Effect of 2D Random Field with Spatially Variable Soil Parameters on Site Effect

• Main Authors: Yen-Hsiang Chang, 張硯翔
• Other Authors: Chi-Chin Tsai
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/a24gd8

碩士 === 國立中興大學 === 土木工程學系所 === 107 === It is generally known that soil is a heterogeneous material. However, in the engineering problems, soil is often treated as homogeneous material for the convenience of analysis. This study simulates the spatial variability of soil properties through the method of the random field to explore its effect on the site response analysis. In addition, this study also analyzes the coherency of ground motion propagated through the simulated soil layer and compare with that of the actual observed seismic record, thus the variability of the real soil layer is estimated. In the past, 1D analysis were mostly used for the analysis of site response. In this approach, the horizontal variability of the soil could not be considered. Therefore, this study establishes a method to simulate 2D soil profiles, considering the spatial variability of shear wave velocity (Vs) and nonlinear parameters. The 2D profile is simulated according to the random field theory. The autocorrelation function and correlation length (CL) is used to establish the correlation between soil elements, and then the coefficient of variation (COV) is used to control the variability of the soil material itself. Numerical analysis software (FLAC) was used for site response analysis using the simulated 2D profiles. The results show that site response by 1D analysis is greater than that by the 2D analysis. In all cases, as COV of the Vs increases, the analysis results become more dispersion, and the chances of an extreme results are higher. As the CL of the Vs increases, the mean of Fourier spectrum is lower. In the linear analysis, as COV of the Vs increases, the mean of Fourier spectrum is lower. However, in the nonlinear analysis, the mean of the Fourier spectrum is higher when COV of the Vs increases. In addition, the COV of Vs has a great influence on the ground motion correlation. The larger the COV is, the lower the ground motion correlation is. Ground motion correlation is less affected by CL of Vs and nonlinear parameters than the coefficient of variation.