Optimization of buffer allocation in mixed production systems with multiple classes, variable batch sizes and worker fatigue considerations

PurposeWith rapid market customized demand and short development cycles, mixed production with multiple classes and variable batches has been popular, and its buffer allocation problem has become a new challenge. The mixed production cannot be analyzed based on the assumption of a stationary demand process which was typically used in previous studies. Furthermore, mixed production is still in human-machine cooperation mode where dynamic working efficiency because of workers' fatigue causes uncertain processes. Therefore, the purpose of this study is to solve the buffer allocation problem in mixed production systems with multiple classes, variable batch sizes and worker fatigue considerations.Design/methodology/approachA dynamic modeling method of mixed production with multiple classes and variable batches is improved, which uses nonstationary demand processes to model the dynamic nature of multiple classes and variable batches. Human working efficiency decreasing due to fatigue is modeled as the time-varying service rate to represent human-machine cooperation. Furthermore, a parallel evolutionary algorithm that combines global and local search strategies parallelly is developed to solve the buffer allocation problem in mixed production for the first time.FindingsNumerical examples demonstrate the efficacy of the proposed algorithm. The proposed algorithm achieves better solution quality than the state of art algorithms.Originality/valueThis study improves the dynamic modeling of mixed production to consider human factors and develops a hybrid algorithm to effectively solve the buffer allocation problem in dynamic mixed production.