Multi-agent Modeling and Simulation of a Hub Port Logistics Houda Zouhaier * x , Fahem Kebair *§ , Fr´ ed´ eric Serin xx , Lamjed Ben Said * x * Strat´ egies d’Optimisation et Informatique intelligentE x Higher Institute of Management Le Bardo, Tunis, Tunisia § Higher Institute of Computer Science Ariana, Tunis,Tunisia xx LITIS. Laboratoire d’Informatique, de Traitement de l’Information et des Syst` emes University of Le Havre Le Havre, France Abstract—A hub port logistics deals with thousands of daily transfers thanks to a multimodal transport. A crucial challenge in this domain is to avoid delays in the global supply chain and to ensure deliveries on time by optimising container management. This involves engines operating planning, resources management and container storage optimisation. This paper addresses this challenge, in which a multiagent approach aiming to modelise and to simulate container management is presented. The proposed model is based-on an individual-centred approach and has as objective to enable engines to meet container demands while ensuring their satisfactions. First experimentations of a multia- gent simulator measuring container satisfactions are presented. Results about the effectiveness of the multiagent approach in container management and handling are then discussed. Index Terms—ATN, container management, individual- centered approach, multi-agent system, MESSAGE/UML method, satisfaction. I. I NTRODUCTION Container management requires in a container terminal a continuous and relevant monitoring operations to optimize planning process and freights and to deliver containers on time. This is accomplished according to the resources capacity and availability and the containers priority. According to [16], a container terminal is a dynamic system. Dynamicity is defined by the movements multiplicity that can occur, the ship’s irregular arrival, the unexpected appearance of breakdowns and the appearance of a new missions although planning is al- ready established, etc. [14]. Centralized management systems showed their ineffectiveness when an important dynamicity is noted (transport flows) or required (cost-effectiveness of turn over). Therefore, we propose in this paper a distributed approach using the multiagent system (MAS) technology [17]. The aim of this approach is to optimize resource management with a decentralized manner and to build an innovant simulator that helps define management strategies. Several works in the domain of container terminal manage- ment using MAS have been carried out. Indeed, Henesey in [9] uses an agent-based simulator for evaluating operational policies for containers transshipment in a container terminal. Lawrence in [10] describes an approach (supported with MASCommonKADS) enabling decision makers to simulate various port policies and analyze the multitude of “what if” scenarios to model the system architecture and the role of different parts which compose it. Degano in [?] who uses the petri network to model the behavior of each agent in order to define a hierarchical structure intended to monitor real-time fault diagnosis in an intermodal terminal. We cite also the work of Kefi in [11] who uses an informed algorithm to de- scribe the container agents behaviour introducing intelligence degree to reduce significantly the total number of unproductive movements. The container management requires implementation of a new strategy for management and decision making. To ensure an appropriate management, it is necessary to optimize the supply chain in a container terminal. In this paper, we use multi-agent systems to model and to simulate container management with the aim to deliver goods to customers in the right time in order to satisfy them. We intend to get through the simulation the evaluation of different planning policies for decision making on strategic choices. This allows to plan and to control the management of containers in a way to minimize the waiting time and the parasite shifts and to insure the consistency of the performed tasks sequence. To achieve this goal, we opt for an individual-centered approach [4] to design the containers life cycle by modelising it with a software agent. This structure enables the container to be an active agent of the management by assigning it the individualism properties as autonomy, rationality, socialization and the ability to communicate with other entities of the environment. The paper starts by describing, the container terminal, its different components and operations, as well as the containers life cycle. Next, it discusses the encountered problems in the container management domain through a literature re- view. Then, it presents the proposed multiagent model of the simulator using the MESSAGE/UML methodology [1] and the agent paradigm. It provides after that an implementation to experiment and to validate the model using the Netlogo simulation platform. Finally, it concludes with a discussion