POET: towards power-system-aware e-taxi coordination under dynamic passenger mobility

Yukun Yuan, Yue Zhao, Shan Lin

Abstract: Electric vehicle (EV) fleets, e.g., electric taxis, buses, and trucks, have been increasingly implemented in cities. Compared with internal-combustion vehicles, the operation of EV fleets requires resilient charging infrastructures. Notably, the charging of EV fleets can bring significant reliability and efficiency challenges to the operation of the underlying power system. Based on the real data collected from New York City (NYC), our analysis shows that an increased power load ramping can result from various charging behaviors of electric taxis (e-taxis). Additionally, the increase of the peak load can further overload the local power distribution network. To address these problems, we exploit the possibility of utilizing e-taxi fleets' mobility to improve the reliability and reduce the cost of the power system, while maintaining the taxi service quality. Specifically, we design POET, a POwer-system-aware E-Taxi coordination algorithm, and evaluate the solution with comprehensive datasets for taxis and power distribution systems from NYC. These data include (i) more than 13,000 taxis with more than ten million taxi trips per month, (ii) a city local power distribution network with 38 local area substations and nearly 45 GWh overall power consumption per day, and (iii) deployed EV charging stations. The extensive data-driven evaluation demonstrates that, compared with the existing e-taxi charging solution that focuses on optimizations of taxi service quality, our solution decreases the power load ramping of local regions by 22.3% and reduces the daily peak charging load by 44.2% while achieving almost the same taxi revenue.