Stochastic analysis of multiply-supported piping systems under correlated ground excitations.

Piping systems are among the most important systems in modern industrial facilities, which are also highly affected by events like earthquakes, hurricanes, flood, etc. Like nerves in our body they nurture the industrial plants and hence require protection from these events. For dynamic and specially seismic loading there is no universally accepted design code, hence research is still underway in this area. The response of a system excited by ground motions has two components: (1) pseudo-static which is due to the ground displacement at support points and (2) dynamic which is due to the vibrational excitation. The current methods of estimating that response are typically based on the assumption of a uniform ground motion. Some authors have recently attempted to modify the methods for the spatial variations of ground motion. In this work a different method is presented which also considers the non-uniformity in support excitations. The treatment is based on the random vibration principles and can develop the pseudo-static time history of the system and also the maximum dynamic response at each degree of freedom. For a typical piping system, the correlated ground acceleration and displacement time series are developed at the excited support points. The time series are used to develop the pseudo-static component, and also for the dynamic component by means of a classical step-by-step integration method. Dynamic responses are also developed by the stochastic method and some current methods and are compared, in each case, with the results from the integration method. It is verified that the stochastic results are comparable with the current method results. Comparing the results in case of a uniform and a non-uniform ground motion, it is shown that the cross correlation of support inputs does affect the components of the response significantly and the effect is not predictable. It is also shown that neglecting the non-uniformity in ground motions can lead to an underestimation of results, in some cases, while in some other cases to an overestimation