Simulation of an electric vehicle fleet to forecast availability of grid balancing resources

Abstract

Vehicle-to-Grid (V2G) technology utilizes the batteries of electric vehicles (EV) to provide ancillary services to the electric power grid on the order of milliseconds to seconds. An e-commerce market for frequency regulation (FR) will be served into the future by EV fleets that would otherwise remain idle for much of the day. Fleet operators typically utilize vehicles on predictable schedules. When idle and appropriately charged, EVs are able to interact with the grid. However, bidding into the FR market requires hourly forecasts of EV resource availability. This paper describes an effort to model FR via V2G on a large scale, with an emphasis on forecasting the availabilities of the vehicles at hour-long intervals, including route locations, and states of charge. The effort is in three parts: agent-based simulation, route data analysis, and empirical validation. With EV-fleet operations data and experiments in a small municipality, a simulation model incorporates both fleet and FR parameters and random events to characterize and distinguish the performance of various bidding and operations strategies. The results suggest how fleet operators should bid in the FR market to maximize revenues and compatibility with logistics services, with schedule demands and the available vehicles and chargers. The paper concludes with recommendations for future research that will advance the implementation of V2G in the FR market, addressing real-world perspectives of technology, business, information, and operations.