| Summary: | 碩士 === 國立臺灣科技大學 === 工業管理系 === 103 === A hybrid flow-shop (HFS) is a general manufacturing system in which the number of machines in a stage should present multiple levels because of maintenance, partial failures, unexpected failures, etc. In other words, it is suitable that the capacity of each stage is regarded as stochastic (i.e. multistate). From the decision-making viewpoint, a practical issue in industry is whether the demand enables to be completed within a desired due date. That is, makespan for a specific amount of demand satisfies due date. To respond above issue under the stochastic capacity of each stage, network reliability can assess the performance of a HFS facing uncertain demand. This paper first transforms the HFS with stochastic capacity into a multistate hybrid flow-shop network. Since the makespan for all jobs cannot be computed directly, an algorithm is then proposed in chapter 3 to derive an estimated interval in terms of a pair of capacity vectors generated from two estimated demand levels. Two practical cases, including a tile manufacturing system and a footwear manufacturing system, are presented to demonstrate how estimated interval is obtained and to investigate efficiency and accuracy by conducting experiments. In order to obtain the exact network reliability, an algorithm is further developed in chapter 4 to find the exact-lower boundaries that satisfy demand and due date and network reliability is evaluated. In addition, a practical example of an IC card manufacturing system is utilized to illustrate the proposed algorithm. From the decision-making viewpoint, supervisors can evaluate network reliability by proposed algorithms and treat it as a capability indicator to know whether HFS satisfies demand within a desire due date and make decisions.
|
|---|
