Today's industrial companies increasingly face challenges to adjust capacity to meet varying demand while maintaining cost-efficiency of production. These challenges are typical for make-to-order (MTO) assembly production of customized products. Efficient production of variable, large MTO products often requires parallel, independent assembly stations. A feature that is typical of such production is labour-intensiveness, and thus the role of workers is highlighted. Workers have to both learn new tasks and, if necessary, move to where the current workload is. The dynamic management of production requires simple enough worker coordination policies (WCPs) which allocate workers to tasks, and tasks to workers, in real time. Alongside this, one should consider the resulting labour productivity when multiple workers share the same task. This dissertation investigates the MTO assembly production described above. The first objective is to find out how different WCPs affect the performance of parallel station assembly systems. The WCPs that are studied are based on workers helping each other, which is a novel approach to workforce coordination. The second objective is to find out how worker collaboration practices (the number of workers and ways of working) and learning affect the performance on a single assembly product. To this end, experimental studies were conducted with a case product that has, more than previous studies, elements similar to industrial assembly products. The research methods used in the five articles comprise laboratory experiments, video analysis, Markov models, and simulation. The results from the laboratory experiments showed that, for novice workers, instructions are crucial to learning new tasks. Productivity per worker decreased as the number of workers on the product increased. However, the significance of the number of workers decreased through repetitions of assembly. Assistance by another worker was seen as beneficial, especially with large parts, and hence two workers per product were considered most appropriate. A larger number of workers resulted in permanent productivity losses as a result of difficulties associated with the spatial and temporal coordination of the workers. The results of the simulation study of a parallel station system showed that with stochastic demand and manufacturing conditions, workers helping each other in fixed pairs is effective enough compared to a policy in which everyone can help everyone. In fixed pairs, workers also learn each other's ways of working and may become very productive. However, attention should be paid to the selection of pairs of workers. Arbitrary selection may lead to major differences in performance between different pairs, which, on average, weakens the reactivity of production. Greater flexibility in helping compensates for variation resulting from differences in skills but highlights efficiency in collaboration on a shared task. When making decisions on worker allocation, workers' opinions and preferences and other worker-related factors should also be considered.
|Publication status||Published - 2018|
|MoE publication type||G5 Doctoral dissertation (article)|
- manual assembly, make-to-order, worker coordination, collaboration, learning