The Cost of Wind Power:
Learning Rates and Drivers
Michael Taylor
Senior Analyst
4 November 2014
Costs are falling….
Wind Power
Higher
capacity
factors from
improved
technology
Wind turbine
cost
reductions
Cost data for 3462 wind farms
3
Why look at wind costs
and learning rates?
We need to fill gap in understanding of true costs
of wind
We have good data for non-OECD and know average
LCOE is falling, but don’t have good time series data
Understanding past cost reductions and their drivers are
critical to understanding future cost reduction scale and
drivers
Learning rate analysis for wind is out of date and/or not
comprehensive – can mislead policy makers
4
Wind turbine costs
Turbine prices and installed costs since 2002/03 aren’t
captured in learning curve literature
H2 2012
Chinese
turbine prices
Source: IRENA and BNEF
5
Learning curve studies
Early analysis covered few countries, but was
comprehensive (data until mid- to late 90’s)
Later analysis often covered few countries with
similar shared goals, but was not comprehensive
Global analysis, has used only data to 2006, and
has often not been transparent or uses limited
sample
No analysis of LCOE learning rates
6
Why is this a problem?
•
Current learning curve analysis is easily open to
criticism
• Lack of accurate learning curve analysis may:
• Compromise scenario analysis
• Provide misleading cost reduction expectations
• Allow criticism of renewable competiveness
• We are missing an opportunity to highlight winds
increasing competiveness with robust data
7
IRENA’S PROJECT:
LEARNING RATE
ANALYSIS FOR WIND
8
IRENA’s project
Comprehensive new analysis of
global learning curves for wind
1990 - 2013
Total installed cost and LCOE
Decomposition of LCOE learning curve
9
The importance of LCOE
learning curves and decomposition
Allow us to show higher installed costs have been
However
combined with lower LCOE
Decomposition
allows us
Part of the reason
thisto:analysis has not been done
• Explain drivers
costisreduction
beforeofnow
the huge data needs
• Accurately assess contribution of technology
improvements
• Look at potential issues around BoS and O&M costs
10
Data requirements for a robust analysis
Country
China
United States
Germany
Spain
India
Italy
France
Canada
United Kingdom
Portugal
Denmark
Rest of the world
Cumulative installed capacity - 2013
90.984 MW - 29%
61.136 MW - 20%
33.730 MW - 11%
22.954 MW - 7%
20.150 MW - 6%
8552 MW - 3%
8254 MW - 3%
7803 MW - 3%
6850 MW - 2%
4722 MW - 2%
3501 MW - 1%
42781 MW - 14%
No. of wind farms
1512
1610
4217
1375
7287
817
810
334
618
540
3614
22734 wind farms account for 85%
of cumulative installed capacity
11
Learning curve data needs
Data needs are large…
we need representative data by country for:
12
Decomposition analysis
Provides new insights for policy makers
Hypothetical learning rate decomposition
Total
-12
O&M
-5
-15
Technology improvements
Wind resource
2
Balance of system
1
Turbine cost
-20
-15
-10
-5
5
0
5
10
Data collection example:
Denmark
Item
Number of
data points
Percentage
Wind farms
3614
100%
Total Capacity
3612
99.94%
Manufacturer name
3492
96.62%
Number of turbines
3175
87.85%
Data collection
for
learning
curve
for
installed
cost
Geographical location
1812
50.14%
and LCOEWind
time
consuming but feasible
resource
1789 in near
49.50%future
Average annual output
1657
45.85%
Hub height
1633
45.19%
Rotor diameter
1633
45.19%
Capacity factor
1615
44.69%
Total Investment
0
0.00%
14
How to get statistically robust results
Only representative sample is needed for each
country and year
Of 11 countries, not all are statistically necessary in a
given year, allows for some data gaps
Comprehensive data for pre-2000 mostly available
Some data can be calculated, just time consuming and
expensive
15
VALUE ADDED
AND
NEXT STEPS
16
What the project will achieve
The most comprehensive wind learning curve and
cost reduction analysis to date
Understanding of the drivers of cost reduction and their
share of in total cost reduction
Powerful messages about competiveness of wind
Comprehensive database to base future cost reduction
potentials analysis
Some insights into future cost reduction potentials and
their relative importance already
17
Next steps
Continue data collection for wind turbine details,
wind resource, investment costs
Calculation of wind resource characteristics and capacity
factor based on power curves
Global and country analysis of learning curve for
installed costs and LCOE
Decomposition analysis by country and global
Write up the project and publish
18
IRENA’s Cost Analysis
Bringing Our Future Forward
Thank you!
[email protected]
www.irena.org/costs
Decomposition analysis and
where CF is missing
Even more detailed data needs…
we need representative data by country for:
20
Study
Year released
Dataset
Methodology
Results
Coverage quality
Marsh, G. (1998)
1998
1980 - 1985
Cost of electricity ECU(1990)/kWh plotted
against cumulative electricity production
Learning curve - 18%
Regional. Germany, Denmark,
United Kingdom and Netherlands
Durstewitz, M. and Hoppe-Kilpper, M.
(1999)
1999
1990 - 1998
Wind turbine price (DM 1995/kW) plotted
against cumulative installed capacity
Learning curve - 8%
Germany only.
Kouvaritakis, N., Soria, A. and Isoard, S.
2000
(2000)
1995 -
Two factors learning curve. Investment cost
1995 (US$1990) & R&D Investment plotted Learning curve - 16%
against cumulative installed capacity
POLES model. Costs source and
coverage not mentionned.
K. Ibenholt, Explaining learning curves for
wind power, Energy Policy 30 (2002) 1181- 2002
1189
1991 - 1999
Investment costs plotted against
cumulative installed capacity.
Germany only. 41 data points
2004
1971 - 1997
Two factors learning curve. Electricity cost
(USD90 per watt) R&D plotted against
cumulative installed capacity
Klaassen, G., Miketa, A., Larsenb, K. and
2005
Sundqvist, T., (2005)
1986 - 2000
Kobos, P., Erickson, J. and Drennen, T.
2006
(2006)
1981 - 1997
G.F. Nemet, Energy Policy 37 (2009)
2009
825–835.
1981 - 2006
European Wind
(EWEA), (2009)
2009
1986 - 2006
Cost of electricity by turbine size plotted
against cumulative installed capacity
2014
1980 - 2010
Investment cost plotted against cumulative
Learning curve - 6%
capacity
Miketa, A., Schrattenholzer, L., (2004)
Energy
Association
Patrick Criqui, Silvana Mima, Philippe
Menanteau, Alban Kitous. Elsevier, 2014,
pp.1-18
Learning curve - 11%
Learning curve for
investment costs plotted
against cumulative installed
capacity - 9.73%
Two factors learning curve. Investment cost Learning curve for
(US$98/kW) & Research and Development investment costs plotted
plotted against cumulative installed
against cumulative installed
capacity
capacity - 4.8 to 5.8%
Two factors learning curve. Electricity
Learning curve for
(USD90 per watt) & Research and
investment costs plotted
Development plotted against cumulative
against cumulative installed
installed capacity
capacity - 14.2%
Investment costs (wind turbine capital cost
2006) USD/W plotted against cumulative
Learning curve - 11%
installed capacity
Learning curve - 9 - 17%
Denmark, Germany, United
Kingdom
Denmark, Germany, United
Kingdom
Worldwide
Worldwide
European countries, mostly
Germany, Denmark and United
Kingdom
The dataset appears to be global.
There is no mention of the
coverage.