Risk-based assessment of robustness: what can it do and what can’t it do?

A guiding principle in the design of structures is to maximize the reliability of a structural system (that is, to minimize the probability of failure, or perhaps more generally to minimize the risk associated with potential failures). Robustness has been recognized as a valuable property to incorporate into systems, in support of that principle. A discussion of possible methods for quantifying robustness is presented here, along with an argument as to why riskand reliabilitybased methods may be useful for quantifying robustness and for evaluating robustness-related design criteria. The second part of the paper discusses the concept of robustness from the perspective of Normal Accident Theory and the System-Action-Management framework for risk assessment. These concepts both recognize that complex systems have difficult-to-identify feedback loops that may cause failure mechanisms that are difficult to identify a priori. Redundant systems are often coupled, so that backups fail in the same event that causes failure of a primary system. Establishment of backups can also increase system complexity and encourage risk-taking by designers or operators. Design of high-rise residential buildings in the seismically-active western United States is used an example to discuss these concepts. While there are methods available to deal with these challenges, quantitative measures of robustness should be treated as only a partial solution if they are computed without recognizing the potential for complex failure sequences.