Transcript MEASURING DEGRADATION RATES WITHOUT IRRADIANCE DATA

MEASURING DEGRADATION
RATES WITHOUT
IRRADIANCE DATA
Steve Pulver
Daniel Cormode
Alex Cronin
U of A
Dirk Jordan
Sarah Kurtz
Ryan Smith
NREL
Outline
• Goal: Measure degradation rates for
systems that don’t have irradiance data
• Approach: Use 20 PV systems from TEP
• Step 1: Quantify relative degradation
rates
• Challenge: Relative vs. absolute rates
• Step 2: Bayesian statistics to derive
absolute degradation rates
TEP solar test yard
Missing POA irradiance sensors
Variation in daily insolation
Daily Final Yield, 2005-2009
Daily and annual insolation can vary by
more than the degradation rate
Removing irradiance variation
Daily Final Yield
Daily Relative Final Yield
Define Daily Relative Final Yield as:
Daily Final Yield divided by field average final yield
Relative rates of change
Provides an easy comparison between systems, and can be
calculated with low uncertainty
Example: only two systems…
relative
absolute
Rate of change (percent per year)
2 systems
Adding more systems…
2 systems
5 systems
10 systems
There appears to be a higher probability of low degradation rates
Translating to absolute…
relative
Assume the best system
doesn’t degrade?
absolute
Rate of change (percent per year)
17 systems
In a large dataset, possibly there is a system that changes
very little
Challenging scenarios
Outlier at 2.0 (percent / year)
Best performing system degraded by 0.5 (percent / year)
Normalizing by the best system may not be the best solution
Our analysis
Starting point: a PDF (Probability Distribution Function) to
describe the absolute degradation rate.
Comparison to histograms
Histogram of our dataset
(relative rate of change)
Histogram of a much larger
data set
(absolute rate of change)
Exponential PDF appears to be capable of handling both our
dataset and the larger dataset
Amount to shift relative rates
Results depend on the function used
Absolute rates of change
Consistent results between rates measured w/ and w/o irradiance
data, and similar uncertainties between the two methods.
Conclusions
• We determined degradation rates without
irradiance data.
• Relative degradation rates provide a
useful comparison between systems
• Absolute degradation rates can be
obtained from relative rates