Vessels served by a container port can usually be classified into two types: deep-sea vessels and feeders. While the arrival times and service times of deep-sea vessels are known to the port operator when berth plans are being devised, the service times of feeders are usually uncertain due to lack of data interchange between the port operator and the feeder operators. The uncertainty of feeder service times can incur long waiting lines and severe port congestion if the service plans for deep-sea vessels and feeders are poorly devised. This paper studies the problem of how to allocate berths to deep-sea vessels and schedule arrivals of feeders for congestion mitigation at a container port where the number of feeders to be served is significantly larger than the number of deep-sea vessels, and where the service times of feeders are uncertain. We develop a stochastic optimization model that determines the berth plans of deep-sea vessels and arrival schedules of feeders, so as to minimize the departure delays of deep-sea vessels and schedule displacements of feeders. The model controls port congestion through restricting the expected queue length of feeders. We develop a three-phase simulation optimization method to solve this problem. Our method comprises a global phase, a local phase, and a clean-up phase, where the simulation budget is wisely allocated to the solutions explored in different phases so that a locally optimal solution can be identified with a reasonable amount of computation effort. We evaluate the performance of the simulation optimization method using test instances generated based on the operational data of a container port in Shanghai.

集装箱港口服务的船只通常可分为两种：深海船只和接驳船。在制定泊位计划时，尽管深海船只的到达时间和服务时间是港口运营商已知的，但由于港口运营商与支线运营商之间缺乏数据交换，因此支线船的服务时间通常是不确定的。如果对深海船只和支线船的服务计划设计不当，支线船服务时间的不确定性可能会导致漫长的等待线和严重的港口拥堵。本文研究了如何为深海船舶分配泊位并安排支线船到达港口以缓解拥堵的集装箱港口的问题，在该港口，支线船的数量明显大于深海船舶的数量。支线的服务时间不确定。我们开发了一种随机优化模型，用于确定深海船的泊位计划和支线船的到达时间表，从而最大程度地减少深海船的离港延误和支线船的排班时间。该模型通过限制馈线的预期队列长度来控制端口拥塞。我们开发了一种三相仿真优化方法来解决此问题。我们的方法包括全局阶段，局部阶段和清理阶段，其中将模拟预算明智地分配给在不同阶段中探索的解决方案，以便可以通过合理的计算工作量来确定局部最优的解决方案。我们使用基于上海集装箱港口运营数据生成的测试实例评估模拟优化方法的性能。