The reliability study of structures under random vibration involves the performance of Monte Carlo simulation in time domain. One important part of such analyses is the generation of load time histories that must accurately represent the characteristics of the stochastic loads. There are several methods to generate time histories from a power spectral density (PSD) function. One common approach is based on the linear combination of harmonic functions with random amplitudes and fixed frequencies. This method requires considering numerous harmonic terms to simulate the target PSD function. In this paper, a new approach is presented to improve the performance of Monte Carlo simulation under random vibration when the excitation is represented by a PSD function. This goal is accomplished by refreshing the frequencies for every new time history of excitation. This can considerably reduce the number of replications that are required to perform a Monte Carlo simulation. The method is demonstrated through an example which involves a linear quarter car model.
Keywords: Monte Carlo simulation, random vibration, Gaussian process, power spectral density function, probability of first-passage