Quantitative analysis of imprecise system models with configurable fault trees
 
Authors: Christine Jakobs, Peter Tröger, Matthias Werner

Fault trees are a well-known and widely used method for dependability predictions in engineering sciences. They enable a structured analysis of error propagation chains, which is especially important in safety critical systems.  
Classical fault trees are based on complete knowledge about the modelled system. This kind of information is available only in late phases of the development process. Before that, the system architecture is defined incompletely, what makes the usage of fault trees among a lot of other methods impossible. 
In this talk, we discuss the concept of configurable fault trees by Tröger et al. that addresses the problem with new event and gate types. We show a new mathematical representation of such fault trees in closed structural equations. This enables the quantitative analysis of large imprecise models. The talk discusses the practical relevance of the approach for new types of uncertainty importance measures.