Evaluation of 1-D seismic site response modeling of sand using centrifuge experiments

• Nonlinear site response analyses can reliably predict soil behavior in 1-D shaking.
• Class C site response analyses of dry sand compared well with centrifuge results.
• A proper selection of dynamic soil properties is critical in 1-D site response.
• Strength-corrected modulus reduction and damping curves improve analysis results.

The response of an equivalent 26 m-thick deposit of dry, medium-dense, Nevada Sand with a relative density of 60% is measured in the centrifuge under six 1-D, horizontal earthquake motions applied to the base of the centrifuge container. Several 1-D site response analysis techniques are employed to simulate the experiments, including (a) equivalent linear analyses, (b) nonlinear analyses using a multi-degree-of-freedom, lumped mass model, and (c) finite element analyses of a soil column using a pressure-dependent, multi-yield, plasticity soil model. An average Vs profile was estimated using empirical correlations. Soil dynamic properties included published generic modulus reduction and damping curves with implied strength correction as well as recommended plasticity model parameters based on soil index properties. Computed and measured lateral displacements, accelerations, shear strains, spectral accelerations, and Arias Intensities are presented and their differences are quantified in terms of mean residuals and variance. The comparisons demonstrate that 1-D seismic site response analyses using the available strength corrected, generic, pressure-dependent modulus reduction and damping curves for medium-dense dry sand can reliably compute soil response under 1-D wave propagation using any of the three methods, with an absolute mean residual of less than 0.5.