Peel pack planning: a comparison of decomposition approaches
研究了通过引入剥离包装减少手术器械浪费的优化问题，提出了两种分解方法（基于聚类的分解法和列生成分解法），并通过模拟数据比较了它们的计算性能。
Surgical trays are assembled with surgeons’ preference cards to satisfy the demand of multiple surgery types, leading to reprocessing waste when unused instruments are exposed in an open container. We explore a novel probabilistic tray optimisation problem on reducing surgical instrument waste by strategically incorporating peel packs (PPs). The problem involves multiple surgery types, instruments, and expected PP configuration design under expert guidance. Our primary goal is to minimise total tray processing costs via the probability of instrument usage while ensuring all required sterile instruments are available during procedures. The decision to open a custom PP hinges on the availability of its assigned instrument. Failure to immediately access the required instrument necessitates retrieving an entirely new surgical tray from the Sterile Processing Department, which is a costly scenario we aim to prevent. We formulate the tray optimisation problem (TOP) as a mixed-integer program (MIP) and propose two distinct solution approaches to tackle its computational complexity: (1) a Clustering-Based Decomposition Approach (CBDA) designed for computational efficiency, and (2) a Column Generation Decomposition Approach (CGDA) providing near-exact solutions. Through numerical experiments using simulated data, we analyse the scalability of the proposed methods. Our results indicate that an increase in the number of instrument types has a significantly impressive impact on the computational time of the CBDA compared to the CGDA. We demonstrate that the CBDA exhibits robust performance, effectively providing high-quality solutions, making it a practical method for implementation in hospital settings to achieve substantial reductions in instrument tray waste.