Content uploaded by Andrew Francis Thomson
Author content
All content in this area was uploaded by Andrew Francis Thomson on Dec 13, 2015
Content may be subject to copyright.
Andrew Thomson
Prospecting solar energy in Australia: accounting for temperature losses
Andrew Thomson1, Ingrid Haedrich1, Marco Ernst1, Luke Johnson2, Andrew Blakers1, and
Sachin Surve1
1Centre for Sustainable Energy Systems, ANU, Canberra, Australia
2SunPulse KK, Japan
E-mail: andrew.thomson@anu.edu.au
Abstract
In this paper, we prospect the solar potential of 5 varieties of commercially available modules
in 15 locations around Australia, accounting for regional temperature and irradiance. We
employ irradiance datasets, from the Australian Solar Energy Information System (ASEIS).
Through our analysis, we categorise regions around Australia, by their impact on the
performance of different solar module technology. From this comparison we find coastal DNI
on average is lower in the mornings owing to the high relative humidity and daily temperature
variation. These irradiance conditions, slightly alter the optimum installation direction and tilt.
The best performing modules are the premium back-contact c-Si modules, and the worst is the
standard mc-Si module. Importantly, the impact of a module technology on yield must be
determined with site-specific irradiances and ambient temperatures. We find temperature losses
correlate most strongly correlated with average mean monthly temperature. An additional
interesting finding is that coastal locations have lower direct normal irradiance in the morning,
which infers the optimum orientation is slightly West of North.
1. Introduction
Solar panels are often marketed heavily on parameters other than their $/kWh performance.
Features often spruiked on datasheets are ascetics, warranty, low-light performance, and
temperature performance. In particular low-light performance and the module temperature
coefficient impact the modules performance ratio (PR) which we define as:
,
the ratio of the module output power to the measured standard test condition power
adjusted for the real irradiance . To evaluate IReal we extract the direct-normal and
global-horizontal irradiance (DNI and GHI) data sets, from the Australian Solar Energy
Information System (ASEIS). We do this for 15 locations around Australia, depicted in Figure
1.
This paper is set out as follows: we present our methodology for the calculation of PR, in
particular, the temperature performance ratio PRtemp. In the results section, we present location
dependent PRtemp and discuss the location specific effects. Finally, we make discuss and
conclude our findings.
Figure 1: Map of locations studies for the impact of temperature on PR.
2. Methodology
The monthly hourly averaged DNI and GHI was downloaded from ASEIS online (ASEIS 2015)
for the locations depicted in Figure 1. We summarise the: location, distance to the coast, a
categorization of the location, the yearly integrated DNI and GHI and the DNI/GHI ratios in
Table 2. We account for the module ambient temperature by using Bureau of Meteorology
(BOM 2015) monthly mean min and max to give the ambient temperature Tamb, which we
assumend varied sinusoidally with the ambient minimum and maximum temperatures aligned
with the minimum and maximum GHI. We defined 5 categories of modules to model the
PRTEMP and summarised the pertinent performance parameters Table 2. The performance
parameters are typical of the specific module varieties available from datasheets available
online. The 5 module categories studied are representative of commercially available panels,
include 3 c-Si and two thin-film modules.
To compute the irradiance on the module, first the incident angle of the direct radiation was
determined from the relative position of the sun in the center of each recorded hourly irradiance.
The direct beam on a tilted plane was then determined to calculate the direct irradiance of the
solar panels; this computation considers the dot-product of the direct radiation vector with the
normal vector of the module. We note the normal vector of the module results from the cross
product of its tilt vector with the vector representing directional orientation. For this work, all
modules were position facing North at a tilt equal to the latitude. The total irradiance was given
by
The sum of the direct, diffuse and reflect irradiance on a tilted plane. The respective irradiances
were determined using the equation
where is the measured direct normal irradiance, is the tilted cross product of the
direct radiation vector and the module vector, is the diffuse irradiance, is the measured
global horizontal irraciance is the module tilt, and R is the reflectivity surrounding the
module. The reflectivity in this work is consider to be moderate 0.2. The diffuse radiation on a
horizontal is determined by subtracting the DNI projected on a horizontal plane from the GHI
and in
where zen is the Zenith angle of the direct irradiance. We calculated all required parmeters from
the measured and and the geometric setup of the modules with respect to the sun.
For modelling the commercial module performance, we used the parameters in Table 1. In this
instance, we only considered the effect of temperature loss where we determine the module
temperature from the equation
where is the datasheet listed nominal operating cell temperature (NOCT), is the
global tilted irradiance, computed as outlined above, in kW/m2, is the ambient
temperature, is the module power output per unit area and γ is the power loss coefficient
form STC. We note Tmod is solved iteratively as it is an implicit equation. The relative
temperature power loss is then determined hour by hour, considering the relative power loss
coefficient. The impact of the temperature losses is integrated over a year to give the
temperature performance ratio PRTEMP.
Table 1: Modules where the temperature PR was compared across Australia.
Manufacturer/Source
Type
(W)
/m2
(W/m2)
(°C)
(%/°C)
A
CdTe
90
125
45.0
-0.25
B
c-Si
(premium)
345
212
41.5
-0.30
C
mc-Si
(standard)
260
159
45.0
-0.40
D
CIGS
150
138
46.5
-0.31
E
Heterojunction
240
190
44.0
-0.29
Table 2: Summary of the locations and yearly integrated DNI and GHI energy in terms
of kWh/m2
City
Long.
Lat.
Distance to coast
I_DNI
I_GHI
DNI/GHI
Canberra
149
-35.3
112
inland
2005
1768
1.13
Sydney
151
-33.9
6.5
Coastal
1822
1701
1.07
Melbourne
145
-37.8
5.2
Coastal
1615
1572
1.03
Adelaide
139
-34.8
4.8
Coastal
2019
1804
1.12
Perth
116
-31.9
9.6
Coastal
2231
1965
1.14
Broome
122
-17.8
3.8
Coastal
2386
2223
1.07
Darwin
131
-12.4
1.5
Coastal
1979
2105
0.94
Mount Isa
139
-20.5
344
inland
2490
2214
1.12
Alice Springs
134
-23.5
925
inland
2574
2196
1.17
Cairns
146
-16.9
0.6
Coastal
1821
1999
0.91
Brisbane
153
-27.1
15.8
inland
2012
1912
1.05
Moree
150
-29.3
331
inland
2287
1979
1.16
Broken Hill
141
-31.8
549
inland
2467
2015
1.22
WA
114
-22.8
623
inland
2590
2254
1.15
Hobart
147
-42.4
0.8
Coastal
1596
1496
1.07
3. Results
Prior to evaluating module performance ratio we investigate the solar resource at 3 key location,
Canberra, Alice Springs and Brisbane plotted in Figure 2 respectively. We see that the DNI is
typically higher in the afternoon compare to the morning for all three locations, but this effect
is the strongest for Brisbane. We note this changes the optimum orientation, for Brisbane it is
13° West of North. Improving the orientation would yield and additional 0.5% available energy.
Figure 2: ASEIS measured monthly, hourly average hourly DNI, for Canberra A), Alice
Springs B) and Brisbane C). Monthly modelled PRTemp for the mc-Si modules D).
Figure 3 plots PRTEMP for the locations studied in this work. We rank in terms from lowest to
highest yearly irradiance. The dependence of PRTEMP on module type is clear, the best
performing module is the IBC premium (B) owing to the high efficiency, low NOCT and
reasonable temperature coefficient. The performance of the CdTe (A) and c-Si heterojunction
(E) modules are similar and rank equal second. Although the CdTe module has the lowest
temperature coefficient, its comparatively low efficiency and moderate NOCT limit its
temperature performance. The CIGS (D) module is forth, and the standard c-Si (C) module is
fifth. The difference between the standard c-Si (C) and the high performing modules is
significant ranging from 1.1% to 3.3%
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 5 10 15 20
DNI (kW/m2)
Relative sunlight hours, local time
jan feb mar
apr may jun
jul aug sep
A)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 5 10 15 20
DNI (kW/m2)
Relative sunlight hours, local time
jan feb mar
apr may jun
jul aug sep
B)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 5 10 15 20
DNI (kW/m2)
Relative sunlight hours, local time
jan feb mar
apr may jun
jul aug sep
C)
0.88
0.9
0.92
0.94
0.96
0.98
1
1.02
1.04
jan
feb
mar
apr
may
jun
jul
aug
sep
oct
nov
dec
PRtemp
Canberra
Alice Springs
Brisbane
D)
Figure 3: PRTEMP plotted for the locations listed in Table 2 and depicted in Figure 1. The
PRTEMP is plotted for each module variety listed in Table 1.
To further investigate the root drivers of temperature loss, the PRTEMP is plotted as a function
of average monthly mean maximum temperature in Figure 4. A strong correlation is observed.
Of the obvious independent parameters available such as annual yearly irradiance, longitude,
latitude, we found the PRTEMP is most strongly correlated with the average mean monthly max
temperature. This effect was universal for all module types.
Also, we plot the location specific annual energy generation accounting for temperature losses
on the left vertical axis, versus location with the location specific annual energy irradiance on
the right axis. Again ranked from lowest to highest annual irradiance. Although the irradiance
and the energy yield are similar in magnitude, they have different units. The annual energy yield
is in terms of MWh/kWp. It is normalised to the kWp rating of the installed system, therefore
normalising for module efficiency. The annual irradiance is in MWh/kWp. From this analysis,
we see the normalised energy generation ranges from 1.13 to 1.77 MWh/kWp. We note there is
a reasonable correlation with reducing PRTEMP and total insolation, as the average mean
monthly max temperature are also correlated.
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
1.02
PRTEMP
A (CdTe)
B (c-Si premium)
C (c-Si standard)
D (CIGS)
E (c-Si heterojunction)
Figure 4: PRTEMP plotted as function of the average mean monthly temperature. The
PRTEMP is plotted for each module variety listed in Table 1.
Figure 5: Total annual energy yield for different module types and annual insolation
plotted against location.
4. Discussion and Conclusions
We have used freely available online resources to evaluate the temperature losses for 5 different
module varieties at 15 different locations around Australia. While the PRTENP of the different
module technology ranged between 0.907 and 0.996, the maximum variation due to module
type was 3.3%. The temperature losses are more strongly correlated with the location than the
module variety.
0.9
0.91
0.92
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1
1.01
15 20 25 30 35
PRTEMP
Average mean monthly maximum (°C)
A (CdTe)
B (c-Si premium)
C (c-Si standard)
D (CIGS)
E (c-Si heterojunction)
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
Insolation (MWh/m2)
Yearly energy generation (MWh/kWp)
A (CdTe)
B (c-Si premium)
C (c-Si standard)
D (CIGS)
E (c-Si heterojunction)
Annual insolation
By far the worst performing module variety were the standard c-Si technology. This low
performance owes to the high temperature coefficient and NOCT, and moderate efficiency.
Through the PV-Mate project, we will see to develop modules with lower NOCT and hence
better temperature performance in Australia.
In addition to the impact of module type on PRTEMP we have noticed that the insolation in many
of the cities studied is skewed. That is there is significantly more direct irradiance in the
afternoon compared with the morning. The higher morning insolation infers that a module
orientation of North is not optimum, and that module should be orientated slightly to the west.
This effect was fairly strong for all coastal cities; it is likely related to evaporation and
precipitation moisture in the mornings which is burnt off throughout the day. The climate date
indicates that the insolation is strong and more direct in the afternoons. The effect is not overly
large at best it could be used to enhance yield by 0.5%. However when choosing between a
Northeast and Northwest orientation, the effect is significant, in Brisbane they yield from a
Northwest orientation is 5.5% higher than a Northeast orientation.
References
ASEIS (2015). "http://www.ga.gov.au/solarmapping/?".
BOM (2015).
"http://www.bom.gov.au/jsp/ncc/cdio/weatherData/av?p_nccObsCode=36&p_display_type=d
ataFile&p_stn_num=066196."