Two-Stage Stochastic Programming for a Reconfigurable Unrelated Parallel Machine Scheduling Problem

Production managers are faced with uncertainties like machine failures, demand uncertainty, and workers' availability. Reconfigurable Manufacturing Systems (RMSs) provide a flexible solution to these challenges. Considering worker availability, this paper presents a two-stage stochastic programming model for optimizing unrelated parallel machine scheduling in the RMS. The model adopts a resource flow-based framework to ensure that workers with different skills and availabilities are allocated efficiently throughout job sequences. The objective of the mathematical model is to minimize the expected makespan across multiple scenarios. Eighteen experiments were conducted to evaluate the model, in which key parameters were varied at various levels, including worker's availability, reconfiguration time, and processing time. The Analysis of Variance (ANOVA) reveals that processing time had the most significant impact on the makespan, followed by worker's availability and reconfiguration time. The results demonstrate the model's ability to generate flexible schedules under uncertainty, offering valuable insights for enhancing the adaptability and efficiency of RMS operations.