Abstract
TimeNET is a software package for modeling and performance evaluation with non-Markovian Petri nets. Concepts and implementation of the simulation component of this tool are introduced. The paper focuses on a reliable statistical analysis and the application of variance reduction techniques in a parallel, distributed simulation framework. It examines the application of variance reduction with control variates and shows an approach for their automatic selection in Petri net models. An example demonstrates the gain of variance reduction and parallelization.<>