Abstract:Earthquake ground motions can significantly affect buildings when they are amplified in surface soil layers.Therefore,in order to estimate the behavior of buildings during severe earthquakes,it is essential to evaluate those characteristics of amplification.Although investigations have been carried out,many problems remain unsolved with regard to those characteristics.Ground response analysis typically involves predictions of surface ground motions,developing a spectrum of design responses,determining dynamic stress and strain,and evaluating the potential for liquefaction.Methods used to calculate the ground response are commonly divided into linear,nonlinear,and equivalent linear concepts.The linear analysis assumes that the shear modulus as constant.However,soils often exhibit nonlinear behavior,even at low levels of strain.The nonlinear method is time consuming.The equivalent linear method,represented by the SHAKE model,is one of the most frequently used methods for analysis of ground responses to earthquakes.This method is used primarily for approximations because it uses effective strain to define the material property for analysis.However,the equivalent linear method remains the most popular method of seismic response analysis.It has been 20 years since the safety evaluation program LSSRLI-1 was developed and was used for seismic response analysis.At the time,LSSRLI-1 represented the most advanced method and contributed greatly to earthquake preparedness and disaster reduction for China.However,some deficiencies in the model were identified.Based on its focus in the frequency domain with a harmonic wave incident,the exact linear solution was deduced in the time domain,and compiled into a program in the Matlab environment.For the study,eight sites with medium-hard soils were chosen.Seismic responses were calculated using three different methods:the LSSRLI-1,exact solution,and SHAKE2000.The results show that the ground response spectrums and distribution of soil shear strain,calculated using the exact solution and SHAKE2000 methods,are in agreement with each other.The ground response spectrums,calculated using LSSRLI-1,were consistent with the other two methods.However,in some cases,the distribution of soil shear strain,calculated using LSSRLI-1,showed a relatively large deviation from the SHAKE2000 and exact solution methods.The extent of these differences may significantly affect the estimates of ground motion,and cannot be ignored.