供應鏈協同式之生產規劃架構

Abstract

隨著半導體供應鏈的發展，各家企業廠商之間的關係錯綜複雜，形成動態網絡，各節點的決策行為是以自身企業利益目標進行規劃，一個節點的生產派工決策、產能、排程資訊與其他節點的活動存在互動牽連的關係，因此形成企業從本身的生產規劃延伸到與其它企業間的協調及合作機制；同時，造成上中下游整合的趨勢以完成產品製造之使命。因此本文針對晶圓前端製程流程，製造廠及針測廠間之生產特性，發展一生產規劃架構，用以規劃各階段最適宜生產之產品等級別及數量。 本論文包含兩大規劃程序，首先為晶圓製造廠的生產規劃程序，在此根據企業所規劃各產品之產出目標，考量晶圓製造各機台之產能水準、各製程機台之機能及同產品族生產數量等限制，在設定績效下，求算出晶圓製造廠生產之產品別以及數量。接著，配合生產週期時間的估算，推估出各工單之可交貨日程表等資訊，以利晶圓針測廠之生產規劃程序之運作。 接著在晶圓針測廠生產規劃程序中，則是彙整晶圓製造廠之生產規劃機制所推得的各工單之預計可交貨日、產品別及產出量等資訊，考量針測廠可提供之產能水準、機台製程規格能力等限制，求算針測廠各機台應分擔各類產品之負荷量，以及規劃期間內的最終機台設定位置。利用上述結果，再將機台指派規劃工作產出的資訊---工單對機台的派令(assignments)，作為依據的決策方法與規劃目標產生排程結果。 本研究選擇B-shop機台分派的設定轉換次作為績效考量，作為與F-shop協同生產排程規劃的指標。透過本研究構建之生產規劃架構，來進行協同式生產排程的分析，並以實驗討論有進行協同規劃比未進行協同規劃所帶來效益較佳。

Following the development of semiconductor industry, the relationship between firms in supply chains is very complex. The performance of supply chains depends much on the integration of business processes, decision-making and information systems. The production planning of an enterprise thus is extended for cooperating with others. Moreover, there is a trend to integrate the whole semiconductor factories for achieving the mission of producing final products. This study will focus on wafer fabrication and probing process and will develop a collaborative production planning system. Two core mechanisms are addressed in this thesis. The first is master scheduling system for the wafer fabrication. According to the throughput plan of the enterprise, we derive product mix with the considerations of the capacity limitation of each site, in order to suit the performance. Next, based on the estimated cycle time, we can predict the due date of each order that is an input data for the production planning system of probing process. The production planning mechanism for the wafer probing factory aggregates planning results of wafer fabrication, such as predicted due date of each order, throughput target of each product type, capacity supply and capability constraints of every workstation. With this result, this thesis then assign job-order of each product type to the workstation and provide last setup configuration of machines. In this study, the interaction between two nodes is discussed about B-shop setup changeover rate as a performance index. We developed a framework to integrate scheduling and planning processes and obtain better performance evaluation.