| Automobile assembly lines originally were designed for repetitive production of a single product, but they are now used to assemble a variety of models simultaneously. The models differ in their processing requirements, and the model mix changes daily. This creates an enormous challenge in line balancing--deciding which tasks to assign to each station on the line, which in turn affects the processing time of each job at each station--and in determining the daily job sequence. Historically, line balancing procedures have been concerned only with equalizing the workload across stations, but in a mixed-model context, this may not guarantee good short-term performance. Instead, it is also important to aim for uniformity of processing times of the various models within each station, and uniformity of processing times across stations for a given model. With this greater uniformity, it is easier to construct job sequences that provide a smooth flow of work to each station, which reduces labor requirements and helps to improve product quality.;This dissertation focuses on developing a line balancing methodology to facilitate the construction of good job sequences. In Chapter 2, we survey the literature on job sequencing for mixed-model assembly lines. We also perform an exploratory study which indicates that simple statistical characterizations of the processing times are not enough to predict short-term performance. This motivated our development of a new metric.;In Chapter 3, we survey the literature on assembly line balancing. We then propose a new line balancing metric which considers traditional measures of workload balance, but also incorporates the uniformity concepts described earlier. A heuristic procedure is developed to solve the problem for this new objective.;Chapter 4 reports computational results which show that using the proposed objective as the performance metric leads to line balancing solutions that not only facilitate the construction of good job sequences, but also make the system more robust to daily changes in the model mix. In Chapters 5 and 6, we outline other approaches that can be used to mitigate problems resulting from product diversity, and discuss promising areas for future research on mixed-model assembly lines. |
