Regression-based Evaluation of Novel Intensity Measures under Alternative Ground Motion Record Sets

Abstract

Abstract Intensity measures are generally considered as the fundamental properties of strong ground motion records and are widely utilized in performance-based seismic design methodology to relate the seismic hazard levels to the structural damages, or seismic response in general. Recent two decades have exhibited several studies proposing new alternatives that have intended to reduce the variability in seismic demand predictions. Although there exist several simple-to-advanced scalar and vector ground motion intensity measures; the literature is limited in the number of comparative studies investigating the efficiency of these parameters, especially in the entire response range of structures. This study aims at evaluating the correlation of major engineering demand parameters with novel simple-to-advanced intensity parameters following a regression-based approach to calculate efficiency metrics. For a group of low-to relatively high-rise reinforced concrete frames, alternative ground motion record sets have been formed considering different simple scalar intensity measures including peak ground acceleration, peak ground velocity, acceleration spectrum intensity, velocity spectrum intensity and spectral acceleration at the fundamental period of the structure of consideration, and utilizing these record sets, nonlinear time history analyses have been performed using OpenSees software to obtain key engineering demand parameters of the multi-degree-of-freedom systems. The correlation performance of novel scalar and vector intensity parameters have been quantified by evaluating regression models formed in between demand parameters and intensity measures. The regression-based approach has assisted to rank the ground motion intensity parameters according to their efficiency in terms of reducing the variability in response. The results of this comprehensive study display the relative correlation performance of scalar and vector forms together and mark particular scalar parameters as ‘best candidates’ for reliable loss estimation studies, despite their simplicity.