Why forecast?

Accurate forecasts have the potential for the following benefits:

• Improved market trading

• Optimised scheduled maintenance

• Enhanced plant scheduling by system operators

Garrad Hassan have been predicting the long term energy production of wind farms on a commercial basis for more than 20 years, and internationally GH have analysed more than 50,000 MW of projects. GH's short term forecasting techniques build on the experience gained from this work.

In the above diagram:

1. GH Forecaster incorporates input data from an appropriately scaled Numerical Weather Prediction (NWP) source.

2. Individual wind farms supply continous feedback of meteorological and power signals.

3. Online, adaptive regression algorithms convert NWP input to site-specific meteorology and ultimately power forecasts.

GH Forecaster adopts a combined physical/statistical approach.

Garrad Hassan has close relationships with a number of physical meteorological forecasting suppliers and for any given wind farm an optimal model is selected. These forecasts are combined with real-time data fed back from the wind farm using statistically adaptive techniques to forecast the meteorological conditions and subsequently the power of the wind farm.

Forecasts are updated regularly, and are typically supplied via e-mail or FTP on an hourly basis for 365 days of the year. No major setup costs are required from the client, the service is provided for a simple monthly fee.

Accuracy levels which are typically achieved 12 hours in advance and beyond, are improvements over persistence forecasting of 40 - 60 %, and a MAE (Mean Absolute Error) of 12 - 15%. Further information on accuracy can be supplied on request.