper we
investigate the performance of a variety of modules
(crystalline, polycrystalline and single and multi junction
amorphous silicon and cadmium telluride) installed in
Loughborough in the UK. The paper investigates the
reasons for the significant difference in performance of
modules of the same technology and efficiency. The
analysis is carried out on the basis of influence of device
temperature, magnitude of irradiance, incident spectrum
and age of device.
INTRODUCTION
1
Photovoltaic devices are typically sold on the basis of
standard test condition efficiency. This is not necessarily a
good measure for the realistic energy production. For
example, it was shown that in the UK climate, modules of
the same material and efficiency could have a very
different energy production [I]. The variation between
different systems is more often than not analysed as being
due to the different materials types. This paper shows that
the photovoltaic material is only one factor affecting the
power production of a PV system, while manufacturing
issues are equally important.
The data used for this study were collected using the
outdoor measurement system installed at CREST in
Loughborough, UK. Loughborough is at latitude of 53"
and experiences sunny skies for less than 10% over the
course of a year. The temperature is very temperate and
lacks extremes. This system measures a comprehensive set
of data in 10 minute intervals: full device I-V
characteristic, ambient and device temperature, irradiance
and incident spectrum. The latter is measured using a
custom made spectroradiometer, which measures the
spectral irradiance for wavelengths from 300 to 1700 nm.
Currently a variety of modules is monitored employing
different photovoltaic materials and technologies. This
paper investigates crystalline (c-Si) and polycrystalline
silicon (p-Si) devices, single, double and triple junction
amorphous silicon devices (a-Si) and cadmium telluride
(CdTe). The ages of the devices range from months to
years.
In order to understand the device performance, four
factors are considered: operating temperature, irradiance,
incident spectrum and age of the device.
SEASONAL PERFORMANCE
2
The measurement system at CREST was re-developed
and has been operating continuously since October 2002.
Data presented here is based on the new system only. The
work concentrates on the analysis of this new set of data.
The first parameter to be investigated is the monthly
variation of the efficiency, as shown in Figure 1. In order
to make the changes within the devices comparable, the
efficiency was plotted against the time. Ignoring the data
for September, which was generated in the last 3 days only
and thus is not representative, two trends are apparent, the
CdTe device and the c-Si device exhibit an increased
efficiency in winter, while all amorphous samples exhibit
a decrease in efficiency. Similar behaviour was reported
elsewhere [2-41. The aim of this work is to investigate the
reasons for this behaviour. The second parameter to be
investigated is the short circuit current, which is shown in
Figure 2. Here a similar trend is observed, with an
increased magnitude for the a-Si devices. This clearly
shows that some changes in the spectrum are occurring.
The CdTe device is not benefiting from this, most likely
due to a relatively weak blue response of these devices.
The c-Si device exhibits hardly any variation.
Another indicator of the reasons for the behaviour is
the fill factor, which is investigated in Figure 3. There is
no significant change for any of the devices. There is,
1.15
1.15 ........................................................................................
1.1
...........................
P
//
i
;
1.1
".."
-
Sep42
a-SI (Tnple Junction)
CdTe
1
.
1
Oct-02
Dee42
Jan43
Mar43
May43
Time
Figure 1: Variation of the Relative Monthly Efficiency
2070
Poster
,
Sep-02
Oc142
Jan43
Des42
Mar43
May43
Time
Figure 2: Variation in the Relative Short Circuit Current
May 11-18, 2003 Osaka. Japan
3rd World Conference on Photovoltaic Energy Conversion
"."
,
Sep42
Oct42
Dec-02
Jan43
Mar43
May43
Time
Figure 3: Variation of the Relative Monthly Fill Factor
however, an increase of the single junction a-Si for the
warmer months, and a slight decrease for the double
junction. This particular double junction is still in the
initial phase of degradation, hence this behaviour is
expected. The behaviour of the single junction is as seen
in other locations [5].
3
DETAILED ANALYSIS
The variation of the short circuit current shown in
Figure 2 could be due either to thermal changes or to
changes in the spectrum.
3.1
Spectral Effect
The effects of the spectrum can be investigated by
plotting the spectrally useful fraction of the total irradiance
relative to the average of all values. This is done in Figure
4 where the cut-off wavelengths are taken to be 780 for aSi, 900 for CdTe, 1100 for c-Si and 1360 for CIGS. The
fraction is then plotted against the averages of all values,
in order to investigate the magnitude of this effect.
The useful fractions for the different technologies have,
as demonstrated in Figure 4, a significant seasonal change.
The variation is most pronounced for a-Si, where it varies
+5 to -9% with respect to the overall average. This
magnitude tallies with the results of Hirata and Tani [6]
for a Japanese location, who reported a similar magnitude
of the variation. The reduction of 9% in winter will not
affect the annual production too significantly, as it is the
time where hardly any irradiance is available; the
maximum for a given day can be below 50 W/m2. The
positive effect in summer time will, however, have a
significant impact on the annual production, as this
coincides with the maximum irradiance available. The
effect is less pronounced for CdTe (+4 to 4%),
c-Si (+2%
to -3%) and hardly noticeable for CIGS (+1.5% to 1.5%). The observation of this effect will, obviously,
depend on the quantum efficiency of specific devices, and
thus their manufacturing technology. However, it does
give an indication of the magnitude of this effect. On top
of the primary effect, which is directly linked to the device
efficiency, a secondary effect is noticeable for multijunctions. This secondary effect depends on the spectral
composition of the light and will cause a mismatch when
the spectrum deviates from the design spectrum. This
effect can be seen when correcting the short circuit current
for the intensity, i.e. dividing by the irradiance. This can
be done using the useful irradiance (UI), i.e. irradiance
within the spectrally useful range. The results of doing this
for a double junction a-Si device are shown in Figure 5
(diamonds). A second order polynomial was fitted in order
to guide the eye through the shape of the measurements.
All measurements here are presented with respect to the
average of all measurements. The average photon energy
(APE) is used as an indicator for the blueness of the
incident spectrum. The current mismatch will cause a
bending of the generated short circuit current for
increasing blueness of the light. Depending on the exact
method this can increase the efficiency up to a design
spectrum and then generally fall again. This effect can
cause several percent difference in the performance, but
will be dominated by the primary effect. This is
demonstrated in Figure 5, where as a comparison the
spectrally uncorrected irradiance is used (crosses),
resulting in a much stronger dependence on the blueness
of the light, as could be expected.
1.2
1.3
14
1.5
16
1.7
APE [eV]
-
1.8
1.9
2
2.1
2.2
Figure 5: Comparing Primary and Secondary Spectral
Effects for a Double Junction Device
3.2
Thermal Effect
The thermal effect generally speaking is well
I .04
understood, except for a-Si devices. The effect of
1.02
temperature is often cited as the main influence for a
performance increase of a-Si devices in summer time, as
5 '
Y
there are sometimes positive temperature coefficients
.a0.98
reported. A positive temperature effect is found for some
2 0.96
of the modules operated at CREST before correcting for
0.94
primary spectral effects. After correcting for primary
0.92
spectral effects, the effect of device temperature is
0.9 t,~FM~~,~,~AsoNd,~~M~M,~,~*.~~N.d,~P;II~,~~E.dN.dl'FM*MI1I~*SoN~I~F;;*MI'I~*.I.OLh~
negative, as could be expected. Hence, the efficiency was
1998
1999
2000
2001
2002
Time
corrected to the efficiency of AM1.5 equivalent irradiance
in a first step. In the second step, the temperature effect
Figure 4: Seasonal Variation of the Useful Fraction of
was determined by filtering for irradiance values higher
Different Device Technologies
-
Poster
2071
3rd World Conferenceon PhotovoltaicEnergy Conversion
May 11-18,2003 Osaka, Japan
than 950 W/m2 and less than 1050 W/m2. The results of
this procedure are shown in Figure 6 . This treatment of the
data reduces the effects of other environmental conditions
on the temperature effect, as the range of spectra as well as
irradiance levels is very narrow.
1.00 ...............................................................................
~~
0.90
j
0.80
:
0.70
0.60i
0.50 4,
0.40
0.30
0.20 j
0.10
o,oo
j
' :.m
:...+:
~
+;.:.:.>
Figure 6: Thermal Factors for some Devices
A single junction a-Si (aSi3), two double junction
devices (aSi4, aSi5), one CdTe are compared with a single
crystalline silicon device. It is apparent that these wider
band-gap materials have a much smaller thermal effect
than c-Si. There is also a difference within the a-Si
devices, as one double junction (aSi5) is nearly twice as
affected as the other (aSi4).
3.3
Irradiance Effect
The irradiance is probably the property changing the
most. In the UK, a relatively high percentage of energy is
generated at low intensities, while in other locations low
light conditions are not very significant. There also
appears to be an inherent difference between conventional
cells (c-Si, p-Si) and thin film cells. Conventional cells
typically benefit from increases in irradiance, while thin
films often suffer from a decrease in operating efficiency.
The low light irradiation response of thin film devices was
identified by Eikelboom and Jansen [3]as the major
contributor to the good performance of thin film devices in
particular. In contrast to this, a simulation carried out for
the UK [7] identified low irradiance losses as the
dominating loss mechanism for one certain a-Si material.
The reason for this apparent difference is discussed below.
The variation of the spectral aperture efficiency is
plotted against the irradiation in Figure 7. The spectral
efficiency is used because otherwise the spectral changes
would distort the results. In order to make the spectral
-
o.y;
p
-
G~l~Iw/m21
Figure 7: Influence of Irradiance on Efficiency
2072
Poster
irradiation levels for the different materials comparable,
the AM1.5 irradiation with the equivalent spectral
irradiation is used as the x-axis. The aperture efficiencies
shown here were determined in the first month of the
operation of these devices, excluding degradational effects
but resulting in limited irradiance ranges in some cases.
The use of the aperture area obscures the absolute
efficiency, but the intention is to show trends and not
compare devices and thus this definition is acceptable.
This use of aperture area efficiency is the reason for the
overall magnitude of the efficiency, as most of the devices
are small area samples with significant amounts of
inactive glass around them. The difference in the
technologies is quite apparent. Some devices (aSi5,
CdTe2) improve with increasing irradiance, while others
exhibit a reduction in efficiency. The reasons for this are
not immediately apparent and are investigated further. The
conventional devices in this test exhibit an increase in
their performance. There are several effects which could
result in a decrease of efficiency with increasing
irradiance, some due to production effects, some due to
measurement arrangements. The latter could be e.g. the
result of measuring devices with relatively high
capacitance too fast, so that the dynamic current will be
affected. This was, however, excluded in this study - the
measurement of 200 I-V points takes about 4 seconds,
which is certainly not too fast. It does not appear to be
technology specific, as one double junction (aSi4) exhibits
a slight decrease with intensity, while another (aSi5) has
an increased performance.
The difference in the behaviour could be due to
differences in the resistances apparent within the devices.
ASi5 is a device which has a relatively low parallel and
series resistance (RP and Rs, respectively), while aSi4 has
an extremely high, nearly infinite parallel shunt resistance
but exhibits a relatively high contact (series) resistance.
Thus the effects of series and shunt resistance on a
simulated device are investigated. A single diode model
was assumed for these simulations; the required
parameters were extracted from dark laboratory
measurements. Based on these results, the parallel and the
series resistance were varied in order to identify the
influence of these properties on the device performance.
The energy production was then divided by the energy
production in the ideal case, i.e. no series resistance and
infinite shunt resistance. The results were plotted against
the short circuit current, which relates to irradiance. The
results of these simulations are shown in Figure 8.
There is a difference in the influence of Rp and Rs on
the performance, which could explain the difference in the
performance of similar devices illustrated in Figure 7. The
areas of influence are very different for RP and Rs. The
series resistance is significantly more important for high
irradiance conditions, while its influence is marginalized
for low irradiance conditions (it will be zero for a current
of zero). The parallel resistance, on the other hand, has a
diminishing influence on the energy losses at high
irradiance conditions, but can have a devastating effect for
low irradiance conditions. Generally speaking, Rp can be
measured using the slope of the I-V characteristic at short
circuit conditions, thus the denominator in this graph
should be the resistance at short circuit current Rsc. This
point is important because in the case of a-Si this is not
only a measure of the shunt resistance but it is also
strongly influenced by the lifetime of charge camers of
the material. Hence it is an a-priori indicator for the
3rd World Conference on Photovoltaic Energy Conversion
May 11-18, 2003 Osaka, Japan
quality of a device production. Using these results and
investigating the I-Vs of aSi4 and aSi5 indeed shows that
the Rsc is relatively low (while the Rp extracted from the
dark measurements is reasonable). Hence the difference in
device behaviour of aSi4 and aSi5 is related to production
quality rather than material specifics. The loss of
efficiency towards higher irradiances can, on the other
hand, be attributed to a relatively high series resistance,
which is partially due to the use of a transparent oxide as a
front contact.
\\
O
0.5. t
t
0.41
0.2
'._
...
physically unavoidable losses, such as availability of
spectrally useful irradiance or reduction in the efficiency
in multi-junctions if the incident spectrum is not equal to
the design spectrum.
The material specific effects are important and can
explain a significant proportion of the device behaviour
but other, technological effects are equally important. In
this work, the effect of series and shunt resistance was
investigated in detail, outlining the importance of these
parameters in different operating environments. It was also
shown that the thermal behaviour of devices can be
substantially different even within one category of
devices.
5
R. Gottschalg is supported by the Engineering and
Physical Science Research Council through contract No.
GR/N04232.
..,
6
---- Rr=0.075Il
,/,
t
031
7
Q 0.2
2
kc [AI
-
Figure 8: Influence of Series Resistance (top graph) and
Shunt Resistance (bottom graph) on Power Production
This discussion explains the differences in the
statements given earlier, that although the low light
performance of thin film cells is one reason for
outstanding performance, it still can be a significant loss
factor. The magnitude of these effects will depend
strongly on the location, and the simulation identifying
low light conditions as the dominant factor in this case
was carried out for Loughborough. In winter time, the
maximum irradiance seen on any particular day can be as
low as 45 W/m2. Even a device performing well in low
light conditions will have losses in this region, and
depending on the magnitude of the short circuit resistance,
this will result in more or less significant losses.
4
ACKNOWLEDGEMENTS
CONCLUSIONS
REFERENCES
[ l ] C. Jardine, G. J. Coniber, and K. Lane, "PVCOMPARE: Direct Comparison of Eleven PV
Technologies at Two Locations in Northern and
Southern Europe," Proceedings of the 17th European
Photovoltaic Solar Energy Conference, Munich,
2001).
[2] J. A. del Cueto, "Comparison of Energy Production
and Performance from Flat-Plate Photovoltaic Module
Technologies Deployed at Fixed Tilt," 29th IEEE
Photovoltaic Specialists Conference, New Orleans,
(IEEE, 2002).
[3] J. A. Eikelboom and M. J. Jansen, Characterisation of
PV Modules of New Generations. Results of Tests and
Simulations, (ECN, Petten, NL, 2000).
[4] M. Camani, N. Cereghetti, D. Chianese, and S.
Rezzonico, Comparison and Behaviour of PV
Modules, (Stephenson, Felmersham, 1998).
[5] R. Gottschalg, J. A. del Cueto, T. R. Betts, S. R.
Williams, and D. G. Infield, "Investigating the
Seasonal Performance of A-Si Single- and Multijunction Modules," Proceedings of the 3rd World
Conference on Photovoltaic Energy Conversion,
Osaka, (2003).
[6] Y. Hirata and T. Tani, "Output Variation of
Photovoltaic Modules with Environmental Factors - I.
The Effect of Spectral Solar Radiation on Photovoltaic
Module Output," Solar Energy, 55,463 (1 995).
[7] M. J. Holley, R. Gottschalg, A. D. Simmons, D. G.
Infield, and M. J. Kearney, "Modelling the
Performance of a-Si PV Systems," European
Photovoltaic Solar Energy Conference, 16th, Glasgow,
(London, 2000).
The energy production of photovoltaic devices is
dependent on a number of different parameters. The
material employed is only one effect of many and should
not be over-interpreted as the dominant factor. There are
some effects which can be attributed to material specific
properties such as the utilisation of spectrum and the effect
of temperature on device performance. However, even in
these properties there is a significant difference between
devices of the same material. These will result in
Poster
2073