This paper studies multistage serial manufacturing systems with the integrated consideration of machine failures, process defects, multiple rework loops, etc. In particular, multiple rework loops and product polymorphism lead to a more complex conversion of internal material flows, and therefore it's difficult to model and analyse such manufacturing systems. A modular modeling method based on Generalized Stochastic Petri Nets (GSPN) is presented to characterize the material flows, it is capable of representing the processing differences resulting from product polymorphism comparing with traditional Markov model or Queuing network model. By analysing the model, the processing ratio of each workstation is inferred. Using 2M1B (two-machine and one-buffer) Markov cell model as the building blocks, which is obtained based on the GSPN models for their isomorphism, an overlapping decomposition method is then developed for evaluating the performance of the multistage serial systems with rework loops. Numerical experiments and a case study of a powertrain assembly line illustrate the efficiency of the proposed method.
展开▼
