Multi-Agent Systems for Container Terminal Management

Sammanfattning: This thesis describes research concerning the application of multi-agent based simulation for evaluating container terminal management operations. The growth of containerization, i.e., transporting goods in a container, has created problems for ports and container terminals. For instance, many container terminals are reaching their capacity limits and increasingly leading to traffic and port congestion. Container terminal managers have several, often conflicting goals, such as serve a container ship as fast as possible while minimizing terminal equipment costs The focus of the research involves the performance from the container terminal manager’s perspective and how to improve the understanding of the factors of productivity and how they are related to each other. The need to manage complex systems such as container terminals requires new ways for finding solutions, e.g., by applying novel methods and technologies. The approach taken in this thesis is to model the decision makers involved in the container terminal operations and various types of terminal equipment, e.g., cranes, transporters, etc., as software agents. The general question addressed in this work is: can the performance of a container terminal be improved by using agent-based technologies? In order to evaluate the multi-agent based systems approach, a simulation tool, called SimPort, was developed for evaluating container terminal management policies. The methods for modelling the entities in a container terminal are presented along with the simulation experiments conducted. The results indicate that certain policies can yield faster ship turn-around times and that certain stacking policies can lead to improved productivity. Moreover, a multi-agent based simulation approach is used to evaluate a new type of Automated Guided Vehicles (AGVs) using a cassette system, and compare it to a traditional AGV system. The results suggest that the cassette-based system is more cost efficient than a traditional AGV system in certain configurations. Finally, an agent-based approach is investigated for evaluating the governance structure of the stakeholders involved in a transport corridor. The results of the research indicate that the performance of a container terminal can be improved by using agent-based technologies. This conclusion is based upon several studies, both conceptual and concrete simulation experiments. In particular, multi-agent based simulation seems to offer container terminal management a suitable tool to control, coordinate, design, evaluate and improve productivity.