Statistical Analysis and Stochastic Simulation of Ground-Motion Data

For many years structural engineers have been concerned with the dynamic response of structural systems subjected to seismic excitations. Ground motions may be caused by natural earthquakes, by underground explosions, or by nuclear air blasts. Structures such as high-rise buildings, nuclear reactor facilities, or sensitive equipment in the vicinity of such events are vulnerable to induced random-type disturbances. Traditionally, deterministic methods of analysis relying on the known earthquake response spectra have been used. 1 These methods have provided valuable information regarding the behavior of structures during earthquakes. However, this procedure has a serious restriction in t h a t only a few strong-motion accelerograms exist which provide ground-motion input. An earthquake is usually initiated by a series of irregular slippages along faults, followed by 2273 2274 T H E BELL SYSTEM TECHNICAL JOURNAL, DECEMBER 1968 innumerable random reflections, refractions, dispersions, and attenuations of the seismic waves within the complex ground formations through which they travel. Since ground motions are generally random, a probabilistic method of analysis appears to be more appropriate than the traditional method of establishing a reliable design basis for structures subjected to ground motions. The simulation of ground motion is undoubtedly a necessary step in performing such a probabilistic analysis. Because earthquakes are unpredictable, some researchers in structural and earthquake engineering have attempted in recent years to simulate earthquakes by using stochastic processes.