3.1 Wind forecasting in grid and market operations

 

During the last decades, wind energy has demonstrated to be an important alternative to the fossil fuels. According to the Crompton greaves consumer share price forecast World Wind Energy Association, the total installed capacity around the world reached 435 GW at the end of 2015, achieving a growth rate of 17.2% during this year.

Furthermore, in some electrical systems, wind power represents a significant percentage of the total energy production which must be taken into account for optimized market and grid operations. However, as opposed to conventional electricity production, wind power is intermittent and difficult to predict, mainly due to the complex dynamics associated to the wind flow.

This intermittency and the distributed production

This intermittency and the distributed production, derived from the geographical spread of wind farms, are two characteristics that make the integration of wind energy in the electrical system difficult. These characteristics convey a diversity of problems as wind power is more present in the electrical system: frequency deviations, balancing problems, transmission efficiency, increase of operating reserves, etc. Market operations associated with wind energy are also affected by wind unpredictability.

Thus, energy bids from wind energy producers involve important uncertainties that must be carefully managed, as deficits in production are strongly penalized.

In these conditions, wind energy producers and market and system operators require accurate wind forecasts in order to optimize the grid management, the plans and schedules of the generation units, and their operations in the electrical market. All these elements have a repercussion in the final cost of the electricity.

Day-to-day integration of wind energy

The day-to-day integration of wind energy in the electrical system is fundamentally supported by wind forecasts with horizons from seconds or minutes to few days. This chapter focuses on the wind forecasts related to these horizons, describing how they are produced and their repercussion on the concerned 1Universidad de Ca´diz, Algeciras, Spain's electrical system management processes.

There is a diversity of interpretations about the expressions associated to the referred forecasting horizons in wind forecasting literature. We have considered recent wind forecasting reviews to illustrate this point. In the expression ‘short-term’ is associated to horizons between 30 min and 6 h. However, in, Yan et al. apply the term to the range of 24–72 h. Other less precise definitions of ‘short-term’ are given by Foley et al. (‘hours up to two days’) or by Zhang et al. (‘from hours to days’). The diversity of definitions in these reviews is also similar in the expressions ‘medium-term’ or ‘longterm’. Hence, for clarity reasons, in this chapter we referred to the forecasting horizons in a quantitative way when necessary.

 

3.1.1 Uncertainty in wind energy production

In this subsection, some information about wind production errors report in different works is give. This information helps to illustrate the uncertainties that the electricity system has to deal with in the integration of wind energy. For instance, it seems that it is assume that the hour-ahead production estimations for a single wind farm involve errors within the range of 10%–15% with respect to the actual wind production. These errors could represent important deviations between the expected energy production and the actually delivered to the power Crompton greaves consumer share price forecast network.

As these deviations

As these deviations are penalize in most of the electricity systems, wind plant owners use conservative estimations in their plans which ultimately involve non-optimized exploitation of the wind farms. Considering regional electricity systems, Weber states that the day-ahead error for the German market is more than 20% with respect to the averaged wind energy production. This error could be significantly higher in specific cases. Thus, Gonzalez-Aparicio and Zucker reported that ‘on an annual average, the maximum range of the error reached up to 70% of wind production’ in the Spanish market during the period 2010–13.

Uncertainty in wind power forecasts highly depends on the forecasting horizon. Weber quantified how prediction errors increase with the forecasting horizon in the German market.

Concretely, the observed RMSE values, as a percentage of the total installed power in Germany, were approximately 5.7%, 4.4%, and 3.1% considering two days ahead, day-ahead, and intraday predictions, respectively.

Another remarkable effect is that wind power forecasting errors tend to be lower if larger areas are consider. In this sense, Borggrefe and Neuhoff explain that the day-ahead forecast error was reduce in Germany when the four transmission system operators were aggregated. The errors associate with the four separated areas were distribute between 6.6% and 8.9%.

After the integration

After the integration, the error was reduce to 5.9%.

However, these rough numbers based on annual averages or installed capacity can hardly describe the structure of these errors and their repercussion on the day-today wind energy integration, because uncertainty in wind energy production depends on the load demand, current wind power production and percentage of wind power of the total production. In this sense, Gonza´lez-Aparicio and Zucker made a thorough characterization of wind power forecasting errors in the Spanish electricity market from 2010 to 2013. They confirm that wind power forecasting Crompton greaves consumer share price forecast errors are spreader as higher loads are demand by consumers. During the study period, the installed wind power increased from 19.5 to 22.8 GW, and the hourly averaged production increased from 4.9 to 6.5 GWh. Despite they conclude that the averaged error remained similar, they corroborated that the uncertainty in predicting the absolute wind power output increases as wind penetration grows.

 

3.1.2 Effects of the wind forecasts uncertainty in the power system

The integration of wind power as an important part of the energy supply in a country or region imposes certain operating characteristics in the electrical system to assure security, as the intermittency affects the grid operation at different temporal and spatial scales. Figure 3.1 summarizes the different elements and processes that can be affected by the insertion of wind energy in the power systems. The maximum level of wind energy integration would be associat to the capacity of the electrical system to absorb this intermittent production while keeping these elements and processes in security conditions. The magnitude of the integration problem is dependent on the proportion of wind power, the dispersion/concentration of wind farms and the correlation between wind production and demand.

Figure 3.1 Processes and elements affected by the wind energy penetration, and their associated temporal and spatial scales

The constraints imposed

The constraints imposed by grid lines and stations are decisive to determine the amount of wind capacity which could be accepted, having into account that energy must be transmitted from windy places which are generally far from the consumption areas.

Thus, in the first steps of regional wind energy development, it is important to check the adequacy of the grid. For this purpose, it is essential to have very long-term estimations (in the range of years) of the wind resource available Crompton greaves consumer share price forecast in the concerned region.

These wind resource maps are use to demarcate zones where wind farms could be install and to design the infrastructure to allow their connection.

Regarding the power networks design, there is a recent concern on the interconnection of different transmission systems as a way to mitigate the problems derive from the unpredictability in power production.

Apart from long-term and network design considerations, wind penetration has an important impact in the daily management of the grid, because power quality, congestion management, or transmission efficiency can be significantly affected by intermittency in wind production.

Impact

This impact can be minimize (or at least smoothed) by the action of the operating reserves, which allows the system operator to make real-time adjustments to guarantee grid security.

In this sense, wind power variability and forecasting uncertainties force the system operator to increase the amount of reserve capacity in order to deal with unexpected variations in wind production.

The operating reserves are divide according to the time required to their full activation. The primary reserve is generally associate to response time from seconds to several minutes.

The primary reserve is apply in frequency control and voltage management, thereby minimizing disturbances and preserving adequate power quality. The secondary reserve is operative in the range of minutes to approximately 1 h, and it is use to balance load and demand and grid congestion problems. Thus, system operators are particularly interest in this range of very short-term horizons (from seconds to 1 h) of wind forecasts, to correct deviations from the planned wind energy and to deal with unexpected fluctuations of wind power, for example, due to ramp events.

Medium-term wind forecasts (in the range of days) are especially interesting for market operations (see next subsection). However, a part of this market is devote to contracting reserves and, in this sense, they also affect the system operation.

3.1.3 Wind uncertainty in market operations

Along with the wind production and the power network constraints, the electrical market is another factor that determines the level of integration of wind energy. Again, wind forecasts are necessary to optimize strategies adapted to the market design and operation. The market operators are interest in forecasting horizons longer than the system operators, as the first finish their operations hours before the energy dispatch, while the second are responsible for the real-time management of the electrical grid.

Electrical market

Each electrical market has its own timelines and characteristics, which ultimately establish the way in which wind power producers sell the electricity. Despite this diversity, in general, markets operations can be classified in three classes: day-ahead operations, intraday operations and reserve operations.

In the day-ahead markets, producers offer energy for the next day from 0:00 a.m. to 12:00 p.m. in an hourly base.

Considering that the closure of the day-ahead market occurs at 12:00 a.m. of the day before dispatch1, wind energy bids are unavoidably based on wind power forecasts with horizons from 12 to 36 h and,  consequently, they involve important uncertainties.

The market price is determine by the marginal pricing of the supply and the demand. The market-clearing mechanism also has into account a set of constraints for security in grid transmission.

As a result of this process

As a result of this process, Crompton greaves consumer share price forecast a preliminary plan, and distribution of electricity prices along the next day are obtain. Wind energy producers have very low short-run marginal costs, and therefore the amount of bidden wind energy affects the final price of electricity, especially in markets with high wind energy penetration. However, the uncertainties in wind power forecasts force to adopt bidding strategies under certain confidence intervals, which ultimately are conservative as deficits in actual productions are strongly penalize.

Note: As said, each electrical market has its own timeline but, for explanation purposes, 12:00 a.m. represents a good generalization of the day-ahead market closure time of the different energy systems.

The day-ahead planning

If the day-ahead planning were perfect for the real-time operation, corrections in plans, and additional market operations would not be necessary.

However, unexpected failures, plant outages, or changes in load and production conditions require short-term adjustments to guarantee the power system security. A part of these adjustments is derive from deviations in weather predictions.

Obviously, these deviations directly affect renewable energy production plans and schedules based on wind and irradiance forecasts, but they also have a relevant effect in the expected load which depends on the temperature, luminosity, or humidity conditions. In this context, market participants can use the intraday market to adapt their estimations or correct unfeasible schedules resulting for day-ahead markets. The intraday market starts after the closure of the day-ahead market and generally closes few hours before the real-time dispatch. Regarding wind energy producers, the intraday market allows re-bidding and correct positions according to forecasts with shorter horizons which, as commented before, are more accurate than those used in the day-ahead market.

The system operator

The system operator is in charge of providing reserves able to face possible contingencies or unbalances in the real-time management of the grid. The provision of these reserves is do in the balancing market. In most of cases, only a part of the contract reserve in the balancing market is finally use in grid operations, as reserve requirements are generally determine considering worse scenarios than actually occurs. The penetration of wind power increases the uncertainties in energy production and, consequently, forces the system operator to contract additional reserves to guarantee security. The reduction of wind power uncertainty (i.e., the availability of accurate wind power forecasts) would reduce the provision of reserves and the final cost of the electricity.

Figure 3.2 Generalization of an electrical market timeline

Aims

Figure 3.2 aims to illustrate all the above said by showing a simplification of a typical electrical market timeline (the specified hours are for guidance only).

It can be see how the composition of the real-time generation is the result of the day ahead of plan (established approximately one day before), the corrections introduce in the intraday market (established a few hours before), and the final use of the reserves according to the real-time necessities (determined in the last minutes).

In this illustration, the processes in which wind power forecasts are involve are delimit with dashe lines. Thus, the day-ahead bids and the reserve provision are based on Crompton greaves consumer share price forecast wind power forecasts with horizons of around a day, the intraday bids are based in forecasting horizons of few hours, and the reserve management in forecasting horizons of minutes to 1 h. For a specific case, Botterud et al. bring a detailed timeline of the Midwest Independent System Operator market (USA).

It would be desirable that most of the short-term adjustments would be assume by the intraday market. Thus, the use of reserve resources would be minimize and the final electricity cost would be benefit. A part of the current research deals with the adequate design of the electrical market. A common conclusion in these works is that flexibility and liquidity in the intraday markets allow a higher integration of wind energy.