A variable screening analysis for the correlation between seismic intensity measures and expected damage in reinforced concrete frame buildings

Abstract

Identification of proper predictors of structural damage in buildings after the occurrence of seismic events is essential for performance-based design. In this study, 4, 6, 8, and 10-story reinforced concrete (RC) special moment-frame buildings are designed and analyzed through nonlinear dynamic analyses (NLDA) to assess structural damage estimated from lateral displacement (maximum and average inter-story drift ratios, MIDR, and AIDR respectively). The structures were designed following the Costa Rican Seismic Code and the NLDA were performed using records from the Costa Rican Strong Motion database. Several seismic intensity measures (IMs) were estimated directly from the recorded time history and the response of a 5% damped elastic single degree of freedom oscillator with a period equal to the fundamental period of the buildings studied. These IMs were contrasted with the MIDR and AIDR obtained from the NLDA to explore their correlation in terms of the coefficient of determination, R². Finally, the coefficients of determination for linear and nonlinear regressions were arranged into four clustered matrices that ranked the influence that each IM showed on the expected damage. These matrices unveiled that spectral acceleration (Sa) and spectral velocity (Sv), both based on the oscillator response, represented the highest damage predictors. For the IMs based on records at the ground level, the peak ground velocity (PGV) and acceleration (PGA), presented the highest and lowest correlations respectively.