This paper investigates the Stochastic Container Relocation Problem in which a flexible service policy is adopted in the import container retrieval process. The flexible policy allows the terminal operators to determine the container retrieval sequence to some extent, which provides more opportunity for reducing the number of relocations and the truck waiting times. A more general probabilistic model that captures customers’ arrival preference is presented to describe the randomness for external truck arrivals within their appointed time windows. Being a multi-stage stochastic sequential decision-making problem, it is first formulated into a stochastic dynamic programming (SDP) model to minimize the expected number of relocations. Then, the SDP model is extended considering a secondary objective representing the truck waiting times. Tree search-based algorithms are adapted to solve the two models to their optimality. Heuristic algorithms are designed to seek high-quality solutions efficiently for larger problems. A discrete-event simulation model is developed to evaluate the optimal solutions and the heuristic solutions respectively on two performance metrics. Extensive computational experiments are performed based on instances from literature to verify the effectiveness of the proposed models and algorithms.

本文研究了随机容器重定位问题，该问题在导入容器检索过程中采用了灵活的服务策略。灵活的策略允许终端操作员在一定程度上确定集装箱的取回顺序，这为减少重新安置的次数和卡车的等待时间提供了更多的机会。提出了一种更通用的概率模型来捕获客户的到达偏好，以描述外部卡车在其指定时间窗口内到达的随机性。作为多阶段随机顺序决策问题，首先将其公式化为随机动态规划（SDP）模型以最大程度地减少预期的重新安置数量。然后，考虑表示卡车等待时间的次要目标，对SDP模型进行扩展。基于树搜索的算法适用于将两个模型求解到最优。启发式算法旨在有效解决大型问题的高质量解决方案。建立了离散事件仿真模型，以分别在两个性能指标上评估最优解和启发式解。基于文献中的实例进行了广泛的计算实验，以验证所提出的模型和算法的有效性。