液晶面板組裝廠批量製程派工法則之設計

Abstract

TFT-LCD產業包含Array、Cell及Module三大主要製程，其中Cell之後段製程係由數個批量工作站所組成。多數的學者皆僅考量如何提昇單一批量工作站之績效，如此將會忽略整體生產線的平穩化。且Cell段製程之批量機台又有許多製程特性上的限制，例如各批量工作站間有等候時間限制、批量機台特殊開工條件之限制等，更加深了其排程的複雜度。因此，本文發展一同時考量前後製程平穩之批量派工法則，以完成下列兩項目標：在不影響產出前提下，縮短批量製程生產週期時間;在不違反製程特性與交期限制下使產出最大化。 故本文利用限制理論(TOC)之觀念，發展一同步化集批法則(Synchronized Batching Heuristic)，首先透過「生產規劃設計流程」求算系統中各工作站之負荷情形，且定義系統瓶頸為具最小可換線次數之工作站，並根據系統瓶頸位置及排程資訊設定批量決策參數。接著，在「非瓶頸批量工作站派工流程」中，先利用「換線評估機制」求算換線與不換線兩種狀況下之最早可載入時點以評估是否值得換線生產，再以「產能負荷評估機制」動態地去衡量是否因載入過小批量或集批時間過久而浪費過多產能，並以「最佳載入時點評估機制」根據下一工件到臨時點及下一製程等候線長度去決定最適載入批量及最適載入時點，期使數個連續批量工作站間達到平穩的生產，完成既定目標。 模擬結果顯示，本文發展之同步化派工法則在各種不同利用率水準之環境中，均能在不影響產出情況下，有效地縮短批量製程生產週期時間。且在高負荷之系統環境中，也能在不影響生產週期時間的前提下使產出量上升。實例驗證顯示，當系統之工件到達批量製程之不確定性愈大時，同步化派工法則相較於其它法則下更具優愈性。整體而言，吾人所發展之批量派工則同時具備有效性及可行性。

The three main process of TFT-LCD are Array, Cell, and Module. The back-end of Cell process is combined by several batch workstations. Most scholars focus on how to improve the single batch workstation performance; however, the stability of the entire production system is ignored. Moreover, there are many process constraints at the batch workstation of Cell process, such as: queue time limitation, the special terms to run machine, etc. All of these will embitter the difficulty of scheduling. This study develops a dispatching rule that consider the production stability for the process contain multiple types of batch machine to accomplish the following two objects：shorten the production cycle time under the hypothesis of reaching the throughput target; increase the throughput under the hypothesis of satisifying due day constraint. This study develops a Synchronized Batching Heuristic rule (SBH) by using the concept of TOC. This rule is composed of two parts. At first part, it calculates the loading of each workstation and defines the bottleneck as the workstation with the minimum number of setup allowable. Then it sets the decision parameters based on the bottleneck location and schedule. At the beginning of second part, it evaluates the necessity of setup by comparing the loading time under setup situation to not setup situation. Second, it dynamic evaluates if load with the current batch size will cause the capacity of workstation being wasted too much. Finally, it determines the proper loading batch size and timing with consideration of the next job arrival time and the queue length of next batch workstation information. Simulation experiments reveal that by using the proposed dispatching rule can shorten the product cycle time under different system loading level. For the production system with high loading level, it can increase the throughput. Furthermore, it has the good performance if the uncertainty of arrival time is severe. Overall, the SBH rule has its effects for TFT-LCD Assembly Factory.