Data driven optimization for electric vehicle charging station locating and sizing with charging satisfaction consideration in urban areas

The lagging development of charging infrastructure seriously restricts the penetration of electric vehicles. Inadequate and unreasonably setting of charging stations (CS) aggravates charging anxiety of electric vehicle(EV) users. Location and sizing model is developed to expand existing CS to maximize EV charging service satisfaction in urban areas. The model captures charging behavior from massive GPS trajectory extraction data of electric taxis. Satisfaction from EV users with charging service is measured by whether the respond time from an electric vehicle driver issuing charging demand signal to completion of charging is within a threshold value. The respond time is considered to include seeking time, queuing time and charging time, which are determined by the decisions of the CS location, capacity and charging types. Heuristic algorithms, including the greedy, greedy-substitute and two-layer genetic algorithms, are designed to solve the problem. Algorithms are simulated in a large scale random computational environment. Finally, a practical case of Shenzhen is investigated. It is found that charger pooling is more effective than decentralized layout. Moreover, it is critical to weigh the cost and charging rate of different types of chargers with budget restriction. High charging rate chargers are not necessarily effective than slow ones in reducing charging respond time.