Wafer fabrication scheduling subject to resource constraints

Abstract

This dissertation focuses on scheduling problems in a semiconductor wafer fabrication (fab) which produces semiconductor products of a large number of different product types in a low-volume and high-variety setting. In this fab, it is not easy to achieve both meeting due dates of orders and increasing production efficiency because the production environment is very complicated due to various resource constraints. Therefore, it is required to develop new scheduling methodologies that consider those constraints effectively. In this dissertation, we develop scheduling algorithms for the production subsystems with the resource constraints in the fab, and then, we develop a production scheduling methodologies for the entire fab based on the dispatching rule-based approaches. First, we consider a two-machine flowshop scheduling problem with limited waiting time constraints with the objective of minimizing makespan. By the limited waiting time constraints, the second operation of each job should be started within a certain time period after the first operation of the job is completed. We develop optimal solution properties, upper bounds and lower bounds, and suggest a branch and bound algorithm that is developed using the properties and bounds. Secondly, two scheduling problems with distinct resource constraints are considered, respectively. We first consider a batch-processing machine scheduling problem with a product-mix ratio constraint with the objective of minimizing makespan. By the product-mix ratio constraint, jobs of multiple product types can be processed simultaneously as a batch according to the predetermined product-mix ratio. We develop an optimal solution algorithm which gives solutions in polynomial time. Then, we consider a batch-processing machine scheduling problem with resource constraints with the objective of minimizing total tardiness. By the resource constraints, a setup operation is incurred to change the resource tank with a new one whenever...