OEM產業聚焦高科技新產品開發，委外生產製造給CM。新產品上市面對快速需求擴散過程和受學習效應影響的生產提升過程，產能的部署是一項挑戰。本論文建立 OEM 的兩階段產能部署模式，對OEM在高科技新產品的二階段產能配置提供決策支持。該模式建構一個數值模擬環境，模擬考慮耐用消費產品生命週期極短及新產品試產後的生產爬坡特徵。動態需求及供應的模擬系統由巴斯擴散需求率(Bass demand rate)和時間常數生產率(Time constant production) 來描述。 需求高峰和生產高原的不匹配為需求/供應交互軌跡帶來了平衡挑戰。生產網絡中不可分割且不可逆的塊狀產能增量加劇了產能部署的難度。為驗證兩階段產能部署模式的可行，細分時間為極小區間以產生離散化數據，提供不同參數組合情境下的數值實驗數據來源。結合數學和圖形分析和計算能力，在 CPLEX 上運行離散優化模型，實驗觀察巴斯擴散、產能提升和成本參數組合情境下的產能部署。 基於啟發式圖解開發優化模型，為面臨需求軌跡和產能軌跡特徵不匹配挑戰的OEM提供了研究成果。通過應用數學和圖形分析以及解算器運算能力，借助優化語言工具，驗證模式可行，進而實現OEM二階段產能部署“何時”和“多少”問題的決策支持。 ;The OEM industry focuses on developing high-tech new products and outsources manufacturing to CM. In the face of the rapid demand diffusion and production ramp-up process affected by the learning effect, the deployment of outsourcing production capacity is a challenge. This thesis develops a model to provide decision support for the OEM′s two-stage capacity deployment in high-tech new products. This model builds a numerical simulation environment, which considers durable consumer products′ very short life cycles and the production ramp-up characteristics after the new product pilot run. The dynamic evolution of the simulated supply and demand system consists of the Bath demand rate and the time constant production rate. The mismatch between production ramp-up and demand diffusion trajectory challenges the balance of interacting demand/supply. The indivisible and irreversible lumpy capacity increments in the production network exacerbate the difficulty of capacity deployment. To verify the feasibility of the two-stage capacity deployment model, the subdivision time is a very small interval to generate discrete data and provide a source of numerical experimental data under different parameter combinations. Combining mathematical and graphical analysis and computing power, discrete optimization models are run on CPLEX to experimentally observe the capacity deployment under a combination of Bass diffusion, capacity ramp-up, and cost parameters. The model is verified feasible to achieve decision support for "when" and "how much" issues of OEM two-stage capacity deployment.
