作業等候時間限制下批量及多產品生產系統控制

Abstract

本論文探討具有作業等候時間限制下，不同製造系統的允入控制方法。在半導體、液晶顯示面板及其他產業中，製品於系統內之等候時間常受限於特殊製程需求，違反時間限制將導致重工或在製品報廢。此類系統若無妥善之生產控管，將導致生產良率及產能利用率降低，並顯著增加整體生產成本。 另一方面，為了增加系統產出率，能同時加工多個產品之批量機台，常被運用於生產系統中，幫助公司減少生產成本及等候時間。而為了滿足多樣性顧客需求，能加工不同種類產品之多產品機台，亦常利用於生產系統中，幫助公司增加整體收益。因此本研究分別針對批量製造系統及多產品製造系統，探討系統中製品具有等候時間限制情況下，尋求最佳之生產控管方法。 本研究基於馬可夫決策過程，在具有時間限制問題情況下，於批量製造系統提出批量允入控制法則(batch process admission control, BPAC)，於多產品製造系統提出動態允入選擇控制方法(dynamic admission selection control, DASC)。這兩種策略決策目標皆為最小化總生產成本，總生產成本為在製品等候成本及報廢成本加總。相較於文獻中其他方法，模擬結果顯示BPAC及DASC兩種策略，於多個重要之系統績效指標皆有傑出表現。不但減少總生產成本及總報廢產品數量，亦可提高總生產產品數量。

This paper studies the admission control method for different manufacturing system with process queue time constraints. In semiconductor, Thin Film Transistor-Liquid Crystal Display (TFT-LCD), other industries, the waiting time of WIP is limited to upper bound in system, the time constraint is called the process queue time constraint (PQT). When the PQT constraint is violated, a significant scrap cost will be accrued. The queue time constraint makes the production control difficult, production managers need to balance the need for achieving production targets and yield quality. To raise the productivity, batch process machine is widely applied in manufacturing system. The batch machine can process several jobs to reduce waiting time and improve system throughput rate, the increased productivity will facilitate industries to gain more revenue. To fulfill the variety customer demand, multiple product machine is also broadly used in manufacturing system. The multiple product machine can process different type of jobs to meet customers’ diversified demand, the overall revenue will be increased. Hence, this research focuses the batch process manufacturing system and multiple product manufacturing system. Under the PQT problem, we develop the optimal production control method for the manufacturing systems aforementioned. We use Markov decision processes (MDP) to formulate a batch process admission control (BPAC) model in batch process manufacturing system under PQT constraint. Moreover, we also propose a dynamic admission selection control (DASC) model in multiple product manufacturing system under PQT constraint. The objectives of our policies are to minimize the sum of inventory holding costs and scrap costs. We also conduct simulation study to verify the robustness of our method. The simulation results reveal that our policies significantly reduce total costs while improving throughput and system utilization.