the effects of spectral evaluation of c-si modules
TRANSCRIPT
PROGRESS IN PHOTOVOLTAICS: RESEARCH AND APPLICATIONS
Prog. Photovolt: Res. Appl. 2011; 19:1–10
Published online 15 July 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/pip.973
RESEARCH ARTICLE
The effects of spectral evaluation of c-Si modules
Michael Simon*,y and Edson L. Meyer
Fort Hare Institute of Technology, University of Fort Hare, Private Bag x1314, Alice 5700, South Africa
ABSTRACT
Outdoor spectral measurements in sub-Sahara, South Africa in particular have not been documented probably due to lack of
data or lack of proper methodologies for quantifying the spectral effects on photovoltaic performance parameters.
Crystalline-Si modules are widely used for system designs in most cases based on the data provided from indoor
measurements or from maritime northern hemispheric conditions. As a result of this, PV systems fail to deliver their
intended maximum power output. In this study, a methodology for quantifying outdoor spectral effects of c-Si modules
commonly found in the African continent is presented. The results of three crystalline-Si modules indicate that these
modules are affected as the spectrum shifts during seasons although these devices are perceived (without outdoor data) that
their performance is not influenced by the seasonal changes in outdoor spectrum. Copyright # 2010 John Wiley & Sons,
Ltd.
KEYWORDS
outdoor spectrum; crystalline-Si modules; spectral effects
*Correspondence
Michael Simon, Fort Hare Institute of Technology, University of Fort Hare, Private Bag x1314, Alice 5700, South Africa.
E-mail: [email protected]
Received 19 May 2009; Revised 12 September 2009
1. INTRODUCTION
It is becoming certain that the impact of PVmodules on our
daily energy needs is increasing tremendously. Their place
of application has also broadened, thereby calling for the
need to understand their performance. Outdoor monitoring
procedure of PV modules should definitely continue since
the operational conditions are different to the conditions on
which rated parameters are based. Understanding the
performance of different PV devices under various
environmental conditions assists in obtaining useful
information that would help system designers in predicting
energy output, making them more reliable and effective.
One such parameter which is not clearly understood and
affects output electrical parameters is the continuous
change of the outdoor spectrum.
2. EXPERIMENTAL PROCEDURE
PVmodules used in this study are placed on a fixed rack on
the roof top of our Research Centre, clear of any
obstruction that might cause shading. The tilt angle of
Copyright � 2010 John Wiley & Sons, Ltd.
328470 facing north is used since this is the latitude of
Alice, in South Africa. Spectral measurements were taken
using the EPP2000 (portable) fibre optic spectrometer. This
instrument is capable of making various types of spectral
measurements in the UV–VIS–NIR (350–1200 nm)
ranges. The EPP2000 provides 2048 wavelengths for each
scan over the pre-configured range. Range and coarse
resolution are determined by the installed spectrometer
grating groove density. EPP2000 is interfaced to a
computer via a digital parallel port using a 25-pin flat
ribbon cable. Outdoor measurements for the current–
voltage data were facilitated by a current–voltage tester
employing a power supply unit as load. Figure 1 illustrates
the module layout and the apparatus for measuring
environmental conditions.
The concept of Useful Fraction (UF) as used according
to Ref. [1] is a useful parameter for quantifying the spectral
influence on PV devices. One major advantage in
comparison with other concepts is its ability to give direct
feedback of a particular device performance since it is
device dependent. One major assumption made in this
concept is that the spectral response (SR) within the device
wavelength range is 100%. In fact not all of the incident
1
Figure 1. PV module layout installed at the Institute of Tech-
nology, South Africa at tilt angle of 328470 facing north.
Spectral evaluation of c-Si modules M. Simon and E. L. Meyer
energy that is within the device’s spectral range will be
converted to useful energy. To compensate for this, it was
therefore necessary to introduce what is referred to as the
Weighted Useful Fraction (WUF) [2]:
WUF ¼ 1
Gtot
ZEg
0
GðlÞdðlÞSRðlÞ (1)
where G(l) is the spectral irradiance encountered by a
given solar cell.
From Equation (1), WUF is defined as an average SR
weighted on a particular solar spectrum. As a quick
example, at 350 nm for an a-Si device, 20% of the incident
energy density (W/m2/nm) received contributes to the
electron–hole (e–h) creation and for mc-Si at the same
wavelength, 60% is used to create e–h pairs. But the UF
considers that at each wavelength, all the energy received
contributes to the e–h, which is one of the shortcomings
observed from this methodology. The idea of using WUF
Figure 2. Illustration of the change in outdoor spectrum at different t
as a % at each w
2 Pr
was to address these short falls which tend to overestimate
the overall device SR. Figure 2 shows the outdoor spectrum
at different times of the day and the SR of the mc-Si
module. Also illustrated is the effective spectrum perceived
(corrected) by the mc-Si module.
From Figure 2 it is observed that the per cent of the
useful energy converted to output energy at each
wavelength differs. This results in the total integrated
energy density within the device spectral range to be lower
than when the % SR at each wavelength is not considered.
The effective spectrum perceived by the module is
calculated with % SR multiplied by the actual spectrum
at that wavelength.
3. RESULTS AND DISCUSSION
The variation of the incident spectrum has an impact on the
modules short circuit (Isc) current [4]. The influence of theback-of-module temperature (Tmodule) and the change in
irradiance also greatly affect the Isc of the module.
Determining the effect of spectral variations requires the
elimination of these factors by plotting each one of them
with the modules Isc. Normalization of Isc was achieved bydividing the module’s Isc with the total irradiance within
the device spectral range (Gspectral range).
3.1. Spectral influence on mc-Si module’sperformance
The commonly adopted correlation existing between the
module’s Isc and back-of-module temperature is of the
imes of the day. Also shown is the mc-Si spectral response curve
avelength [3].
og. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
0.00
0.05
0.10
0.15
0.20
0.25
65605550454035302520
Temperature (oC)
φ spe
ctra
l ran
ge (A
/Wm
-2)
-1
0
1
2
3
4
200400
600800
1000
20
30
40
50
I sc(A)
Irradiance (W/
m²)
Temperature
(°C)
Figure 3. Correlation between irradiance corrected Isc and Tdevice. Insert is a 3D graph of variations of the 3 parameters.
M. Simon and E. L. Meyer Spectral evaluation of c-Si modules
form Isc ¼ ðC0 þ C1TdeviceÞ � GSpectralRange [5]. Firstly
the relationship between Isc=GSpectralRange ¼ fSpectralRange
(irradiance corrected Isc) is plotted against back-of-
module temperature. The empirical coefficients C0 and
C1 are obtained. This correlation does not clearly show
the device spectral variations, but this can be
overcome by fitting the data to equation of the form:
fSpectralRange ¼ ðC0 þ C1TdeviceÞ � f ðWUFÞ. This procedureallows the spectral variations to be clearly analysed. Figure 3
(insert) illustrates the dependence of irradiance corrected Iscwith temperature.
0.00
0.05
0.10
0.15
0.20
0.25
00.8120.810.8080.806
φsp
ectr
al r
ange
(A/W
-2)
Figure 4. Variation of fSpectralRang
Prog. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
The variation of fSpectralRange with temperature is linear
as confirmed in Figure 3. From the Figure 3 (insert), a
linear correlation between Isc, temperature and irradiance
is vivid. Since fSpectralRange is the ratio of Isc to Gspectral range
and both Isc and Gspectral range increases with temperature,
the ratio (Isc/Gspectral range) should exhibit a linear
dependence with temperature. To further understand the
behaviour of fSpectralRange with outdoor spectrum, a plot of
fSpectralRange versus WUF is shown in Figure 4.
Observed from Figure 4 is a direct correlation of
fSpectralRange andWUF. This correlation can be explained as
.814 0.8220.820.8180.816
WUF
e against the module’s WUF.
3
Figure 5. The effects of temperature and irradiance as a function
of fSpectralRange.
Spectral evaluation of c-Si modules M. Simon and E. L. Meyer
follows: An increase in fSpectralRange as WUF increases is
due to the fact that each WUF value corresponds to a
particular irradiance value and distribution.WUF increases
not when the irradiance inside the spectral range does, but
also when the spectral distribution light better matches
with the SR [4]. As WUF is a weighted SR, it is expected
that Isc increases with WUF and that fSpectralRange increases
with WUF.
Analysing how both temperature and irradiance affects
fSpectralRange is of paramount importance for further
understanding the effect of spectral variation. Figure 5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
6004002000
Gglobal
I sc (A
) 0.0
0.5
1.0
1.5
2.0
2.5
3.0
5004003002001000
GSpectral Range (W/m2)
I sc (A
)
Figure 6. Illustrates the effects of Gglobal (measured with pyranom
device
4 Pr
illustrates the variation of fSpectralRange with both irradiance
and temperature. It should be noted that the fSpectralRange
values represented in Figure 5 have been increased by a
factor of 200. This was done purely because of the small
fraction values which require many decimal places to
differentiate them. This would result in the graph to look
rather clumsy.
Notably from Figure 5 is the increase in fSpectralRange as
Gspectral range is reduced. This behaviour contradicts what is
expected if the irradiance used in fSpectralRange had been
global. A linear increase relationship with Gglobal would
have been expected since the irradiance effect would
compensate for each other. The reason for this behaviour in
Figure 5 is due to the fact that the Gspectral range used in
fSpectralRange is very sensitive to outdoor changes as
opposed to Gglobal. Therefore the behaviour of these two
parameters is not the same which results in fSpectralRange to
be high at low irradiance. The change in Isc as outdoor
irradiance varies is well understood. High values of
irradiance result in high values of Isc. This relationship (Iscvs. Gglobal) which is illustrated in Figure 6 illustrates less
scatter as Isc and Gglobal varies simultaneously.
Analysing the insert in Figure 7, the variation of Isc asfunction of irradiance within the device spectral range
(Gspectral range) is not a clear linear relationship as in the case
of Isc versus Gglobal. A large scatter observed as the Iscvaries with Gspectral range of a device illustrates how
sensitive the device is to the sudden changes of the outdoor
environment. This shows that PV module’s Isc respond
greatly to the outdoor spectral changes.
Outdoor spectral effects on crystalline-Si has not been as
greatly documented and studied as amorphous-Si devices.
This has been partly due to the ‘perceived’ larger SR range
140012001000800
(W/m2)
600
eter) and Gspectral range (measured with spectroradiometer) on
Isc.
og. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
High cloud cover
Low WUF regionHigh WUF region
Sunrise/sunset: Longer wavelengths dominating
0.0
0.4
0.8
1.2
1.6
2.0
1400120010008006004002000
Irradiance (W/m2)
φ Spe
ctra
l Ran
ge (A
/Wm
-2)
Figure 7. Illustration of the dependence of fSpectralRange on the total irradiance as measured by the pyranometer.
M. Simon and E. L. Meyer Spectral evaluation of c-Si modules
of c-Si which theoretically translates to a relatively stable
outdoor performance. But analysing the outdoor spectral
data and correlating this to the device performance data, a
different conclusion is reached. Figure 7 illustrates the
dependence of fSpectralRange on irradiance (Gglobal).
Observing Figure 7, the spectrally corrected fSpectralRange
versus irradiance (Gglobal) reveals the effect of outdoor
conditions on device’s fSpectralRange. For irradiances greater
than 800W/m2, the data points are more defined and less
scattered and linear relationships exists. Below Gglobal of
800W/m2 and close to 300W/m2 the linear trend
relationship that has been exhibited at Gglobal> 800W/
m2 disappears. The scatter is more pronounced with no
clear trend. This figure also reiterates the fact that PV
device performance is greatly affected by the changes in
outdoor conditions. For Gglobal> 800W/m2 it indicates
‘clear sky’ condition typically found under full spectrum
and Gglobal< 300W/m2 also indicate spectrum character-
ized by low WUF, a typical spectrum dominated by longer
wavelength (sunrise/sunset) resulting in poor performance
of the device, as indicated in Figure 7 (green curve). Low
WUF values can also be found under high cloud cover with
mostly shorter wavelength being scattered back into the
sky, letting only longer wavelength to pass through. Also
the high cloud cover acts as a filter for the infrared (IR)
light and will shift the spectrum to full spectrum region (as
defined in previous section) resulting in an increase in
device performance indicated by the upper curve (purple).
These two phenomena lead to both high values of Isc andlow values for Isc, as is evident in Figure 7. The change in
outdoor spectrum is influenced by a number of factors, e.g.
time of the year, sun elevation, meteorological conditions
and tilt of the module [4]. The commonly adopted
reference spectrum of AM 1.5 was created using a radiative
transfer model called BRITE [6]. Most of the input
parameters used were obtained from United States
Prog. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
environment, e.g. tilt angle of 37o which is average
latitude in US continent [6]. Air mass (AM) contributes
greatly to the outdoor spectral variations and consequently
device spectral changes as characterized by WUF. Ideally
an increase in AM results in a decrease in deviceWUF. The
correlation between WUF and AM for mc-Si is presented
in Figure 8. Note that the AM values were calculated at the
same time frame as the WUF. This was done so that the
same time frame when WUF measurements were taken
would be the same as the calculated AM values.
Evident from Figure 8 is a functional relationship
between AM and WUF. Large AM values correspond to
larger wavelength mainly found under low irradiance
levels. At these conditions, the WUF of the device will be
lower than its value at AM 1.5. The variation of WUF with
AM for mc-Si module for both summer and winter has the
same trend as is indicated by the gradients of the two
curves. Although the WUF winter values are not the same,
the functional relationship between the two parameters in
both cases is the same. This shows that the WUF versus
AM relationship is device dependent. Depending on the
type of PV module, the gradient of the WUF versus AM
curve will differ. This, however, will be clear only after
having examined the other modules. Figure 9 shows a
typical daily WUF and AM on clear days for both summer
and winter seasons.
During early morning (before 07:30 h) and late after-
noon (after 16:00 h) the low solar altitude is characterized
by high AM values (>8). The spectrum during these times
is associated with longer wavelengths dominating due to
Raleigh scattering. From the figure it is then deduced that
WUF< 0.78 for summer season is indicative of a ‘red’
spectrum while WUF> 0.78 indicates a full spectrum. The
WUF in summer peaks at 10:30 h at a value of 0.8099
corresponding to an AM¼ 2.19, which is different from
AM 1.5 expected at solar noon, 12:30 h. Furthermore,
5
y = -0.0029x + 0.7245R2 = 0.8231
y = -0.0032x + 0.815R2 = 0.9078
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
181614121086420
Air Mass
WU
F
Winter Summer
Figure 8. Influence of the air mass on device spectral variations as characterized by WUF for mc-Si module for both summer and
winter seasons.
Spectral evaluation of c-Si modules M. Simon and E. L. Meyer
repeating the exercise for clear winter days, it was deduced
also that WUF< 0.71 is indicative of ‘red’ spectrum and
WUF> 0.71 indicates a full spectrum. This procedure was
adopted because the summer and winter spectra are
different. The WUF was found to peak at 13:00 h at a value
of 0.7125, corresponding to AM 1.83 during winter
seasons. To isolate the effect of seasonal spectral changes,
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
6:007:12
8:249:36
10:481
WU
F
WUF Summer WUF
10:30 @ WUF= 0.8099 &
13:00 @ W
Red Spectrum
Full Spectrum
Figure 9. Illustrates the seasonal WUF profile again
6 Pr
the influence of WUF on FF was characterized in summer
and winter. Figure 10 illustrates the entire SR range (s2) for
mc-Si during summer and winter. Also illustrated is the
half-width of the Gaussian peaks, s1.
Noteworthy from Figure 10 is the low frequency (or
time spent) under the red spectrum in summer. This implies
that the device preferentially experiences, in summer, a full
2:0013:12
14:2415:36
16:4818:00
-15
-10
-5
0
5
10
15A
ir M
ass
Winter Air Mass
AM2.19
UF= 0.7125 & AM1.83
Full Spectrum
Red Spectrum
st time together with the change in AM values.
og. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.000.950.900.850.800.750.700.650.600.550.50
WUF
Fill
Fact
or
0
20
40
60
80
100
120
Freq
uenc
y (%
)
Summer Winter
σ1winter σ1summer
σ2winter (Device range)σ2summer (Device range)
Figure 10. Effect of season on device junction properties as illustrated by Fill Factor.
M. Simon and E. L. Meyer Spectral evaluation of c-Si modules
spectrum to operate optimally. Conversely the wider
s1winter implies that the device operates optimally over a
wider spectral band. More than half of s1winter corresponds
to red-shifted spectrum implying that the device prefer-
entially experiences, in winter a red-dominated spectrum to
operate optimally.
3.2. Performance of mono-Si
Crystalline silicon modules used in this study (mc-Si,
mono-Si and poly-Si) should ideally behave in the same
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
86420
Air
WU
F
Figure 11. Influence of air mass on device spectral varia
Prog. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
way under outdoor conditions. The effect of the three major
outdoor factors that affect their performance should affect
them in the same manner regardless of their ratings.
Figure 11 illustrates the influence of AM on WUF for
mono-Si module for winter and summer seasons.
Closer analysis of Figure 11 reveals that the gradient of
the two curves for both summer and winter seasons is the
same for the mono-Si module. The gradient exhibited in
Figure 11 is the same with that of mc-Si in Figure 8. This
indicates that the two devices have the same AM
dependence although their WUFs are different.
Figure 12 illustrates the variation in WUF due to seasonal
y = -0.003x + 0.5773R2 = 0.9359
y = -0.0032x + 0.6439R2 = 0.9515
1816141210
Mass
Winter Summer
tions as characterized by WUF for mono-Si module.
7
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
6:007:12
8:249:36
10:4812:00
13:1214:24
15:3616:48
18:00
WU
F
-15
-10
-5
0
5
10
15
Air
Mas
s
WUF Summer WUF Winter Air Mass
14:00 @ WUF= 0.5691 2.58& AM
12:00 @ WUF= 0.6457 & AM 1.83 Full spectrum
spectrum Red
Full spectrum
Red spectrum
Figure 12. Illustrates a typical daily WUF and air mass on clear summer and winter day for c-Si module.
Spectral evaluation of c-Si modules M. Simon and E. L. Meyer
changes in the spectrum. Also illustrated is the AM
variation with day time.
The same procedure adopted in Figure 9 was also
applied in analysing the seasonal spectral effects on c-Si
module. For summer period, it is deduced that for
WUF< 0.61 is indicative of the red spectrum while
WUF> 0.61 indicates a full spectrum. For winter season,
WUF< 0.555 indicates a red spectrum and WUF> 0.555
is indicative of a full spectrum.
For summer seasons, the mono-Si module preferentially
experiences a spectrum characterized by low AM values
(AM 1.83). This spectrum which peaks at 12:00 h noon
corresponds to the WUF¼ 0.6457. For winter season, the
WUF was found to peak at 14:00 h at a value of 0.5691,
corresponding to AM 2.58. As evident in both seasons, the
spectrum that is preferentially experienced by the mono-Si
module is different from AM 1.5.
Table I lists the parameters of both mono-Si module and
poly-Si module during summer and winter seasons.
The device spectral range (sspectral range) for mono-Si
module is wider in summer than it is during winter period.
This results in 71.13% difference as the winter sspectral rangeis reduced. The FF and efficiency still remains the same as
indicated by a slight % difference, which is within
experimental errors. The response of mono-Si module to
Table I. Dependence of Fill Factor and efficiency of both mono-Si a
Useful Fraction in summe
mono-Si
sspectral range WUFmean FFAverage hA
Summer 0.41–0.895 0.6457 0.67 1
Winter 0.49–0.630 0.5791 0.67 1
% Difference 71.13 12.31 0
8 Pr
the outdoor changes due to season is not compromised
although its spectral range has been affected.
3.3. Performance of poly-Si
As mentioned earlier on in Section 3.1, the variation of
WUF versus AM depends on the characteristics of the PV
module. It was considered important to analyse the effect
of AM onWUF for poly-Si module for summer and winter
seasons. Figure 13 illustrates the influence of AM on
device spectral variations for the poly-Si module.
The variation of WUF as a function of AM for both
winter and summer does not vary significantly as indicated
by the slope of the two curves. Comparing Figure 13 with
Figures 8 and 11, the poly-Si module’s gradient is different.
Figure 14 shows the summer and winter profiles in relation
to AM for poly-Si module.
Notable from Figure 14 are the low WUF values for the
poly-Si module as compared to the other crystalline-Si
modules. It is deduced from Figure 14 that WUF< 0.551
for both summer and winter seasons is indicative of a ‘red’
spectrum while WUF> 0.551 indicates a full spectrum.
The WUF in summer peaks at 09:30 h at a value of 0.5541
corresponding to an AM¼ 3.36, which is different from
nd poly-Si with outdoor spectrum as characterized by Weighted
r and winter seasons.
poly-Si
verage sspectral range WUFmean FFAverage hAverage
3.36 0.41–0.70 0.5541 0.713 7.21
3.67 0.47–0.66 0.5813 0.676 7.32
2.27 34.5 4.5 5.2 1.5
og. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
0.30
0.35
0.40
0.45
0.50
0.55
0.60
6:007:12
8:249:36
10:4812:00
13:1214:24
15:3616:48
18:00
WU
F
-15
-10
-5
0
5
10
15
Air
Mas
s
WUF Summer WUF Winter AM
09:30 @ WUF= 0.5541 & AM 3.36
13:00 @ WUF= 0.5813 & AM 1.83 Full spectrum
Red spectrum
Figure 14. Seasonal WUF for summer and winter profiles for poly-Si module.
y = -0.0022x + 0.5729R2 = 0.7657
y = -0.001x + 0.5688R2 = 0.6943
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
181614121086420AM
WU
F
Winter Summer
Figure 13. Variation of WUF as a function of AM for summer and winter seasons for pc-Si module.
M. Simon and E. L. Meyer Spectral evaluation of c-Si modules
AM 1.5 expected at solar noon, 12:30 h. For clear winter
days, it was also observed that the WUF was found to
peak at 13:00 h at a value of 0.5813, corresponding to AM
1.83.
Poly-Si module’s performance parameters are not signifi-
cantly affected by the outdoor seasonal changes. The device
average efficiency has a 1.5% difference which could be
attributed by some errors associated with the measuring
Prog. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip
instruments. Although the device’s sspectral range had increased,
this could not affect both WUFmean and FF greatly.
4. SUMMARY AND CONCLUSION
Evident from the results presented in this study is that the
spectrum received in sub-Sahara is totally different from
9
Spectral evaluation of c-Si modules M. Simon and E. L. Meyer
the standard AM 1.5 spectrum and from that received from
the northern hemisphere regions. Also noted is that
crystalline-Si modules are also affected by the spectrum
shift during seasons although these devices are perceived
(without outdoor data) that their performance is not
influenced by the seasonal changes in outdoor spectrum. It
has been showed that for summer season, mc-Si
preferentially experiences to operate optimally at
WUF¼ 0.8099, which is an indication of a full spectrum
as defined in this study. For winter season the mc-Si
module responds optimally at WUF¼ 0.7125 which
corresponds to AM 2.19. It is evident that the change in
outdoor spectrum due to seasons affects the module’s
spectrum to which it best preferentially experiences to
operate optimally and also the time of the day at the
corresponding spectrum. Results for mono-Si module
showed that the device performs best at WUF¼ 0.6457
which corresponds to AM 1.83 during summer season,
while it operates optimally under a winter spectrum
indicated by WUF of 0.5691 (AM 2.58). The seasonal
changes resulted in the shift in day time corresponding to
the preferentially experiences spectrum. This shift
indicates that these devices should be rated using AM
values that correspond to the WUF values under which the
device operates optimal. For poly-Si, it was also observed
the WUF values are lower than the other two crystalline-Si
counterparts. The poly-Si was observed to preferentially
experiences a low AM spectrum indicated by
WUF¼ 0.5813 during winter season while for summer
it preferentially experiences a spectrum characterized by
WUF¼ 0.5541 at AM 3.36.
10 Pr
5. ACKNOWLEDGEMENTS
Authors acknowledge the Govan Mbeki Research and
Development Centre (GMRC) and Eskom Tertiary
Education Support Programme (TESP) for their financial
support.
REFERENCES
1. Gottschalg R, Betts TR, Infield DG, Kearney MJ.
Experimental study of variations of the solar spectrum
of relevance to thin film solar cells. Solar EnergyMaterials and Solar Cells 2003; 79: 527–537.
2. Simon M, Meyer EL. Spectral distribution on photo-
voltaic module performance in South Africa. Presentedat the 33rd IEEE Photovoltaic Specialist Conference,2008.
3. Kenny PR, Ioannides A, Mullejansa H, Zaaiman W,
Dunlop ED. Performance of thin film PVmodules. ThinSolid Films 2006; 511–512: 663–672.
4. Parretta A, Sarno A, Vicari LRM. Effects of solar
irradiation conditions on the outdoor performance of
photovoltaic modules. Optics Communications 1998;
153: 153–163.5. Christian NJ, Gottschalg TR, Infield DG, Lane K. Influ-
ence of spectral effects on the performance of multi-
junction amorphous silicon cells. PV Conference andExhibition, Rome, 2002.
6. Riordan C, Hulstrom R. What is an air mass 1.5
spectrum? IEEE Photovoltaic Specialist ConferenceProceedings, 1990; 1085–1088.
og. Photovolt: Res. Appl. 2011; 19:1–10 � 2010 John Wiley & Sons, Ltd.DOI: 10.1002/pip