在進料檢驗過程中,進貨品質檢驗數量多寡將會影響生產成本與作業時間。當抽取過多的樣本可能造成檢驗時間過長,而影響後續整體生產作業時間。反之,抽樣減少,則可能誤判退貨,造成商譽受損或因退貨斷料反而無法如期交貨。另外,為有效控制抽樣成本,必須制定合宜之抽樣檢驗計畫。本研究探討當抽驗發生判退時,在不影響壞批被允收情況下,使好批被批退風險降低,建立適當的抽樣計畫。從MIL-STD-105E及MIL-STD-1916檢驗準則衍伸,建立合理及適當的第一階段抽樣數目,並決定接受此批或拒絕此批;若拒絕則進行第二階段之全檢計畫。進而能在最短的檢驗時程中做出最大的品質把關以縮短作業時間及人力成本。提出二階段特殊抽樣計畫,在第二階段之檢驗個數則從1500個降至120個,作業成本因此而降低92%,作業時間因此減少23小時。
In a feed-testing process, sample size of quality inspection will affect the production cost and operation. A long inspection time will be required with large sample size, while a high risk of being short of material or failing to deliver products on schedule if rejecting a good batch of material falsely when a small sample size is applied. Hence, handling sampling costs consisting laber and time effectively, an appropriate sampling scheme should be considered. In this study, we propose a different sampling way, two-stage special sampling plan, based on the MIL-STD-105E and MIL-STD-1916 standard sampling schemes. In our proposed method, we determine the most suitable sample size at the first stage based on the AQL and LTPD in an acceptance sampling. If the inspecting decision is to reject the batch ,then full inspection plan starts at the second stage. In this way, we expect to reduce the sampling cost but without lowering the inspection quality. We implement our sampling plan to a set of real date and analysis result shows that the initial sample size can as small as 120 instead of 1500. Meanwhile, operational cost reduces 92% and operational time shortens 23 hours. Key words: Wafer Fab, Sampling scheme, Acceptance sampling .