Abstract
This study focuses on a lot sizing and scheduling problem in the RGB process which is part of the TFT color filter manufacturing processes. The RGB process produces multiple product types, and there are multiple parallel machines. Also, sequence-dependent setup time is required between different operations, and all product types have reentrant flows. We develop a two-stage method to solve this difficult scheduling problem. In the first stage, we use EOQ model to determine lot size for every product and modify the lot size if the lot size is infeasible. In the second stage, we solve the scheduling problem by ATCS (Apparent Tardiness Cost with Setups) rule. Whenever one machine becomes free, we compute the priority index for the set of unscheduled jobs, if any. The job with the largest index is selected. If more than one machine is free at same time, we choose one arbitrarily. Experimental designs are carried out to test the performance of twelve two-stage methods which are combined with two lot sizing methods and six dispatching rules. Results reveal that the two-stage method EOQ+ATCS is easy to use and not worse than the other two-stage methods in terms of the total tardiness of jobs. Keywords: color filter, parallel machines, sequence-dependent setup time, reentrance, lot sizing and scheduling