A DATA-DRIVEN METHODOLOGY FOR ASSESSING IMPACT OF EARTHQUAKES ON THE HEALTH OF BUILDING STRUCTURAL SYSTEMS DIONISIO BERNAL 1 * AND ERIC HERNANDEZ 2 1 Civil and Environmental Engineering Department, Center for Digital Signal Processing, Northeastern University, Boston, MA, USA 2 Civil and Environmental Engineering Department, Northeastern University, Boston, MA, USA SUMMARY A data-driven approach for post-earthquake posting of buildings is presented. The approach is based on the analy- sis of residuals obtained as differences between measured responses and reference signals computed using a set of observers. The observers are particular to every output channel and can be described as mathematical struc- tures that offer a reference signal for the channel in question when response at all the other channels is prescribed. The basic feature of the reference signals is that when there is no damage they closely match the actual meas- urements. The time histories of the residuals are used to calculate two metrics: the first one measures the extent of nonlinearity and the second captures how much the nonlinear response leads to permanent loss of stiffness. A suggested mapping from the selected metrics to the current ATC-20 posting categories is provided. The approach is tested, with encouraging results, using a number of buildings taken from the CSMIP database. 1. INTRODUCTION An item that has come to the forefront of the earthquake engineering agenda is assessing the state of health of structural systems after violent ground motion. The matter is of significant practical and eco- nomic importance given that assurance of structural safety is required before structures can be reoc- cupied following a major earthquake. At the present time post-earthquake assessment of structural health is based on visual inspections (ATC-20). Although use of sensor data to assess the impact of earthquakes on structural systems is appealing, the idea has proven difficult to implement successfully. Work on using instrumental data to asses the impact of earthquake motion has been mainly focused on looking at the evolution of ‘effective fun- damental period’ (Naeim, 1997; Chen, 1980; Beck, 1978), the basic premise being that elongation of the ‘effective period’ during the motion is an indication of softening and, therefore, of damage. The approach, which suffers from the fact that the feature used cannot be objectively defined (since there is no ‘effective period’ at a given point in the response of a nonlinear system), has not proven suffi- ciently robust for general application. This paper presents a different strategy to characterize the impact of earthquake motions in build- ings. The approach is based on contrasting response signals measured during a potentially damaging event with reference signals computed in such a way that they coincide with the measurements only when the structural response is quasi-linear. Specifically, the reference signals are computed by means Copyright © 2006 John Wiley & Sons, Ltd. *Correspondence to: Dionisio Bernal, 427 Snell Engineering Center, Northeastern University, Boston, MA 02115, USA. E-mail: bernal@neu.edu THE STRUCTURAL DESIGN OF TALL AND SPECIAL BUILDINGS Struct. Design Tall Spec. Build. 15, 21–34 (2006) Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/tal.341