Overview of Field Experience - Degradation Rates ... · NREL is a national laboratory of the U.S....
Transcript of Overview of Field Experience - Degradation Rates ... · NREL is a national laboratory of the U.S....
NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.
Overview of Field Experience - Degradation Rates & Lifetimes
Solar Power International Anaheim, CA
Dirk Jordan, Sarah Kurtz 9/14/2015
NREL/PR-5J00-65040
2
Outline
Definitions
Degradation rates • Study type (using nameplate rating vs. multiple
measurements)
• Climate
• Module vs. system
Degradation curves - non-linearity
Failure modes
3
Merriam-Webster: “The quality or state of being reliable”
Degradation no sudden failure
“Quality over time”
PV: high reliability application
Failure (sudden) Degradation Failure
Lifetime distribution
Accelerated tests
• Tracked systems • Warranty returns • Maintenance
records
Need to agree
What is reliability?
Field data
4
> 30 system studies of 20+ years field exposure
Geographic distribution of degradation rates (Rd)
20
25
30
35
Year
s1995 2000 2005 2010 2015
Ref year
Fiel
d ex
posu
re (y
ears
)
Algeria
Brazil
Canada
China
Germany
Greece
India
Italy
Korea
Libya
Saudia Arabia
Spain
Sweden
Switzerland
Tunisia
USA
Publication year
Studies of systems > 20+ years in field
Increased interest in recent years in long-term performance & degradation
5
Presidential election 1948: Harry Truman defeated Thomas Dewey
6
It is difficult to obtain sample w/o bias. Presidential election 1948: Harry Truman defeated Thomas Dewey
Chicago Tribune conducted poll by telephone. 1948 telephone existed only in affluent homes, which voted overwhelmingly for Dewey. The sample was biased because the telephone was used!
7
0.0
0.2
0.4
0.6
0.8
1.0
N
0 1 2 3 4 5
Median(Midpoint Rd+/Pmax+)
Nor
mal
ized
Freq
uenc
y
Jordan et al., “Degradation Rates – An Analytical Review”, Progress in PV, 2011
0
50
100
150
200
250
Freq
uenc
y
Degradation Rate (%/year)
Median: 0.5 %/yearAverage: 0.8 %/year# reported rates = 2128
Degradation rate (%/year)
2011 2015
Literature survey
• Few studies with > 1000 modules • Aggregated distribution is dominated by particular module, system, mounting, method etc. • Counteract overrepresentation analyze in different ways • Median per study & system second peak disappears • High quality data (multiple measurements, calibrations etc.) second peak disappears
Sampling bias is present – representative of the population?
8
1 2 continuous discrete
17%20%
15%48%
7%
10%9%
74%
Percent of studies Percent of degradation rates
Measurements
Majority of Rd are determined by single measurement.
9
Majority of Rd are determined by single measurement.
60
70
80
90
100
NP
(%)
0 5 10 15 20 25
Year
How are Rd determined with 1 measurement?
Perc
enta
ge o
f N
amep
late
Rat
ing
1 2 continuous discrete
17%20%
15%48%
7%
10%9%
74%
Percent of studies Percent of degradation rates
Measurements
10
Majority of Rd are determined by single measurement.
60
70
80
90
100
NP
(%)
0 5 10 15 20 25
Year
How are Rd determined with 1 measurement?
Perc
enta
ge o
f N
amep
late
Rat
ing
1 2 continuous discrete
17%20%
15%48%
7%
10%9%
74%
Percent of studies Percent of degradation rates
Measurements
11
Majority of Rd are determined by single measurement.
60
70
80
90
100
NP
(%)
0 5 10 15 20 25
Year
How are Rd determined with 1 measurement?
Perc
enta
ge o
f N
amep
late
Rat
ing
Did it start at 100%? Distribution?
1 2 continuous discrete
17%20%
15%48%
7%
10%9%
74%
Percent of studies Percent of degradation rates
Measurements
12
0.2
0.4
0.6
0.8
1
STC
diff
1980
1985
1990
1995
2000
2005
2010
2015
Year
M
Measurement
Indoor IV
Outdoor IV
0.2
0.4
0.6
0.8
1
1980
1985
1990
1995
2000
2005
2010
2015
Ti ( )
Climate
Desert
Hot & Humid
Moderate
Snow
Continent
Africa
Asia
Europe
North America
South America
100
60
80
40
20
RD using nameplate rating may have significant error. Pe
rcen
tage
of N
amep
late
ratin
g
Studies using nameplate rating Historical deviation from nameplate rating
• Indoor IV seem closer to nameplate rating (though limited data)
• Outdoor seem farther away trending towards nameplate rating. Outdoor data may include light-induced degradation (LID)
13
Med
ian(
Rd
+/P
max
+)
-1
0
1
2
3
4
5
6
2000
s
2010
s
1980
s
1990
s
2000
s
2010
s
2000
s
2010
s
1990
s
2000
s
2010
s
1990
s
2000
s
2010
s
1980
s
1990
s
2000
s
2010
s
µc-Si a-Si HIT CdTe CIGS x-Si
D d h ithi M d l T 3
Rd+
/Pm
ax+
-1
0
1
2
3
4
5
6
2000
s
2010
s
1980
s
1990
s
2000
s
2010
s
2000
s
2010
s
1990
s
2000
s
2010
s
1990
s
2000
s
2010
s
1980
s
1990
s
2000
s
2010
sµc-Si a-Si HIT CdTe CIGS x-Si
D d h ithi M d l T 3
Degr
adat
ion
rate
(%/y
ear)
All degradation rates Median per study & system
Measurements
Technology
1
2+
µc-Si
a-Si
HIT
CdTe
CIGS
x-Si
Significant difference between all Rd & median per study & system a-Si Rd > 1 %/year Hetero interface (HIT) Rd similar to a-Si than x-Si CIGS around 0.5 %/year
Reducing sampling bias x-Si median 0.5 - 0.6, mean 0.8 %/year
14
Rd+
/Pm
ax+
-1
0
1
2
3
4
1 2+ 1 2+ 1 2+ 1 2+
Desert Hot & Humid Moderate Snow
D t Pt 2 ithi Cli t 3
Measurements1
2+
Desert
Hot & Humid
Moderate
Snow
Climate
Rd+
/Pm
ax+
-1
0
1
2
3
4
1 2+ 1 2+ 1 2+ 1 2+
Desert Hot & Humid Moderate Snow
D t Pt 2 ithi Cli t 3
Degr
adat
ion
rate
(%/y
ear)
Climate category: simplified Köppen-Geiger
All degradation rates Median per study
1-measurement studies show statistically significant higher rates Climate statistically not significant … but…
Hotter climates appear to show higher Rd
15
Rd+
/Pm
ax+
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Solarex Siemens Solarex Siemens Solarex Siemens Siemens Solarex
Rack Roof Rack Roof Rack Roof Rack
Desert Hot & Humid Moderate Desert Moderate
1 2+
Manufacturer
Siemens
Solarex
Degr
adat
ion
rate
(%/y
ear)
Measurements
Climate
Mounting
Manufacturer
2.5x
Some modules more susceptible to hotter climates & mounting.
Show low median Rd across climates Desert (1-measurement) shows higher Rd 1-axis tracker & 2.5x concentration
Solarex:
Siemens: Roof mounting shows higher Rd even for moderate climate Hot & Humid shows higher Rd than Solarex
16
System vs. module degradation?
Performance and Aging of a 20-Year-Old Silicon PV System, Jordan et al., J. PV, 2015.
East
West
20 year old mono-Si system at NREL
Siemens M55 modules Size: 7.4kW
17
• The worst string appears to determine the subarray and array performance
• System degradation matches module degradation if modules degrade similarly, System degradation > average module degradation when large spread
System & module degrade similarly
18
0.2
0.4
0.6
0.8
Cum
ulat
ive
Prob
abili
ty
0 1 2 3
Degradation (%/year)
Modules, all (1552) Systems, all (385) Modules, median (61) Systems, median (71)
Cum
ulat
ive
prob
abili
ty
Degradation rate (%/year)
At median systems & modules degrade similarly
High –quality, x-Si data
At median modules degrade a little less than systems but in general very similar For less well-performing products more significant gap between modules & systems
19
60
70
80
90
100
Pmax
_sca
led
1990
1995
2000
2005
2010
2015
2020
Year2
90
80
70
60
70
80
90
100
Dat
a_sc
aled
1990
1995
2000
2005
2010
2015
Year
0
20
40
60
80
100
Dat
a
1990
1995
2000
2005
2010
2015
2020
Year
0
20
40
60
80
100
Dat
a_sc
aled
1985
1990
1995
2000
2005
2010
2015
Year
Perc
enta
ge o
f Ini
tial M
easu
rem
ent
Modules measured: 191 Modules measured: 8
Modules measured: 288
Modules measured: 54
Year
Northern CaliforniaColoradoSwitzerlandItaly 60
80
70
Appears linear for most modules, worst modules non-linear
1st measurement taken after 11 years
Studies that measured sample modules several times in 20+ years
Distribution skews towards low end worse performing modules show some non-linearity Central tendency & better modules fairly linear
ARCO ARCO
ARCO Siemens
20 years
30 years
20
( )( ) ( )
( )( )∑
∑
=
=
+−⋅
+−
+−⋅
+==
N
nn
nD
N
nnn
rRoductionInitial
rValuesidual
rRateTaxtMOCostInitial
oductionEnergyLifetimeTotalCostCycleLifeTotalkWhLCOE
1
1
11Pr
1Re
11cos&
Pr)/($)1(
80
85
90
95
100
Y
0 5 10 15 20 25
Year
Linear, 0.5 %/year Exponential 2-step Linear, 0.16 %/year
Nam
epla
te ra
ting
(%)
Year
Simulation assumption for LCOE impact
Parameter Distribution Mean Range Unit Initial production
Normal 159,437 ± 2% kWhr/year
Initial cost Triangular 2.6 2.4 – 2.8 $/W Discount rate Uniform 7.5 6 – 9 % Residual value Uniform 5000 0 – 10,000 $ O&M cost Triangular 20 17 – 25 $/kW/year O&M escalation
Triangular 2 0 – 4 %
Hypothetical commercial PV system near NREL: Size: 100 kW, tilt: 40º, azimuth: 180º PVWatts Tax rate: 30%
Input degradation curves
21
LCO
E M
ean
15.5
16.0
16.5
17.0
17.5
18.0
18.5
19.0
19.5
20.0
0 20 40 60 80 0
Discount rate
Initial cost
OM escalation
OM_total
Production_0
Residual value
1
2
3
4
5
6
LCO
E (¢
/kW
h)
Percentile
20.0
Mean: 17.3
Sensitivity analysis: Spider plots
See most important factors, can read the impact directly of axis. Different trends: ⇓ Productioninitial⇑ LCOE ; ⇓Costinitial ⇓LCOE
Most important factor: Discount rate ≈ 2.5 ȼ/kWh; Costinitial ≈ 2 ȼ/kWh; Productioninitial ≈ 1.5 ȼ/kWh; O&M < 0.5 ȼ/kWh
22
LCO
E M
ean
15.5
16.0
16.5
17.0
17.5
18.0
18.5
19.0
19.5
20.0
0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 0 20 40 60 80
Percentile2
Discount rate
Initial cost
OM escalation
OM_total
Production_0
Residual value
1
2
3
4
5
6
LCO
E (¢
/kW
h)
Percentile
20.0
Mean: 17.3 Mean: 18.4 Mean: 18.2 Mean: 16.7
Sensitivity analysis: Spider plots
Degradation curve has significant impact on LCOE
See most important factors, can read the impact directly of axis. Different trends: ⇓ Productioninitial⇑ LCOE ; ⇓Costinitial ⇓LCOE
Most important factor: Discount rate ≈ 2.5 ȼ/kWh; Costinitial ≈ 2 ȼ/kWh; Productioninitial ≈ 1.5 ȼ/kWh; O&M < 0.5 ȼ/kWh
Degradation curve type ≈ 1.75 ȼ/kWh
23
Discoloration & delamination most common
0 25 50 75 100
Percentage of Studies
Discoloration/Browning
Delamination
Corrosion
Glass breakage
J-box
Cell breakage
Cracked backsheet
Hot spot
Soling
Mismatch
LID
Discoloration/Browning Delamination
Corrosion Glass breakage
J-box Cell breakage
Cracked backsheet Hot spot
Soiling Mismatch
LID
Majority of studies report discoloration & delamination
Pareto chart of failure modes in literature (2012)
24
Summary
1-measurement studies show significantly higher Rd because of nameplate rating deviation
Increased Rd in hotter climates & mounting for some products
x-Si has median in 0.5 – 0.6 %/year range, mean in 0.8 %/year range
Non-linearities have significant financial impact, need more accurately measure curves instead of relying solely on rates.
25
Tornado in Germany, May 2015
Roofs have substantial damage but modules appear for the most part undamaged.
Lifetime goal?
26
Acknowledgments
Thank you for your attention National Renewable Energy Laboratory 15013 Denver West Parkway, MS 3411 Golden, CO 80401, USA [email protected]
This work was supported by the U.S. Department of Energy under Contract No. DE-AC36-08-GO28308 with the National Renewable Energy Laboratory