Safety Systems research studies failure in a new breed of highly complex systems built by man. Existing scientific and engineering techniques have been unable to describe or control these phenomena.
For example, it was previously thought to be the case that non-trivial software systems were untestable, in the sense that testing could not provide meaningful quantitative assurance of software function. We have demonstrated that this not the case for a significant class of software, and have applied new statistical reliability testing techniques to critical nuclear systems, including reactor protection systems.
Socio-technical systems also exhibit complex systematic failure behaviour, the result of which can be seen in repeated large industry accidents (Bunsfield, Macondo, Hatfield, Piper Alpha, etc.) The UK National Security Strategy classes these accidents as 'Tier 1' risks to the UK, alongside terrorist and cyber attacks. Safety Systems is developing new ideas for the control of dangerous accident precursors in large organisations.
Within the University of Bristol, Safety Systems research is part of the:
Research focuses on a broad appreciation of safety issues across various industry sectors such as:
These safety issues are tackled with novel holistic methods for satisfying real industry dependability requirements.
Thames barrier, London, UK