A model-based data-interpretation framework for post-earthquake building assessment with scarce measurement data

Recent earthquake events throughout the world have once again exposed the vulnerability of buildings with respect to earthquakes. It is unlikely and unsustainable to design and - especially in regions with low-to-moderate seismic hazard – to retrofit all buildings to remain within elastic displacement ranges during earthquakes with high return periods. Therefore, post-earthquake assessment plays a fundamental role in the resilience of cities, given the potential to reduce time between an earthquake event and the clearance for (renewed) occupancy of a building. In this paper, a framework for model-based data interpretation of measurements of earthquake-damaged structures is presented. The framework allows engineers to combine ambient-vibration measurements and visual inspection to reduce parametric uncertainty of a high-fidelity model using the error-domain model-falsification methodology. For building types that have limited stiffness contributions from non-structural elements (i.e. shear-wall buildings) and for which non-ductile failure modes (such as out-of-plane failure) can be excluded, reduction in natural frequency and damage grades derived from visual inspection provide global measurement sources for structural identification. The application of the proposed methodology to a shear-resisting building tested on a shake table illustrates that vulnerability-curve predictions provide accurate damage estimates for subsequent earthquakes with probabilities between 50% and 100% for five measured scenarios. In complete absence of baseline information regarding the initial building state and the earthquake signal, parametric uncertainty is reduced by up to 76%. This study thus demonstrates usefulness for certain building types to enhance post-seismic vulnerability predictions.