Scheduling Mass Customized Large Product Assembly Line Considering Learning Effect and Shifting Bottleneck
MetadataShow full item record
The large product assembly industry is a complex and heavily manually based assembly manufacturing process. Examples are commercial airplanes, ships, and wind turbines. These manufacturing enterprises are increasingly striving to satisfy the individual needs of customers by providing customized products. Under this circumstance, production completion times are not easy to predict precisely and likely to be variable. This leads to problematic scheduling decisions since scheduling precision depends on reliable job processing time projection. Mass customization has been a contemporary manufacturing industry management technique, enabling a company to compete with their competitors. It is described as the ability to provide individually designed products and services to satisfy every customer through high process agility, flexibility and integration with acceptable production cost (Davis, 1989). However, implementing the concept of mass customization into the production system is not free and increases both the complexity and the difficulty of the production process compared to mass production. Thanks to the characteristic of learning effect, which is a phenomenon that people improve operational efficiency when they do the same task repeatedly (Wright, 1936), mass production can simply reach steady production and operational efficiency by scheduling a series of standardized process consecutively. However, it is a challenging task to improve the operational performance in mass customization by utilizing learning effect due to a large variety of different jobs. In order to achieve mass customization at near mass production efficiency, eliminating the bottleneck is one of manufacturing tactics to improve production efficiency. Most manufacturing systems' performance is constrained by one or more bottlenecks and the critical bottleneck may shift from one work station to another. Therefore, this dissertation addressed the linking of learning effect consideration in a bottleneck selection scheduling heuristic for the mass customized large scale product industry to improve production performance measures such as the completion time, throughput, or due date assignment. A systematic scheduling heuristic was developed to allocate the same task consecutively to reduce processing time on the critical bottleneck stage and further improve the whole production performance. The scheduling heuristic is based on the classic job shop scheduling problem, the Shifting Bottleneck Procedure, and learning curve consideration. The objective of the sought scheduling heuristics is to provide manufacturers better scheduling decision support for not only satisfying their customer's demand but also improving the performance of the production system. Through the presentation of a simulation model carried out in a wind turbine assembly line, the impact of the proposed scheduling heuristic to the large product assembly performance under mass customization is evaluated and validated.