半導體一元化代工服務之整合生產與運輸規劃問題

Integrated Production and Transportation Problem (IPTP) for Semiconductor Manufacturing with Turnkey Service

鍾一平(I-Ping Chung)

指導教授 ： 彭文理

國立交通大學/管理學院/工業工程與管理系所/博士(2015年)

https://doi.org/10.6842/NCTU.2015.00327

摘要

半導體生產一元化代工服務係由晶圓製造廠整合產業之生產資源，同時提供晶圓針測(Circuit Probing Testing)、封裝(Assembly)與最終測試(Final Testing)的一元化半導體生產加工服務，讓客戶省下繁複的訂單與跟催作業，並降低半成品在各階段廠商與客戶間運輸的前置時間，此舉大大地提高客戶下單的意願，進而創造晶圓代工廠的競爭優勢。所以，如何充分利用產能，滿足顧客訂單需求來增加競爭力及獲利，為一重要之研究課題。半導體後段製造具有多樣產品別、多製程階段，以及跨多廠區的特性，從生產製造的面向來看，需考量多家代工廠的設置成本、製程能力差異、代工廠的產能規模，以及其生產不同產品樣式的加工成本；從運輸配送的面向來看，則須考量到多台車輛資源限制、車輛運輸距離限制，以及單位運輸成本。然而，單獨追求生產或運輸成本的最佳化，並沒有辦法帶來最大的整體效益。因此，本研究將深入探討整合生產運輸規劃問題(Integrated production and transportation problem, IPTP)，以解決提供半導體一元化代工服務的晶圓製造商所面臨的實務問題。
本論文探討兩種IPTP問題，首先為考量外包廠一般產能及無車輛運輸距離上限的IPTP問題(IPTP with normal outsourcing factory capacity and unlimited vehicle traveling length; nIPTP)，另外則是考量外包廠備用產能及有車輛運輸距離上限的IPTP問題(IPTP with backup outsourcing factory capacity and limited vehicle traveling length; bIPTP)，此二種生產配送規劃問題皆以最小化生產及運輸成本為目標。
於本論文中，我們針對nIPTP與bIPTP兩個生產配送規劃問題分別建立其混合整數規劃模式(MILP)，並求取最佳解。在nIPTP問題研究中，我們提出三個啟發式解法(VRSCA、OFIA與OFDA)來解決問題，由運算結果及績效比較中得知，本研究所提出之啟發式法有很好的求解品質。在一個包含30個產品別與19個工廠的大型實務問題當中，OFIA平均能在1.15秒鐘找到近似最佳解，其解題的品質與效率均優於其他兩個演算法；在bIPTP問題研究中，由於考慮了代工廠商的備用產能，以及增加了運輸車輛的運輸距離限制，本文提出一套基因演算法模式(Genetic Algorithm, GA)作為求解的工具。從實驗設計的數據分析中得知，GA在低產能負荷及低產能變異的題型之下，其解題品質相對較好。在小型與中型的問題中，GA的求解品質相對穩健，因此，當面對MILP無法在合理時間找到最佳解的大型問題時，GA足以勝任解題之工作。

關鍵字

整合生產與運輸問題 ； 半導體生產一元化代工服務 ； 混合整數規劃 ； 基因演算法

並列摘要

Solving an integrated production and transportation problem (IPTP) is a very challenging task in semiconductor manufacturing with turnkey service. A wafer fabricator needs to coordinate with outsourcing factories in the processes including circuit probing testing, integrated circuit assembly, and final testing for buyers. The jobs are clustered by their product types, and they must be processed by groups of outsourcing factories in various stages in the manufacturing process. Furthermore, the job production cost depends on various product types and different outsourcing factories. The IPTP is very complicated since it is a dimensional problem which involves job clusters, job-cluster dependent production cost, factory setup cost, process capabilities, and transportation cost with multiple vehicles, with a minimal total cost criterion. Therefore, the development of efficient algorithms is critical to solve the IPTP for semiconductor manufacturing with turnkey service provider.
In this dissertation, we considered IPTP with normal outsourcing factory capacity and unlimited vehicle traveling length (nIPTP), and IPTP with backup outsourcing factory capacity and limited vehicle traveling length (bIPTP). The objective of both problems is to minimiza total production and transportation cost. We first formulate the problems as two mixed integer linear programming models (MILP) to obtain the exact solutions. We present three heuristic algorithms (VRSCA, OFIA and OFDA) based on the network algorithms with some modifications to accommodate the nIPTP. From the computational tests, the performances of the proposed model and heuristic algorithms are quite satisfactory. In a large size real-life problem with 30 product types and 19 outsourcing factories, the OFIA outperforms the other two algorithms, obtaining the near optimal solution within 1.55 CPU seconds. The bIPTP considering outsourcing factory back-up capacity and limited vehicle route traveling length is too complicated to solve efficiently and effectively. Therefore, an efficient genetic algorithm (GA) is proposed to tackle the problem. The GA performs well under the problem characters of low capacity loading and small capacity variance. Although GA may not outperform the MILP for small-size or medium-size problems, it can obtain near-optimal solutions for large-size problems while the MILP model may not obtain a solution after a long computational time.

並列關鍵字

integrated production and transportation problem ； semiconductor manufacturing with turnkey service ； mixed integer linear programming ； genetic algorithm

參考文獻

[1]Aliev, R.A., Fazlollahi, B., and Guirimov, B.G., and Aliev, R.R. (2007). Fuzzy-genetic approach to aggregate production–distribution planning in supply chain management. Information Sciences, 177, 4241–4255.

[2]Aytug, H., Khouja, M., and Vergara, F.E. (2003). Use of genetic algorithms to solve production and operations management: a review. International Journal of Production Research, 41, 3955–4009.

[3]Babu, A.J.G. and Sureshb, N. (1996). Project management with time, cost, and quality considerations. European Journal of Operational Research, 88(2), 320–327.

[4]Benjamin, J. (1989). An analysis of inventory and transportation costs in a constrained network. Transportation Science. 23(3), 177–183.

[6]Boese, K.D. and Kahng, A.B. (1994). Best-so-far vs. where-you-are: implications for optimal ?nite-time annealing. Systems & Control Letters, 22, 71-78.