Risk-Based Optimal Design of Seismic Protective Devices for a Multicomponent Bridge System Using Parameterized Annual Repair Cost Ratio

Abstract

A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of Master of Engineering.

Enabling efficient regional seismic fragility assessment of multi‐component bridge portfolios through Gaussian process regression and active learning

“…Six EDPs are selected to assess the seismic fragility of different bridge components, and their corresponding seismic capacity models are shown in Table 2. As two primary components, pier columns and deck unseating are considered to have four damage states (i.e., slight, moderate, extensive, and collapse damage) as defined by HAZUS, 9,42 whereas the remaining secondary components have two or three damage states 43,44 . The seismic capacities of each bridge component are assumed to follow lognormal distributions with median values listed in the table, and dispersions of 0.25 for slight and moderate damage and 0.47 for extensive and collapse damage.…”

“…In numerical simulations, the required seismic demand data is generated using three commonly applied approaches—incremental dynamic analysis, multiple stripes analysis (MSA), and cloud analysis 7 . The cloud analysis approach is most widely used for bridge seismic fragility models, 8 where high‐fidelity nonlinear structural models are subjected to a single time‐consuming set of nonlinear response history analyses (NRHAs) to develop the probabilistic seismic demand models (PSDM) 9 . The classic PSDM provides a connection between the seismic demands of bridge components through engineering demand parameters (EDP) and the ground motion IM based on three common assumptions—the median EDP‐IM logarithmically linear relationship, the constant variance (or homoskedasticity), and the lognormal distribution of a given EDP at a single IM level 10,11 …”

“…Moreover, it remains unclear how many NRHAs are sufficient for the data and the resulting surrogate model to cover the entire solution space without overfitting. These issues become more salient when the parameterized fragility models are developed at the bridge component level, being more useful for seismic repair cost assessment and post‐earthquake inspection and repair planning 9 . In such cases, care must also be taken to scrutinize the model for each individual bridge to address seismic demand correlations in multiple components with different damage modes 10 .…”