System imbalance forecasting and short-term bidding strategy to minimize imbalance costs of transacting in the spanish electricity market

Abstract

Energy imbalances can represent a significant cost for agents transacting in markets that penalize participants’ imbalances. In markets with increasing penetration of intermittent renewable sources of energy (RES‐E), system imbalances can not only be costly, but also increase, as is the case for the Spanish power market. Market participants, especially those trading non‐dispatchable energy, are therefore interested in minimizing this cost while simultaneously maximizing their profits. A lot of work has been developed around the forecast accuracy and uncertainty of RESE production to determine bidding strategies that minimize imbalance costs, especially for wind power trading. Challenges inherent to agents specialized in power trading and/or retailing activities, especially wind power trading of energy produced by third parties or retailing to small consumers means that applying strategies that rely on production forecasts may not be sufficient. In this master thesis we considers those challenges by developing an optimized bidding strategy that reduces the expected imbalance cost for a real case‐study of a Spanish energy trader/retailer based on a forecast of the system´s imbalance volume and past imbalance costs, while using new information available after the day‐ahead market gate closure for participation in the intra‐day market to influence the imbalance volume of the agent’s portfolio towards the direction that reduces their potential imbalance cost. This strategy does not replace accurate forecasting but considers the practical aspects of energy traders/retailers with numerous small clients who cannot operate production units. The strategy can be applied from the perspective of both a trader and retailer. We have developed an advanced model based on random forest technique to forecast the system imbalance and used a genetic algorithm to apply the bidding strategy that minimizes the imbalance costs based on system imbalance forecasts and past imbalance costs. The proposed strategy application using new information available after the dayahead gate closure outperforms its application in the pre‐day‐ahead market.