Hybrid Stochastic Model for Simulating Non‐Stationary Ground Motion Records

ABSTRACT This study presents an improved hybrid (empirical‐stochastic) model for simulating non‐stationary accelerograms based on filtered Gaussian white noise. A linear filter for the frequency domain and a modulating time function are used to generate a stochastic accelerogram based on target parameters. Arias intensity, strong‐motion duration, and the lag time between S waves and P waves are used for the modulating function calibration. Universal empirical equations linking the time domain strong motion parameters and that of the modulating function are developed based on mathematical properties (inflection points) of the modulating function. The cumulative energy of the resultant modulation functions is assessed and compared to the cumulative energy of the target accelerogram. The central frequency and frequency bandwidth, determined from the physical spectrum of the target, are utilized for identifying the parameters of the linear filter. Empirical equations governing the frequency and damping ratio of the linear filter are proposed based on the specified spectral parameters. Stochastic accelerograms are simulated using proposed equations, resulting in notable improvements in the regeneration of target frequency characteristics, response spectra, and cumulative energy, as evidenced by comparisons to target records.