Abstract:

It is found that the conventional equivalent linear method underestimates the peak ground acceleration of the soil layer earthquake response under the strong earthquake input in practice. The phenomenon is proved by calculating and analyzing the record data from the site effect borehole station array. The reason is that the artificial value of the equivalent shear strain is quite larger than the real shear strain in high frequency band and it reduces the contribution of the high frequency components of the site soil layer earthquake response. Studies show that the shear strain changes with the frequency; the shear strain-dependent shear modulus and damping was proved in the experiments; so the shear modulus and damping are supposed to be frequency-dependent. Based on this assumption, how the high frequency component of the earthquake input motion is filtered by the soil layer caused by the equivalent shear strain reduction factor is discussed. Meanwhile, the frequency-dependent stiffness and damping are applied to the equivalent linear method and the feasibility is discussed by numerical calculation. The accuracy of the proposed method is examined by comparing with the record borehole data from large earthquakes, for example, the Port Island borehole record data in Hanshin Earthquake of 1995. The proposed method always gives much better results than the conventional equivalent linear method; so that the high frequency component can be reasonably considered.

Key words: nonlinear soil response, equivalent linear method, frequency dependent, stiffness, damping