Uncertainty Analysis for a Seismic Warning System

Following the collapse of the Cypress structure, which killed 42 people in the 1989 Loma Prieta earthquake, the Washington State Department of Transportation (WSDOT) initiated a series of studies of the seismic vulnerability of the Alaskan Way Viaduct (AWV). The AWV, which carries nearly 100,000 vehicles per day, is one of only two north–south highways through downtown Seattle. The studies showed that this double-deck structure is vulnerable to extensive damage during and shortly after strong earthquake shaking. The structural vulnerabilities, associated with reinforced concrete details typical of 1950s construction, are exacerbated by the presence of highly liquefiable soils beneath the AWV, and by the seismic vulnerability of a seawall that is located immediately west of the AWV. To make decisions about closing the viaduct during or after an earthquake, WSDOT will need to have reliable and timely information about the likelihood of potential damage. To this end, a study was undertaken to develop design requirements for an instrumentation system that would detect the onset of soil liquefaction and provide input to a collapse warning system. Given the uncertainties associated with predictions of ground-shaking intensity, site response, liquefaction, and structural performance, it is difficult to predict levels of damage accurately. Therefore, an important part of this study was the identification and quantification of sources of uncertainty that needed to be considered in the development of a criterion for collapse potential. A framework based on the total probability theorem was developed to evaluate the probabilities of failure. For each source, a conditional probability distribution function was developed across the anticipated range in values. The results take the form of probabilities of liquefaction-induced collapse as a function of triggering criteria.