Batch scheduling problems for two-stage assembly shop

• Main Authors: CHUNG-HUA LIN, 林宗華
• Other Authors: Ching-Jong Liao
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/35601029398409794068

博士 === 國立臺灣科技大學 === 工業管理系 === 100 === In this thesis, we consider the batch scheduling problems for two-stage assembly shop, which are frequently encountered in the machinery manufacturing. In the first stage, a fixed number of jobs are assembled on a batch machine with a common processing time. Then the assembled jobs are moved to the second stage to perform system integration with different processing times on a discrete machine. In this thesis, two cases are considered: (1) Batch scheduling for an assembly line, and (2) Batch scheduling for assembly islands. In Case (1), all jobs in a machinery factory share a common assembly line. A constant batch setup time is needed to form a batch at the first stage and fixed family setup times are required on both stages when starting the processing of a new family or switching to a job in a different family. The objective is to find a schedule that minimizes the weighted sum of makespan, total completion time and total tardiness. In Case (2), a machinery factory is equipped with fixed-position assembly islands where each job must remain in an assembly island during its entire processing. A family batch setup time is needed to form a batch at the first stage and a fixed removal time is incurred when a finished job is moved from an assembly island to a temporary storage area if the payment has not been made. The objective is to minimize the total completion time. Since the two considered problems have been proved to be NP-hard, we develop Mixed Integer Programming (MIP) models to find the optimal solutions for small-size problem and propose some heuristic methods for each large-size problem. We also conduct computational experiments to compare MIP models with the corresponding heuristic methods. The computational results show that the proposed methods (FBFS and P-FBFS) outperform the current scheduling methods and can be applied to resolve real-world problems similar to those in this thesis.