In this paper, we propose an autonomous decentralized scheduling method with an optimization function of lotsizing and scheduling for multi-stage aluminum production line. The proposed method repeats the generation of schedule at each production process and data exchange among scheduling subsystems until a feasible schedule for the entire plant is derived. Two types of lotsizing and scheduling methods are compared for various size of example problems. The results demonstrate that the simultaneous optimization of lotsizing and scheduling must be considered when intermediate storage cost is embedded in the objective function.