Designing and organizing a high-performance divisional seru considering learning and forgetting effect under stochastic processing requirements

Seru production systems (SPSs), known for their flexibility and responsiveness, have drawn considerable attention from both academics and practitioners. Owing to the costs associated with worker training, divisional serus, which require only partially cross-trained workers, are the most fundamental and widely adopted organizational forms. As a human-centered system, characterizing a divisional seru necessitates accounting for human factors to avoid discrepancies between model evaluation and realistic performance. This study considers a divisional seru with learning and forgetting effect and stochastic processing requirements. We try to explore the impact of learning and forgetting effect on seru performance in various production situations as well as its interplay with the role of work-in-process (WIP) buffers. By establishing the underlying Markovian model, three performance measures are evaluated, including throughput, flow time and WIP on the seru. Numerical experiments are conducted by varying several parameters, such as workers’ initial processing rates, the number and locations of overlapped stations, and the number of workers. The results indicate that learning can enhance system performance in terms of throughput and flow time without causing a substantial increase in WIP level. Moreover, learning effect can compensate for imperfect initial seru settings and expedite the stabilization of performance measures. Buffer arrangement should be adapted to the specific production environment and prioritized operational objectives. Furthermore, our findings reveal that buffer usage and learning effect interact synergistically to improve seru efficiency. A series of practical guidelines are provided to help design and organize a high-performance divisional seru.