A global experience curve for wind energy - Download Now PDF by xfo14057


									                             A global experience curve for wind energy
                                      M. Junginger*, A. Faaij
Department of Science, Technology and Society. Copernicus Institute, Utrecht University. Padualaan 14,
                                 3584 CH Utrecht, The Netherlands.
      * Corresponding author. Tel. 030-2537613/00, fax 030-2537601, M.Junginger@chem.uu.nl


In order to forecast the technological development and cost of wind turbines and the production costs of wind
electricity, frequent use is made of the so-called experience curve concept. Experience curves of wind turbines are
generally based on data describing the development of national markets, which cause a number of methodological
difficulties when applied for global assessments. To analyze global wind energy price development more adequately,
we compose a global experience curve. We present and discuss a new approach of establishing a global experience
curve and thus a global progress ratio for the investment cost of wind farms. Results show that global progress ratios
for wind farms may lie between 77-85% (with an average of 81%), which is significantly more optimistic than
progress ratios applied in most current scenario studies and integrated assessment models. Combining the progress
ratios with three different growth scenarios for global cumulative wind capacity results in possible cost reductions of
43-75% in 2020 compared to current turnkey wind farm investment prices. This is equivalent to a decrease in
electricity production costs from 6.6 to 2.1 – 4.0 €ct/kWh. Thus, it can be concluded that a large fraction of the
electricity produced by wind farms may be able to directly compete with electricity from fossil fuels by 2020.

Keywords:        Experience curve, wind energy, technological learning, wind farm investment cost, cost of

Introduction and objective

Over the last twenty years, the specific turnkey prices of wind farms have dropped significantly as the global installed
wind capacity has increased from a few hundred MW to about 32 GW. However, wind electricity is not yet entirely
competitive with electricity from fossil fuel sources. Thus, it is of interest whether (and at what speed) further cost
reductions may be expected. One way of analysing the cost reductions is the experience curve concept. In this top-
down approach, costs decline by a fixed percentage with every cumulative doubling of capacity (e.g. a decline of
15% with a cumulative doubling of capacity equals a progress ratio of 85%). The concept of experience curves has
been applied widely within the energy technologies area. Recent examples are PV modules [1], combined cycle gas
turbines [2] or fuel cells [3]. An overview of studies concerning energy technologies is given by McDonald and
Schrattenholzer [4]. Especially for the wind energy sector, experience curves have been devised for Denmark [5],
Germany [6], the United States [7], and other countries [8-10].

However, in contrast to other renewable energy technologies, the cost reductions of wind turbines, wind farms and
wind electricity have so far only been measured on a national basis, i.e. against the national installed capacity. This
approach involves a number of possible pitfalls. For example, national policy measures may influence both local
prices and the annually capacity added [11]. Both effects may strongly influence the progress ratio, but are not
directly related to technological learning. Both methodological problems may possibly be overcome when using
worldwide data.

The objective of this work is to set up a new approach to establish a global experience curve and determine a global
progress ratio for wind farms. In combination with different growth scenarios for global cumulative capacity, we also
indicate possible cost reduction margins for investment costs and subsequently cost of wind electricity.

Experience curve theory and methodological setup

Typically, the costs of a technology come down with increased experience, i.e. increased penetration of the
technology. This phenomenon has been observed numerous times in history, starting with serial production of
airplanes at the beginning of the last century. One special empirical observation is that costs tend to decline a fixed
percentage with every cumulative doubling of capacity. One way of analyzing the cost reductions is the experience
curve concept. In this methodology, costs decline by a fixed percentage with every cumulative doubling of capacity.
A basic experience curve can be expressed as:

CCum     = C0 * Cumb                     (1)                  PR        = 2b                            (3)
log CCum = log C0 + b log Cum            (2)                  LR        = 1 – PR                        (4)

CCum      Cost per unit                                       C0        Cost of the first unit produced
Cum       Cumulative (unit) production                        b         experience index
PR        Progress ratio                                      LR        Learning rate

The progress ratio (PR) is a parameter that expresses the rate at which costs decline each time the cumulative
production doubles. For example, a progress ratio of 0.8 (= 80%) equals a learning rate of 0.2 (20%) and thus a 20%
cost decrease for each doubling of the cumulative capacity.

The definition of the ‘unit’ may vary: in many cases a unit is a product (for example a car or an airplane). In relation
to energy technologies, more often the unit is the capacity of an energy technology (e.g. the capacity of solar modules
or steam turbines) or the amount of electricity produced by a technology. In the remainder of this article we focus on
installed wind farm capacity, i.e. the turnkey prices of wind farms against the global cumulative installed capacity of
wind farms.

For such a global experience curve, we must make sure that it actually concerns a reasonably homogeneous learning
system, i.e. that technological innovations are exchanged between different actors, and that the resulting technology is
available in the entire geographical area.

Technological development: The global wind market is dominated by seven turbine manufacturers: Vestas, Enercon,
NEG Micon, GE Wind (formerly Enron Wind), Gamesa, Bonus and Nordex. With the exception of Enercon1, these
manufacturers also use similar technology (horizontal axis wind turbines, variable speed, pitch regulated, utilizing a
gearbox and a asynchronous generator).

Global diffusion: In the year 2002, the seven largest wind turbine producers (see previous paragraph) together held a
global market share of approximately 78%. Over 80% of this worldwide wind-based power generation capacity is
currently installed in Germany, Spain, Denmark, the U.S.A. and India [12]. The remaining 20% is installed in at least
50 countries all over the world, mainly in Europe.

Thus, as the big manufacturing companies deliver turbines all over the world and use the same basic technology
concepts, we can conclude that the knowledge and technology present in these companies is applied on a global level,
and that there is to a large extent “global learning”. The additional investment costs of wind farms (e.g. foundations
and grid-connections) are more subject to local learning. Yet, many wind farms are built by the wind turbine
producers (i.e. delivering turnkey wind farms) and thus ‘export’ knowledge on these components as well.

Furthermore, in order to qualify as suitable, the data should meet the following requirements [11]:
     •   Price of wind farms should originate stem from competitive market environments to avoid effects of price
         distortions (caused e.g. by governmental support measures)
     •   Markets should also be open for imports from all major manufacturers, with little or no obligations to
         involve local producers
     •   Data must be available over sufficiently long time period (at least five years or longer)

Data selection

UK data for small wind farms: Given the criteria set in the previous section, data from the UK seemed to be well
suitable. First of all, in the UK, the NFFO/SRO-system required the electricity supply companies in the UK to
provide a proportion of their supply from renewable energy sources. From 1991-1999, several bidding rounds were
held, where project developers could subscribe. Bids were assessed on a competitive basis and an upper threshold
was selected after a predetermined deadline. While this system failed to stimulate a large capacity expansion (in
contrast to the German feed-in tariff system), the cost of electricity dropped strongly from 1991-1999 [13], [14]. Due
to this competitive bidding system, it is likely that turbine manufacturers offered turbines at low profit margins.
Second, the United Kingdom does not have a domestic wind turbine industry, apart from some plants manufacturing
components of wind turbines from international manufacturers. Most turbines had to be imported, with the possible
exception of certain components, e.g. towers. From 1991-2001, more than twelve international manufacturers
installed wind turbines in the UK, Vestas having the largest share [15]. Thus, it is likely that the turbines reflect world
market prices. Third, over the last decade about 500 MW of capacity have been installed (with annual additions

  Enercon and a few other minor producers use a direct-drive ring generator, eliminating the requirement of a
gearbox. A drawback of this concept is the higher weight of the ring generator.
varying between 10-80 MW) [15], thus providing sufficient data. Most existing wind parks built between 1995-1999
utilize wind turbines between 500-700 kW. More recently, also large turbines up to 2.5 MW have been installed. The
average size of the wind farms is rather small (4-7 MW), and thus the data from the UK may be seen as representative
for small wind parks.

Spanish data from large wind farms: A shortcoming of the UK data is the relatively small average plant size. Specific
investment costs tend to decline with increasing wind farm size as the share of grid-connection, project planning and
other overhead costs become smaller. Therefore, we also attempted to find data for large-scale wind farms. Possible
data sources for relatively large-scale wind parks may be either the USA or Spain, which both have an number wind
farms of over 100 MW capacity. As not enough wind farm price data over an extended period of time could be found
for American wind farms, Spain was chosen as a second source of data.

In contrast to the UK, Spain hosts some of the largest wind farms in the world, while the average wind farm size is
about 20 MW. The operating capacity in Spain at the beginning of 2003 was about 4100 MW [12], while the total
technical potential is estimated to be approximately 15,000 MW [16]. However, the requirements regarding
competitive and open market are not entirely met. To support renewable energy technologies, Spain uses a feed-in
tariff. Feed-in systems have been known to discourage cost reductions in case the tariff paid is higher than the actual
costs of electricity production. Especially since 1998, rather favorable feed-in tariffs have been in place in Spain. We
therefore chose only to regard data up to 1998. Second, a number of national turbine manufacturers dominate the
market. While the market is open to foreign manufacturers, there are over 40 facilities for the production of turbines
and components in Spain, which makes adaptation of the national production capacity to new technological
developments more difficult and more costly [17].

As little other data on large wind farms are available, we used Spanish data for incorporating large-scale wind farms
in composing a global experience curve, taking into account the limitations described above. For both experience
curve based on British and Spanish data, the conversion from nominal to real prices was carried out using the
advanced economies GDP deflator of the IMF [18]. The choice for the advanced economies deflator was deemed the
most appropriate when looking at global wind prices development.

Results and discussion

Using data from the UK and Spain as substitute for the worldwide cost reduction of wind power plants, two
experience curves for small-scale and large-scale wind farms were devised (Figure 1). Both result in progress ratios
of 81-82%. As expected, the Spanish investment costs are lower than British investment costs. This is likely due to
the larger average wind farm size in Spain, allowing on average for cheaper turbines and a relatively smaller share of
other investment costs.

Figure 1 Global experience curves for wind farms, using data from British and Spanish wind farms. All data
are adjusted for inflation using the advanced economies GDP deflator [18]. The Spanish data runs from 1990-
1998. The data set for the UK runs from 1992-2001.
Sensitivity analysis revealed that when the national GDP-deflator is used to correct for inflation (instead of the GDP
deflator of the advanced economies), progress ratios may be lower, between 77-80%. Also, the chosen time frame
and possible policy measures can be of influence. When using Spanish data from 1990-2001, the progress ratio
becomes 85%. Thus, it is clear that the progress ratio may vary somewhat depending on the chosen calculation
method. Therefore we chose to use the range of 77-85% as uncertainty interval, with an average of 81%.

These progress ratios are combined with three different growth scenarios (see Figure 2). In the strong growth
scenario, a continuing increase in global installed wind farm capacity is assumed, identical to the growth rates in the
‘wind force 12’ study by EWEA [19]. This capacity would be sufficient to satisfy 12% of the world’s electricity
demand in 2020. On the other hand, in the stagnant growth scenario, annual growth rates decline to 4%, indicating an
almost saturated market in 2020. The difference in installed capacity between these two scenarios is a factor four. The
medium growth scenario is chosen in between these two.

Next, the most three scenarios are combined with the three progress ratios, thereby revealing two extreme
combinations of 250 and 575 €/kW, and a best guess of 405 €/kW (see also Figure 3). Cross-combining the strong
growth scenario with a pessimistic PR and vice versa yields investment costs of 422 and 408 €/kW respectively.
Compared to current turnkey investment costs (of approximately 1000 €/kW), this indicates an average possible price
reduction of about 60%, with margins of 43-75% reduction.

                                            10000                                                                                                               1100

                                                                                                                   Wind farm turnkey investment prices (€/kW)
                                                       Realized capacity
Cumulative installed global capacity (GW)

                                                       Strong growth, 25-15% / year
                                                       Moderate growth, 25-10% / year
                                                                                                                                                                                     Stagnant capacity growth
                                                       Stagnant growth, 25-4% / year                                                                                                              PR = 85%
                                                                                                                                                                                                     Moderate capacity growth
                                                                                                                                                                          Strong capacity                         PR = 81%
                                               10                                                                                                                         growth, PR = 77%


                                                1                                                                                                                200
                                                1990   1995       2000         2005           2010   2015   2020                                                   2001   2006          2011            2016           2021
                                                                               Time (years)                                                                                           Time (years)

       Figure 2 Growth scenarios for global installed wind                                                             Figure 3 Development of wind farm turnkey
       capacity.                                                                                                       investment costs over time, depending on the
                                                                                                                       average global growth rate of installed capacity
                                                                                                                       and on the progress ratio.

Finally, based on these turnkey investment costs and assuming an economic lifetime of 15 years, an average capacity
factor of 25%, and a real discount rate of 10%, electricity production costs may decrease from current 6.6 to 2.1 –4.0
€ct/kWh in 2020.

It must be emphasised that this analysis can only be used and interpreted for general cost trends. Specific cost of wind
electricity do depend on site conditions such as average wind speeds, accessibility, varying O&M costs et cetera.
Also other issues have not been taken into account, such as the potential additional costs associated with the large-
scale introduction of intermittent energy sources (e.g. electricity storage capacity or grid fortifications). In addition,
the results depend to large degree on the chosen assumptions regarding progress ratios and capacity growth scenarios.
Yet, even under low capacity growth rates and pessimistic progress ratios, significant cost reductions are achieved.


Regarding the methodology, it was shown that the choice of time frame, geographical area, GDP-deflator et cetera
can cause significant differences in the resulting progress ratios. Even though we narrowed down this range to 81-
82% using the most appropriate GDP-deflators and time frames, we prefer not to determine a single PR for wind
farms. Rather we would like to suggest to limit the range between 77-85% (with an average of 81%) and to use this
uncertainty interval for different scenarios.

Regarding the outcome of the scenario analysis, it is evident that estimates may range substantially depending on
capacity growth rates and progress ratio. Yet, even under low capacity growth rates and pessimistic progress ratios,
significant cost reductions may be achieved. We conclude that by 2020 a large fraction of the electricity produced by
wind farms may be able to directly compete with electricity from fossil fuels.

This research was carried out within the framework of the ‘Stimuleringsprogramma Energieonderzoek’, which was
set up by the Netherlands Organization for Scientific Research (NWO) and Novem (Nederlandse Onderneming voor
Energie en Milieu). The financial support from NWO / Novem is gratefully acknowledged.


1. Harmon C, Experience Curves of Photovoltaic Technology. IIASA: Laxenburg, Austria, 2000; p. 22.
2. Claeson Colpier U and Cornland D. The economics of the combined cycle gas turbine. An experience curve
    analysis. Energy Policy 2002; 30(4):309-316.
3. Tsuchiya H. Fuel Cell Cost Study by Learning Curve. In: Proceedings of the International Energy Workshop
    2002. Stanford University, USA, 2002.
4. McDonald A and Schrattenholzer L. Learning rates for energy technologies. Energy policy 2001; 29(4):255-261.
5. Neij L, Cost dynamics of wind power. Energy 1999; 24(5):375-389.
6. Durstewitz M and Hoppe-Kilpper M. Wind Energy Experience Curve from the German "250 MW Wind"-
    Programme. in IEA International Workshop on Experience Curves for Policy Making - the case of energy
    technologies. 1999. Stuttgart, Germany.
7. Mackay RM and Probert SD. Likely market-penetrations of renewable-energy technologies. Applied Energy
    1998;. 59(1):1-38.
8. Ibenholt K. Explaining learning curves for wind power. Energy Policy 2002; 30(13):1181-1189.
9. Lund PD. An improved market penetration model for wind energy technology forecasting. in EWEA special topic
    conference, The economics of wind energy. Finland. 1995.
10. Klaassen G, Larsen K, Miketa A, and Sundqvist T. The impact of R&D on Innovation for wind energy in
    Denmark, Germany and in the United Kingdom. Submitted to Environmental and Resource Economics, 2002.
11. Junginger M, Faaij A, and Turkenburg WC. Global experience curves for wind farms. Energy policy. Accepted
    for publication, 2003.
12. Milborrow D, Tishler C, Harrison L, O'Bryant M. The Windicator. Windpower Monthly, April 2003:78.
13. Hill A, Personal communication on the Non-Fossil Fuel Obligation, British Wind Energy Association, 2002.
14. International Energy Agency. IEA Wind energy annual report 1999. 1999.
15. British Wind Energy Association. Operating Wind Farms in the UK & capital cost breakdown of a typical 5 MW
    plant, available at: http://www.bwea.com/map/list.html, accessed in December 2002.
16. Ayuso Ortiz JR, Romero Zamora JJ, and Garcia VO, Wind Energy in Spain, in European Wind Energy
    Conference and Exhibition. Hamburg, Germany, 2002.
17. McGovern M and Harrison L, Spain's structural faults. Wind Power Monthly, December 2002.
18. International Monetary Fund. IMF World Economic Outlook (WEO). Recessions and Recoveries. Statistical
    appendix. Washington, D.C., U.S.A. 2002.
19. EWEA and Greenpeace, Wind Force 12. A blueprint to achieve 12% of the worlds electricity from wind power by
    the year 2020, 2003.

To top