Researchers at the Division of Human-Centered Systems condact research on complex safety-critical systems where resilient behaviour is desired. Lack of resilience in such systems is also one of the new views of accident causation in our research on accident investigations.
Resilience Engineering is a new approach to safety and risk management. Whereas conventional approaches to system safety are dominated by hindsight and emphasise error tabulation and probabilistic risk analysis, Resilience Engineering emphasizes an organisation's ability to adjust its functioning, prior to or following changes and disturbances, so that it can sustain operations even after a major mishap or in the presence of continuous stress. Resilience has previously proven to be a useful construct in analyzing the persistence, stability and flexibility of ecological systems, for many decades. Applied to engineering of safe systems, the focus is on the following three abilities:
Accident investigations both regard explaining what happened, and proposing remedial actions. When an accident has happened, the investigation often manages to clarify some of the causes, however, the remedial actions should prevent a broader ranged of future accidents than identical negative events. A complicating factor is that usually, the more serious the accident is, the less important the triggering event is. Serious accidents often have many latent conditions, creating a dangerous environment, formed over a long time. These factors may combine with a rather innocent act or event, to trigger the events leading to an accident. Over the years, many different accident models have been used, to describe general principles of causation and defence, for incidents, accidents, and disasters. The models have been developed to explain causation in accidents in systems with widely differing characteristics. From systems of one machine operated by one operator, to nuclear power plants with complex connections and dependencies between parts and actors. Our research in this area focuses on how effective remedial actions can be designed, and how that work is affected by the explicit or implicit accident models of the designers and analysts.