| Summary: | 碩士 === 國立交通大學 === 工業工程與管理系所 === 95 === As the semiconductor process technology progress, the inevitable trend of IC product developing is toward mini-size. Hence, in order to meet new manufacturing technology and the precise alignment, wafers must be processed on the machines that satisfy process specification. As a result, the process window and machine dedication restrictions have appeared within photo and etching workstation. The different machines capabilities will probably affect the variance of product cycle time.
In this thesis, we first develop two mix-integer programming model to allocate capacity loading between machine groups of photo and etching workstation for minimizing the variability of product cycle time. Then, the utilization rate of each machine group within photo and etching workstations is derived. Next, we also develop a cycle time estimation methodology to estimate the cycle time of each product considering different utilization rates of machine groups existed in photo and etching. Furthermore, in the controlling maximum utilization difference module, a value is set as the maximum utilization difference between machine groups for photo and etching workstations to avoid the product cycle time being lengthened due to the too large machine utilization and hence lose industry competitiveness. In this module, we also develop production plan adjustment models for the too low average utilization rate and the too long product cycle time scenarios to revise the production target. Finally, we proposed a material release planning module to balance the machine groups utilization within a workstations by setting a fixed a release sequence,adopting CONWIP release rule and developing dispatching rule. Following the above procedure, a reasonable master production schedule under the constraint of process window and machine dedication is obtained.
Experiment reveals that comparison the cycle time, estimated by the block based cycle time estimation methodology under various utilization, with that of simulation result, the maximum error is within 8% and the average error is about 5%. Besides, by controlling of maximum utilization difference, production plan adjusting models can generate suitable MPS quickly. Also, based on the material release planning module, the utilization rate of photo and etching workstations can be maintained at planned level and the system production performance can be achieved. Consequently, the mechanism proposed in this thesis can be used as reference for master production planning under the constraint of process window and machine dedication.
|
|---|
