Monthly Report 5 on NREL 99034

advertisement
Subcontractor: Dr. Alex Cronin, University of Arizona
NREL Contract 99043, “Study Degradation Rates of Photovoltaic (PV) Modules Deployed in Arizona”
NREL Technical Monitor: Dr. Sarah Kurtz
Description: Fifth Monthly Report
Authors: Steve Pulver and Alex Cronin
Date: October 27, 2009
Our next conference call is scheduled for 3:00 pm Colorado time Oct. 29.
Alex Cronin’s office number: (520) 626-3348.
Continuing from last month’s comparison of methods for calculating degradation rates, we are
comparing some additional methods. We revisit “Method I” which uses a linear fit of the daily
performance ratio, skipping the step of dividing by an average of multiple systems. We add a
method where we first estimate sunny day irradiance by fitting the multi-system average to a
sine wave, and then use this sine wave to normalize the daily performance ratio. We again use
this estimate of sunny day irradiance to add filtering for cloudy days to all of the methods.
Last month’s report shows degradation rates without including plots of the fitted curves to the
underlying data. We review some of those plots here in order to verify the results are not being
skewed by small subsets of the data. We attempt to eliminate this type of data by limiting date
ranges used, or completely removing systems from the analysis that may have an insufficient
amount of data.
The system numbers have again been reassigned this month, this time to group data by the
type of PV module used in the system (These numbers may change once again next month as
we decide to remove some of the datasets from further analysis). The table on the next page
lists the system number with the construction type of the solar modules.
Material
Make (Model)
a-Si
System
No.
1
Sticker
Rating
60*50 W
Relative Annual
Degradation Rate
+ 1.7 %
CIS/CIGS
2
32*45 W
- 1.1 %
3
Global Solar GSE (GG112,13309)
Shell Solar (ST40)
38*40 W
- 1.4 %
HIT (Si)
4
5
Sanyo (HIP-G751BA2)
Sanyo (HIP-J54 BA2)
8*167 W
8*180 W
+ 0.4 %
+ 1.8 %
MJ-Si
6
7
8
9
BP Solar (MST50MVHS)
BP Solar (MST50 MVHS)
BP Solar (MST50 MVHS)
UniSolar (US-64)
30*50 W
150*50 W
30*50 W
24*64 W
- 2.9 %
not enough data
0.0 % (noisy)
+ 2.4 % (noisy)
px-Si
10
11
12
13
BP Solar (BP 3150U)
BP Solar (SX140S)
Kyocera (KC150G-A)
Schott ASE (ASE-300DGF/17)
Schott ASE (ASE-300DGF/50)
Schott ASE (ASE-300DGF/50)
Schott ASE (ASE-300DGF/50)
Sharp (NE-Q5E2U)
10*150 W
9*150 W
4*315 W
+ 0.7 %
+ 1.6 %
+ 2.1 %
not enough data
4*300 W
+ 0.5 % (noisy)
72*315 W
- 0.2 %
72*315 W
+ 0.4 %
18*165 W
not enough data
AstroPower (API-165MCB)
9*165 W
+ 0.9 %
14
15
16
17
x-Si
18
Solarex (MST-43MV)
Systems not located at the TEP Test Yard:
System No.
19
20
21
22
Distance from TEP
test yard
10.5 mi
6.5 mi
6.5 mi
2 mi
Make
Sticker
(Model) Rating
21.6 kW
108 kW
108 kW
1.2 kW
Relative
Degradation Rate
- 0.6 %
- 1.5 %
- 1.4 %
+ 0.6 %
Degradation rates for 22 systems using 6 methods. The methods named in the key are
discussed in this report. Since the average decays at 1.1% per year, it is expected that
normalizing by the average will cause increased rates by this amount.
AVG kwh/kw vs date.
The average daily (kwh/kwrated) shows 2297 sunny days (black squares) 2912 days total (red and
black points). An empirical function (blue curve) helps to identify sunny days.
The empirical function used here to mimic sunny days is:
sunnyempirical = 4.85 + 0.6*sin(day*0.03442 - 1.508), where day is day since jan1 2004.
The black points are within 1h of sunnyempirical.
Raw data systems 1-8
Raw data systems 9-16
Raw data systems 17-22
Sunny days only systems 1-8
Based on our identified start and stop dates, most systems have a 3 to 5 year long duration
which we plan to study. Systems with short durations will be excluded from the final analysis.
Next we show all the individual data sets with just the sunny days.
Sunny days only systems 9-17
Sunny days only systems 18-22
Sunny days only, divide by AVG, systems 1-8
Sunny days only, divide by AVG, systems 9-17
Sunny days only, divided by AVG, systems 18-22
normalized to EMPIRICAL
normalized to Empirical sys 9-17
normalized to Empirical sys 18-22
In the following, we try to justify eliminating some date ranges by estimating average values for
each segment of 365 data points, doing a line fit to the underlying points, and removing
segments that are more than 4% from the linear fit. Also, we remove any datasets that don’t
include at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System1
System 1 has 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System2
System 2, final year of data falls outside of 4% range.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System3
System 3 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System4
System 4 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System5
System 5 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System6
System 6 has at least 3 years of usable data, with 1 year falling outside of the 4 % range, may want to
remove data prior to 2002.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System7
System 7 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System8
System 8 has at least 3 years of usable data, may want to remove data after 2008.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System9
System 9 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System10
System 10 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System11
System 11, final year of data falls outside of 4% range.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System12
System 12 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System13
System 13 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System14
System 14, appears that data prior to 2004 and after 2007 would need to be discarded in order to fit
within a 4% range.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System15
System 15 has at least 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System16
System 16 has at least 3 years of usable data. One year falls outside of the 4 % range.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System17
System 17 has less than 3 years of usable data.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System18
System 18 has less than 3 years of usable data
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
System19
System 19, removing data after 2008 would allow a fit within a 4% range.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
System20
System 20, appears that data prior to 2003 and after 2006 would need to be discarded in order to fit
within a 4% range.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
System21
System 21, appears that data prior to 2003 and after 2007 would need to be discarded in order to fit
within a 4% range.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
System22
System 22, appears that data prior to 2003 and after 2008 would need to be discarded in order to fit
within a 4% range.
For the following calculations, the data from systems 14, 17, 18, 20, 21, and 22 were not
included; and the final year of data was removed from systems 2, 11, and 19. Using a cutoff
percentage larger than 4% at systems not located at the TEP Test Yard may be desirable.
14
12
10
8
6
4
2
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
dat
6
5
4
3
2
1
0
1/1/2002
1/1/2004
dat
Number of systems on and daily average after the above systems were removed or edited.
The above plots presented again, using the edited data and the new calculated average.
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System1
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System2
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System3
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System4
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System5
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System6
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System7
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System8
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System9
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System10
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System11
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System12
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System13
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System15
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System16
2.0
1.5
1.0
0.5
1/1/2002
1/1/2004
System19
Fitting this data to w[2]+w[1]*t+w[0]*cos(2*pi*t/(365*24*60*60)+2.46)
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System1
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System2
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System3
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System4
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System5
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System6
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System7
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System9
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System10
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System11
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System12
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System13
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System15
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System16
1.6
1.4
1.2
1.0
0.8
0.6
0.4
1/1/2002
1/1/2004
System19
The resulting degradation rates using a line fit, and a line + cosine fit. Systems 6, 8, and 9 were then
calculated again using reduced range of dates, this time determine by visual appearance
System 9: July 8, 2005-May 16, 2009
System 6: Nov 4, 2003-May 16,2009
System 8: May 22, 2004-Jan 16, 2008
1.2
1.0
0.8
0.6
0.4
0.2
1/1/2002
1/1/2004
1/1/2006
System8
1/1/2008
1.4
1.2
1.0
0.8
0.6
0.4
0.2
1/1/2002
1/1/2004
1/1/2006
1/1/2008
1/1/2006
1/1/2008
System9
1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
1/1/2002
1/1/2004
System6
rate of change per year
0.05
0.00
sunny ; norm to avg; start-stop
sunny ; norm to avg; start-Nyrs
all weather ; norm to avg
sunny only ; norm to empirical
all weather; norm to empirical
linecosinefit
linefit
-0.05
-0.10
0
5
10
system number
15
20
Download