Prediction errors and balancing costs for wind power production in Finland

Hannele Holttinen, Pirkko Saarikivi, Sami Repo, Jussi Ikäheimo, Göran Koreneff

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review


In Finland, as there is no feed-in tariff, the wind power producers have to make contracts on how the electricity is sold and also make contracts for balance settlement. For this paper, wind power production is estimated 1–38 hours ahead for wind turbines located in 4 small clusters about 380 km apart on the West coast of Finland. Wind speed forecasts from the very-high-resolution ETA meteorological forecast model run by Foreca Ltd are used together with two production forecast models, which employ neural networks developed by Cybersoft and time series modeling by VTT. To show the effect of even more spread wind power production in Finland, prediction errors are determined for both one site, one producer and for the whole West Coast of Finland. Balancing costs resulting from day-ahead forecasts are estimated with year 2004 price data. Some example weeks are taken to examine the possibilities of improving the meteorological prediction for wind speeds.
Original languageEnglish
Title of host publicationProceedings of 6th International Workshop on Large-scale Integration of Wind Power and Transmission Networks for Offshore Wind Farms
Subtitle of host publicationDelft, Netherland, 26-28 October 2006
Publication statusPublished - 2006
MoE publication typeA4 Article in a conference publication


  • Forecasting
  • neural network applications
  • time series
  • wind energy
  • weather forecasting
  • electricity markets
  • balancing costs


Dive into the research topics of 'Prediction errors and balancing costs for wind power production in Finland'. Together they form a unique fingerprint.

Cite this