A factory-level dynamic operator allocation policy: the bubble allocation

In this paper, we propose a factory-level dynamic operator allocation policy called the bubble allocation policy. This policy is commonly implemented in labour-intensive industries in the presence of different operator speeds, high labour turnover, and learning effects. We prove the optimality of bubble allocation in several typical scenarios under deterministic and exponential processing time and different operator speed assumptions. When labour turnover and learning effects were considered, the effects of the bubble allocation were verified through simulation. Bubble allocation had a more significant positive effect on system throughput than the passive operator allocation policy. The positive effects of bubble allocation are enhanced with a larger production system scale (more parallel lines and more stations), higher turnover rate and slower learning process. Compared with active allocation policies such as work-sharing policy, bubble allocation policy has no requirements for additional cross-training and is not sensitive to the switching time of tasks. The bubble allocation policy stands out when the system is large and the flow line is designed in a balanced way.