Suitability of equivalent linear soil models for analysing the seismic response of a concrete tunnel.

Current methods of analysis for the seismic response of tunnels rely on linear elastic soil constitutive behaviour. This has obvious benefits in terms of minimising the number of soil parameters required and the complexity compared to more sophisticated soil models. However, it has recently become possible to parameterise sophisticated soil models using only routine data from boreholes or in-situ testing. This paper will therefore review the effectiveness of seismic analyses using an equivalent linear soil constitutive model, by comparison of 2D Finite Element simulations with those using an advanced non-linear elastic model with isotropic hardening plasticity. In the elastic case, the parameters have been estimated using Equivalent-linear Earthquake site Response Analyses software (EERA) given a specific amount of sublayering required to match the variation of soil properties with depth. The tunnel considered is of horseshoe shape and sprayed concrete construction (New Austrian Tunneling Method), based on metro tunnels in Santiago, Chile, subjected to the Takarazuka/000 ground motion from the 1995 Kobe Earthquake. The results will focus on the differences in the induced structural forces within the tunnel lining and modification to the ground motion in the near-field of the tunnel, and discuss the implications of this for tunnel design.