Tag Archives: forecast error

25-32 A. Annuk, H. Tammoja, H. Agabus, K. Toom and T. Tamm
Possibilities for Correcting Forecast Errors by Cutting off Production Chart Peaks
Abstract |
Full text PDF (142 kB)

Possibilities for Correcting Forecast Errors by Cutting off Production Chart Peaks

A. Annuk¹, H. Tammoja², H. Agabus³, K. Toom¹ and T. Tamm¹

¹ Department of Energy Application, Institute of Technology,
Estonian University of Life Sciences, 56 Kreutzwaldi Str., EE51014 Tartu, Estonia
e-mail: andres.annuk@emu.ee
² Department of Electrical Power Engineering, Tallinn University of Technology,
Ehitajate Rd. 5, EE19086 Tallinn, Estonia; e-mail: heiki.tammoja@ttu.ee
³ Nelja Energia LLC, 1 Regati pst., EE11911 Tallinn; e-mail: hannes@4energia.ee


In this paper we describe a conception for the mitigation of wind power fluctuations by cutting off production chart peaks.
The rapid growth of wind energetics has been induced by several factors. Although the government support may be the main incentive, other important motives include the increasing network access fees and strict requirements set for ensuring the balancing capacity of production.
However, such capacity has the tendency of being underdeveloped. The possibilities of the operating oil-shale plants for providing the capacity to balance the wind parks are running out. Sudden changes in the oil-shale plant output contribute to additional CO2 emissions, increased fuel consumption and decreased boiler efficiency. Under the circumstances, the transmission system operator (TSO) can face the need to reduce the power output of the wind parks. The operators of the wind parks integrated into the transmission network are responsible for presenting to TSO a 24 h forecast of their output power.
The forecast error is mainly specified in terms of Mean Absolute Percentage Error (MAPE), which for Estonian wind parks is about 20% on average. For forecast error estimation we have also applied the notion of Mean Percentage Error (MPE). Estimation of Pakri wind park data shows divergent actual forecast errors for different values of output power. For the values approximating the rated power of the wind park, the actual output power is larger than predicted. This situation clearly arises when the proportional output power is over 80% and MAPE is quite evenly distributed around 19.2%. In good wind conditions, for the relative output power value of 80%, the share of energy lost by cutting off production chart peaks amounts to 8.6% of the total energy production. The share is rapidly decreasing with declining wind conditions. Nevertheless, the average share of energy lost does not exceed 5%. The cut off energy might be applicable for heat production in boiler houses, although it is cheaper than the energy supplied to the electrical network.

Key words:

, , ,