When most people check the weather forecast they want to know if it’s going to rain or whether they need to put on an extra layer. Wind farm operators and utility operators are more interested in a different kind of weather. They need to know exactly how much wind they can expect where and for how long in order to produce the maximum amount of electricity and ensure grid stability.

But what if you want to do more than just nowcasting?

Wind forecasting is crucial both for wind farm operators and utility operators. Accurate forecasting allows operators to achieve favorable trading performances on the electricity markets. The further in advance an operator can make a reliable estimate about how much electricity he will produce, the more profit he can make. Network operators require reliable and accurate data to absorb the growing share of wind power and anticipate shortages due to rapid changes in wind speed and direction. In Denmark, the country with the highest share of electricity from wind energy, a change in wind speed of one m/s can result in a change of 450MW in national electricity production.  Improvements in wind forecasting, in increasing the profitability of wind farms and allowing easier grid integration, would greatly facilitate the development of future wind projects.

Wind power continues to grow and it is widely seen as the renewable energy source best able to compete with fossil fuel electricity generation. Last week China announced that it had supplanted the US as the country with the most installed wind-power capacity. In 2010 China installed 16 GW of new wind-power capacity, representing 45 percent of the world’s total. China hopes to install an additional 100GW by 2020 to meet its growing energy demand. In China and the US, the two fastest growing markets for wind power, wind farm operators can still take advantage of untapped coastal areas with high, sustained winds. In contrast, several European countries have already developed the majority of these prime coastal wind-harvesting areas.

Faced with the lack of suitable geographic terrain on-shore and prohibitively high costs offshore, Europe’s wind farm operators rely heavily on short-term wind forecasting to maximize the yield of their existing turbines. Similar to the “measure, correlate, predict” (MCP) analysis employed to evaluate prospective wind-farm sites, short-term wind forecasting relies on a mixture of long-term averages, geographic terrain, and real-time measurements to make accurate predictions.

Wind power forecasts are generated in two steps. First, forecasters employ numerical weather prediction (NWP) to predict a number of meteorological values such as velocity and direction of the wind, temperature, and humidity. These data points are then used to calculate the wind power forecast.

Forecasters have developed a wealth of different approaches to converting the meteorological variables into accurate wind power forecasts. The standard technique for predictions over the next hour remains persistence forecasting. Here forecasters assume that wind power generation will be the same one hour into the future as the last measured value. This approach is often referred to as “what you see is what you get.’ When it comes to predicting wind power beyond that, forecasters employ either a physical or statistical approach. Each method has its advantages and drawbacks. Physical forecasting is heavily reliant on meteorological information and the data from the NWP. Forecasters have an abundance of data they can work with. The grid resolution of meteorological data however is often-times several miles across and thus not detailed enough to incorporate the effects that local variations in topography or terrain can have on wind speeds and directions. The physical approach is also hampered by the fact that the wind speeds measured a few feet above the ground can differ significantly from wind speeds 250 feet in the air where the turbines are located. In order to rectify these drawback forecasters can increase the resolution of the NWP for areas of importance, such as mountains or coastlines. By combining data sets with various resolutions researchers hope to incorporate all the important variables in their forecast.

Statistical forecasting is less dependent on the accuracy of NWP, but relies more heavily on historical wind speed data at each site. Historical data sets are coupled with real-time observations from the wind turbines to calculate the most accurate wind energy prediction. In contrast to physical forecasting, the statistical method suffers from a lack of available data. Wind farm operators consider their historical wind data “trade secrets” and are thus reluctant to share it with forecasters.

Over the past 10 years improvements in both forecasting methods have helped to reduce the next-day forecast error from about 40 percent to below 20 percent. By comparison, we can make standard weather forecasts for the next 24 hour period with over 95 percent confidence.

Danish grid operator Energinet.dk and American wind farm operator Xcel Energy have, among others, been at the forefront of improving the reliability of wind forecasts and to turn wind into a highly reliable source of electricity. The availability of faster computers will allow forecasters to generate more detailed NWP maps and plug larger historical data sets into their models. Forecasters plan to incorporate the site-specific biases for each turbine into the models and allow for more accurate predictions in the future. Ultimately operators want to be able to exactly predict the wind energy output for each individual turbine. A major step towards this goal could be the combination of the physical and statistical methods of wind forecasting.

The profitability of wind power and the extent to which it can function as base load electricity in the future will depend on the ability of wind forecasting to make more accurate and reliable predictions. Wind power is and will remain a highly fluctuating resource, but wind forecasting can help to reduce uncertainty and allow countries to rely heavily on wind energy to meet their targets for electricity from renewable sources.

Related Posts with Thumbnails
denmark, Energinet.dk, grid integration, numerical weather prediction, offshore wind, renewable energy, United States, weather, wind forecasting, wind power, Xcel Energy