A slight variation in wind energy production can have a significant financial impact.

Wind can be elusive and even very small changes in the atmosphere can make a difference in wind speed and direction. Significantly, there are certain time periods when the wind is highly variable.

In the past, analysis of the impacts of extreme weather events on wind energy generation have also indicated an unrecognised cause of rapid wind power changes. Such variability has posed challenges to wind forecasting systems over the years.

To the extent wind energy puts serious demands on the grid – either its generating or transmission resources – it can be challenging to manage.

For instance, in West Texas, where a huge amount of wind capacity has been built, it is clear there isn’t enough transmission to send the power where it is needed.

This has led to frequent curtailments of wind generation to avoid overloading the lines. Or, in some regions there is so much wind generation at night that it is difficult to keep the baseload units which normally meet this load – and which prefer to be operated at constant output – running steadily.

“Such issues cannot be solved by forecasting,” says Michael Brower, chief technical officer, AWS Truewind, LLC.

Where forecasts can help is in predicting changes in the amount of non-wind generation and transmission that will be required from one time to another.

“Having such information helps optimise the day-to-day system operation by minimising the number of units kept running unnecessarily and making sure the right units are on hand to respond to the need,” according to Brower. 

Also, wind forecasting can help system operators anticipate when there is unusual stress on the system that might pose a threat to reliability – for example, when large numbers of wind plants may shut down abruptly because of too much wind.

“The cardinal rule in grid operations is to keep power flowing to consumers. State-of-the-art wind forecasts help ensure this happens,” said Brower.

Weather variables

The difficulty in predicting when and how strongly the wind will blow at the wind farms is certainly a major hurdle for the industry.

Wind is definitely among the most difficult weather variables to forecast.
 
Factors like topography, ground cover and temperature inversions, that affect wind, make wind forecasting more difficult than ordinary weather forecasting, admits Brower.
“Wind speeds vary far more with the local terrain and thermal structure of the boundary layer than most other meteorological parameters. From a hilltop to a valley, the wind speed may vary by a factor of three or four; temperature not so. The presence or absence of temperature inversions, nocturnal jets, and other phenomena can likewise radically change the available wind resource,” he added. 

It has already emerged that researchers are trying to pinpoint breezes in the vicinity of wind turbines, which are generally about 200 to 400 feet above the ground and arrayed in tightly clustered wind farms.

Brower says effective analysis of diagnostic system information can help flag turbines or turbine systems that are in trouble and likely to fail.

“Depending on the time frame – is this a gradual degradation or an imminent event? – multi-day forecasts can help the plant operator choose the optimum time to carry out the preventive maintenance to maximise total power production,” added Brower.
 

Methods

Depending on the form of meteorological forecasts, methods for power curve modeling or statistical downscaling, considering the relationship between wind power and local forecasts for meteorological variables, are developed.

Several different methods are used to convert from a meteorological forecast to a plant output forecasts.

Brower explains that some forecasters take what might be called a “deterministic approach”, where one enters the predicted wind speed and direction and possibly other parameters  in a model of the wind farm (similar to models used in designing projects).

“This type of approach may be necessary, in fact, when you don’t have any output data from the wind plant,” said Brower.

Others take a statistical approach, in effective constructing a whole-plant power curve based on the observed relationship between the plant output and predicted wind. This curve may vary depending on direction, time of year, and other factors.

“A good forecaster will be flexible and adapt whatever method he uses to the circumstances at hand,” said Brower.