The influence of random soil damping on linear-equivalent seismic response analysis

Small-strain stiffness and damping ratio are the key parameters to evaluate dynamic behaviour of soils even if they are often problematic and onerous to obtain. This is particularly true for the attenuation characteristics. Common predictive seismic response analysis needs the input parameters as a selection of accelerograms and realistic subsoil models. The models need to be populated by physical properties such as the layers structure, seismic waves propagation velocities, materials densities as well as a their capability of dissipating energy. This last parameter takes the various forms of an attenuation coefficient a [dB/m], a quality factor Q [dimensionless] or Damping ratio D [dimensionless]. The damping ratio defined in this manner is, in general, frequency dependent. However at small strain level the material damping ratio is considered to be a constant and therefore frequency independent. Laboratory measurements of the damping ratio are very well established, even if they suffer from the effect of specimen disturbance on the measurements results. On site soil parameter characterization is getting more and more common, particularly on the basis of surface wave dispersion. Many recent works claim that in situ measurements shall be used also to gain a solid understanding of attenuation characteristics in soil deposits. Despite these progresses, only limited applications of these techniques is observed. The scarce interest towards the identification of damping characteristics is partly caused by a lack of understanding of the importance of damping. In this work we preliminary try to identify the cases where an accurate identification of the damping characteristics is needed, as opposed to cases where literature values can be used with no major detriment of the prediction accuracy. Our investigation is based on 1D linear-equivalent Shake predictions based on synthetic soil stiffness profiles and a real input seismogram. A Monte Carlo sampling of damping values takes into account several damping literature data for our soil deposits.