晶圓製造廠多等級工單最適比率決策模式之構建

Abstract

現今半導體產業為了面對多變的市場環境，不僅要利用製造彈性之優勢來擴大市場，以因應顧客化程度的提高，更需要縮短生產週期時間以減短先進製程或新產品之研發時間，期能領先潮流並創造未來市場。為了縮短特定產品或工程實驗批之生產週期時間而設定的高等級工單，是目前晶圓製造產業為了達到上述目的所應用之手法。本文因應高等級訂單對半導體產業的重要性，於總體規劃階段提出一套決定多等級訂單最適比率之決策模式，以期能同時考量高等級工單之正反面影響及未來效益，選取最適之多等級訂單組合。 本文所提之多等級工單最適比率決策模式，乃根據企業規劃各產品族之產出目標及現有產能狀況，利用瓶頸不允許漂移之假設及緊急工單不受集批法則限制的前題下，發展一「高等級訂單比率上限分析機制」。接著，求取各等級工單不同比率之情境組合，並透過模擬系統，產生在各情境組合下系統之生產績效資訊。接續，依本文所發展之「生產績效評估機制」，針對多等級工單將造成之生產績效降低，評估其所產生之損失，並作篩選刪除不具競爭力之情境。此外，本文建構之「最適比率選取機制」將利用「立即收益標準化模組」為各可行情境依其所提供之收益值，產生一無尺度之客觀指標值。並利用「無形效益評估模組」，透過分析階層程序評估各可行情境之未來效益，產生一主觀指標值。最後，依本文所發展之「不同等級偏好尺度」，綜合主、客觀指標值為各可行情境求得一最終評選值。 實驗結果顯示，本文建構之決策模式除了對高等級訂單在財務上之影響有更深入之分析外，亦考量了不同比率組合之高等級訂單所能帶來之未來效益，透過一完整之規劃與決策流程，提供管理者更多攸關且完整之資訊，決定最適高等級訂單比率，提供未來生產系統承接及管理訂單等級之依據。

In today’s versatile environment, semiconductor industry not only needs the flexible production advantage to expand the market by customization, but it also needs to decrease R&D time in advanced process or engineering lots by shortening production cycle time in the attempt to create future market. In order to shorten the production cycle time of specific products or engineering lots, a higher priority is often given to these lots in current wafer fabrication. To meet the importance of multiple-priority orders in the semiconductor industry, this research proposes a decision model in the aggregate production planning stage. The decision model simultaneously considers the positive and negative effects and future benefits of multiple-priority orders in the selection of the optimal priority mix. The optimal priority mix decision model first develops analytic mechanism for setting upper-limit of higher priority orders. With such a limit, we set all the available priority mixes under the same throughput tartget. Then we use simulation model to obtain production performance foe each available priority mix based on the current capacity and the assumption that bottleneck cannot be shifted and hot lots are not restricted by batch size constraint. Next, a production performance assessment mechanism is developed to appraise the losses due to the reduction of production performance under each multiple-priority mix, and then each of the uncompetitive mixes is deleted. Next, an optimal priority mix selection mechanism first utilizes an immediate profit standardization module to generate a no-scale objective measure for each feasible priority mix based on its profit. Aslo an invisible benefit assessment module is developed based on analytical hierarchy process to produce a subjective measure for each feasible priority mix based on future benefit of each mix. Finally, the objective and subjective information is integrated so as to select the optimal priority mix which will lead to the maximum benefit for the entire system. Experiment result shows that the proposed decision model supplies a complete planning and decision process for managers to select the optimal priority mix. This model not only has a comprehensive analysis on the financial impact by multiple-priority orders, but also considers the future benefit that higher-priority orders can offer under different priority mixes. Through the decision process, managers can get relevant and complete information to determine the optimal priority mix. The result can provide production system with reference for accepting and managing orders in the future.