Abstract

The increasing scale and complexity of safety critical systems pose significant challenges in the safety assessment of such systems which becomes increasingly more expensive, error prone and difficult to complete. To address these challenges, the past fifteen years research has focused on automating the synthesis of predictive models of system failure from design representations. In one approach known as compositional safety analysis, system failure models such as fault trees and Failure Modes and Effects Analyses (FMEAs) are being constructed from the topology of a system and component failure models using a process of composition.

In this seminar, I focus on Hierarchically Performed Hazard Origin and Propagation Studies (HiP-HOPS) – one of the more advanced and well supported compositional safety analysis techniques. I discuss the principles that underpin this technique and show how these principles have been effectively used to deliver a number of scientific and practical contributions which include:

- Fast algorithms for automatic synthesis of Fault Trees and multiple failure mode FME As from design representations of a system.

- An extension to the Fault Tree notation and Fault Tree Analysis(FTA) technique that enable assessment of the effects of sequences of faults.

- A language for specification of inheritable and reusable component failure patterns.

- Pareto-envelope genetic algorithms that enable automatic optimisation of system models in respect of dependability (i.e. safety, reliability, availability) and cost via application of automatic model transformations.

I summarise those contributions and demonstrate application of HiP-HOPS on small examples discussing the strengths and limitations of the method. HiP-HOPS offers less automation that formal safety analyses techniques, e.g. those using model-checking. However, it is computationally less expensive, not prone to combinatorial explosion, scales up and can be easily iterated throughout the design lifecycle. For the same reason, it enables exhaustive assessment of combinations of failures and design optimisation using computationally greedy meta-heuristics. I outline current work towards harmonisation and integration of HiP-HOPS with EAST ADL and AADL two architecture description languages emerging in the Avionics and Automotive industries, and conclude by pointing to future work.