Selection of Random Vibration Procedures for the NGA-East Project

Pseudo-spectral acceleration (PSA) is the most commonly used intensity measure in earthquake engineering as it serves as a simple approximate predictor of structural response for many types of systems. Therefore, most ground-motion models (GMMs, aka GMPEs) provide median and standard deviation PSA using a suite of input parameters characterizing the source, path, and site effects. Unfortunately, PSA is a complex metric: the PSA for a single oscillator frequency depends on the Fourier amplitudes across a range of frequencies. The Fourier amplitude spectrum (FAS) is an appealing alternative because its simple linear superposition allows effects to be modeled as transfer functions. For this reason, most seismological models, i.e., the source spectrum, are developed for the FAS. Using FAS in conjunction with random-vibration theory (RVT) allows GMM developers to superimpose seismological models directly, computing PSA only at the end of the process. The FAS-RVT-PSA approach was first used by the Hollenback et al. team in their development of GMMs for the Next Generation Attenuation Relationships for Central & Eastern North-America (NGA-East) project (see Chapter 11 of PEER Report No. 2015/04). As part of the NGA-East project to support the Hollenback et al. team and similar efforts, the current report summarizes a systematic processing algorithm for FAS that minimizes computational requirements and bias that results from the RVT approximation for median GMM development.