Understanding the Spatiotemporal Charging Patterns of Battery Electric Vehicles: A Case Study in Shanghai, China

Understanding the spatiotemporal patterns of electric vehicle (EV) drivers' charging behavior is an essential component of both charging demand management and public charging infrastructure planning. However, most existing studies rely on relatively small samples of EV drivers' charging behavior, potentially missing the complexity and heterogeneity inherent in these patterns. To address this limitation, our study proposes a two-step framework that uses trajectory and state-of-charge data from over 200,000 battery electric vehicles (BEVs) operating in Shanghai, China over a month. This comprehensive data set represents all actively used BEVs in the city. The initial step involves using a density-based clustering algorithm to differentiate between mobility service BEV (MBEV) drivers (traditional taxi and ridesourcing service drivers) and private BEV (PBEV) drivers. The algorithm bases this distinction on criteria such as daily travel distance, variance in charging locations, and charging frequency. Then, Latent Dirichlet Allocation models are estimated separately for MBEV and PBEV drivers. These models identify 36 unique spatiotemporal charging patterns for each type of BEV driver, encompassing 6 spatial and 6 temporal patterns. The study results underscore major differences between MBEV and PBEV drivers, and within these groups themselves, in terms of their travel behavior, charging behavior, and public charging station choices. These differences likely reflect their distinct travel needs and preferences. This study not only offers a comprehensive analysis of the charging patterns of MBEV and PBEV drivers in a megacity, but also lays the groundwork for developing more driver-centered strategies for charging demand management and public charging infrastructure planning.