A model-based approach for the collection of Human Reliability data
Massaiu, S.ESREL 2011 Conference, Troyes, France, 2011-09-18--09-22. Advances in safety, reliability and risk management / Bérenguer, Grall & Guedes Soares (eds.). CRC Press, 2011 - ISBN 978-0-415-68379-1
- Publ. year
- Publ. type
- One of the major criticisms against Human Reliability Analysis (HRA) is the lack of empirical support for its quantitative estimations. While there is increased confidence around the quantification of executions failure probabilities, there is still a great deal of uncertainty when it comes to less well-defined situations and higher-level activities, like diagnosis and decision errors: that is, the conditions that characterize the history of industrial accidents. This paper proposes a new approach for empirically supporting quantification in these conditions: The idea is to use a domain-specific modeling approach to convey empirical evidence into the expert-judgment processes of HRA. The model is the Guidance-Expertise Model (GEM) of crew cognitive control, which provides the classification system for collection and retrieval of domain-specific data. In this paper we present a test of the ability of the model to identify regularities between environmental conditions (procedures), crew expertise (teamwork) and crew behaviors. The analysis of four nuclear power plant control room crews simulated emergency scenarios shows that the approach is capable of retrospectively identifying crew behaviors of interest for HRA (e.g., unexpected progressions in the procedures set, responses to cues and unexpected events) and relating these to observable performance conditions. The identification of such environment-cognition-behavior regularities can empirically inform assumptions made in predictive analysis of scenario evolutions and of system failure.