Assessment of structural damage detection methods for steel structures using full-scale experimental data and nonlinear analysis

Abstract

Rapid structural damage assessment methodologies are essential to properly allocate emergency response and minimize business interruption due to downtime in the aftermath of earthquakes. Within this context, data-driven algorithms supported by sensing capabilities can be potentially employed. In this paper, we evaluate an extensive number of damage indicators computed based on nonmodel-based system identification techniques and wavelet analysis. The efficiency of these indicators to infer the damage state of conventional steel moment-resisting frames (MRFs) and concentrically braced frames (CBFs) is evaluated through the utilization of landmark full-scale shake table experiments that examined the inelastic behavior of such frames at various seismic intensities. The same data is complemented with numerical simulations of multi-story steel MRFs and CBFs with the overarching goal to identify potential limitations and propose refinements in commonly used damage indicators for rapid seismic risk assessment. It is shown that wavelet-based damage sensitive features are well correlated with commonly used story-based engineering demand parameters that control structural and non-structural damage in conventional steel frame buildings.