Tunnelling-induced surface settlement: on the choice of soil constitutive model
Résumé
Increasing tunnel excavations in dense urban areas faces the issue of accurately predicting tunnelling-induced surface settlement. The quality of those predictions closely depends on the choice of soil constitutive model. In this context, some widely used soil models in engineering practice not only fail to reproduce some characteristics of soil behaviour on the constitutive level but also lead to shallower settlement. This article studies the influence of the choice of the soil model on tunnelling-induced displacements through a case study. Three different soil models are used to simulate soil behaviour, namely the hardening soil model, the modified cam clay model and a proposed enhanced critical state model incorporating deviatoric and volumetric hardening. The proposed model is first validated on different triaxial tests and its features are especially highlighted for overconsolidated soils. This model is then compared to the other two models on tunnelling simulations. The results show great predicted settlement sensitivity to the choice of soil model. The proposed model engenders deeper settlement and, consequently, enables a more conservative design. Furthermore, parametric studies demonstrate significant disparities in the predicted displacements by the different soil models depending on the initial conditions and the stress relaxation ratio.