The discussion research of overall equipment effectiveness in semiconductor factory -A case study of semiconductor testing house

• Main Authors: Jiun-Bin Chiou, 邱俊斌
• Other Authors: Tai-Yue Wang
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/79818150726732392332

碩士 === 國立成功大學 === 工業與資訊管理學系碩博士班 === 94 === Abstract How to fully utilize the equipment to improve the production efficiency is very important to the semiconductor industries. Besides it is a heavy burden for the industry to invest on newer equipment because of the constantly changed demand and the continuously advanced technology in the industry. Overall Equipment Effectiveness (OEE), a common metric to measure the production efficiency, has been adopted completely in the semiconductor industry for calculating the production efficiency. However, it is difficult for the companies to estimate the OEE correctly and they usually use a fixed OEE rate, 70 %, as the standard OEE. Therefore, this study aims to construct OEE estimation model and find the factors influencing OEE in order to find ways improving OEE. And this model will help to make the production schedule better and to have better customer service. In this study, we first model the relationship between OEE and manufacturing cost to show the importance of OEE estimation. Then, the latent factors influencing OEE can be discovered by factor analysis. Finally we construct OEE estimation model by three methods: multiple regression analysis, back-propagation neural network, and compound estimation model. The verifification of model’s estimation performance via a case study of an IC testing house is performed. Through empirical analysis, the ratio of the decreasing manufacturing cost and the increasing contribution profit to total manufacturing cost is about 2.74 % when OEE rate increases 1 % under the current OEE level. Furthermore, five latent factors are presented by factor analysis to provide the improving directions of OEE for managers. Finally, we show the estimated OEE model has significant improvement compared to the current OEE estimation.