Bidding strategies for renewable energy generation with non stationary statistics

The intrinsic variability in non-dispatchable power generation raises important challenges to the integration of renewable energy sources into the electricity grid. This paper studies the problem of optimizing energy bids for a photovoltaic (PV) power producer taking part into a competitive electricity market characterized by financial penalties for generation shortfall and surplus. To this purpose, an optimization procedure is devised to cope with the intermittent nature of PV generation and maximize the expected profit of the producer. Since the optimal offer turns out to be a suitable percentile of the PV power cumulative distribution function (cdf), we investigate two approaches to properly take into account the effects of seasonal variation and non stationary nature of PV power generation in the estimation of PV power statistics. The first one normalizes the generated power with the power obtainable under clear-sky conditions. The second approach estimates a time-varying PV power cdf using only power data in a moving window of suitable width. A numerical comparison of the different bidding strategies is performed on a real data set from an Italian PV plant.