Matthias Volk: Dynamic Fault Trees: Semantics, Analysis and Applications

 

Matthias Volk’s dissertation project focused on the reliability analysis of safety-critical systems. He considered dynamic fault trees (DFTs), Dugan’s extension of the well-known static fault trees. Whereas DFTs exist since the early 1990s, their expressiveness leaves room for various possible interpretations. Volk (jointly with Junges) defined a formal semantics of DFTs using generalised stochastic Petri nets. This semantics is parametric and covers all different existing DFT interpretations in the literature so far. Based on this semantics, Volk has developed an efficient state-space generation algorithm for DFTs yielding continuoustime Markov chains. The crux of his algorithm is to use partial-order reduction and symmetry reduction techniques combined with modularisation and ignoring DFT fragments that are irrelevant. As DFTs can be huge in practice, Volk developed a lazy analysis techniques that obtains upper- and lower bounds on reliability metrics by analysing partial state spaces. Experimental results on an industrial autonomous driving setting as well as on the analysis of the criticality of infrastructural elements in railway stations (joint work with UNRAVEL doctoral researcher Weik) showed the effectiveness of these analysis techniques.