Modeling logistic systems with an agent-based model and dynamic graphs

This paper is about modeling and simulation of logistic systems. We define them as corridors established between a gateway port, where goods are imported, and urban areas, where the final distributors are located. The efficient management of the flow of goods operated on these corridors requires a structured territory and organized actors. Decentralized decisions of actors and interactions between them make it possible to provide consistent logistic services despite the numerous system constraints (legal, environmental, economical, ... ). Our goal is to reproduce the behavior of logistic systems through simulation. Our approach consists of describing the dynamics of such a system at a micro level. Therefore, we first enumerate the local properties, constraints and behaviors of each main actor and the infrastructures of this territory in order to extract the essential elements' that will be part of the theoretical model. A major aspect of the model is the description of the interface between maritime dynamics (schedule on a day-basis) and metropolitan dynamics (scheduled on an hour basis). This interface is self-organized: macro characteristics emerge from local properties and rules. It is revealing of a complex system, working on different scales, that we model with agents and dynamic graphs. Each actor and infrastructure is represented with agents. The transportation network is a multi-modal dynamic graph that makes possible to model the traffic and topology evolution. This approach enables users, like public authorities, to modify local parameters and observe their effects at the macro level. Thus users can identify levers to control the whole system. We execute some simulations with data on the Seine axis to confront our results with a real case study. We provide some measures (e.g. number of vehicles and quantity of goods) to show that the simulation reproduces the atomization process of logistic flows. We propose a spatial analysis of the goods traffic within the transportation network and compare the effects of two replenishment strategies on the stock shortages.