Markov decision process for multi-manned mixed-model assembly lines with walking workers

Product customization and frequent market changes force manufacturing companies to employ mixed-model instead of simple assembly lines. To well adjust the line’s capacity production requirements, line can benefit from concept reconfigurability. Our study deals with a reconfigurable where tasks be dynamically assigned stations at each takt, workers move among end order entering product models is infinite unknown. The equipment assignment occurs design stage, duplication allowed. dynamic task workers’ movements Markov Decision Process (MDP) that translated as Linear Program (LP). As result, problem formulated Mixed-Integer (MILP) integrates MDP model. We propose some reduction rules decomposed transition process reduce new MILP taking into account stochastic parameters are built solve robust problems, objectives expected total cost minimization in all takts worst respectively. Computational experiments benchmark generated instances demonstrate performance proposed models. managerial insights provided paper show superiority over model-dependent fixed assignments usually studied literature.