| Abstract |
Highlights: What are the main findings? Shipment delivery time is not only affected by the type of goods but also by the receiver function, shipment size, shipment distance, and zoning type of the receiver’s location. The proposed modeling approach using GPS data can be used to replicate the heterogeneity of shipment delivery time. What are the implications of the main finding? Understanding various observable characteristics of receivers and shipments leads to infer time of day delivery demand distribution. Readily available GPS data can be used to develop a simulation model of delivery times for urban freight analysis. Developing policy instruments related to urban freight, such as congestion pricing, urban consolidation schemes, and off-hours delivery, requires an understanding of the distribution of shipment delivery times. Furthermore, agent-based urban freight simulators use relevant information (shipment delivery time distribution or vehicle tour start time distribution) as input to simulate tour generation. However, studies focusing on shipment delivery time-period selection modeling are very limited. In this study, we propose a method using GPS trajectory data from the Tokyo Metropolitan Area to estimate a shipment delivery time-period selection model based on pseudo-shipment records inferred from GPS data. The results indicate that shipment distance, size, and destination attributes can explain the delivery times of goods. Moreover, we demonstrate the practicality of the model by comparing the simulation result with the observed data for three areas with distinct characteristics, concluding that the model could be applied to urban freight simulation models for accurately reproducing spatial heterogeneity in shipment delivery time periods. This study contributes to promoting smart city development and management by proposing a method to use big data to better understand deliveries and support the development of relevant advanced city logistics solutions. © 2025 by the authors. |
