Optimal Structural System Design for Catastrophic Unforeseen Events

Johns Hopkins University, Baltimore MD

Investigators

Abstract

The events of 9/11 demonstrate the need for consideration of a decision-making framework for structural building safety in response to unforeseen events. This research project will provide such a framework for decision-making and constrained optimization of building structural safety under unforeseen catastrophic events. Unforeseen events (i.e., terrorism damage, corroded members) occur with a surprisingly short return period and break the traditional link between environmental loads and the probability of structural failure (Pf). Current design, using only component-wise reliability, ignoring redundancy, and lacking a formal decision-making framework, is inadequate. By extending recent ideas in structural sensitivity, redundancy, and probabilistic earthquake engineering we propose a characterization for Pf in unforeseen events through two stochastic variables: an intensity measure (IM) and an engineering demand parameter (EDP). The IM describes the magnitude and correlation of the unforeseen damage. The EDP provides a link to Pf for a given IM. The use of a redundancy index (R) to provide a coarse, but computationally cheap indicator of Pf, will also be considered. Using Pf and/or the stochastic variables IM, EDP and R we will explore decision-making for low-probability, high-risk, catastrophic events. Such events force us to reconsider the applicability of the von Neumann Morgenstren utility theory. Recent research in catastrophic risk insurance will be utilized to develop an appropriate objective function for optimization of building safety. Whether one views the behavior of the World Trade Center during the attacks of 9/11 as a design success, because so many people were able to get to safety, or a design failure, after all the buildings collapsed - one issue that may have been lost in the resulting discussion is that the building's behavior is not a direct result of our "building codes" nor our current design methodologies. The proposed work combines new findings in structural sensitivity, redundancy, probabilistic earthquake engineering, optimization under uncertainty and catastrophic risk insurance. This research will lead to better decision-making in the design of buildings and insure a higher level of building structural safety, within realistic economic constraints, when considering the possibility of catastrophic damage events.