Modern systems involve more and more complex interactions leading to very large models. In the area of Stochastic Petri Nets, standard approaches are to use High Level Stochastic Petri Nets and/or some kind of compositionality to cope with this increasing complexity. In this paper we present an experimental implementation of the asynchronous decomposition method for Stochastic Well formed Nets (SWN), a class of Stochastic High level Petri nets. The method combines the Multi-valued Decision Diagram methods for structured Markov chains with the theoretical results for decomposable SWN. This implementation allows us to compute performance indices for very large and very symmetric systems. We apply our tool to the analysis of a complex Manufacturing System.