Research On Nonlinearity In Different Soil Site Effect Based On KiK-net Strong Motion Records

The seismic site effect has always been the hot topic in the field of earthquake engineering, and the soil nonlinear response in strong motion has been the difficulty in earthquake and geotechnical engineering. There are two main methods to analyze the site effect:one is numerical simulation and theoretical calculation analysis, and the other is to analyze the site effect based on the strong motion records on different sites. Recently, with great development in strong motion observation technology, large numbers of strong motion records have been obtained, and analyzing the site effect based on strong motion records has been more widely used. The KiK-net surface and vertical arrays in Japan provide abundant data for the research on site effect.Based on strong motion records in KiK-net, the site effect on different soil sites in near and far earthquake are analysed systematically in this paper. Firstly, several hundred strong motion records from two bedrock stations and two soil stations are used to analyze the effect of shallow surface soil layer on ground motion. Secondly, thousands of strong motion records from eight stations on different soil sites are used to analyze the site effect. PGA amplification coefficient and the distribution of predominant frequency of different sites are analyzed. Lastly, SBFSR (Surface to Borehole Fourier Spectral Ratio) method and HVFSR (Horizontal to Vertical Fourier Spectral Ratio) method are used respectively to identify the nonlinearity in site response. The main work and achievements are as follows:(1) Analyzing the effect of shallow surface soil layer on ground motion, it is found that, whether on bedrock sites or soil sites, the fourier spectrum predominant frequency of borehole records decreases with the increase of magnitude or distance. The predominant frequency of surface records almost has the same values and distribution with that of borehole records on bedrock sites; but on soil sites, the fourier spectrum dominant frequency of surface records and borehole records have obviously different distribution and values, it is also different between different sites. The results indicate that the hard bedrock has little effect on ground motion, but the surface soil layer has great and complex effect on ground motion.(2) Analyzing the PGA amplification coefficient of different sites, it is found that, no matter how far the epicentral distance, the PGA amplification coefficient decreases with the increase of PGA in the borehole. The PGA amplification coefficient of three components is always the same in near earthquake(A< 100km) on soft sites, but in distant earthquake(250<A<500km) the PGA amplification coefficient of two horizontal components is larger than that of vertical component. However, on hard sites, whether in near or distant earthquake, the PGA amplification coefficient of two horizontal components is significantly larger than that of vertical component. The Vse has greater effect on the PGA site amplification coefficient than the thickness of overlying layer. In near earthquake, the PGA amplification coefficient on hard site is larger, while in distant earthquake the PGA amplification coefficient on soft site is larger. The larger the gradient of Vs profiles in the deep borehole, the larger the PGA amplification coefficient.(3) Comparing the predominant frequency obtained from spectral analysis of different sites in near and distant earthquake, it is found that, in near earthquake the predominant frequency of three components in borehole is scattered and high, but it is low on surface. However, in distant earthquake the predominant frequency of three components in borehole is concentrated and low, while it is high on surface. The predominant frequency of soft site on surface is close in near and distant earthquake, while the predominant frequency of hard site on surface in distant earthquake is smaller than that in near earthquake.(4) SBFSR method and HVFSR method are used respectively to identify the nonlinearity in site response. It shows that, the site response nonlinearity on soft and thick sites is more obvious than that on hard and thin sites. Even though the input ground motion PGA in borehole is really small (only 9-15gal), the nonlinearity on softer sites can be identified obviously by spectral ratio method. The input ground motion PGA in borehole directly affects the site response nonlinearity, when the underground input PGA is relatively small, due to the nonlinearity in site response is also small, both SBFSR method and HVFSR method are not ideal for identification of the nonlinearity; when the underground input PGA is large enough, both methods are ideal for identification of nonlinearity.(5) SBFSR method and HVFSR method are compared in the determination of predominant frequency and the identification of nonlinearity in site response. The predominant frequency of two spectral ratio methods on bedrock sites is really scattered, but on soil sites, predominant frequency obtained by SBFSR method is more accurate than that from the HVFSR method whose result is uncertain in some soil sites. The predominant frequency obtained from surface records and SBFSR method and HVFSR method are close in most sites, but the HVFSR frequency is slightly smaller than the SBFSR frequency, which is more obvious on hard sites. To identify the nonlinearity in site response, SBFSR method is generally better than HVFSR method. HVFSR method is sensitive to find the frequency moving to low-frequency, but for the decrease of spectral ratio, it is not as good as SBFSR method, as well as the H/V spectral ratio significantly less than S/B spectral ratio, which is due to the ignoreness of the amplification of the vertical component in ground motion in HVFSR.