作者: Osman M. Anlõ , Michael C. Caramanis
DOI:
关键词:
摘要: The purpose of this report is to explore ideas that we plan research further with support from NSF Grant DMII-0300359. To end, present a computationally tractable frame- work for decentralized, yet coordinated, management and control supply chains (SCs) capturing the salient stochastic dynamics facilities (production cells, warehouses, transportation links, test- ing, processing like) making up Supply Chain. In fact, SC modeled as network super nodes representing such facilities. Each facility queueing workstations operated by general material flow/routing policy. facility's dynam- ics are fast (they vary hourly) relative planning horizon's deterministic weekly). This enables time scale based decomposition algorithm assigns facility-specific perfor- mance sensitivity evaluation tasks decentralized sub-problem layer consisting performance evaluating sub problems. production assigned centralized determin- istic mathematical programming master problem layer. We optimize weekly schedules minimize inventory backlog costs subject non-linear constraints on imposed varying dynamic lead-times inter-facility quality service driven hedging policies. Computational experiments demonstrate it possible obtain significantly faster chain velocities static lead-time state art industry practice.