- 學位論文
探討投料、緩衝及瓶頸資源派工於半導體封裝製程
Dispatching of Bottleneck Workstation Combined with Releasing and Buffer Control in Semiconductor Packaging Process
摘要
本研究以IC後段加工製程之半導體封裝廠為例，以限制資源為基礎進行滾動式生產計畫。主要架構包含投料計畫與現場管理，在已知訂單資訊下，考量到廠內生產特性-訂單加工之拆批行為及瓶頸站中機台型號之限制。此外，在瓶頸站加工時，因IC種類多且複雜，機台在轉換不同的產品進行加工時，須調整加工方式、程式設定等，會造成大量的設置時間(Setup Time)，導致單位時間的產出量下降，進而影響整體系統產出；另外，為了考量訂單在系統中的流程時間(Flow Time)，控制在製品水位亦為生產中不可忽視的議題。故本研究以最小化機台總設置時間及系統平均流程時間為績效目標，求得最佳派工、投料及緩衝控制之決策計畫。 本研究透過投料計畫與現場管理探討何種派工法則、投料法則及緩衝控制之組合，對於不同訂單與產能環境下，能有效地降低機台總設置時間及系統平均流程時間。半導體封裝廠為一混合流線型(HFS)生產系統，各站機台數量眾多，加上加工時間與瓶頸站設置時間具有隨機性及機台當機，需重複模擬以得績效評估值，因此本研究建構模擬模式評估並利用最佳模擬預算分配法(OCBA)有效分配有限模擬資源，同時節省模擬時間。 本研究經模擬分析結果得證，派工與投料法則會顯著影響系統績效，透過有效搭配可減少系統總設置時間及系統平均流程時間；總設置時間與系統平均流程時間有制衡關係，且不同績效權重會影響最佳方案之選擇；另外，因問題特性，確定性環境下所求之最佳方案與隨機性環境不同，但不影響派工與投料法則對於系統績效的影響程度。因此，派工與投料法則之選擇對於半導體封裝產業相當重要，最後本研究將系統流程架構及最佳方案之結果提供給半導體封裝業參考。
並列摘要
In this paper, a framework of rolling production plan based on bottlebeck resource is introduced. The framework consists of two parts- releasing plan and shop floor control. Semiconductor-packaging production system where a bottleneck workstation exists is used as the case study. In the bottleneck workstation, plenty of setup time is needed for changeover due to the complexity of products. Setup time in bottleneck workstation decreases the throughput per unit of time and thus has impact on the whole system. The methodologies to control the amount of work-in-process (WIP) are provided and mean flow time are used to measure their performance. To minimize total setup time and mean flow time, an adequate production plan must be adopted. The proposed framework is to discuss the multiple decisions comprising dispatching rule, releasing rule and buffer contorl must be well made to decrease total setup time and mean flow time effectively under different scenarios. Due to stochastic processing time and setup time in the bottleneck workstation, replications of simulation have to be conducted to estimate the performance of each design. In this article, an evaluative model is constructed. Moreover, optimal computing budget allocation (OCBA) is applied to efficiently allocate limited simulation budget and reduce the total simulation time. It is proved that dispatching and releasing rule have significantly impact on total setup time and mean flow time. Furthermore, the relationship of trade-off between total setup time and mean flow time are obvious and different weight will affect the choise of alternatives. Due to the characteristics of the problem, the consequence are different between the deterministic and stochastic environment, but the dispatching and releasing rules won’t make change the effect of the performance of system.