供應鏈動態事件之前攝應變與分析

Abstract

許多生產供應鏈在產品需求及製造面臨極高的不確定性，當供應鏈中產能或是存貨傾向精實狀態時，動態事件的發生，不僅對發生事件的工廠造成影響，同時也會影響供應鏈中其他成員。半導體製造無論是在內部製程或是外部需求皆有很大的變異，況且產能投資成本高，無法以過多的生產餘裕應付不確定事件，相關事件發生，往往對供應鏈造成很大的衝擊。動態事件根據其變動或影響的程度，可以分為三個等級，即偏移、擾亂以及災難。供應鏈通常能處理偏移。本研究著重在擾亂事件發生於滿載情況下供應鏈的反應行為，經由系統設計的方式將動態事件的影響控制在偏移程度。本文供應鏈模型的建構包含兩個部分，一為單一節點的生產函數，另一為動態連結兩生產節點運作的機制。文獻中有兩種觀點的生產函數，時間遞延模式以及投入產出模式，但都受到產能的侷限或是只考慮非滿載的穩態系統。本文以投入產出模式為基礎，建構滿載生產函數，並發展數學規劃模型，處理動態事件下協調多個節點的生產策略並求解最佳的回復方式，以提升供應鏈績效及可控制性。

High level of uncertainties in product demand and manufacturing is a major characteristic facing many manufacturing supply chains. If a supply chain is lean in buffer capacity or inventory, an event that arises from the uncertainties will have an impact not only on the operation of the plant in which the event takes place but also on other plants of the chain. In semiconductor manufacturing there are also many internal or external uncertainties and capacity is lean due to the extremely high cost of equipment. When dynamic events occur, they will have a serious impact on the whole chain. Based on their impact on the nominal operation, uncertainties can be distinguished in three levels: deviation, disruption and disaster. Supply chains are usually designed to cope with deviation in operation parameters and they undergo irreversible changes when disasters strike. The focus of this paper is on the behavior of supply chains under disrupting events in full-load states. To manage dynamic events, a supply chain operation model is described which comprises a production function for production units under full-load and a dynamic model linking the operation of multiple units over a certain duration. In the literature, there are two approaches to modeling the behavior of production units which are time-delay function and input-output control. The production functions in the literature are for regular load scenarios, so a full-load production is derived by using of flexible capacity through alternative routing machines. Finally, a mathematical solution method based on non-linear programming is described for optimizing a recovery path. This method enables supply chains to treat disrupting events as controllable deviation event and to enhance the controllability of the chain.