Wind power bidding in a soft penalty market

In this paper we consider the problem of offering wind power in a market featuring soft penalties, i.e. penalties are applied whenever the delivered power deviates from the nominal bid more than a given relative tolerance. The optimal bidding strategy, based on the knowledge of the prior wind power statistics, is derived analytically by maximizing the expected profit of the wind power producer. Moreover, the paper investigates the use of additional knowledge, represented by wind speed forecasts provided by a meteorological service, to make more reliable bids. The proposed approach consists in exploiting wind speed forecasts to classify the day of the bidding into one of several predetermined classes. Then, the bids are represented by the optimal contracts computed for the selected class. The performance of the optimal bidding strategy, both with and without classification, is demonstrated on experimental data from a real Italian wind farm, and compared with that of the naive bidding strategy based on offering wind power forecasts computed by plugging the wind speed forecasts into the wind plant power curve.