Cost-Effective scheduling and flexibility provision of EV aggregators under departure uncertainty

Rapid electrification of transport presents both challenges and opportunities for distribution system operators (DSOs) seeking to balance supply and demand. In particular, the inherent uncertainty in electric vehicle (EV) user behaviour, in particular arrival and departure times, and state-of-charge requirements, complicates day-ahead scheduling and may lead to suboptimal grid utilisation or unmet charging needs. This paper makes two key contributions. First, it formulates a day-ahead robust optimisation model that accounts for early departure uncertainty via a min-max framework, ensuring guaranteed feasibility under worst-case conditions. Second, it proposes a rollinghorizon dynamic optimisation strategy that re-solves convex scheduling problems in real time as EV availability and market prices are revealed. Together, these approaches are compared under identical local energy and flexibility market constructs. The methodology integrates vehicle state-of-charge dynamics, market participation constraints, and flexibility limits into a unified convex optimisation problem. The robust model utilises pessimistic departure times, while the dynamic model updates schedules hourly based on actual EV behaviour. Case studies with a fleet of 50 EVs demonstrate that the proposed methods can achieve cost savings between 18% and 55% for EV owners. Moreover, the dynamic approach yields more efficient outcomes when compared to robust optimisation. These results highlight the impact of real-time adaptability in EV aggregation, providing DSOs and aggregators with actionable insights on scheduling strategies that balance economic performance and reliability in emerging flexibility markets.