On Two Ways To Consider Site Effect In Random Synthesis Of Ground Motion

In the widely adopted random synthesis approach of ground motion,in addition to the source spectrum and propagation path items,a "site item" is also included,which is expressed as the product of the near-surface site amplification factor Amp(f)and the high frequency-cut filter P(f).However,the site term doesn't include the effect of local site condition as usually referred.A common view in the earthquake engineering society worldwide is that site condition influences ground motion strongly,and there are many actual cases from the past earthquakes.This dissertation studies the effect of site soil condition that can be expressed by a transfer function(TF)of shear velocity structure of the site soil layers.Incorporate the effect into the frame of the random synthesis of ground motion and then develop an integrated random synthesis procedure,it requires a deeper analysis and comparison with the above mentioned “site term”.On a basis of general review on state of the art of the site effect expressions,the TF and quarter-wavelength(QWL)as well as a way to combine them are analyzed and compared in order to be taken into account the site effect in random synthesis of ground motion in this dissertation.The derivations of both of the TF and the QWL methods are systematically reorganized in an unified symbol system,and the quantitative relationship between the two methods is studied and discussed from a detail comparison.It is pointed out that the two methods are both developed for effect of site soil layers on ground motion as filtering and amplifying,and the latter is a simplification.For single soil layer covered site,the natural frequency of the site estimated by the two methods is concordant to each other,and the maximum value of the QWL amplification could be estimated from the corresponding amplitudes of TF.The amplifying curves by the two methods are getting close to each other as the frequency band under considered increases.For the multi-layer soil site,the difference between the TF and the QWL to express the site effect is numerically compared.The results show that the fundamental frequency by TF is in between the frequencies at inflection points on QWL amplifying curve,and the half of the maximum amplitude of the QWL curve is larger than the square root of the peak of TF curve.With various combinations of surface layer and inter layers,the corresponding relation of the two methods for the sites with thickness of soil layers extended to several thousand meters by assuming the shear wave velocity getting larger with depth site,will vary and be quite complicated.The UHF noise in ground motion could be filtered out effectively by taking into account a viscous damping in TF calculation,then the high frequency cutoff filtering usually adopted in QWL is not needed to be considered.The procedure to consider the site effect in random synthesis of ground motion by TF,QWL and a combination of them is studied further from two earthquake cases.In the case of the 1994 Northridge earthquake in US,the ground motion at the top of each soil layer is synthesized by frequency-wave number method(FK)as the expectations,and the corresponding results of the two methods from the real shear wave velocity structure and input of motion at the buried bedrock surface by FK are compared with them.The fact is validated that the upward wave should be separate from the motion at the buried bedrock surface as the input to calculate the site effect,there must be an overestimation of ground motion if input full wave of the motion,and then an approach to separate upward wave is proposed based on TF calculation.The numerical results show that wave forms of the ground motions calculated from the upward wave input by the two methods to consider the site effect are similar to the result of FK,and the motion amplitudes are even closer to those by FK.In the case of the 2016 Kumamoto earthquake in Japan,the ground motions observed at two stations with borehole records are taken as the expectations,the motions calculated by the two methods and combination of them with inputs of upward waves of the down hole records and of random synthesized motions,are compared with the corresponding ground records respectively.The results show that the difference among the ground motions at the two stations calculated by the three procedures,TF,QWL and the combination of the two,to consider the site effect with the real shear wave velocity structure,is not obvious.The PGA and response spectra are close to each other,and are also very similar to the records.The order of the three results close to the record vary with station site,inputs,as known from a more careful comparison.From the limited case studies in this dissertation,suggestion is presented as that QWL could be adopted with a high frequency cutoff filter together to take into account the site effect in random synthesis of ground motion if there is no detail local velocity structure data;TF with viscous damping could be adopted,and no need a high frequency cutoff filter if there is detailed data of the site velocity structure;and the series combination of the two is feasible for the random synthesis if there is no detail site velocity structure data at deep and there is detail velocity data in shallow.