In this thesis, we study three emerging applications of OR/MS, namely, (i) disease spread modeling, intervention strategies, and food supply chain management during an influenza pandemic, (ii) the practical applications of production planning and scheduling in the commercial lithographic printing industry, and (iii) packing/placement problems in chip design in the semiconductor industry. In the first part of the thesis, we study an emergency response planning problem motivated by discussions with the American Red Cross, which has taken on a responsibility to feed people in case of an influenza pandemic. During an emergency such as an influenza pandemic or a bioterror attack, regular distribution channels of critical products and services including food and water may be disrupted, or some of the infected individuals may not be able to go to grocery stores. We analyze the geographical spread of the disease and develop solution approaches for designing the food distribution supply chain network in case of an influenza pandemic. In addition, we investigate the effect of voluntary quarantine on the disease spread and food distribution supply chain network. Finally, we analyze the effect of influenza pandemic on the workforce level. In the second part, we study a real life scheduling/packing problem motivated by the practices in the commercial lithographic printing industry which make up the largest segment of the printing industry. We analyze the problem structure and develop efficient algorithms to form cost effective production schedules. In addition, we propose a new integer programming formulation, strengthen it by adding cuts and propose several preprocessing steps to solve the problem optimally. In the last part of the thesis, motivated by the chip design problem in the semiconductor industry, we study a rectangle packing/placement problem. We discuss the hardness of the problem, explore the structural properties, and discuss a special case which is polynomially solvable. Then, we develop an integer programming formulation and propose efficient algorithms to find a ``good' placement.

Accurate demand forecasting is a key capability for a manufacturing organization, more so, a semiconductor manufacturer. Many crucial decisions are based on demand forecasts. The semiconductor industry is characterized by very short product lifecycles (10 to 24 months) and extremely uncertain demand. The pace at which both the manufacturing technology and the product design changes, induce change in manufacturing throughput and potential demand. Well known methods like exponential smoothing, moving average, weighted moving average, ARMA, ARIMA, econometric methods and neural networks have been used in industry with varying degrees of success. We propose a novel forecasting technique which is based on Support Vector Regression (SVR). Specifically, we formulate ν-SVR models for semiconductor product demand data. We propose a 3-phased input vector modeling approach to comprehend demand characteristics learnt while building a standard ARIMA model on the data. Forecasting Experimentations are done for different semiconductor product demand data like 32 & 64 bit CPU products, 32bit Micro controller units, DSP for cellular products, NAND and NOR Flash Products. Demand data was provided by SRC(Semiconductor Research Consortium) Member Companies. Demand data was actual sales recorded at every month. Model performance is judged based on different performance metrics used in extant literature. Results of experimentation show that compared to other demand forecasting techniques ν-SVR can significantly reduce both mean absolute percentage errors and normalized mean-squared errors of forecasts. ν-SVR with our 3-phased input vector modeling approach performs better than standard ARIMA and simple ν-SVR models in most of the cases.

有關既有廠商面對技術改變時為何會失去領導地位，現有文獻多以技術和廠商能耐互動的供給面切入討論，其原因可能是領導廠商無法及時轉換技術、調整能耐或組織惰性等因素；對於少數新興國家的後進廠商成功挑戰既有領導廠商，成為全球領導廠商的研究，也大多依循這種供給面觀點，認為建立並不斷提升技術能耐是成功的關鍵。 深入檢視既有廠商面對技術改變失去領導地位的原因，不完全是來自供給面的問題，市場需求面也是重要的影響因素。在討論市場需求面的相關理論中，斷裂性創新架構是最典型的觀點，該觀點認為新興廠商推動斷裂性創新會持續破壞現有市場遊戲規則，造成競爭基礎的改變與產業結構的變化，此觀點主要在說明市場顧客需求的變化對新舊廠商競爭的影響，並據以補充說明技術供給面的不足。 至今斷裂性創新研究並未發展出一個有系統的架構來檢測某種創新情境是否為斷裂性創新？因而，本研究根據斷裂性創新架構觀點及個案研究法中描述性理論架構的要求，發展出斷裂性創新描述架構的相關特性要件，並用此架構來檢測台灣積體電路公司及其所促進高度發展的全球晶圓代工產業，檢測結果發現過程符合斷裂性創新的特性。後續研究將可由斷裂性創新的觀點切入，重新詮釋台積電成功成為全球領導廠商的原因。 本研究發展出來的描述架構是一個檢測斷裂性創新的良好基礎，可以作為測試其他產業是否存在斷裂性創新的參考。 / Why established incumbents loosed their leading positions facing the disruptive technology, explanation of current literatures tended to focus on the supply-side interaction of technologies and firms’ capabilities. The research on latecomer firms catching up technological gap in newly development countries also focused on the supply-side perspective and argued that consecutive advancing technological capabilities was the key to success. Closer examination of technology competition, however, reveals that technology or market transitions are not necessarily due to the supply-side difficulties, like incumbent technology’s inherent limit, incumbents’ inability to master new skill, or organizational inertness. The market demand-side is also an important factor offering a complementary set of explanations that highlight the influence of consumers’ need on technology trajectory. The most influential expression of a demand side role in technology competition is the disruptive innovation that will break the market rule, shift the bases of competition and change the industry structure. Current researches, however, did not develop systematically analytical tool or framework to identify disruptive innovation. This study followed the major perspectives of disruptive innovation and the rules of case study to develop the descriptive framework for testifying the phenomenon of disruptive innovation. The semiconductor industry development in Taiwan was identified and described under the conditions of this framework. This descriptive framework is a systematic tool for testing disruptive innovation in any industry.

Vacuum technology is ubiquitous in the high tech industries and scientific endeavors. Since vacuum pumps are critical to operation, semiconductor manufacturers desire reliable operations, ability to schedule downtime, and less costly maintenance services. To better cope with difficult maintenance issues, interests in novel fault diagnosis techniques are growing. This study concerns model based fault diagnosis and isolation (MB-FDI) of dry vacuum pumps in the semiconductor industry. Faults alter normal operation of a vacuum pump resulting in performance deviations, discovered by measurements. Simulations using an appropriate mathematical model with suitably chosen parameters can mimic faulty behavior. This research focuses on the construction of a detailed multi-stage dry vacuum pump model for MB-FDI, and the development of a simple and efficient FDI method to analyze common incipient faults such as particulate deposition and gas leak inside the pump. The pump model features 0-D thermo-fluid dynamics, scalable geometric representations of Roots blower, claw pumps and inter-stage port interfaces, a unified pipe model seamlessly connecting from free molecular to turbulent regimes, sophisticated internal leakage model considering true pump geometry and tribological aspects, and systematic assembly of a multi-stage configuration using single stage pump models. Design of a simple FDI technique for the dry vacuum pump includes staged fault simulations using faulty pump models, parametric study of faulty pump behaviors, and design of a health indicator based on classification. The main research contributions include the developments of an accurate multi-stage dry pump model with many features not found in existing pump models, and the design of a simple MB-FDI technique to detect and isolate the common faults found in dry vacuum pumps. The proposed dry pump model can pave the way for the future development of advanced MB-FDI methods, also performance improvement of existing dry vacuum pumps. The proposed fault classification charts can serve as a quick guideline for vacuum pump manufactures to isolate roots causes from faulty symptoms. / text

Thesis (Ph.D)--Industrial and Systems Engineering, Georgia Institute of Technology, 2010. / Committee Chair: Keskinocak, Pinar; Committee Member: Ergun, Ozlem; Committee Member: Goldsman, David; Committee Member: Hupert, Nathaniel; Committee Member: Swann, Julie. Part of the SMARTech Electronic Thesis and Dissertation Collection.

L'industrie des semi-conducteurs est caractérisée par une production de forte variabilité et de faible volume, des flux de production ré-entrants ainsi que d'un processus de fabrication complexe. Au sein de ce contexte industriel complexe, a été considéré un problème de planification à capacité finie. C'est le problème de projection des encours de production et des commandes clients à capacité finie. Il s'agit d’estimerles dates de début, les temps d'attente et les dates de fin de chacun des steps des différents lots ainsi que la charge accumulée sur les équipements. Cette projection doit tenir compte des contraintes de capacité et qualifications des équipements et des dates d'échéance de livraison des lots. La contrainte de qualification définit l'éligibilité d'un équipement à traiter un produit. Ainsi, l'objectif de cette étude consiste à établir un plan de production réalisable à moyen terme. Afin de réaliser cet objectif, des méthodes exactes et approchées sont proposées. Des résultats en termes de complexité, et d'algorithmes de résolution, ont permis une application industrielle, dans la mesure où un logiciel de planification de la production à capacité finie a été développé. / In this study, we consider the problem of production planning in the semiconductor industry characterized by high mix low volume production, reentrant flows and complex manufacturing process.The aim of this work is to establish a feasible production schedule that takes into account the limited capacity of the manufacturing system, equipment qualifications constraints and delivery due dates. In this context, we have formulated the objective and constraints in a mixed linear program (MIP). The objective of the MIP is to minimize delivery delays to guarantee on-time delivery. While executing different tests of the MIP, we have reached a limit of resolution in a reasonable time. Thus, we use an approximate method to solve the problem. The results show the effectiveness of the heuristic established as solution quality and time resolution.The obtained results led to an industrial application and a software that provides feasible schedules in reduced execution time in a specific fab.

伴隨知識經濟時代的來臨，企業在全球化競爭下與產業高速變動的環境中，對於資訊與知識的質與量與日俱增，而為了因應全球經濟競爭與產業變遷，企業興起一股以知識管理為基礎的競爭優勢，作為發展與提升效能的利器。 知識管理對於傳統製造業生產流程效能的影響力已可由相關研究驗證，然而智慧型工程產業，如資本與技術密集的半導體資訊產業，是否在採取知識管理後能夠普遍提升生產流程的效能，尚待驗證，本研究係嘗試建立知識管理與半導體資訊企業生產流程效能之間的關聯性。 研究發現，半導體企業在削減變更成本、降低產品報廢率及重工率、產能擴充時間縮短、產品精密度與複雜度、以及彈性生產組態等生產流程效能上，都與知識管理中的知識管理策略、知識管理工具、知識管理網絡架構、以及知識管理流程有顯著的連結。知識管理策略與知識管理工具確實對半導體企業在削減變更成本、降低產品報廢率及重工率、產品精密度與複雜度、以及產能擴充時間縮短上，有顯著的效能表現 / The epoch knowledge economies bring high competitiveness in business market, and as well as stimulate the importance of information qualities and knowledge capacity. To face global economic developments and the transformation of industry, knowledge management thus has become the basic measure to improve the effectiveness of the developments of company. Evidently that knowledge management has been proved for the advantage of traditional industry in the effectiveness of production process, semiconductor industry however needs more verifications. The study hence is to scrutinize the connectedness between knowledge management and the effectiveness of production process in semiconductor industry. The study revealed that the overall effectiveness of production process indeed was correlated with the adoption of knowledge management system, and in detail, the strategies of knowledge management and the tools of knowledge management would enhance the effectiveness of production process in semiconductor industry.

Comment assurer un développement à risques limités alors que marchés et technologies sont inconnus? Ces situations ne peuvent pas être gérées par les stratégies existantes comme market pull ou techno push; et le processus type « trial and learning » apparaît trop coûteux et imprévisible. Or certaines industries comme les semi-conducteurs doivent organiser des développements technologiques à coûts et risques contrôlés, alors que les inconnues techniques et commerciales sont très importantes. Dans ces situations le sucés coïncide avec l'existence des technologies génériques. Mais comment concevoir ces technologies génériques? La thèse modélise la conception des technologies génériques et étudie des modelés d'action collective associés.Ce travail débouche sur quatre résultats principaux : 1) Une stratégie de gestion du risque original, consistant à identifier un « common unknown » ; 2) Un travail de modélisation et de simulation précisant les critères de contingence de cette nouvelle stratégie; 3) La caractérisation de concepts génériques et de processus de conception de généricité a l'aide des théories de la conception les plus récentes; 4) Une nouvelle figure managériale, le « cross-application manager», indispensable pour le succès de la conception de technologies génériques. / Can a company organize development with controlled risk while promising markets and eligible technologies are yet unknown? These situations cannot be handled with classical techno push or market pull strategies; and "trial and learning" processes are too unpredictable and too costly while facing with unknowns in high-tech. However, in some industries such as semiconductors, one has to organize exploration while controlling costs and risks in double unknown. In this situation the success of technological development in double unknown is correlated with the existence of generic technologies. Though, it remains unclear how to account for the design of generic technologies. This thesis models the design of generic technologies and investigates the associated models of collective action.This work is based on four main results:1) an original model of risk management that consists of identifying “common unknown” for several potential applications; 2) the modeling and simulation defining the techno-economic conditions in which the introduced “common unknown” strategy is relevant; 3) The characterization of the notion of generic technological concepts and the process of genericity design built upon the most recent design theory formalisms ; 4) a new managerial figure, the «cross-application manager », essential for the success of generic technologies design.

台灣半導體產業發展成熟，產業群聚效應明顯，供應鏈完整。基於晶圓製造過程中成本花費考量，若能尋找設備、零組件、材料原廠以外的第二來源，對於台灣半導體廠商而言，除了有機會降低成本，也能不受到原廠的牽制。另外，台廠期望能透過自主開發生產設備來取代設備的高額進口成本，然而目前本土設備廠商自主開發前段製程設備能力有限，向外尋求關鍵技術之合作為良好的解決方案。筆者思考這兩個方面是否能透過與俄羅斯合作，成功突破目前的情況。 本論文針對俄羅斯半導體產業過去發展、現況與未來發展進行研究，同時了解台灣與全球半導體產業現況及發展前景。運用文獻檢閱法搜集全球、台灣、俄羅斯半導體產業現有資料。筆者也使用訪問法，自行設計訪談問題，以台灣及俄羅斯半導體產業專家為對象進行訪談，並將搜集到的第一手資料資料進行整理，以Rugman 與 Verbeke的「雙鑽石模型」，分別為「生產因素條件」、「需求條件」、「相關及支援性產業」、「企業策略、結構與競爭」，以及第五個因素「政府」作為理論架構，分析台灣與俄羅斯半導體產業發展現況，並提出雙方可能合作機會與策略之建議。 研究發現俄羅斯科學基礎研究實力豐厚，晶片應用著重在軍事、航太等方面，晶片生產朝向少量多樣方向發展。從俄羅斯政府許多政策上可以其支持國內微電子產業的發展，並期望微電子產業的振興能帶動國內其他產業的升級。台灣晶圓製造能力強，供應鏈完整，應用領域主要專注於資通訊方面，雙方競爭可能性低。台灣在基礎研究與設備開發所需技術方面，實力較為不足，與俄羅斯進行技術合作可謂實現設備國產化的策略之一。經由理論架構分析，台灣與俄羅斯進行技術合作，有望能縮短設備開發所需的時日。台灣供應鏈及先進製造能力，能為俄羅斯特殊晶片提供生產製造、封裝檢測等服務。台俄雙方在半導體領域的合作機會大，可互相補足缺口，促成產業升級。 / The semiconductor industry in Taiwan is well developed. Because of the high cost of wafer manufacturing process, having a second source of equipment, components, and materials would be beneficial for Taiwanese semiconductor companies. The cost of production and dependency on original equipment suppliers will decrease. In addition, Taiwanese companies seek the opportunity to be self-reliant on developing production equipment in order to reduce the high cost of importing equipment. However, the capacity of Taiwanese equipment manufacturers is limited. Cooperate with foreign companies on key technology is the preferable solution. This thesis focuses on the cooperation with Russia companies and check whether Taiwanese companies can have a breakthrough at current situation. This study includes the past, present, and the future development of Taiwanese and Russian semiconductor industry. Literature review and interview method are two main research methods. Collecting current reference of global, Taiwanese, and Russian semiconductor industry and primary sources from the interview with semiconductor industry experts, the author conducts an analysis on the development of the semiconductor industry in Taiwan and Russia with the “Double Diamond” model of Rugman and Verbeke. The analysis factors include “Factors conditions”, “Demand conditions”, “Related and supporting industries”, and “Government.” Further, this research proposes advice for cooperation and strategies. Through theoretical analysis, technical cooperation between Taiwan and Russia is expected to shorten the develop time of equipment. Industrial supply chain and advanced manufacturing capacity in Taiwan can provide manufacturing, packaging and testing service for Russia. Cooperation between Taiwan and Russia in the field of semiconductor can complement each other and will promote industrial upgrading.

We have developed a placement-aware 3-D floorplanning algorithm that enables additional wirelength reduction by planning for 3-D placement of logic gates in selected circuit modules during the floorplanning stage. Thus it also bridges the existing gap between 3-D floorplanning and 3-D placement. To reduce the solution space of 3-D floorplanning which is known to be an NP-hard problem, we derive a set of feasibility conditions on the topological representation of a floorplan. In addition, we have designed a fast module packing algorithm that satisfies a set of constraints for placement-aware 3-D floorplanning. Furthermore, we have designed an efficient evolutionary algorithm that is used in the proposed 3-D floorplanning algorithm for multi-objective combinatorial optimization. Our results show that the proposed placement-aware 3-D floorplanning algorithm is very fast, and it reduces the system level total wirelength by 9.8% compared to existing state-of-the-art floorplanning tools that do not plan for 3-D placement of floorplanning modules.