A Fault-Tree Semantics to Model Software-Controlled Systems

Fault Tree Analysis is a very popular tech-nique to assess safety and reliability of technical sys-tems. However, being a combinatorial model, Fault Trees can only express which combinations of failures contribute to a certain hazard or accident. There is no means to model sequences of actions and temporal or-ders of states and events. Since today technical systems are often controlled by software that executes over time, the traditional Fault Tree model is no longer sufficient. Moreover, integration of software modelling techniques with safety assessment techniques is hampered by the lack of semantically equivalent entities in both do-mains. To overcome these drawbacks, we propose an extended Fault Tree semantics that distinguishes events that happen at a point in time from states that last over a period of time. Typed Fault Tree gates are introduced and calculation rules for quantitative analysis are given. We address the integration of this new concep