Penetrating Insights | IEEE Power & Energy Magazine
IEEE Power & Energy Society

Penetrating Insights

Lessons Learned from Large-Scale Wind Power Integration

Wind power is one of the most cost-effective sources of renewable energy. In 2010, Denmark received 22% of its annual electric energy from wind generation; 17% of Portugal’s load was served by wind, and the U.S. state of Texas (the center of the U.S. oil industry) produced 6.4% of its electricity from wind power. Ireland now receives more than 10% of its annual electrical energy from wind, and there are periods when wind supplies more than 50% of Ireland’s total load. Each year, wind power is contributing increasing portions of world electricity production. Is there a limit to what portion of a grid’s total energy can come from wind? What technical characteristics of power grids would limit the amount of wind energy that could be accommodated? What changes in technical design or operating practices could increase a grid’s ability to accommodate more wind energy?

Over the past decade, much has been learned about integrating high levels of wind power into bulk power grids while successfully managing the increased variability and uncertainty inherent in variable energy sources such as wind. These lessons have been learned from renewable integration studies conducted by numerous regional power grids and from the actual operating experience of power grids leading the world in wind energy utilization (see “Wind Integration Costs”). This article summarizes some of the key lessons. We first describe a set of “universal truths” that are generally applicable to all power grids seeking to integrate more wind energy. We then provide a few specific examples, emphasizing the unique perspectives and experiences of systems in Europe, China, and North America.

Lessons Learned: Wind Plant Interconnection

The first and perhaps most basic aspect of successful wind power integration relates to the wind plant itself. Early designs of wind plants were very simple, typically with induction generators that delivered power to the grid when the wind blew, consumed reactive power in proportion to real power generated, and tripped off-line whenever the terminal voltage dipped due to system disturbances. Such plants caused unacceptable voltage fluctuations during normal operation and caused major loss-of-generation events in response to otherwise minor system disturbances. It’s impossible for a power grid to accommodate very many wind plants of this type without reducing system reliability and performance. Grid codes have made substantial progress in resolving these problems by requiring that wind plants meet a set of minimum performance criteria. Several wind plant design features that contribute to the reliability of the power grid and enable increased levels of wind penetration are summarized below.

Low-Voltage Ride-Through (LVRT)

Figure 1. LVRT and ZVRT requirements.

Figure 1. LVRT and ZVRT requirements.

This design feature enables wind turbines to remain in operation during and after grid faults and other events that may momentarily reduce grid voltage. These requirements have been evolving and vary from region to region, but they share two basic characteristics: turbines must remain in operation during a severe voltage drop for several hundred milliseconds, and they must be able to withstand a recovery period with depressed voltage for a few seconds. Figure 1 shows a pair of typical voltage ride-through requirements. The ability to tolerate a voltage of zero—called zero-voltage ride-through (ZVRT)—is becoming the industry norm. The ability of certain wind plants to tolerate the high voltages that sometimes occur following the clearing of grid faults has led to development of similar requirements for high-voltage ride-through. These requirements are less mature and are the subject of ongoing discussions within the industry.

Reactive Power and Voltage Regulation

Conventional thermal and hydro generation facilities have traditionally provided -reactive power to support the voltage of the power grid. Such facilities have synchronous machines, typically capable of operating in power factor ranges of -approximately ±0.90 or ±0.95. Voltage regulators on the excitation systems of these plants provide the primary voltage control function for the power grid. State-of-the-art wind turbines are also able to provide reactive power and regulate voltage. And to the extent that wind generation displaces thermal generation when the wind is blowing, wind generation also needs to provide reactive power to support and regulate system voltage. Many existing wind plants have been interconnected without these very basic capabilities, and power grids in some regions have suffered from depressed voltages, excessive voltage fluctuation, and in some cases, inability to deliver full rated power. This situation has improved dramatically with the introduction of interconnection requirements and grid codes that require minimum levels of reactive power and voltage control capabilities in wind plants.

Wind Integration Costs

Figure S1. System cost versus renewable energy penetration.

Figure S1. System cost versus renewable energy penetration.

It is common to hear questions such as “What is the limit of wind power the system can handle?” In practice, there is never a single power or energy level beyond which integration becomes impossible. Rather, the integration of wind energy is more and more challenging as the penetration level increases. Figure S1 qualitatively shows that the marginal cost of accommodating each increment of renewable energy invariably increases. “System cost” represents numerous factors that have variously been ascribed to “integration,” including increased cost of ancillary services, higher marginal energy costs for conventional thermal generation, higher O&M expenditures due to increased unit cycling, increased curtailment of renewable energy resources, and so on. Of course, this figure doesn’t reflect the benefits, such as lower emissions and reduced fuel costs. In every system, there are physical and institutional impediments to the integration of renewable energy. These factors, which can include inadequate transmission resources and inflexible operating rules, tend to increase system costs and make integration of renewable energy difficult, even at lower levels of penetration. Conversely, grids that utilize wind forecasts for unit commitment, increase the flexibility of thermal generation resources, adopt flexible and rapidly adjustable exchange schedules with neighboring systems, and utilize demand response for reserves and other ancillary services can successfully accommodate higher renewable energy penetrations with only modest increases in system cost.

Multiplant Coordination

Wind plants tend to grow in bunches. This is not surprising, since multiple project developers seek to install wind plants in areas with the best wind resources. With multiple wind plants having requirements to provide reactive power and control grid voltages, problems have emerged when the volt/var controllers in nearby plants have been incompatible, resulting in poor voltage control and counterproductive var flows. There have been numerous reports of one or more plants at maximum var output while neighboring plants are absorbing reactive power. The industry is migrating toward two solution methods:

  • The first is passive coordination, whereby each plant regulates voltage with a droop-based controller. The droop function enables the independent voltage regulators in nearby plants to coexist and balance voltage-support duties. This is often the preferred approach, as it lets the plants operate independently. It is similar to the approach commonly used in multiunit thermal or hydro plants.
  • The second is centralized control, whereby several plants are coordinated by a single voltage control unit that delivers individual volt/var control commands to each plant so that all plants share reactive power and voltage-support duties in a balanced manner. This -approach is more attractive when the plants have different control capabilities or have older generations of wind turbines.

Lessons Learned: Grid Operations

Figure 2. Example of time series of normalized power output from (a) a single wind turbine, (b) a group of wind power plants, and (c) all wind turbines in Germany (data recorded 21–31 December 2004; source: Holttinen et al., 2009).

Figure 2. Example of time series of normalized power output from (a) a single wind turbine, (b) a group of wind power plants, and (c) all wind turbines in Germany (data recorded 21–31 December 2004; source: Holttinen et al., 2009).

Wind power variability decreases as the geographical area over which wind turbines are dispersed increases and the power they generate is aggregated (see Figure 2). This is due to the fact that spatial variation in the wind field increases with geographical distance. The so-called spatial smoothing is very dominant for fast changes in wind power output, i.e., second-to-second and minute-to-minute changes. For slower variations the effect is less strong, and dispersed wind power installations exhibit significant variations from hour to hour, especially under the rare condition of a storm front passing over an area and causing the wind turbines to shut down.

Likewise, the predictability of wind power is improved when forecasts consider dispersed wind power production rather than only localized wind power production. This is because the forecast errors at different sites cancel each other out to some extent. As should be expected, forecast errors increase with forecasting horizon. Two different forecasting regimes can be distinguished:

  • For short-term forecasts up to six hours ahead, the real-time measurements of the wind power production play a significant role, and the wind power forecast errors show a steep increase with forecast horizon.
  • For forecasts from six hours ahead to 72 hours ahead, the numerical weather predictions used in the forecasting tools play a dominating role, and the wind power forecast errors increase slowly with forecast horizon.

Because forecast horizons shorter than six hours have significantly fewer forecast errors, power markets need to facilitate the rescheduling of power production so it is close to the actual operating hour. One effective method for minimizing the operational impacts of short-term forecast errors involves intraday (and intrahour) markets that can make use of flexible generation or demand-side resources to respond quickly when actual wind power falls significantly below forecast wind power.

Commercial forecasting tools have achieved significant reductions in forecast errors during the last ten years. For example, the average day-ahead forecast error in the E.ON Netz control area in Germany, given as the rms error expressed as a percentage of total installed wind power, was reduced from 10% in 2001 to 6.5% in 2006. Usage of these wind power forecasts in power system operation is also improving. Transmission system operators (TSOs) and independent system operators (ISOs) each day estimate the amount of different types of reserves needed to ensure stable operation. Traditionally, this is done by means of N–1 criteria, ensuring reserves to cover the outage of the largest power in-feed in the balancing area. With increasing shares of wind power, the reserve criteria need to include the effect of wind power forecast errors. As shown by recent academic papers such as those by Matos and Bessa, this should be done using probabilistic wind power forecasts that are dynamically updated during the day. Probabilistic wind power forecasts not only produce the future expected wind power production (a so-called “point forecast”) but also information about the probability distribution of the future wind power (e.g., in the form of forecasting different confidence bands for the wind power production). Ideally, the algorithm for estimating reserves should combine forced outages, load forecast errors, and wind power production errors in an estimation of the probability distribution of the total forecast error for a certain forecast horizon.

Locations with high wind speeds suitable for installing wind turbines are often far away from the urban areas with high electricity consumption. An example is the Great Plains area in the United States, with significant wind resources in thinly populated areas with low electricity consumption. Exploiting wind resources in such locations requires significant reinforcements of the transmission grid between the wind resource and the load centers.

The distance between wind resources and load centers, combined with spatial smoothing of the wind power variability and wind power forecast errors, is a strong incentive for creating large balancing areas with few internal transmission grid bottlenecks in power systems with relatively large penetrations of wind power. This can be achieved either by merging smaller balancing areas into bigger ones or, alternatively, by creating well-functioning power exchange schemes between the balancing areas. The latter possibility requires some harmonization of power market rules in the different balancing areas, as follows:

  • The clearing of the day-ahead markets in each balancing area needs to take the power exchange with neighboring balancing areas into account. In Europe, a so-called “market-coupling scheme” is used for this purpose. Europe is moving very quickly at the moment to create a day-ahead market covering the northern part of Western Europe by using market coupling between regional day-ahead markets.
  • It should be possible to procure some amount of the required power reserves (“primary,” “secondary,” and “tertiary” power reserves, in the terminology of the UCTE grid code) from neighboring balancing areas. Great care must be taken when designing such schemes not to jeopardize the security of supply.
  • Intraday and even intrahour rescheduling of the power exchange between balancing areas must be possible to enable the sharing of flexible production resources and merging of forecast errors across balancing areas. This requires harmonization of the rescheduling schemes (e.g., intraday markets) in place in each balancing area.

The variability and partial predictability of wind power creates an increased demand for flexible production, storage, and demand resources in the power system. A few “smart grid” concepts are promising for this purpose. The idea is to use information and communication technologies in combination with advanced monitoring and control to make the electricity consumption more flexible in the face of fluctuations in the renewable power production. Suitable candidates for flexible demand-side units include electrical space heating and cooling, electrical water heating, refrigerators and cooling houses, water pumping, and—in the future—electric vehicles. The electricity consumption of such units can be time-shifted (within limits imposed by consumer preferences and technical restrictions). Hence, electricity consumption will be moved to time periods with the lowest power prices, which to a large extent will be the periods during the day with the lowest forecast net load (load minus wind power production). Many of the flexible demand-side units will only be flexible within the space of a day (i.e., they can time-shift consumption from day to night but not from Monday to Wednesday). Wind power production sometimes has periods with high or low wind power production covering several days. Handling such periods might require production from thermal or hydro power plants or possibly large-scale electricity storage.

Day-ahead power prices decrease when wind power is increasing due to the near-zero marginal costs of wind power production. Hence, the energy revenues of the thermal generators are reduced. The role of thermal power generators will therefore shift in the future from selling energy to selling flexibility, i.e., back-up power, up- and down-regulation, and reserves. There will be an increased focus on the technical flexibility of the thermal units, especially quick start-up, high ramp-up and ramp-down rates, low minimum stable production levels, and high part-load efficiency. Some generation designs offer high flexibility without sacrificing high production efficiency.

The socioeconomic value and profitability of different types of electricity storage have been analyzed widely in the academic literature. Most analysis is based on model studies of power systems with different amounts of wind power and electricity storage. Apart from the few geographical locations offering the possibility of large-scale hydropower with hydropower reservoirs, it is very difficult to find a convincing business case for electricity storage in a power system with little wind curtailment, due to the significant storage losses and high investment costs of electricity storage. A study of Ireland by Tuohy and O’Malley recently found pumped hydro storage becoming economically attractive at an average annual wind power penetration of approximately 50%. Short-term energy storage facilities (flywheels and batteries) are finding some near-term applications to provide ancillary services, including regulation and reserves.

The European Experience

Figure 3. European wind energy penetration in 2010 (total for all of the EU is 5%).

Figure 3. European wind energy penetration in 2010 (total for all of the EU is 5%).

In Europe, the first balancing area to reach high penetration levels of wind power was West Denmark, where the wind penetration has been more than 20% of yearly energy production since 2002. All of Denmark reached 20% wind penetration in 2007. In 2010, the Iberian peninsula approached the 20% level: Portugal produced 17% of its electricity consumption by means of wind power, and Spain 16%. Figure 3 illustrates recent wind power penetration in Europe.

Unlike North America and China, where most wind generation is from large plants, much of Europe’s wind generation is connected to the distribution level. This means that the TSOs do not have direct access to the real-time wind plant information. As there is a need to monitor the output level of wind power plant fleets in real time, this will require some extra effort on the part of the TSOs. In Denmark and Germany, the TSOs use “up-scaling tools” to estimate total wind generation, based on regional measured wind power plant information (in Germany, about 20% of the wind power plants are measured directly). In Spain and Portugal, the problem has been solved by requesting all wind generation (and other renewable distributed generation) to connect to centers that will communicate with the TSO. The centers will collect all the generation data in real time and will also be able to control the output of wind plants. The TSOs will then have access to data and also the means to control the output in critical situations. Curtailments so far have remained at a moderate level, but some signs of increasing curtailments have been reported.

The grid codes for wind power plants have also evolved based on experience. In the beginning, turbines were required to disconnect from the grid in the case of grid disturbances such as voltage or frequency drops. This kept the fault clearance process clearly defined for grid operators. With more than 80 GW of wind capacity installed in Europe, however—more than 20 GW of which is concentrated in one country—a grid disturbance could trip off several gigawatts of wind power in one balancing area, even when wind power is operating at partial load. The maximum instantaneous trip-off (largest contingency) is 3,000 MW in the synchronous grid of continental Europe (UCTE). The need to have wind turbines equipped with LVRT was recognized in the late 1990s and early 2000s. The German Energy Agency (DENA) wind integration study in Germany revealed that the wind power installed in 2005 was already jeopardizing the largest permissible loss-of-source contingency. In Spain, the grid integration study by the TSO Red Eléctrica de España (REE) showed that in order to reach the wind power targets of the energy policy, LVRT would be needed for most plants. In both countries, retrofitting existing wind turbines with LVRT has been encouraged or required, and the grid code now requires LVRT in all new wind plants.

The experience of wind integration in Denmark has highlighted the benefits of operating with good interconnection capabilities in a larger market with flexible hydropower resources. The day-ahead market is used for most wind power and other production in Denmark and other Nordic countries. The imbalance price arising from forecast errors has typically been 2–3h/MWh of wind power produced. The transmission upgrades needed to accommodate wind power have been highlighted in several countries, including Germany, Ireland, and the United Kingdom. Potential barriers to reaching the current renewable energy targets in Europe include low public acceptance of transmission lines as well as lengthy planning processes. Offshore grids are being considered as one possible way to overcome these barriers.

Developments in Ireland are also stimulating a high level of interest. The feasibility of up to 37% wind penetration is being studied for this small island system with very limited interconnections to its neighboring island, Great Britain. The current 14% annual wind energy penetration has so far required some curtailments to keep the instantaneous penetration level (during a single hour) at a maximum of 54%. The All Island Grid Study as well as the follow-up studies on enabling high penetrations have shown that the instant penetration level of asynchronous generation can go as high as 70% when certain technical measures are taken at the distribution level (such as changing rate-of-change-of-frequency relay settings to avoid premature tripping of generation).

The Chinese Experience

Figure 4. Cumulative installed wind capacity in China (source: CWEA).

Figure 4. Cumulative installed wind capacity in China (source: CWEA).

China’s wind generation began growing rapidly with the development of concession wind projects from 2003 and the issue of the Renewable Energy Law in January 2006. The cumulative installed wind capacity was 1.3 GW in 2005. Since then, it has doubled every year, and it reached 44 GW in 2010 (see Figure 4). According to the government’s plan, wind generation capacity will be more than 150 GW by 2020.

The capacity of most wind plants in China is 50 MW or more, and many of them have capacities of more than 100 MW. In addition, wind resources are not evenly distributed. It is very common for a wind-rich area to have many wind plants. The total installed wind capacity in one area can easily be more than 1,000 MW. For example, the Jiuquan district of Gansu Province has more than 5 GW of installed wind capacity, and the western Inner Mongolia region already has more than 10 GW. To make the situation even more challenging, most Chinese wind resources are located in remote areas where the load is small and the grid network is relatively weak, and the wind energy must be transported to load centers by means of long-distance transmission lines.

Before 2005 there was no specific grid code for the interconnection of wind plants, since wind generation capacity was tiny and its impact on the grid was negligible. In 2005, the first national-level grid code on wind grid-interconnection (GB/Z 19963-2005) was issued. This grid code was not mandatory, however, and was only used as a reference. Furthermore, it was quite weak and included no actual requirements for grid-friendly features such as reactive power capability, active and reactive power control, voltage regulation, and fault ride-through, all of which are quite common in the grid codes of other countries.

Due to these weak grid code requirements, most wind plants were built without the grid-friendly functionality mentioned above, and the overall performance of the system has not been as good as expected. Many wind plants in some areas performed poorly, were not able to deliver all available wind energy, and caused operational problems such as voltage collapse and cascaded tripping during system contingencies.

For example, one event in February 2011 resulted in the tripping of 598 wind turbines and the loss of 840 MW of wind power in the Jiuquan district. This was all due to a fault in one wind plant. One month later, another incidence of cascaded tripping was reported in the same area: 702 wind turbines tripped and more than 1,000 MW of wind power were lost. On the same day, a similar accident occurred in another region. If these wind plants had voltage regulation and fault ride-through capability, these problems could have been avoided.

Such operational problems caused grid operators a great deal of concern. In order to mitigate the adverse effects of wind generation, State Grid Corporation of China (State Grid) issued its own grid code (Q/GDW392-2009) on grid interconnection of wind plants in late 2009. Compared with GB/Z 19963-2005, the requirements of this code are much more stringent. It includes new requirements for grid-friendly features, as summarized below:

  • Active power control: It requires wind plants to have an active power control system and be able to accept the active power set-point command from the grid operator and control active power and ramp rate accordingly.
  • Reactive power and voltage regulation: Wind plants are required to compensate all the reactive power loss and line charging inside the wind plant and outgoing transmission lines. Wind plants are also required to be capable of regulating the point of interconnection (POI) voltage using their own dynamic reactive power, in accordance with instructions from the grid operator.
  • LVRT: The LVRT requirement is shown in Figure 5. Wind turbine generators must be capable of staying online for 625 ms when the POI voltage is 0.2 pu. The grid code also requires postfault active power recovery at a rate of 10% of rated power per second.
Figure 5. State Grid’s LVRT requirement for wind turbines in China.

Figure 5. State Grid's LVRT requirement for wind turbines in China.

In addition, this code requires wind plant modeling data and parameters, testing for plant interconnection performance and grid compliance, and wind power forecasting.

Measures have also been taken by national authorities to address the operational problems caused by wind generation due to the lack of grid-friendly functions. In 2010, a revision of national standard GB/Z 19963-2005 was initiated. It is expected that this new version of the standard will be more stringent and the features mentioned above will be included.

In addition, at the end of 2010 the National Energy Administration (NEA) imposed a regulation that requires a grid interconnection function test for wind turbines in order for the wind plants to obtain grid access. This took effect on 1 January 2011. The test examines active and reactive power control, LVRT, grid compatibility, the wind model, and power quality.

Newly installed wind plants will soon have grid-friendly features with the adoption of the revised grid code from State Grid, NEA’s regulation on mandatory grid interconnection testing, and the more stringent national grid code. Some of the previously installed wind plants are also being required to retrofit with new features. With these measures, it is expected that the performance of wind plants will improve and that many of the operational problems described above will be avoided in the future.

The North American Experience

Despite an enthusiastic beginning with wind power in the early 1980s (mostly in California), the growth of wind power in North America was not significant until 2000–much later than in Europe. As such, many of the lessons learned by the Europeans served as valuable guidance for policies and practices for large-scale wind integration in North America. Requirements for reactive power, LVRT, and SCADA communications with grid operators were implemented early in the growth cycle, thereby avoiding some major issues faced by the Europeans. A few recent lessons learned in North America relate to the benefits of balancing area cooperation, the need for adequate transmission, and a having a proper grid frequency response.

Balancing Area Cooperation

Accommodating the variability of wind generation is much easier when that variability is balanced over a larger area. Geographic diversity cancels out some of the variability, leaving less to be balanced by the grid operator. Portions of North America have very small balancing areas with only a few inflexible power plants and area interchange (import/export) schedules that can only be adjusted hourly. Balancing even a small amount of wind generation in such a system is very challenging and inevitably leads to excessive curtailment (spilling) of wind energy.

Some utilities in the western United States are developing a promising solution: an energy imbalance market (EIM). An EIM is a voluntary tool that enables participants to share subhourly energy imbalances over a wide area covering multiple balancing areas. Areas with imbalances can submit their needs to the EIM. Owners of generation or demand-side resources can submit offers to the EIM. The market operator determines the net imbalance for all participating areas, performs security-constrained economic dispatch (and adjustments if congestion occurs), and sends dispatch commands to generators or demand-side resources. This process enables smaller balancing areas to achieve some of the operational flexibility and benefits of larger operating areas and ISOs while retaining their small size and jurisdictional independence. Studies of this concept by E3, the National Renewable Energy Laboratory (NREL), and others show promising results. Operational tests are expected soon.

Transmission Adequacy

Transmission (or lack thereof) has been another major challenge in North America, and lessons have been learned there as well. The roots of the problem lie in a classic “chicken and egg” situation: Transmission owners would not build new lines unless there were generating plants to utilize them, while plant developers could not get funding to build new plants in areas without adequate transmission. As a result, some of the best wind resource areas were not developed. One emerging concept, the renewable energy zone (REZ), has broken through that barrier. The concept involves a regional planning process that identifies geographic areas with high potential for renewable resources. Several large-scale transmission expansion plans are developed to enable delivery of a desired level of energy (say, to meet a regional renewable energy requirement). Alternative plans are compared, and the best plan is selected for development. Once the transmission system design is approved and scheduled for construction, wind plant developers are able to select locations for new plants and begin construction. Texas has had ground-breaking success with this approach, implementing its competitive REZ (CREZ) transmission plan with 2,300 mi of new 345-kV lines to facilitate 11.5 GW of additional wind energy. Other regions within the western United States are learning from the Texas experience and pursuing REZ plans of their own. In another related development, the Federal Energy Regulatory Commission (FERC) recently released its Order No. 1000, which imposes new requirements for transmission planning and cost allocation. The order requires transmission providers to participate in regional planning processes and coordinate with each other to produce efficient and cost-effective solutions. It further orders a regional cost allocation method that satisfies six principles. It is anticipated that this order will smooth the path toward the new transmission development necessary for public policies that could include harvesting new wind energy resources in remote locations.

Grid Frequency Regulation

The primary frequency response of several North American grids has been declining for many years. This was the subject of a FERC-sponsored report prepared by Lawrence Berkeley National Lab in late 2010. While declining frequency performance (as measured by deeper frequency excursions for loss-of-generation events) predates the recent growth of wind generation in North America, the addition of large amounts of wind (or photovoltaic solar) generation could potentially exacerbate the problem. The concern is most acute during light load conditions with wind plants at high output, where economics dictates that fewer synchronous generators will be operating, and overall grid inertial response will consequently be reduced.

Although wind turbines have a very large rotating inertia, their natural response is not sensitive to changes in grid frequency. That is, the power output does not change when grid frequency declines. In recognition of the grid’s need for frequency response, however, wind turbine manufacturers have developed turbine control functions that temporarily increase power output when frequency declines by withdrawing energy from the rotating inertia of the turbine. If wind plants are equipped with such functions, one contributing factor to the decline in North American grid frequency response can be eliminated.

Looking Ahead

Wind power has grown to astonishing levels over the past decade, and that growth in penetration has introduced significant challenges to power grid operation–from increased variability and uncertainty in operations to new types of dynamic events and phenomena that could threaten grid reliability and security. Some regions of the world have paved the way by learning valuable lessons from either forward-looking studies or the pains of unfortunate operational experiences. And thankfully, they are sharing the benefits of their learning with those who are following in their footsteps. So far, each challenge has been answered by one or more possible solutions, enabling the continued growth of renewable energy from wind. May the learning, the sharing, and the growth continue!

For Further Reading

G. Gregor, R. Browsword, G. Kariniotakis, M. Denhard, and C. Draxl, “The state-of-the-art in short-term prediction of wind power: A literature overview,” Deliverable D-1.2, 2nd ed. AnemosPlus project, 2011.

H. Holttinen, P. Meibom, A. Orths, F. van Hulle, B. Lange, M. O’Malley, J. Pierik, B. Ummels, J. O. Tande, A. Estanqueiro, M. Matos, E. Gomez, L. Söder, G. Strbac, A. Shakoor, J. Ricardo, J. C. Smith, M. Milligan, and E. Ela. (2009). Design and operation of power systems with large amounts of wind power: Final report, IEA WIND Task 25, Phase one 2006–2008 (VTT Tiedotteita—Research Notes 2493). VTT Tech. Res. Centre of Finland, Helsinki, p. 229. Available Online

M. A. Matos and R. Bessa, “Setting the operating reserve using probabilistic wind power forecasts,” IEEE Trans. Power Syst., vol. 26, no. 2, pp. 594–603, May 2011.

A. Tuohy and M. O’Malley. (2011, Feb.). Pumped storage in systems with very high wind penetration. Energy Policy. Available Online


Richard Piwko is with GE Energy Consulting in Schenectady, New York.

Peter Meibom is with the Technical University of Denmark.

Hannele Holttinen is with the Technical Research Centre of Finland, VTT.

Baozhuang Shi is with GE Energy Consulting in Shanghai, China.

Nicholas Miller is with GE Energy Consulting in Schenectady, New York.

Yongning Chi is with China Electric Power Research Institute in Beijing, China.

Weisheng Wang is with China Electric Power Research Institute in Beijing, China.

In This Issue

Feature Articles

Departments & Columns

Upcoming Issue Themes