Seminar - Stuart Reid - Risk acceptance criteria based on uncertain estimates of structural failure probabilities

Wednesday 6 April 2011, 4.00 pm - 5.00 pm
Civil Engineering Conference Room

Associate Professor Stuart G Reid
School of Civil Engineering
University of Sydney

Abstract:
Calculated probabilities of structural failures commonly have significant uncertainties associated with the possible estimation errors relative to the 'true' failure probabilities. The 'true' failure probabilities may be defined as the failure probabilities that arise from the real (aleatory) variability of structural loads and resistances, whilst possible estimation errors may arise from epistemic uncertainties associated with imperfect knowledge and imperfect understanding of the load and resistance mechanisms. For uncertain probabilities of failure a measure of 'probabilistic confidence' has been proposed to reflect the concern that uncertainty about the true probability of failure could result in a structure that is unsafe and could subsequently fail. The talk describes how the concept of probabilistic confidence can be applied to evaluate and appropriately limit the probabilities of failure attributable to particular uncertainties such as design errors that may critically affect the dependability of risk-acceptance decisions. This approach is illustrated with regard to the dependability of structural design processes based on prototype testing with uncertainties attributable to sampling variability.


Biography
Associate Professor Stuart Reid graduated with BE and ME from the University of Canterbury, NZ, and then worked as a structural engineer for about 4 years, based in Wellington, NZ. He then went to McGill University in Canada, where he obtained a PhD for work on probabilistic structural design. He subsequently worked for the CSIRO (Building Research Division) for 3 years as leader of a risk analysis project, before moving to a position at the University of Sydney. Professor Reid’s research interests include structural reliability modeling, risk acceptance criteria and the development of risk-based design standards.