多工單等級下晶圓廠生產週期時間估算模式

Abstract

晶圓製造廠所追求的生產目標不外乎是高產出率、短生產週期時間、高良率及準時交貨等，而上述目標均與生產週期時間有關，故生產週期時間的掌握在晶圓製造廠中佔有舉足輕重的角色。有鑑於此，本研究應用區塊基礎式週期時間演算法(Block Based Cycle Time Forecasting Algorithm; BBCT)與產能分割之概念，推估產品在各工作站之生產週期時間。 首先，因應多工單等級之環境，本文考量工作站不同之加工批量特性，及前後機台產出速率相異等糾結而成之交互的影響，將產品等候時間之成因分成：（1.）負荷因素－以non-preemptive priority queueing model估算、及（2.）批量因素－以修正之區塊基礎式週期時間演算法估算之。 批量因素等候時間主要是來自於工件在批次工作站前所引發之集批時間與批次工作站之釋料造成下游序列工作站的負荷瞬間增加，形成之尖峰負荷等候時間。因此吾人以BBCT為基礎，推估各製程區塊之批量因素等候時間，並加總負荷因素等候時間，推估出產品週期時間。接者再以各製程區塊為基礎，依照區塊特性之不同，推算出各製程步驟之週期時間，得到各工作站之週期時間。 經實例驗證顯示：本文模式所求得之產品生產週期時間與模擬結果作比較，平均誤差僅為1.74%。工作站週期時間之平均誤差為14.98%，除了加工批量為4與機台數為1之工作站未能達到較準確的估算外，其他之工作站皆有不錯成效。此外本模組之運算時間僅需數分鐘，故本文所構建之模式可快速的提供生產規劃及控制系統做為參考。

The production targets for the wafer fabrication include high throughput、short cycle time、high yield and on time delivery,…,etc. All these targets are related with the cycle time. Precise and quick estimating of product’s cycle times and estimating workstation’s flow time are important. In this thesis, the average waiting time for a product waiting at a workstation is separated into two parts. First part is the waiting time resulted from the workload of the workstation, which is estimated with the non-preemptive priority queuing model. Second part is the waiting time related to the batch issue and is derived by the BBCT algorithm, in which two batch-operations and all in-between operations are defined as a block. To fit the multiple-priority orders, BBCT algorithm is modified with the concept of capacity splitting, and named BBCT-MP. The waiting time incurred from the batch issue are the batch forming time for lots waiting before the batch workstation and the peak load digestion time caused by lot releasing from a batch workstation to its downstream serial workstation. Thus, the variety of batch size on workstations, the difference in throughput rates among workstations and the characteristics of material flow existed in a block are considered for each block. The product’s cycle time is estimated by summing up all the block cycle time, calculates by BBCT-MP and queuing theory, of the corresponding process plan. Then, based on the characteristics of each block, the flow time for each process step is derived so as to get the workstation’s flow time. The case study shows that there only exists about 1.74% difference in product’s cycle time and 14.98% difference in workstation flow time when compared to simulation results. Most kinds of workstations have acceptable accuracy of flow time estimation except for the workstation with batch size being four lots and for the workstation containing single machine unit. Meanwhile, the calculation takes only a few minutes instead of a few hours for the simulation model. Thus, the model developed for the cycle time estimation can quickly provide the good enough reference digits to production planning and production activity control system. 圖目錄 vi 表目錄 vii 符號一覽表 viii 第一章 緒論 1 1.1 研究背景動機 1 1.2 研究目的 1 1.3 研究範圍與限制 2 1.4 研究方法 3 第二章 文獻探討 6 2.1 緊急訂單對週期時間之影響 6 2.2 週期時間之估算方法 7 第三章 模式構建 12 3.1 問題分析與定義 12 3.2 多等級生產週期時間估算法之原理與架構 15 3.3 負荷因素等候時間之求算方法 18 3.4 產品之生產週期時間估算法 23 3.4.1 BBCT演算法[6] 23 3.4.2 BBCT-MP估算法 30 3.5 工作站之生產週期時間估算法 31 3.5.1 推估工作站週期時間之基本概念 32 3.5.2 瓶頸工作站與集批時間 33 3.5.3 SB型區塊 34 3.5.4 BB型區塊 36 3.5.5 BS型區段 39 3.5.6 BSB型區塊 39 3.5.7 工作站週期時間之估算 40 第四章、實例驗證 41 4.1 系統環境說明 41 4.2 執行方法過程 42 4.2.1 產品別生產週期時間之估算 43 4.2.2 工作站別生產週期時間之估算 46 4.3各等級下產品生產週期時間與工作站利用率之驗證 47 4.4 各工作站之生產週期時間之驗證 54 第五章、結論與未來研究方向 66 5.1 結論 66 5.2 未來研究方向 67