Barge processing policies in container terminals

In this thesis, we develop a simulation program which can be used to evaluate policies in a port. We use this program to examine the use of a multi-agent system for making appointments between barge operators and container terminal operators. In an existing model, terminals reserve their quay cranes for specific barges at specific times based on appointments. Since it is hard to predict the arrival times of barges at terminals, it often occurs that terminals keep their quay cranes reserved, while it could already handle other barges. We suggest that this results in more total waiting time for the whole system. Therefore, we propose an unreserved policy, where appointments are still made, but are less strict than with the reserved policy. The simulation experiments show that using an unreserved policy results in significant lower average sojourn times in the port, especially in busy ports. The explanation for the lower sojourn times is that the terminals are less idle, which makes the throughput higher. In addition, we examine the implications for individual barges by measuring their satisfaction with the waiting time and the provided information. The results suggest that barges are more satisfied with the waiting time in case of using an unreserved policy, and that barges are only somewhat satisfied with the provided information for both policies, since the predicted sojourn times in the port lack precision.