Petri Nets are widely used to model discrete event dynamic systems such as computer systems, manufacturing systems, communication systems, etc. A general analysis method using Petri nets is to compute the set of reachable states and deduce the different properties of the system. But when a Petri net contains a large number of tokens, the number of reachable states explodes and this is the major limitation of the application of Petri nets to complex system. Array process in TFT-LCD manufacturing system is such a system that is features of the capacity-oriented productions emphasizing high utilization of machines, and reduction of loss of capacity. Due to equipment uncertainty, product diversity, process intricacy, and ever-improving technologies, it is far more complicated than other conventional manufacturing process. Considering the limitation of the ordinary Petri nets, in this research, hybrid Petri nets, a hybrid model that combines fluid and discrete event dynamics, is taken as the basis for modeling the dynamic concurrent activities of a complex manufacturing systems, in particular, the array process in TFT-LCD production lines. The entire FOHPN model of array process is constructed from several sub-modules such as machine and buffer models in a bottom-up way. The net complexity is also analyzed in terms of the number of nodes in the net model. Due to the fluid approximation characteristics of FOHPNs, the considered system can also be efficiently simulated. To facilitate the simulation, a simulation framework has been designed and an algorithm based on a unified linear dynamic model is implemented as the core of the simulator. Extensive simulation have been conducted using different system operation parameters. According to some given performance indices such as maximal throughput and machine utilizations, one might choose the optimal working configurations of the system. Results show the applicability of the proposed methodology and demonstrate the modeling power of hybrid Petri nets. Furthermore, a hierarchical framework for monitoring and control of given system is proposed. Since the hybrid Petri net model of the system is constructed via continuous approximation of discrete event dynamic, which might results in a slight difference between the expected optimal behavior described by the FOHPN model and the actual behavior of the process observed. It is necessary to adjust the system operation parameters according to the feedback from the actual process behavior at the beginning of the macro-period by using such framework.
