In this paper we show how First-Order Hybrid Petri nets, an hybrid positive model that combines fluid and discrete event dynamics, may be efficiently used to simulate the dynamic concurrent activities of manufacturing systems. In particular we deal with the performance analysis via simulation of a mineral water bottling plant according to the variations of the production controlling input parameters. The model allows a simulation of the productive line behavior through changes in the production capacity of the producing bottles and PET prototype machines, of the filling machines, of the volume and type of the bottles, of the silos dimensions, and so on.