Input ground motions for earthquake site response analyses: deep sediment column with cementation

Earthquake ground motions (GMs) are developed for input to structural response analyses of a highway bridge in Las Vegas, Nevada (southwestern US), an area of arid climate and moderate seismicity. A target response spectrum for rock-outcrop (` bedrock') motions at the site is produced from a probabilistic seismic hazard analysis and GMs are found that are compatible with the target. Three approaches to generate input GMs are tested: a set of unscaled real (natural) GMs, a set of scaled real GMs, and a single, spectrally-matched GM. The input motions are filtered through a representative sediment column, nearly 400 m deep, to produce surface GMs. Both nonlinear and equivalent-linear analyses are conducted. The sediment column tended to deamplify short-period motions while amplifying long-period motions and shifting peak response to longer period, as expected. For nonlinear analysis, all three approaches to generating the input GMs yielded acceptable results. The single, spectrally-matched GM yielded the highest spectral accelerations. To explore effects of a heavily cemented layer (caliche) on site response, an alternative profile is tested. The 30-m depth-averaged shear wave velocity is unchanged, but a thick caliche layer is inserted just beneath it. In a third test, the caliche profile is truncated at the stiff inclusion. Only the nonlinear analyses showed effects of the caliche. The shallow profile predicted the highest amplification by far, demonstrating how consideration of only the upper 30 m of a sediment column can significantly overpredict short-period response.