MAIN PUBLICATION :

• APPENDIX
• ABREVIATIONS
• GLOSSARY
• REFERENCES
• EVENTS

Understanding variable output characteristics of wind power: variability and predictability

Wind power: a variable output source embedded in a variable electricity system

Since wind energy is a technology of variable output, it needs to be considered as just one aspect of a variable, dynamic electricity system. At modest penetration levels, the variability of wind is dwarfed by the normal variations of the load. It is impossible to analyse wind power in isolation from other parts of the electricity system; and all systems differ. The size and the inherent flexibility of the power system are crucial aspects in determining the system’s capacity to accommodate a large amount of wind power.

The variability of wind energy needs to be examined in the wider context of the power system, rather than at the individual wind farm or wind turbine level. The wind does not blow continuously at any particular site; and there is little overall impact if the wind stops blowing in a certain area, as it is always blowing elsewhere. This lack of correlation means that at system level, wind can be harnessed to provide stable output regardless of the fact that wind is not available all the time at any particular site. So, in terms of overall power supply, it is largely irrelevant when a wind power plant produces zero power for a time, due to local wind conditions. Moreover, until wind becomes a significant producer (supplying around 10% of electricity demand) there is a negligible impact on net load variability.  

Wind power varies over time, mainly under the influence of meteorological fluctuations. The variations occur on all time scales: seconds, minutes, hours, days, months, seasons and years. Understanding these variations and their predictability is of key importance for the integration and optimal utilisation in the power system. Electric power systems are inherently variable in terms of both demand and supply. But they are designed to cope effectively with these variations through their configuration, control systems and interconnection.

Variable output versus intermittency Wind power is sometimes incorrectly considered as an intermittent energy source – however, this is misleading. At power system level, wind power does not start and stop at irregular intervals (a characteristic of conventional generation), as is suggested by the term intermittent. Even in extreme conditions, such as storms, it takes hours for wind turbines in a system area to shut down, Moreover, periods with zero wind power production are predictable and the transition to zero power is gradual. Also worthwhile considering is the technical availability of wind turbines, which is at a very high level (98%), compared to other technologies. Another advantage of wind power in this respect is its modular and distributed installation in the power system. Breakdown of a single unit has a negligible effect on overall availability. So, the term ‘intermittent’ is inappropriate for system-wide wind power and the term ‘variable-output’ should be used instead.

Short-term variability

The analysis of data available from operating wind farms and meteorological measurements at typical wind farm locations allows us to quantify the  variations in net wind power output that can be expected for a given time period (within the minute or hour, or during the course of several hours). The distinction between these specific time scales is made, since this type of information corresponds to the various types of power plants for balancing. The results from analyses show that the power system can handle this short-term variability rather well. System operators only need to deal with the net output of large groups of wind farms, and the wind power variability is viewed in relation to the level and variation in power demand.

• Within the minute: The fast variations (seconds to minute) of aggregated wind power output (as a consequence of turbulence or transient events) are quite small, due to the aggregation of wind turbines and wind farms, and are hardly felt by the system.
• Within the hour: The variations within an hour are much more significant for the system. However, they should always be considered in relation to demand fluctuations. Local variations are largely equal to geographical diversity, and will generally remain inside ±5 % of installed wind power capacity at regional level.

The most significant variations arise from the passage of storm fronts, when wind turbines reach their storm limit (cut-out wind speed) and shut down rapidly from full to zero power. However, due to the averaging effect across a wind farm, the overall power output takes several minutes to reduce to zero. In general, this is only significant in relatively small geographical areas, since in larger areas, it takes hours for the wind power capacity to cease during a storm. For example, in Denmark – a small geographical area -- on 8 January 2005, during one of the biggest storms for decades, it took six hours for the installed wind power in the West Denmark area to drop from 2000 MW to 200 MW (5 MW/minute). The passage of a storm front can be predicted and technical solutions are available to reduce the steep gradient, such as the provision of wind turbines with storm control  .

These intra-hour variations will be an issue for power system reserves used for balancing, when wind power penetration reaches the point at which variations are equal to variations in demand (when 5-10 % of annual electricity demand is produced by wind power).

• From hour to hour variations: The variations several hours ahead affect the scheduling of the power system. For system operation, the variation in itself is a problem; it is the uncertainty of how accurately the variation can be predicted that is significant. The uncertainty of wind power predictions should always be considered in relation to the errors in demand forecasts. There is much work being conducted in this area and it is clear that solutions are available.
  
  
  
  

Long-term variability

The slower or long-term variations of wind power relevant for integration in the power system include the seasonal and inter-annual variations, caused by climatic effects. These are not particularly important for the daily operation and management of the grid, but play a role in strategic power system planning.

• Monthly and seasonal variations: These variations are important for electricity traders that have to deal with electricity forward contracts, where wind power volume has an influence on price. They are also important for power system planning. However, it appears that for both electricity trading and system planning purposes, these deviations, resulting from annual statistics of wind power produced, can be sufficiently hedged.
  
  
  
  
• Inter-annual variations: These variations are relevant for long-term system planning, rather than daily power system operation. The annual variability of long-term mean wind speeds at sites across Europe tends to be similar, and can be characterised by a normal distribution with a standard deviation of 6%. The inter-annual variability of the wind resource is less than the variability of hydro inflow. In addition, on a power system level, the annual variations are influenced by the market growth of wind power and the projected onshore/offshore ratio.

Effects of aggregation and geographical dispersion

Due to the wide regional distribution of wind plants, short-term and local wind fluctuations are not correlated and therefore largely balance each other out. As a result, the maximum amplitudes of wind power fluctuations experienced in the power system are reduced. This phenomenon has been extensively studied throughout Europe.

Whereas a single wind farm can exhibit (hour to hour) power swings of up to 60% of capacity, the maximum hourly variation of 350 MW of aggregated wind farms in Germany does not exceed 20% (ISET). For larger areas, such as the Nordel system, which covers four countries, the largest hourly variations would be less than 10% of installed wind power capacity, if the capacity was distributed throughout all the countries. The geographical spread of wind farms across a power system is a highly effective way to deal with the issue of short-term variability; so the more widespread the wind farms, the lower the impact from variability on system operation.

The effect of reduced wind power variability increases with the size of the area considered. Ideally, to maximise the smoothening effect, the wind speeds occurring in different parts of the system should be as uncorrelated as possible. Due to the typical sizes of weather patterns, the scale of aggregation needed to absorb a storm front is in the order of 1500 km. By aggregating wind power over large regions of Europe, the system can benefit from the complementarities of cyclones and anticyclones over Europe (Fig 2.3). The economic case for smoothing wind power fluctuations by utilising transmission capacity (rather than by other means), is an important area of investigation, for example in the TradeWind project.

In addition to the advantage of reducing the fluctuations, the effect of geographically aggregating wind farm output is an increased amount of firm wind power capacity in the system. This will be explained further in Wind power contribution to system adequacy.

Figure 2.3. Example of smoothing effect by geographical dispersion. The figure compares the hourly output of wind power capacity in four situations, calculated with simulated wind power. The simulations are based on December 2000 wind speeds and wind power capacity estimated for the year 2020. (www.trade-wind.eu).

Figure 2.4. Combined wind energy production from Europe and Northern Africa (Morocco) produces a monthly pattern that matches demand in Europe and Norway (Source: Czisch 2001).

Load duration curve

One method of representing the smoothing effect of aggregation on system scale is the load duration curve of wind farms, which gives the frequency distribution of the partial load states of generated wind power (see Figure 2.5). The effect of aggregating wind power is a flattening of the duration curve. This means that when wind power is aggregated over a large area:

• the effective number of hours when wind power is available increases;
• the number of hours with zero or low power diminishes, while the maximum value of instantaneous aggregated power produced is decreasing.

As part of the TradeWind project, a simulation was made for the EU-27, with an assumed wind capacity distribution for the year 2020. The effect of geographical aggregation means that the maximum aggregated instantaneous wind power is only 6O% of the total capacity of 200 GW (Tande et al, Madrid 2008)

 

Figure 2.5. Duration curves for the ‘wind year 2000’, Denmark and Nordic countries, assuming equal wind capacity in each of the 4 countries, Source: Holttinen

The need for interconnection

It is impossible to optimally aggregate large-scale wind power without a suitably interconnected grid. In this context, the grid plays a crucial role in aggregating the various wind farm outputs installed at a variety of geographical locations, with different weather patterns. The larger the integrated grid – especially beyond national borders - the more pronounced this effect becomes. This effect is equivalent to using the grid to aggregate demand over interconnected areas. In order to make best use of this effect, the present transmission system in Europe needs to be upgraded. Ideally, the interconnection capacity should be increased, and the rules governing the power exchange between countries should be adapted to ensure that interconnectors are always available for physical flow.