An Agent-Based Approach to Simulating Urban Parcel Logistics

Abstract

The rapid growth of e-commerce has significantly increased demand for last-mile parcel delivery, leading to heightened urban traffic congestion, emissions, and operational inefficiencies. Traditional logistics models often overlook the fragmented nature of this sector, where multiple parcel service providers (PSPs) operate in isolation. This paper introduces an agent-based simulation model designed to provide a comprehensive, quantitative framework for analyzing urban last-mile parcel logistics. Leveraging MATSim and Jsprit, our model explicitly represents multiple PSPs, each with independent depots, fleets, and personnel, allowing for the capture of heterogeneous behaviors and their interactions. By simulating the flow of parcels from distribution centers to recipients in a multi-carrier environment, the model extends beyond single-provider models. Initial results demonstrate that PSPs with broader facility coverage and larger market shares achieve greater operational efficiency, characterized by higher parcel drop factors and shorter average tour distances. A comparative experiment further revealed that models relying solely on single-provider scenarios can significantly underestimate overall mileage and overestimate parcel drop factors, underscoring the critical impact of market structure on system performance. This research highlights the crucial role of spatial distribution of facilities and PSP market share influences in accurately modeling urban last-mile parcel logistics.