考慮用餐效應之晶圓廠直接人力規劃

Abstract

在半導體製造廠，晶片的產出量在作業員用餐時可能突然銳減，本研究欲找出在適當的工作中心增加人力，使作業員用餐效應影響最小的方法。首先先計算每個工作中心之人力工時，配合模擬估算出每個工作中心的平均人力需求。然後利用邊際配置法與田口實驗法進行用餐時段增加作業員之實驗，兩種實驗之結果皆顯示，考慮用餐效應的結果比不考慮用餐效應的結果較好，且在考慮用餐效應下，不僅僅具有瓶頸機台的工作中心增加人力會對產出有影響，其他工作中心加人對產出也會有影響。本研究發現使用邊際配置並考慮用餐效應來進行實驗，結果會比較具正確性。

In semiconductor manufacturing, the fab throughput may suddenly drop due to the meal breaks of operators. This research investigates methods for adding operators to cost-effectively absorb the meal-impacts. The operator times required for each work center are first analyzed. The average demand of operator for each work center, obtained by simulation, is taken as the initial staffing decision. The number of operators at each workstation is then added, by two systematic methods, to see if the fab throughput can be increased. The two methods are the marginal allocation method and the Taguchi methods. Simulation experiment results show that considered meal-impacts outperform non-considered meal-impacts, and the result also show that not only added operators in bottleneck machines, but also added operators in non-bottleneck machines. This research found that using the marginal allocation method and considered meal-impacts would be the better solution in the operator staffing decision.