| Abstract: | 面臨世界金融危機的衝擊，在各行各業不景氣的情況下，勢必要開源節流。所
謂節流，就公司經營而言，即降低成本或避免資源浪費。現今我國企業為了在不景
氣中能夠生存，各產業體系均致力於強化上中下游成員或夥伴間之多層次分工和協
同合作，提出有效的解決方案，故如何透過有效的供應鏈運作以便產品能夠快速地
佔有市場、降低庫存風險及成本壓力，快速回應顯然已成為企業生存之必要條件。
「供應商管理存貨」(Vendor-Managed Inventory；VMI)，即是企業為了使生產和銷
售目標順利達成而採用之一種有效的快速回應管理利器。在運作時企業會特別注意
訂單和排程的變動，供應商連續自動補貨，可防止缺貨的情況發生，並使生產排程
更穏定，結果可降低存貨量和長鞭效應、減少運籌成本、提升物料供應之服務水準。
去年本計畫以台灣TFT-LCD 廠商為例，著重於找出產業協同供應鏈VMI 系統
中影響VMI 績效之重要因素，作為本計畫第二年建構產業協同供應鏈整合性VMI
模式的重要基礎與內涵。今年本計畫考慮與實際狀況較為契合之供應鏈，即一家核
心廠(例如，友達光電)與供應商之多階層(例如，三階層)VMI 系統，首先定義操作
性變數之符號與提出基本假設，建構數學規劃模式，發展有效的求解方法，並進行
實驗設計及求解效能分析。今年本計畫欲達成的目的為求得供應鏈各成員在計畫期
內的最適訂購、製造或供料次數及最適補貨週期時間，使供應鏈VMI 總成本(包括
第一層核心廠之存貨持有成本、訂購成本、資訊分享成本；第二層或第三層供應商
之存貨持有成本、生產設置或訂購成本、補貨成本、運送成本、資訊分享成本和供
應商管理存貨成本等)極小化，以達到降低供應商及核心廠庫存量，減少供應鏈長鞭
效應，提升供應商供料的服務水準，降低核心廠缺貨危機。最後，本計畫將以台灣
TFT- LCD 廠商進行實務個案研究。
本計畫可分為下列七大部分：
1. 研究目的與問題描述。
2. 相關文獻探討。
3. 利用斐氏圖建構多階層供應鏈成員間之VMI 流程模型，再由改善後VMI 流
程圖發掘影響VMI 運作的重要因素。
4. VMI 重要因素操作性定義、數學符號說明與基本假設。
5. 建構數學規劃模式和發展有效的求解方法。
6. 實驗設計及求解效能分析
7. 台灣TFT-LCD 廠商實務個案研究。
In the face of the seriously economical crisis in today’s world, businesses in
each industry should save costs and make an effort to find profitable opportunities in
the situation where the global economics is in a state of deep depression. As far as
business operations are concerned, cost savings are the cost reduction and the waste
elimination. To achieve the goals of survival and growth, each industry must endeavor
to strengthen the relationships among the partners in the upper, middle, and lower
streams of the supply chain. They should closely cooperate together, propose effective
strategic and action plans, and organize a collaborative team to rapidly penetrate
markets, considerably reduce the stock level, and significantly improve their
competitive advantages through effective supply chain operations. Obviously, quick
response to customer requirements has become one of the prerequisites for business
survival. For this reason, vendor-managed inventory (VMI) that is one of the useful
tools to respond quickly has been widely used in various industries. Businesses must
place special emphases on the variations of customer orders and schedules, capacity
commitment, shortage prevention, and vendor automatically continuous
replenishment when operating a VMI system. Consequently, the whole supply chain
of the industry can lower inventory level, reduce the impact of the bullwhip effect,
decrease logistics cost, and increase the service level of product or material supply.
In the last year, this project focused on studying the case of three TFT-LCD
companies in Taiwan from the viewpoint of industrial collaboration and explored
important factors that significantly affect the performance of VMI systems. The
results can be used as a preliminary basis to construct an integrated VMI system in
this year. The VMI system (e.g., a three-level supply chain) includes a core TFT-LCD
company (i.e., AU Optronics Corporation (AUO)) and its up-stream vendors.
In this year, this project will first define the notation of operational variables
and make the basic assumptions. Then, an integrated VMI model will be formulated
for the industrial supply chain according to the last year’s research results (i.e., VMI
performance influential factors and crucial problems) and it will be solved by an
optimization package (e.g., LINGO) for the small-scale problems since it has been
proven to be NP-hard. For the reason of difficultly obtaining an optimal solution, an
effective heuristic method will be developed and genetic algorithms including
modified and hybrid search methods will be used to solve the large-scale problems. In
addition, an experiment and a computational efficiency analysis will be also
conducted. Finally, the proposed mathematical programming model and methods will
be applied to a well-known Taiwanese TFT-LCD company and its up-stream vendors
in order to verify the aptness of these approaches.
This study can be divided into the following seven parts:
1. Set the research purpose and describe the problem.
2. Review the related literature.
3. Use Petri-Net approach to construct and improve a multi-level supply chain VMI
flow model and to find out important factors that have a significant influence on
the performance of the VMI system.
4. Define the notation of operational variables and make the basic assumptions.
5. Formulate a mathematical programming VMI model, develop an effective heuristic
method, and use genetic algorithms to solve the problem.
6. Conduct an experiment and a computational efficiency analysis for the proposed
model and methods.
7. Apply the proposed model and methods to a well-known Taiwanese TFT-LCD
company (i.e., AUO) and its up-stream vendors to demonstrate the aptness of these
approaches. |
