Abstract: Automated container terminals have been more and more emphasized all around the world. Automated container terminals are facilitated with complex infrastructures designed specifically to handle a large number of containers, and they are playing important roles in international freight transport. In the automated container terminals, there are automated stacking cranes (ASCs) that move containers into and out of the container yard blocks, and automated guided vehicles (AGVs) that move containers inside the terminal yard. The layout and facilities in automated container terminals are different from those in traditional terminals, which bring more challenging operational problems.
Container terminal operators must deal with a wide variety of interrelated logistic problems every day, and the container space allocation in the terminal yard will impact the efficiency and productivity of the terminal yard. In the container terminal yard, the space allocation of outbound containers directly affects the loading efficiency and the berthing time of vessels in the port. Nevertheless, in the existing researches on container allocation problems, the outbound container arriving sequence and the stacking strategy are not considered as the key factors during the space allocation. Management strategies considering these factors are therefore necessary to increase the efficiency and productivity, and thereby reduce the operational costs of related container terminals.
This paper studies the space allocation of outbound containers in an automated container terminal. The random berthing of vessels and random arrivals of outbound containers at the terminal are considered and an integer programming model is established. Some existing researches based on optimization algorithms have been proposed to tackle this problem but they do not pay attention to the uncertain arriving sequence of containers and block balance during the space allocation.Considering these characteristics, we provide a mathematical formulation where the objective is to minimize three parts: The working time of AGV and ASC in the future vessel loading process, and the imbalance between different blocks during the space allocation. We propose a so-called “far-destination container above near-destination container” policy to locate the containers in a certain bay, and a simulation-based genetic algorithm is proposed to solve the problem. In the existing research, the number of rehandles in the container retrieval process has not been well estimated. In the proposed algorithm, a simulation module is established to evaluate the rehandle number during in the future container retrieval process. The output of the simulation module is returned to the genetical algorithm as the fitness measure. The algorithm overcomes the flaws of slow convergence, falling into the local optimum, and premature convergence of the genetic algorithm.
In the computational experiments, different instances are generated to test the effectiveness and efficiency of the algorithm. The experimental results show that the proposed algorithm can solve the problem efficiently. We conduct the sensitivity analysis considering different initial block layouts, different outbound container arriving sequences, and different weight coefficients of objective function. Through the computational experiments, it is found that the average container rehandle number and the imbalance of the blocks increase with the number of allocated containers. The proposed “far-destination container above near-destination container” policy proposed in this paper has good performance in reducing container rehandles.
The research results of this paper provide management insights for the practical space allocation in automated container terminals, and contribute to the efficiency improvement of other related operations in the terminal. The simulation-based genetic algorithm proposed in the paper provides a new solution for the related optimization problems with random factors. This research also enriches the container terminal operation literature.
There are some future research directions: 1) To consider the rolling time horizon model for the container space allocation, in which both space and time factors could be involved in container allocation, and 2) to consider the AGV waiting time in the water side, specifically AGV waiting time for the quay cranes to unload the container from the AGV in the water side.

Key words: automated container terminal, outbound container, space allocation, simulation-based optimization, genetic algorithm

摘要： 在集装箱码头堆场中,出口集装箱的箱位分配直接影响集装箱的装船效率以及船舶的在港停留时间。研究主要探讨自动化集装箱码头出口集装箱的箱位分配问题,并将船舶靠泊随机性和出口集装箱集港顺序综合考虑到整数规划模型中。针对模型的特点,设计了基于仿真优化思想的启发式算法求得集装箱贝位分配量,并提出“长途箱压短途箱”的贝内具体落位策略。多组实验结果表明,提出的基于仿真的遗传算法能够有效解决自动化集装箱码头出口箱的箱位分配问题。通过多组算例对比实验发现贝位平均翻箱次数和箱区间作业不平衡度随集装箱数量增加而增加,且“长途箱压短途箱”堆存策略能有效减少未来取箱的翻箱次数。研究结果为智能港口、全自动化集装箱码头的运营提供了思路和方向,有助于实现集装箱码头各子系统一体化效率的提升。

关键词: 自动化集装箱码头, 出口集装箱, 箱位分配, 仿真优化, 遗传算法