Impact of Forecast Uncertainty on Wind Farm Profitability

Abstract

Abstract The great amount of support schemes that initially fueled the fast and often uncontrollable, renewable energy sources (RESs) growth have been strongly reduced or revoked in many countries. Currently, the general trend is to try to equate RESs to traditional power plants. From the energy market point of view, this entails exposing RESs to market competition and mechanics. For example, it could be requested that RESs submit a production schedule in advance and are financially responsible for any deviation from it. This could push the wind farm (WF) operators to make accurate forecasts, thus fostering the electric system resiliency and an efficient use of balancing resources. From the forecasting point of view, this is not a trivial problem since the schedule submission is often due 10–12 h before the actual delivery. Since forecast errors are unavoidable, the submitted schedule could turn out to be infeasible, thus forcing the WF to adopt correcting actions, which are generally costly. This study estimates the revenue reduction that would affect a WF operating in the energy market due to forecast errors. To do this in a realistic way, a case study is selected, and realistic forecast scenarios are generated by using a copula approach. Relevant forecast error features, like autocorrelation and dependency on forecasted power level and forecast lead time, are modeled. The revenue reduction due to balancing actions is calculated on an annual basis, by using typical days. These were derived through a clustering procedure based on production data. Losses ranging from 5% to 35% have been found, depending on the days and market prices. A sensitivity analysis to the costs of balancing actions is performed. The effect of different market architectures and different RESs penetration level is considered in the analysis. Finally, the effectiveness of two techniques (i.e., curtailment and batteries) to reduce forecast error impact in highly penalizing market environments is assessed.