| |
Customer login
|
|
GMS Yield Study
There is a high demand in the wind energy industry for detailed and
accurate wind and weather forecasts for multiple applications:
- The growing “free” energy market has opened new market opportunities for
operators but requires and accurate wind and/or yield forecasts to optimize
financial returns.
- Effective grid management requires realistic planning of the feeding
capacity of each single wind farm and is heavily dependent on the wind and
weather conditions at the wind farm site.
- Down time during the construction of wind farms increase costs and can b
directly related to weather conditions on site.
Accurate forecasting is essential for the operation and maintenance of wind
farms and is crucial in resource scheduling during wind farm construction.
Precise wind forecasts also substantially improve daily farm yield
predictions.
One main difference in requirements between a “standard” weather forecast
and a forecast service specialized for wind energy is that the latter has to
be as precise as possible at one particular location – the wind farm.
Another key requirement is the need for detailed forecast information in the
atmospheric boundary layer, in particular wind and gust values.
Mesoscale weather forecasting methods typically have a resolution of several
kilometres. This resolution is not precise enough, especially in complex
terrain conditions were very small scale variations in wind flow have a
significant impact on wind farm yield.
GMS has been designed to deliver hourly resolved wind, energy yield and
other weather forecast information with the highest possible accuracy for
several days ahead of the present.
A study was developed to validate the quality of GMS, to further develop the
GMS key features and to improve their accuracy. A three month study was
carried out during the first three months of 2010. A total of 13 wind farms
located throughout Germany participated. The results showed that GMS
forecast quality is very impressive and provided keys to areas were
improvements can be made to the system. Among others GMS was able to provide
yield forecasts with a relative standard error of approximately 13 % for one
day ahead after a training period of only 2 month for GMS SMART LEARNING!
This result is the most remarkable as it was achieved in a period where most
of the wind farms were producing, but very rarely or never reached rated
power, which is the most difficult range to forecast. The forecast quality
is impressively shown in the following graph displaying the time series for
the Twistringen wind farm:
The diagram impressively shows the overall quality of the forecast and in
particular the improvement of GMS SMART LERNING.
The report of the study contains the analysis results, quantifies the
accuracy of the GMS wind speed and wind energy yield forecasts and the
capability of various GMS features to improve those forecasts. To download
the report, click here.
|
|
