Spectrally matching pulse‐like records to a target RotD100 spectrum

This article evaluates two different approaches to generate pulse‐like spectrally matched earthquake records pairs compatible with a target RotD100 spectrum. One is the conventional approach of separately matching each horizontal component to the target RotD100 spectrum, the other approach simultaneously modifies both horizontal components to tightly match the target RotD100. The characteristics of the resulting records are evaluated using amplitude scaled pulse records as baseline for comparison. It is shown that, using a target spectrum that includes a narrow‐band modification to accommodate the presence of a pulse, largely increases both methodologies likelihood of generating spectrally matched records that preserve the pulse‐like nature of the seeds. However, the records generated through independent component matching exhibit RotD100 spectral amplitudes that largely surpass the target amplitudes, and instantaneous power (IP) and peak ground velocities (PGV) substantially larger than the observed for the amplitude scaled records. Consequently, these motions generated unrealistically large inelastic demands when used as input for nonlinear response history analyses (NRHA), with mean peak inelastic demands up to 60% larger than the expected from amplitude scaled records. Conversely, the approach based on simultaneous modification of the two horizontal components generated the largest number of successful matches and the resulting records exhibited IP and PGV values close to the obtained for the amplitude scaled records. When used as input for NRHA, the mean peak inelastic demands were very close to the demands imposed by the amplitude scaled set. These results suggest that current 10% spectral amplitude penalization for spectrally matched records in ASCE7 is unnecessary for pulse‐like motions.