Impact of Public Aggregate Wind Forecasts on Electricity Market Outcomes

Abstract

Abstract-Following a call to foster a transparent and more competitive market, member states of the European transmission system operator are required to publish, among other information, aggregate wind power forecasts. The publication of the latter information is expected to benefit market participants by offering better knowledge of the market operation, leading subsequently to a more competitive energy market. Driven by the above regulation, we consider an equilibrium study to address how public information of aggregate wind power forecasts can potentially affect market results, social welfare as well as the profits of participating power producers. We investigate, therefore, a joint day-ahead energy and reserve auction, where producers offer their conventional power strategically based on a complementarity approach and their wind power at generation cost based on a forecast. In parallel, an iterative game-theoretic approach (diagonalization) is incorporated in order to investigate the existence of an equilibrium for various values of aggregate forecast. As anticipated, variations in public forecasts will affect market results and, more precisely, under-forecasts can mislead power producers to make decisions that favor social welfare, while over-forecasts will cause the opposite effect. Furthermore, energy and reserve market prices can also be affected by deviations in aggregate wind forecasts altering, inevitably, the profits of all power producers.

Machine Learning Techniques for Improving Self-Consumption in Renewable Energy Communities

“…• onshore wind power forecasts at the national level, made available publicly by the Belgian Transmission System Operator (TSO) Elia [43] with the objective to benefit market participants and improve the electricity market outcomes [44], sampled at a quarter hourly scale.…”

“…The Python libraries Keras [45] and TensorFlow [46] were employed for implementing and training the neural networks. The Adam optimization algorithm [47], a state-of-the-art variant of stochastic gradient descent, was selected as the training algorithm for estimating the neural network weights.…”

“…The performance of the five day-ahead wind power forecasting models described in Section 2.1.1 is studied in this section. The neural networks models were implemented in Keras [45], using the original training procedure presented in the previous section, and the tree-based models (RF and GBDT) were implemented in Python using the Scikitlearn library [50]. The output of the ENSEMBLE model was simply coded in Python by computing the average of the outputs of the four other models.…”

Incentive-Compatibility in a Two-Stage Stochastic Electricity Market With High Wind Power Penetration

“…In practice, the assumption of a perfectly competitive market does not usually hold and electricity markets are challenged by the presence of strategic producers [27], which may offer at prices different than their actual production costs. Thus, an increasing number of research efforts has been focusing on investigating market power in electricity markets under various setups, e.g., [28]- [31].…”

“…The solution of the aforementioned bilevel model identifies the strategic price offers for each producer. Then, an iterative diagonalization approach is followed, to identify the equilibrium of the game among all producers [25], [31]. A more detailed presentation of "Strategic LMP" model is available in [26].…”