The most important issue at a container terminal is to minimize the ship’s turnaround time which is determined by the productivities of quay cranes (QCs). The tandem lift quay cranes have 33% higher productivities than single lift QCs. However, the tandem lift operations bring new challenges to the vehicle dispatching at terminals and this has become a big issue in the application of tandem lift QCs. The vehicle dispatching at terminals is to enhance the QCs’ productivities by coordinating the QCs’ operation schedules and the vehicles’ delivery schedules. The static version of the problem can be formulated as an MILP model and it is a combinational optimization problem. When the type of QC is tandem lift, the problem becomes more complicated because it requires two vehicles side by side under the QC. Thus, the alignments of vehicles have to be considered by coordinating the delivery schedules between vehicles. On the other hand, because the containers are operated alone by the yard cranes, the vehicles could not be grouped and dispatched in pairs all the time. This dissertation investigates the static and dynamic version of the problem and proposes heuristic methods to solve them. For the static version, Local Sequence Cut (LSC) Algorithm is proposed to tighten the search space by eliminating those feasible but undesirable delivery sequences. The time windows within which the containers should be delivered are estimated through solving sub-problems iteratively. Numerical experiments show the capability of the LSC algorithm to find competitive solutions in substantially reduced CPU time. To deal with the dynamic and stochastic working environment at the terminal, the dissertation proposes an on-line dispatching rule to make real-time dispatching decisions without any information of future events. Compared with the longest idle vehicle rule, the proposed priority rule shortens the makespan by 18% and increases the QCs’ average productivities by 15%. The sensitivity analysis stated that the superiority of the priority rule is more evident when the availability of vehicles is not sufficient compared with the frequency of releasing transportation requests.

Reinforcement learning techniques offer a very powerful method of finding solutions in unpredictable problem environments where human supervision is not possible. However, in many real world situations, the state space needed to represent the solutions becomes so large that using these methods becomes infeasible. Often the vast majority of these states are not valuable in finding the optimal solution. This work introduces a novel method of using linear programming to identify and represent the small area of the state space that is most likely to lead to a near-optimal solution, significantly reducing the memory requirements and time needed to arrive at a solution. An empirical study is provided to show the validity of this method with respect to a specific problem in vehicle dispatching. This study demonstrates that, in problems that are too large for a traditional reinforcement learning agent, this new approach yields solutions that are a significant improvement over other nonlearning methods. In addition, this new method is shown to be robust to changing conditions both during training and execution. Finally, some areas of future work are outlined to introduce how this new approach might be applied to additional problems and environments.