We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

Institute of Advanced Study

Past Events

Tipping Points in Modelling Seminar - Robust common-cause failure modelling in power networks with non-immediate repair

12th March 2015, 13:00 to 14:00, Seminar Room, Institute of Advanced Study, Palace Green,, Dr Matthias Troffaes (Gent University)

After receiving his MSc degree in physical engineering in 2000 from Gent University, Belgium, Matthias Troffaes joined the SYSTeMS research group at the same university as a doctoral researcher, pursuing research in imprecise probability theory under the guidance of Gert de Cooman, earning the degree of PhD in April 2005. In July 2005, he went to Carnegie Mellon University as a Francqui Foundation Fellow of the Belgian American Educational Foundation, working as a post-doctoral researcher with Teddy Seidenfeld. In September 2006, he became a lecturer in statistics at the Department of Mathematical Sciences, Durham University.

Robust common-cause failure modelling in power networks with non-immediate repair

Power networks are commonly modelled using continuous time Markov chains. Although they are analytically attractive, in practice, the assumptions underlying these models are not necessarily representative of real power networks observed in practice. Moreover, the parameters of the model can be hard to estimate from data. In this talk, we present a model based on so-called imprecise continuous time Markov chains, in which we relax the assumption of transition rates being precise and constant, instead allowing the rates to vary between bounds in a time-inhomogeneous way. For estimation of these bounds, we use a generalisation of the standard alpha-factor model for common cause failures that can deal with asymmetry, in order to apply the model to power networks, which are typically asymmetric. We show how practical bounds on the probability of various failure events can be calculated. Finally, we demonstrate our methodology on a simple yet realistic example.

Contact for more information about this event.