Robust virtual battery optimization for EV fleet aggregation: Operational feasibility in day-ahead and real-time markets

Abstract

The increasing adoption of electric vehicles (EVs) presents both challenges and opportunities for modern power systems. However, accurately aggregating their flexibility is hindered by dynamic mobility and data privacy concerns. This paper proposes a novel robust virtual battery (VB) framework designed to optimize EV fleet flexibility for reliable charge and discharge management and Vehicle-to-Grid (V2G) market participation. Distinguished by its data efficiency, the proposed approach requires only charge demand and connection times, eliminating the need for individual battery capacity or state-of-charge (SoC) forecasts in day-ahead scheduling. By employing time-continuous robust polytope projection, the model guarantees the feasibility of disaggregating the VB schedule to individual EVs. Furthermore, we develop an integrated bidding strategy for day-ahead and real-time markets to minimize imbalance penalties. Simulations conducted over eight months using real-world data demonstrate that the proposed method reduces monthly costs by 8.8–14.9% compared to individual forecasting and achieves scalable, precise disaggregation, thereby enhancing the economic viability of V2G participation. © 2026 Elsevier B.V.