The recent economic outlook has prompted manufacturers to spend a lot of resources towards automation and Cyber Physical Systems (CPS). One of the requisites to successfully deploy CPSs is the availability of up-to-date digital models coupled with the real system, yet this is not always guaranteed in dynamic and complex environments such as production systems. This paper develops a new method that generates the Petri Net model of a manufacturing system starting from an event log with three data labels. The user decides the number of maximum events to be mapped to control the model level of detail. The method has been applied on a test case and it is promising in terms of applicability to real manufacturing systems.