Simulating soil stiffness degradation in transient site response predictions

A series of nonlinear seismic ground response analyses are conducted to investigate the effect of soil degradation on the seismic response of typical sedimentary deposits in Southern California. For this purpose, three downhole array station sites at the Los Angeles basin are subjected to broadband synthetic ground motion excitations. A characteristic strain measure, defined as the averaged maximum shear strain in the top 30 m of the soil profile (γ30γ30), is proposed to enable the correlation between the ground shaking intensity level (here measured by means of the peak ground acceleration on the surface of rock-outcrop, PGAROPGARO) and the average deviation in ground motion response spectral predictions (denoted as eSAeSA) attributed to the effect of modulus degradation due to cyclic loading. It is shown that strong linear correlations exist among log(PGARO)log(PGARO), log(γ30)log(γ30) and log(eSA)log(eSA) for the sites investigated. On the other hand, the number of loading cycles typically employed to indicate the importance of soil degradation is shown to play a much less important role. This becomes more apparent for near-field large amplitude ground motions characterized by a dominant pulse and overall low number of cycles, which nonetheless impose strains large enough to trigger modulus degradation effects. Based on this observation, it is suggested that the soil degradation effect in seismic ground motion is primarily controlled by the ground shaking intensity. The level of PGAROPGARO beyond which soil degradation should be accounted for and incorporated in seismic ground response analysis is also depicted for the sites investigated.