Spatial-Temporal Prediction of Electric Vehicle Charging Demand in Realistic Urban Transportation System of a Mid-sized City in Brazil

Abstract

Increasing the number of Electric Vehicles (EVs) in the urban transportation system will bring a large amount of energy demand, making urban planners construct a large number of charging piles. However, the blind construction of charging infrastructure brings problems such as excessive or insufficient charging facilities in different urban zones and irregular fluctuations in the power grid. These problems can be easily prevented with an accurate forecast of spatial-temporal charging demand in the urban area. The proposed methods in the literature are rarely applicable for charging demand prediction in the urban area because of ignoring the detailed spatial-temporal travel patterns of EVs in actual urban street networks. To obtain accurate spatial-temporal EV charging demand, the detailed travel pattern is modeled in this paper by integrating the actual street network and functional zones of the urban area. The urban street networks are modeled as a detailed graph based on OpenStreetMap data. A novel stochastic trip chain including the destination choice, route choice, speed-flow, and traffic allocation models is developed to simulate the spatial-temporal travel patterns of EVs in the urban street network. The travel patterns are incorporated by charging patterns and preferences of EV users to predict the EV charging demand in different locations of the urban area. The main results of the current research are 1) providing a detailed spatial-temporal model for EV travel patterns in the urban transportation system, 2) obtaining spatial-temporal slow and fast charging demand distributions of EVs in different functional zones, and 3) analyzing the slow and fast charging load demand distributions in different locations to suggest the charging infrastructure construction.