Although automated storage/retrieval systems (AS/RSs) with two depots at opposite ends of each aisle (TD-AS/RSs) have been adopted in I-shaped warehouses (one of the three typical warehouse layouts), class-based storage policy has surprisingly not been investigated in TD-AS/RSs. This paper studies TD-AS/RSs with two storage zones, aiming to determine the optimal first storage zone that minimises the S/R machine’s expected travel time. We first formulate an expected travel time model. The analysis is then performed for two situations: (1) the ratio between the number of storage and total requests (Ratio-ST) is constant and (2) the Ratio-ST changes with time but follows a probability distribution. For the first situation, utilising proposed properties, an efficient optimal search-based algorithm (i.e. Algorithm 1) is developed. The results show that the proposed storage policy can improve system performance by up to about 60% and 40%, respectively, compared with two policies used in practice. Contrary to the literature results, our results indicate that the single command cycle performs better than the dual command cycle in the TD-AS/RS. For the second situation, an algorithm based on Algorithm 1 is developed. A case study using two real-world datasets confirms that our policy outperforms practically used policies in this situation.

尽管在 I 形仓库（三种典型仓库布局之一）中采用了在每个通道的相对端有两个仓库的自动存储/检索系统 (AS/RS) (TD-AS/RS)，但基于类别的存储策略令人惊讶的是，没有在 TD-AS/RS 中进行研究。本文研究了具有两个存储区的 TD-AS/RS，旨在确定最小化 S/R 机器预期行程时间的最佳第一个存储区。我们首先制定一个预期旅行时间模型。然后针对两种情况进行分析：(1) 存储数量与总请求数之间的比率 (Ratio-ST) 是恒定的，以及 (2) Ratio-ST 随时间变化但遵循概率分布。对于第一种情况，利用所提出的特性，开发了一种有效的基于最优搜索的算法（即算法 1）。结果表明，与实际使用的两种策略相比，提出的存储策略可以分别将系统性能提高约 60% 和 40%。与文献结果相反，我们的结果表明单命令周期在 TD-AS/RS 中的性能优于双命令周期。针对第二种情况，开发了一种基于算法1的算法。使用两个真实世界数据集的案例研究证实，我们的策略在这种情况下优于实际使用的策略。开发了一种基于算法 1 的算法。使用两个真实世界数据集的案例研究证实，我们的策略在这种情况下优于实际使用的策略。开发了一种基于算法 1 的算法。使用两个真实世界数据集的案例研究证实，我们的策略在这种情况下优于实际使用的策略。