在半導體供應鏈中，前段製程廠以及後段製程廠的協同是相當重要的，而在目前的文獻中處理半導體供應鏈問題多半是透過混合整數線性規劃解決，而當談及透過模擬的手段 (例如：使用Anylogic、Mosilab等)，大多都是只針對前段製程廠或者後段製程廠進行單一工廠的模擬，鮮少有人針對整體的供應鏈進行模擬。 本研究在供應鏈的背景下，探討供應鏈協同問題設計包含前段製程廠、後段製程廠以及區域集散中心的模擬模型，首先會提出一個混合整數線性規劃模型以提供數學的基礎，接著透過Anylogic進行兩階段的實驗，前置實驗的模擬時間為12個月，目的是為了收集整個供應鏈的數據，並針對工廠進行個別分析。正式實驗的模擬時間為27個月，並且進行50次的測試且每次測試會有五筆訂單，訂單抵達時間呈現指數分布、訂單的訂購量則為常態分配，接著，透過正式實驗取得的數據，分析在各項指標 (例如：訂單達交率等) 表現良好的測試具備何種特質。 最後，透過本研究可以發現在半導體供應鏈中，區域集散中心對前段製程廠所下的訂單對整體供應鏈的表現有顯著影響，另外也發現前段製程廠以及後段製程廠的協同能夠幫助增加訂單的達交率並降低平均在製品存量。 ;In the semiconductor supply chain, the coordination between Front-End and Back-End manufacturing factory plays a crucial role. However, current literature solving semiconductor supply chain problems relies on Mixed-Integer Linear Programming. When it comes to the simulation-based methods (e.g., using AnyLogic, Mosilab, etc.), most studies focus on simulating either the Front-End or Back-End manufacturing factory independently. Few studies have been made to model and analyze the semiconductor supply chain as a whole through simulation. This study investigates coordination problems within the supply chain by developing a simulation model that incorporates Front-End manufacturing factory, Back-End manufacturing factory, and regional distribution centers. Initially, a Mixed-Integer Linear Programming model is proposed to provide a mathematical foundation. Then, a two-stage experiment is conducted using AnyLogic. The first stage is designed to simulate 12-month period and is designed to collect data across the supply chain, along with individual factory performance analysis. The second stage is designed to simulate 27-month period and consists of 50 experimental runs, each involving five orders. The arrival time of each order follows an exponential distribution, while order quantities follow a normal distribution. Based on the data gathered from the second simulation stage, this study analyzes which characteristics are associated with better performance across several key indicators (e.g., order fill rate). The results demonstrate that improved coordination between Front-End and Back-End facilities in the semiconductor supply chain can enhance order fulfillment rates and reduce work in process inventory levels.
