| 摘要: | 如何不增加過多成本又能降低訂單遲交，漸少平均製程的時間，是各工廠都欲持續改善的議題，藉以再提高獲利，降低成本。在目前後疫情時代的下半場，全球產業都面臨獲利緊縮的衝擊，低成長低獲利的情況下，此議題亦突顯重要。本研究的個案公司以往工廠對於生產排程的安排憑藉著經驗，再經由事後檢討來做調整修改來做排程安排。若能藉由一種科學性、系統性、持續性且穩定性的方式，將數據資料經過系統軟體運算，得到有用且具價值的參考資訊，輔以相關使用者可定義的目標數據，將可提供決策者在進行決策時的評估指標，進而選擇排程方案。電腦模擬軟體另一優點，當生產條件變更時，因模型已建立使用，只需調整所需參數值，便能觀察相對應之各項輸出變化，對於系統面的影響供決策參考，適合實務面的應用。
本研究以半導體封裝廠前段製程段為例，使用Arena模擬軟體建構模型與模擬的技術，對於生管人員的排程方法做評估研究，蒐集現有排程的歷史數據，投料為隨機性，將建構模型，並檢測模型差異性，最後模擬生產線原始排程與希望改善的派工法則間的差異作分析，本文使用派工法則中的先進先出法、最短處理時間排程法與最短交期法，三種方式與現行產線的原始排程進行比較。模擬結果顯示，三種排程方法都能改善現有的排程方式的兩個績效評估指標：1.縮短平均製程時間。2.減少訂單延遲。尤其是最短處理時間排程法(Shortest Processing Time, SPT)的改善幅度最大，每訂單平均製程時間減少了12.7%,即28.5小時。平均訂單延遲比率，每個月可改善了4.2%,約當2.6批。
;How to reduce order delays and decrease average process time without increasing costs is an ongoing improvement issue for factories. This is aimed at enhancing profitability and reducing expenses. In the current post-pandemic era, global industries are facing the impact of tight profitability and low growth. Therefore, this issue becomes even more crucial. The case in this study previously relied on experiential-based scheduling arrangements in the factory. If data can be processed through a scientific, systematic, continuous, and stable approach using system software, valuable and useful reference information can be obtained. This information, along with relevant user-defined target data, can serve as evaluation indicators for decision-makers in selecting scheduling strategies. Another advantage of computer simulation software is that once the production conditions change, only need to adjust the required parameter values in the established model allow for the observation of corresponding output variations. The system-level impacts provide decision-maker references on the suitable for practical applications.
This research focuses on the front-end process of semiconductor packaging factory and applied Arena simulation software to construct models and use simulation techniques. The research evaluates the scheduling methods used by production control personnel by collecting historical scheduling data, and the dispatching time is randomness to construct the model. The differences in the models are then examined, and an analysis is conducted comparing the original production line schedule with the desired improved dispatching rules. The research compares three dispatching rules: First-In-First-Out (FIFO), Shortest Processing Time (SPT), and Earliest Due Date (EDD), against the original schedule of the production line. The simulation results demonstrate that all three scheduling methods can improve two performance evaluation indicators of the existing scheduling approach: 1. Reduction in average process time, and 2. Decreasing order delays. The SPT method shows the most significant improvement, with an average reduction of 12.7% in the average process time per order lot, equivalent to 28.5 hours. The average order delay rate can be improved by 4.2% per month, equivalent to approximately 2.6 lots |
