Cross-border electricity trading is becoming increasingly important in the European power system. Unscheduled power flows induce additional costs and may lead to congestion and impair power grid operation. In this contribution we provide a data-centric analysis of unscheduled flows in the Central European power grid. Using methods from explainable machine learning, we identify the main driving factors for unscheduled flows and quantify their impact. Unscheduled flows in the meshed part of the grid can be attributed to transit or loop flows primarily and are well described by a linear model. The performance is substantially worse for unscheduled flows on bridges, with forecast errors being the most important drivers. This performance gap is probably due to data quality issues.
Titz et al. (Wed,) studied this question.