Assessment of spectrum matching procedure for nonlinear analysis of symmetric- and asymmetric-plan buildings

• This study examines the accuracy and efficiency of spectrum matching methods.
• We model 48 single and 9 multi-story buildings with symmetric and asymmetric plans.
• The SM procedure provides more accurate estimates than the ASCE/SEI 7-10 procedure.
• The SM procedure is more efficient than the ASCE/SEI 7-10 procedure.
• The SM procedure decreases the variability of the responses.

Current performance-based seismic design procedures rely on response history analysis (RHA) for estimating engineering demand parameters (EDPs) of base-isolated structures, high-rise buildings and irregular structural systems. Ground motion records for RHAs should be appositely selected in compliance with site-specific hazard conditions, and properly modified either by amplitude-scaling or spectrum matching (SM) to ensure that the modified records provide accurate and efficient estimates of “expected” median demands. While amplitude-scaling techniques change intensity of the ground motion record, SM methods also alter the record’s waveform (in frequency or time domain) to match its response spectrum to a target (or design) spectrum. The research on adequacy of spectrum-matched records for nonlinear RHAs of buildings is not only limited to symmetric-plan buildings subjected to one component of ground motion but also lacks consensus. This study comprehensively examines the accuracy and efficiency of a SM method for nonlinear RHAs under bi-directional earthquake excitations by covering single- and multi-story buildings having symmetric- and asymmetric-plans. For this evaluation, three-dimensional computer models of 48 single-story and nine multi-story buildings were created. Their structural responses were obtained from subsets of seven records modified by SM and separately by amplitude-scaling according to the regulatory ASCE/SEI 7-10 scaling procedure. The accuracy and efficiency of both procedures are examined by comparing their median EDP estimates from subset of records against the median values of EDPs due to a larger set of unscaled records, and by comparing record-to-record variability of the response. It is shown that the time-domain SM procedure provides more accurate median EDP estimates as compared to the ASCE/SEI 7-10 amplitude scaling procedure; however, it critically vanishes the variability of EDPs associated with aleatoric uncertainty in ground motion records. Retaining a certain level of aleatoric variability in EDPs can be an important parameter to be considered for certain projects.