透過您的圖書館登入
IP:18.191.176.66
  • 學位論文

動態排程環境下面板廠製程機台配置方式之研究

Improving Production Performance of Flexible Flow Line with Machine Grouping Strategy - A Case Study of TFT-LCD Cell Manufacturing

指導教授 : 徐旭昇
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


面板為目前最重要之電子產品之一,市場需求是日益漸增,提昇產能及滿足顧客需求以增加市場佔有率為國內各大面板廠之主要目標。本研究針對TFT-LCD之Cell製程各階段機台數量該如何配置,以達到產能及顧客滿意度最大化加以探討,兩者相對應之指標分別為訂單內所有工件總完工時間(Total completion time)及總延遲件數(Number of Tardy Jobs)為最小。研究將以國內某TFT-LCD製造廠為個案,過程中先搜集相關資料包括產品種類,各產品在各製程單機情況下之單位加工時間,每月訂單之各種類產品之批量數及批量中之產品數量,以及製程各階段之機台數。研究中探討三種機台配置方案:(1) 流線型生產系統(Flow shop, FS);(2) 目前工廠之配置,為一個彈性流線型生產系統 (Flexible flow shop, FFS);(3)依據現場經驗而提出之另一種FFS生產系統。三種方案比較先採用工件(單種產品批量)到達時間為靜態(static)情況做探討,靜態排程假設各工件之資訊已知且可在同一時間排程;之後再探討更符合實務之動態(Dynamic)情況,動態排程假設下個工件到達時間、產品種類及其批量大小、交期均未知,這些資訊直到工件到達時刻才知曉。研究結果在靜態情況,以混合式基因演算法做上述三種方案之探討,研究結論為流線型系統為最佳。動態情況運用不同派工法則(Dispatching rule) FIFO、SPT、LPT、EDD、FST等,同時納入設置時間(Setup times),並考慮交期。考量Cmax及總延遲件數雙目標下,並配置不同權重,當同時考慮雙目標之下結論是以方案一流線型生產系統且使用LPT法則為最佳。

並列摘要


TFT-LCD is currently one of the most essential electronic products. Its demand is growing every month. To contend in such a competitive environment, TFT-LCD corporations must focus on productivity and customers’ satisfaction. This study aims to improve the TFT-LCD Cell manufacturing process for a corporation in Taiwan. The Cell process is a flexible flow shop. In this paper, we consider three feasible machine allocation alternatives, and recommend the best alternative based on two evaluation standards: minimizing the total completion time (makespan) and the number of tardy jobs. The analysis involves two cases: static scheduling and dynamic (online) scheduling with unknown release dates. The case study consists of four phases: (1) Data collection, including product types and their demands on each month of recent years, processing and setup time estimations, and the current machine configuration in each Cell stage; (2) Comparing three feasible alternatives, which are flow shop, one-job on one-machine, and current arrangement; (3) Generating test instances using the data collected in phase one for comparison purposes; (4) Evaluating performance based on static scheduling solved with GA, and dynamic scheduling with five dispatching rules: FIFO, SPT, LPT, EDD, and FST. Our experimental results conclude the following: The flow shop arrangement is preferred in a static scheduling environment, whereas the one-job on one-machine excels the other alternatives in a dynamic scheduling environment with the two standards of equal weight.

參考文獻


2.諶承正(2000),「運用系統模擬與基因演算法執行限制理論於流線型生產」,國立台北科技大學生產系統工程與管理研究所碩士論文
3.趙文涼(2000),「基因演算法於單機交期絕對偏差及整備成本最小化排程問題之應用」,元智大學工業工程研究所碩士論文
7.王治平(2003),「實際零工式生產排程問題的派工法則」,國立政治大學資訊管理研究所碩士論文
9.莊文化(2006),「應用基因演算法於彈性流線型工廠排程之研究」,國立成功大學工業與資訊管理系碩士論文
11.Baker, K. R. (1974), Introduction to Sequencing and Scheduling, Wiley, New York.

延伸閱讀