Professor Mark G. Stewart
Director, Centre of Infrastructure Performance and Reliability. The University of Newcastle. New South Wales 2308, Australia.
Probability Neglect, Risk and Public Policy for Extreme Events: What Killer Figs, Terrorism and Climate Change Have in Common?
Terrorism and climate change debates are often characterized by worst-case thinking, probability neglect, risk aversion, and avoidance of the notion of acceptable risk. This is not unexpected when dealing with extreme events. Another case study is a decision by Newcastle council to remove 16 fig trees from an iconic street because the risks were too high – labelled in jest as the “killer figs” by a local newspaper. Yet the trees had survived many storms and were undamaged. However, an arborist consultant calculated that the risk of death was 1 in 19.8 per tree per year – so walking down that street was 10 times more dangerous than downhill skiing! A common denominator is the seemingly inability of public officials to grasp the notion of probabilities. The presentation will describe how decision metrics often involve probabilistic thinking by public officials, and how best to present this type of information in a manner more easily grasped by these decision-makers. The concepts will be illustrated with current public policy research related to (i) killer figs in Newcastle, (ii) aviation security measures, and (ii) climate adaptation strategies for infrastructure.
Professor Michael Beer
Director, Institute for Risk and Reliability, Leibniz Universität Hannover, Germany.
Uncertainty Quantification under Vague Conditions
Engineering infrastructure, formally built by structures and systems, is characterized by a rapid growth in scale, complexity and interconnection, so that uncertainties and risks are involved to a greater extent than ever before. These structures and systems are, to a significant extent, critical for the functionality of our economic and societal life, and thus, require proper approaches and measures to verify and ensure their reliable performance. Reliability and performance analysis, however, become increasingly complicated due to uncertainties and complexity. The realistic quantification of uncertainties and their numerically efficient processing in complex analyses are the two key challenges in this context. This keynote lecture addresses these two challenges. Approaches to deal with epistemic uncertainties are discussed, with focus on imprecise probabilities and in the context of structural and system reliability assessment. The concept of survival signature is presented to analyse complex systems efficiently. Novel pathways to capture interdependencies between systems while estimating their reliability efficiently in a time dependent manner are discussed. Engineering examples are presented to demonstrate the capabilities of the approaches and concepts.
Professor Keith Worden
Professor of Mechanical Engineering, The University of Sheffield, Sheffield, UK.
Probabilistic Reasoning in Structural Health Monitoring
In the early stages of Structural Health Monitoring (SHM), the most common approaches were based on physics-based modelling, and explicit probabilistic reasoning was the exception rather than the rule. With the advent of data-based SHM, following recognition that damage detection could be framed as a pattern recognition problem, the transition to probabilistic methods in the machine learning community followed in the SHM community. This presentation will discuss some of the history of the subject (from a personal viewpoint), highlight some applications of state-of-the-art machine learning, and outline how SHM decision problems can be frame in terms of Probabilistic Risk Analysis.
Professor Alan O’Connor
Professor in Civil Engineering at Trinity College Dublin, Ireland.
Risk, Robustness and Resilience Assessment of Bridges
A common problem among bridge owners/managers is the need to reduce spending while attempting to operate and maintain an increasingly ageing bridge stock. Simultaneously, the increasing propensity of extreme events is creating larger, and more frequent problems among transport systems worldwide. In response to this challenge, the past decade has seen increased interest from bridge owners and managers in the use of probabilistic methods for the design/assessment/management of their bridges. In this regard considerations of risk, resilience (i.e. the capacity to recover following a perturbation) and robustness (a main parameter of resilience, defined as the ability of a structure to withstand extreme events without being damaged to an extent disproportionate to the original cause) become indispensable in analysing systems which are under stress from extreme events, with the final goal being to preserve an acceptable level of performance in harmony with balanced risk and cost. This presentation focuses on consideration of the 3R’s, i.e. risk, robustness and resilience, in the optimal lifecycle management of bridge structures.
Associate Professor Edoardo Patelli
Deputy Director, the Institute for Risk and Uncertainty, University of Liverpool, UK.
Reliable simulation of complex systems and networks under uncertainty
In modern engineering systems, and in particular for critical infrastructures, the assessment of the effect of uncertainty must be performed to assure an adequate level of safety, reliability and resilience. The talk will present first an overview of modern technique to deal with severe uncertainty and powerful approaches for modelling and simulating complex and interconnected systems.
Survival signature and load-flow approach are two powerful approaches recently developed. The benefit of using the survival signature appears if a system has multiple components with exchangeable failure times, we call these ‘components of the same type’, and particularly if there are several types of components. In addition, the survival signature can accommodate lack of information or details on the system configuration. Recent simulation techniques proposed allows to preform system reliability analysis of systems considering common cause failure.
The load-flow approach is a novel and generally applicable simulation approach to reliability and performance evaluation of complex multi-state systems with multiple outputs will be also presented. The approach allows the simulation of repairable and non-repairable complex multi-state systems of any topology without prior knowledge of system path and cut-sets. Owing to its flexibility, it can be easily extended to model cold and warm standby redundant systems, investigate cascading failure models and model systems with maintenance delays.