Marielle Stoelinga: Risk management meets model checking: compositional fault tree analysis via Markov automata
Time: Fri 2016-03-11 13.15
Location: Room 4523, Lindstedtsvägen 5, KTH CSC
Participating: Marielle Stoelinga (Associate Professor of Risk Management for ICT University of Twente)
Abstract
How do we ensure that our railroad, nuclear power plants and medical devices are safe and reliable? That is the topic of risk analysis, and fault tree analysis is a very popular technique here, deployed by many institutions like NASA, ESA, Honeywell, Airbus, the FDA, Toyota, Shell etc.
In this presentation, I will elaborate how the deployment of stochastic model checking can improve the capabilities of fault tree analysis, making them more powerful, flexible and efficient: I will present a compositional framework, where we can analyze a richer variety of questions via stochastic model checking of Markov automata; I will show how we obtain more compact models via bisimulation and graph rewriting techniques, and get more flexibility in the modeling power.
Finally, I will show how one can incorporate maintenance strategies, a crucial aspect in reliability engineering, reporting on our experience with the application and validation of these techniques in industrial practice; in particular in the railroad and nuclear domain.