Method of dynamically determining cycle time of a working stage

Abstract

This paper presents a method of predicting production cycle time of a working stage in semiconductor manufacturing. For Scheduling, cycle time of a stage is usually applied to estimate the cycle time of the entire production line or the remaining production time. It is also often applied to generate the reasonable inventory level that should be queued in front of a stage. Since cycle time is highly affected by equipment loading density, a model is first created by combining Little's formula and Kingman's equation in the queuing theory to relate loading density with it. Cycle time distribution of each stage over the past time is characterized as a parameter set, based on the cycle time model and using a plurality of prior associated data. The equipment loading density of a stage in the next production run is calculated by considering the demanded quantity, mix and process flows of products. The stage cycle time of the next production run is then predicted by applying the equipment loading density to the cycle time model. The optimum amount of inventory kept in front of a stage to prevent equipment staving or WIP (work in process) piling up is also determined by using the data of cycle time and production target of a stage. The cycle time and the optimum inventory level (which is called standard WIP) of a stage are dynamically tuned as the customers' demands vary. With this, cycle time and standard WIP could be more precisely defined and the delivery would be well controlled.