Stochastic Aggregated V2G Optimization for Car-sharing Fleets Using Monte Carlo Simulation

Abstract

This paper presents a macroscopic, data-driven optimization framework for the energy management of electric vehicle (EV) car-sharing fleets providing vehicle-to-grid (V2G) services. Addressing the computational and operational limitations of conventional deterministic models, the proposed framework dynamically decouples stationary and transiting vehicles to ensure scalability without relying on heavy mixed-integer formulations. By employing a Monte Carlo simulation based on 126 real-world GPS mobility scenarios from a 10-node urban network in Tel Aviv, the model captures the inherent stochasticity of fleet connectivity. Results demonstrate that the optimized V2G strategy significantly outperforms baseline operations, mitigating the mean daily operational deficit by 108.2% and providing a P90 firm V2G discharge energy of 5202 kWh. Furthermore, a novel parametric mobility stress-test reveals the physical and economic boundaries of the system, identifying a critical inflection point where, for the analyzed case study, daily mobility demands surpassing 22,000 km monopolize the 30 kW charging infrastructure and abruptly collapse V2G flexibility. Ultimately, this study highlights that stochastic risk assessment and capacity stress-testing are essential for aggregators to submit reliable energy commitments in electricity markets without compromising their primary car-sharing service.