The paper tackles the problem of managing uncertainties during the execution of predictive schedules in a dynamic environment. The dynamic environment in question is represented by a simulation model - connected to a production scheduler system - with flexible modelling capabilities. The paper addresses the simulation module of the proposed architecture highlighting its main functionalities and advantages, compared to former simulation-based solutions. By applying the proposed architecture (using both the proactive and reactive operation modes), the solution methods for stability-oriented rescheduling can be profoundly tested and analyzed. The evaluation of selected scenarios of the timing of rescheduling, rescheduling threshold and human resource management techniques are presented through an industrial case-study.