Quantitative Identification And Simulation Of Far-Field Long-Period Ground Motions

It is observed that super high-rise buildings are vulnerable to far-field long-period(FFLP)ground motions,which excite significant structural and/or non-structural responses to an extent beyond expectations based on their intensity.This means that engineering communities are concerned with issues related to safety and serviceability for super high-rise buildings.However,it is difficult to acquire characteristics of FFLP motions and to conduct performance of super high-rise buildings in sense of engineering,because of complex seismological mechanism and knowledge.Moreover,this hinders a progress by which the strategies to tackle risks for super high-rise buildings are made,when the structures are subjected to FFLP motions.In the context of this,the dissertation discusses the proposed quantitative criterion by which quantitative classifications of FFLP motions are collected in light of behaviour of long-period waves,thus giving an insight into the characteristics associated with FFLP motions.Based on this,the methods to simulate FFLP are proposed for the sake of studying performance of super high-rise buildings subjected to FFLP motions at intensity levels of interest.For detailed conclusions are summarised as follows:(1).Quantitative identification for the FFLP motions.Variations of arrival times for frequencies are presented by phase derivatives.The proposed function of frequency is used to capture characteristics of envelope delays with frequencies.As a result,the quantitative criterion is established upon an equation with corner frequency and energy ratio.The effectiveness of the proposed method is verified through analysing the bias between the manual and identified classifications.It is concluded that the identified FFLP motions are well correlated with later-arriving long-period waves.Furthermore,comparisons of the identification results are made with respect to different quantitative cretirions for FFLP motions.The result shows that the proposed criterion herein is superior to others.(2).Comparisons of long-period characteristics in terms of identified FFLP motions and normal motions.It is expected that the differences in long-period characteristics are presented by calculating temporal and frequency parameters derived from the identified ground motions.Comparisons of averaged spectra with respect to predictors suggest the positive relationship between long-period characteristics of the motions and their predictors.Analysis of non-stationary property shows that the FFLP motions appear to be more obviously non-stationary than normal motions on the basis of correlation coefficients obtained from wavelet packets.Because of these,the proposed criterion for identifying long-period motions is reliable.In addition,the components composed of surface waves are individually extracted from two types of ground motions according to particle polarization.Extent of contribution of long-period surface waves from FFLP motions to their long-period characteristics is greater than those from normal motions,as suggests that long-period surface waves from FFLP motions play a key role in determination of long-period characteristics.(3).Combination of spectral representation and wavelet packets for generation of long-period ground motions.Narrow stationary power spectrum has a potential to capture dominant long-period characteristics,which are presented in the form of later-arriving long-period waves.Meanwhile,modification of Gaussian white noise is performed both in time and frequency domain so as to attain a modulated initial seed motion showing similar non-stationary to a target motion.Simulation of acceleration time history controlled by high-frequency components is obtained as a result of iterations,by which wavelet coefficients of the modulated seed motion are adjusted to match the target acceleration response spectrum and cumulative energy plot.Finally,comparisons of simulated long-period motions and the target motions in terms of similarity for time and frequency domain are conducted,therefore validating the effectiveness of the proposed method.(4).Stochastic simulation for FFLP motions considering envelope delays and amplifications.Envelope delays controlling the non-stationary property of a realization are used to model phase distribution.Site amplifications obtained in empirical statistic sense account for long-period characteristics caused by basin effects.In this respect,the two factors are taken into consideration to modify the point-source program for simulating FFLP motions.Comparisons of simulated and recorded motions validate that the modified model are suitable for the simulation of FFLP motions.Furthermore,after fitting the proposed model of envelope delays to recorded motions,statistical models for the parameters of envelope delays are grouped with respect to the type of ground motions and epicentral distance.In addition,a series of amplification functions of periods are established as a result of extending impedance ratio to shear velocity at shallow depth.This provides an alternative to assess the basin effcts according to site properties at shallow depth in hand.Because of these,the parametric stimution of FFLP motions is proposed.(5).Acceleration response spectra for rock surface with intensity levels are approximately estimated by simplified probabilistic seismic hazard analysis(PSHA).Corresponding long-period spectra conditioned on amplifications for the specific site are determined.Because of this,effectiveness of the proposed methods for simulating longperiod motions is verified on the basis of characteristics of seismic responses of a super high-rise building.Finally,the preliminary study on the influence of energy dssipation devices,including buckling restrained braces or viscous dampers,on seismic responses of the structure subjected to long-period and normal motions is carried out.