News item | 2012-02-20 | 16:00 PM

Challenges in wind power forecasts

Future wind power forecast errors and associated costs in the Swedish power system.

Wind power is one of the renewable energy sources in the electricity system that grows most rapidly, which is in line with both Swedish and EU governmental goals.

There are challenges that need to be addressed with a larger proportion of wind power – variability and predictability. Predictability is important since the spot market Nord Pool Spot requires forecasts of production 12 – 36 hours ahead. The forecast errors must be regulated with regulating power, which is expensive for the actors causing the forecast errors. However, both variability and predictability are connected to the need for the electric power system’s ability to regulate power – that is, to be able to produce what the consumers consume.

Fredrik Carlsson, Senior R&D Engineer at Vattenfall has been investigating a number of scenarios with 10 – 55 TWh of wind power installed in the Swedish system. The focus has been on a base scenario with 10 TWh new wind power consisting of 3,5 GW new wind power and 1,5 GW already installed power, which gives 5 GW. The size of the forecast errors of future scenarios have been assumed to develop to a standard deviation at 13% of installed capacity, which is very good since it is around 20% in Sweden today, however in other countries such as Germany, around 15%.

The scenarios are based on planned wind farms to make the scenarios as true as possible. Eight different actors with balance responsibility have been created that own all the new wind power (not the already installed) in the scenarios. They have been chosen to differ from small to large and from concentrated location to wide spread – which give present owners of wind farms a possibility to find some of the constructed actor that is similar to their targets.

The investigation focuses on the forecast error volumes due to wind power forecast errors for the system as well as for different actors, the costs associated to the forecast errors, the effect of the introduction of four bidding zones in Sweden, and options to reduce costs such as reduction of forecast errors by trading at the intraday market, better forecasts, and changed market design.

The main conclusion that is drawn from the study is that the costs for forecast errors for the wind power owners increase from today’s costs that is about 5 – 10 kr/MWh to future 30 kr/MWh (3 öre/kWh) with 10 TWh new wind power. It should be noted that the present and future cost is based on very high forecast quality, and as the forecast error quality is much lower today, the cost in reality is about 15 - 20 kr/MWh.

So, the increased costs will not be so dramatic in practice. The cost varies between the actors since they have their wind power in different bidding zones and also different concentrations which affect the forecast error volume. It should be noted that these numbers represents higher forecast quality than used today.