TFT-LCD廠生產組合決策評估系統之構建

Abstract

近年來，薄膜液晶顯示器製造商積極擴建新世代廠，且生產之產品種類趨於多樣化，以滿足顧客之需求。然而，薄膜液晶顯示器之製程步驟主要分為三個階段，各階段製程之製程特性並非相同，部分工作站具有換線時間相對於加工時間相當長，因此產品種類數的多寡將影響生產系統之生產績效。此外，由於平均興建一座新世代廠約需新台幣700億，且機台價格動輒上億元的情況下，廠房規模之大小將直接影響財務相關之績效。因此，在同時考量財務與生產現場之績效下，本文提出「生產組合決策評估系統」，在追求整體生產效率最佳之原則下，於建廠前規劃生產組合，亦即決定廠房規模與產品種類數。 對於此課題，本文首先發展「生產組合情境方案設計機制」，設計不同廠房規模與產品種類數並進行配對，以獲得各種生產組合情境方案。接著，「機台規劃模組」針對各生產組合情境方案，在避免各階段製程之瓶頸發生短暫飄移的前提下，利用數學模式規劃各階段製程各工作站之機台數。接著，為能利用模擬蒐集各生產組合情境方案之相關生產績效值，「主生產排程規劃模組」分別建構組立階段製程與組裝階段製程之瓶頸工作站產能配置之線性規劃模式，以規劃各期各產品於各階段製程之加工數量與順序，作為模擬運作之投料方式。最後，「最適生產組合評選模組」運用模擬系統產生各生產組合情境方案之生產績效資訊，利用資料包絡分析法之交叉效率法進行評估，以選出具有最佳整體生產效率之生產組合情境方案。 實驗結果顯示，本文所提出之生產組合決策評估系統，可在因應未來市場需求下，快速且合理地設計各種不同之生產組合情境方案，並且在同時衡量多項績效指標的前提下，利用交叉效率法客觀且明確地評選出整體生產效率最佳之生產組合情境。整體而言，本文所提之方法淺顯易懂，除了方便管理者使用，亦可提供企業一系統化之評選流程，以在建廠規劃前決定最適之廠房規模與產品種類數。

In order to meet customers’ demand, the TFT-LCD manufacturers build new generation factories actively and produce more various products, in recent years. However, the manufacturing process of TFT-LCD mainly consists of three stages, each stage has its distinct characteristics. Notice that the setup time of some workstations are very long, thus the number of product types being produced in a factory simultaneously will influence the system performance. Furthermore, since the capital of a new generation factory costs 70 billion new Taiwan dollars and a machine costs more than one billion new Taiwan dollars commonly, the factory scale will affect the financial performance. Thus, considering financial and production performance simultaneously before building a new factory, this thesis proposes a decision evaluation system for production mix setting to determine the factory scale and the number of product types for achieving the optimal productive efficiency as the principle.

This thesis first develops production mix design mechanism to obtain all available production mix scenarios. Then, in tool planning module, with the principle of avoiding the bottleneck shifting in each manufacturing stage, a mathematic model is proposed to plan tool units needed for each workstation for every production mix scenario. Next, the master production scheduling and planning module constructs two linear programming modules for cell assembly and module assembly respectively, to derive the production quantity of each product type and the production sequence at each time period of each manufacturing stage. Then, the simulation model is run with these results as the input to obtain production performance for each production mix scenario. Finally, the optimal production mix assessment and selection mechanism is built. Based on cross efficiency method of Data Envelopment Analysis (DEA), this mechanism evaluates all available production mix scenarios based on the performance values obtained by simulation system and determines the production mix scenario with the optimal productive efficiency.

Experimental result shows, the proposed decision evaluation system can not only generate all available production mix scenarios fast and reasonably to meet market demand, but also select the production mix scenario with the optimal productive efficiency objectively and explicitly under multiple performance measurements. Overall, this syetem can be easily understood and used by managers, and provides an enterprise with a systematic assessment procedures to determine the factory scale and the number of product types before building a factory.