Performance Evaluation of Algorithms for Flowshop Scheduling in Semiconductor Manufacturing

Chan Geok Cheng

Abstract

Production scheduling is a key process in manufacturing industry. The purpose of production planning is to ensure a smooth flow of supply chain to meet customer delivery date. A good planning will help in boosting company productivity while reduce unnecessary wastage. In this thesis, the research work will surround on production flowshop model in local semiconductor wafer probe card manufacturing. This report outlines different scheduling methods such as optimization approach, heuristic approach and scheduling priority rules for solving n jobs m machines problems. The algorithms are implemented with Mixed-Integer Programming (MIP) model and Python which incorporated with Pyomo and NumPy package. Small, medium and large-scale problems are evaluated with methodologies applied to study the efficiency in achieving minimal makespan with high accessibility. The findings obtained during medium-scale n jobs m machines performance evaluation shows that the percentage error between Minimize Machine Idle Time (MMIT) method and Optimization model is only 4.9%. Computation time for heuristic model is 2546.32% faster than optimal model. Besides, optimal model has limitation when come to large-scale data, yet heuristic model could compute solution within 2 seconds. By considering accessibility and effectiveness, MMIT heuristic model is proposed to be integrated in existing production planning process.