产能过剩支持可实现增加产量和缩短周期时间等目标。此外,各个系统在产能相互支持的影响下,系统外部的干扰加大,其绩效预测将比以前困难得多,产能支持行动的实施也变得困难。然而,对于Gemini晶圆厂来说,实时控制行为更容易实现,主要是因为两个晶圆厂之间的距离相当近,通过传输设备可以实现所谓的实时产能支持控制。例如,当生产线上出现意外的生产状况时,如产品大量破洞/脱模、机器严重停机等;或者由于晶圆制程中逐渐增加的生产约束的影响,如等待时间约束(时间约束紧急工作(Hot lot)【11】。运输系统产能决策的配合和考虑必将成为未来12英寸Gemini晶圆厂管理中必须面对的课题。
针对产能保障问题,涂等人[25]过去提出利用防护能力和机器容量负荷的概念来计算合理的产能保障量。等待理论和利特尔定律的概念构建了一个有效的估计模型。虽然研究环境的考虑是研究两个独立工厂之间的产能支撑问题,但产品备份的设置只是针对单一产品进行研究。另外,没有考虑传输系统,这也导致无法讨论即时响应的问题,而这会带来更好的改善效果。然而,在该研究领域中,缺乏对双电站间传输系统的文献讨论。不仅如此,双工厂之间输电系统的规划、分配或调度考虑因素也必须基于两个工厂的生产能力。规划必须同步协调,否则传导效果无法有效体现。因此,相比之前对单个工厂的传动系统的调查,双星工厂的考虑和调查会更加麻烦和复杂。关于能力支持与AMHS相结合的研究,学者们过去提出了许多概念和方法,例如Toba等人。
研究目的
在当今的大规模生产中,减少和提高吞吐量是 Gemini 工厂的首要任务。鉴于此,本研究将提出一系列方法来解决生产决策问题并评估Gemini工厂在肉类生产中的绩效。
研究方法
DCij:产品所需的生产能力 New:产品投入比例i。 PTij:产品 i 在加工机 j 上加工所需的时间 MCj:加工机 j 可以提供的生产能力。被送回原生产线等待处理,导致传输系统容量损失。对于产能的支持时间,以往的研究大多集中在动态决策上,但这可能会导致对处理系统产能的大量需求。
现阶段,本研究提出两个视角:在制品阈值(threshold)和在制品差值阈值(Difference)来管理和控制产能决策。设置在制品阈值的目的不仅是为了避免在制品过多而导致生产绩效下降,也是为了避免机器短缺。将此值设置得太低不仅会导致您的机器用完材料。在有产能支持的Gemini晶圆厂中,由于其到工作站的到达率会受到外部异常到达的干扰,因此对其性能进行评估。不能使用一般期望模型、均值分析(MVA)等性能评估模型进行计算,必须根据Gemini晶圆厂的特点进行修正。产能支撑模型实施后,会对原有生产系统发生以下两种干预:
結果與討論
Model to determine a general X-factor contribution and use to improve cycle time for wafer manufacturing. It was defined as the total average cycle time of the system divided by the total raw process time (RPT) of the production line. The basic X-factor presents the relationship between normalized cycle time and RPT of the overall system.
Finally, the cycle time of batch workstation is derived from the combination of processing time of a batch product with EWij(Q). To validate the performance difference of the cycle time estimation, a statistical analysis was performed by t-test. H0: The estimate of the cycle time between approximation and simulation does not differ significantly in machine group i (i.
H1: The cycle time estimation between approximation and simulation is significantly different in machine group (i. This statistical analysis also proved that the approximation model can provide an accurate cycle time estimation. At this stage, the improved effects of cycle time and cycle time deviation was validated by various improvement indicators.
At full system load, adding additional capacity will significantly reduce average cycle time and cycle time variance, except for a select group of high AXFC machines. In addition, a different perspective on cycle time improvement is also explored in this work. In order to reduce cycle time and reduce cycle time variation, the location of the machine group with high utilization should be ahead of the machine group with high AXFC in the capacity planning stage.
The authors would like to thank the National Science Council of the Republic of China for the financial support of this research under contract no. 1988) "The shifting bottleneck procedure for job shop scheduling", Management Science, Vol. 1999) "A Block-Based Cycle Time Estimation Algorithm for Wafer Factories", International Journal of Industrial Engineering, Vol. 2006) "Development of a Complete X-Factor Contribution Measurement to Improve Cycle Time and Cycle Time Variability", IEEE Semiconductor Manufacturing, Vol. 2005) 'Availability-adjusted X factor', International Journal of Production Research, Vol. 1995) "A dynamic forecasting model for job flow forecasting and tardiness control", International Journal of Production Research, Vol. 2002) “Optimal batching in a wafer manufacturing plant using a G/G/c multi-product batch processing model”, International Journal of Production Research, Vol. 2001) “Optimized Operations with Extended x-Factor Theory Including the Concept of Unit Clocks”, IEEE Transaction on Semiconductor Manufacture, Vol. 1984) Quantitative System Performance, Prentice-Hall, Englewood Cliffs, NJ. 2002) "An Experimental Study on Intake Scheduling and Bottlenecks for a Semiconductor Manufacturing Line", IIE Transaction, Vol. 1997) "How the Law of Unintended Consequences Can Overturn the Theory of Constraints: The Case of Balanced Capacity in a Semiconductor Manufacturing Line", 1997 IEEE/SEMI, p. Advantages of using short cycle manufacturing (SCM) instead of continuous flow manufacturing. (CFM)', IEEE/SEMI Advanced Semiconductor Manufacturing Conferences, pp Optimizing Furnace Batch Size and Pre-Clean Area Using Dynamic Simulations', IEEE/SEMI Advanced Semiconductor Manufacturing Conference, pp Using Factory Queuing Curve Approximation to Determine Productivity Improvement', conference IEEE/SEMI Advanced Semiconductor Manufacturing Conference, pp Effective modeling of factory flow times', IEEE/CPMT International Electronics Manufacturing Technology Symposium, pp Capacity planning for time-constrained batch-to-batch processes in wafer manufacturing', Global Business and International Management conference. 2006) "Effect of Arrival Balancing Between Batch and Batch Processes on System Performance", Asia-Pacific Conference on Industrial Engineering and Control Systems. 2005) "Effect of machine failure in a queuing system - a simulation study", APIEMS 2005 conference. Based on this concept, the expected waiting time was specified in the audit queue model, which Whitt (1993) modified from the approximation formula EW(M/M /m) into the GI/G/m model. The equations are shown as follows.
W., ―The Development of the Complete X-Factor Contribution Measurement for Improving Cycle Time and Cycle Time Variability,‖ IEEE Transactions on Semiconductor Manufacturing, Vol. This year's conference This year's conference combined with another international conference is called "The First International Conference on Uncertainty Theory".