晶圓代工廠二階段訂單允諾機制

An available-to-promise based on two-phase order fulfillment for IC foundry

吳志武(Chih-Wu Wu)

指導教授 ： 林則孟

國立清華大學/工學院/工業工程與工程管理學系工程碩士在職專班/碩士(2007年)

摘要

面對激烈的競爭環境，晶圓代工業者必須要建立能滿足客戶需求，又能兼顧彈性的訂單允諾系統。本論文目的即在建立一套符合晶圓代工產能特性的訂單可允諾機制，此允諾機制包括客戶需求預測的產能保留及訂單沖銷機制。
本研究首先整理目前晶圓代工業者所使用的需求預測保留及訂單沖銷作法，並針對其缺點，提出改良式二階段訂單可允諾機制。在第一階段，由於晶圓代工廠產能在中、短期不易擴充，所以對於客戶需求預測若採取產能預留方法，最大挑戰是如何在考慮產能限制因素下，對每一客戶需求預測作最好的產能預留分配。而所要考慮的因素包括多廠區的產能限制、客戶指定廠區生產的限制、生產良率的限制、生產成本的限制、以及單一光罩只能在同一個廠區生產的限制等。本研究將針對這些限制建立一改良式數學模型，作為需求預測產能預留的機制。
在第二階段，根據客戶需求預測所提供的產能預留，當客戶實際訂單進來時，訂單與原需求預測間的沖銷方法，對於客戶訂單的可允諾數量及交期也會有重大的影響。目前實務上所採取的先進先出法，對於先來的訂單會優先使用當期最前面產能來作為訂單允諾數量及交期，而不會考慮到客戶或訂單的重要度。本研究提出的改良方法，是根據客戶所獲得的預留產能，將其均攤至各週作為訂單允諾的基礎，也就是說每個客戶每週的可允諾量是固定的，而不會受到訂單早來或晚來的影響，以改善實務上先進先出法的缺點。
另外，本研究探討當實際訂單要求數量與根據原需求預測預留的產能不符時，製造商可以採取的的因應對策，並提出當要求數量大於預留產能時的再分配機制，以符合實務上的需要。最後，藉由實際晶圓代工廠客戶需求與訂單允諾案例，驗證本研究所提出二階段訂單允諾機制的可行性。

關鍵字

可允諾量 ； 產能預留 ； 多廠區 ； 預測沖銷 ； 晶圓代工

並列摘要

Facing the fierce competition environment, the foundry industry has to build up an order fulfillment system that can meet the customer's need. This thesis purpose is to build up an order fulfillment system that matches the foundry industry characteristic. The mechanism includes customer demand forecast reservation and order netting.
This research summarize current working model of demand forecast reservation and order netting for foundry industry first. Then propose an improvement two-phase order fulfillment mechanism. At phase-1 stage, because the foundry industry capacity expansion is not easy in short-term, the most challenge is how to allocate the capacity to each customer demand considering capacity constraint. And the consideration includes multi-plant capacity, customer’s preference allocation, production yield, production cost, and single mask constraint. This research will build up an improvement model to well manage demand forecast capacity reservation.
At phase-2 stage, customer will place an order to consume the reserved capacity. But the placed order quantity may not equal to reserved capacity based on customer demand forecast. Currently actual practice is to use FIFO rule for the order promise and doesn’t consider customer importance. The improvement method is to promise order quantity equally based on the reserved capacity of each customer. And this research also study the issue that orders quantity is different from reserved capacity quantity and propose a management mechanism. Finally, by the actual foundry industry example to proof the mechanism is practicable.

並列關鍵字

available-to-promise ； capacity reservation ； multi-plant ； forecast netting ； foundry

參考文獻

3.張美滿，多廠區跨廠產能規劃之探討，元智大學工業工程與管理學系碩士班碩士論文，民國91年。

5.Barbarosoglu, G. and Yazgax, T.,”A decision support model for customer value assessment and supply quota allocation”, production planning and control 11, 608-616, 2000.

7.Ball, M.O., Chen, C.Y. and Zhao, Z.Y., “A model for batch advanced available -to-promise”, Production and Operations Management 11, 424-440, 2002.

8.Balakrishnan, N., Patterson, J.W. and Sridharan, V., “Robustness of capacity rationing policies”, European Journal of Operation Research 115, 1999.

9.Beamon, B.M. and Ware, T.M. “A process quality model for the analysis, improvement and control of supply chain systems”, Logistics Information Management 11,105-113, 1998