The influence of pre-failure soil stiffness on the numerical analysis of tunnel construction

Abstract

The use of the finite-element method to analyse tunnels is becoming more widespread, but any prediction is dependent (among other things) on the model adopted for the pre-failure soil behaviour. This paper compares and contrasts plane strain predictions of ground movement for both single- and twin-tunnel excavations in stiff clay modelled as (a) isotropic linear elastic perfectly plastic (b) anisotropic linear elastic perfectly plastic (c) isotropic non-linear elastic perfectly plastic with shear stiffness dependent on deviatoric strain and mean effective stress, and bulk modulus dependent on volumetric strain and mean effective stress (d) anisotropic non-linear elastic perfectly plastic employing the model in (c) above (e) isotropic non-linear elastic perfectly plastic with shear and bulk stiffness dependent on deviatoric strain level, mean effective stress, and loading reversals. The analyses model the geometry of the twin Jubilee Line Extension Project tunnels beneath St James's Park (London, UK), and field data are presented for comparison (Standing et al., 1996). By considering the predicted surface settlement, the study shows the importance of modelling non-linear elasticity, and the effect of introducing a soft independent shear modulus. The differences in subsurface displacements for isotropic and anisotropic models are highlighted. The subsequent modelling of an adjacent tunnel excavation exposes more detailed features of all the models. It is concluded (a) that anisotropic parameters appropriate to London Clay do not enhance the plane strain predictions of ground movement as long as non-linear pre-failure deformation behaviour is being modelled; (b) that a soft anisotropic shear modulus significantly improves greenfield predictions but not twin-tunnel predictions; (c) and that accounting for load reversal effects does influence an analysis of this problem (St James's Park twin tunnels). L'analyse des tunnels a de plus en plus souvent recours à la méthode des éléments finis, mais toute prévision dépend (entre autres) de la modélisation du comportement des sols avant la rupture. L'article compare les mouvements de sol de tunnels simples et doubles creusés dans de l'argile rigide, sous l'effet de déformations planes, prédits dans les modèles suivants (a) isotrope, linéaire, élastique, parfaitement plastique (b) anisotrope, linéaire, élastique, parfaitement plastique (c) isotrope, non linéaire, élastique, parfaitement plastique, la rigidité au cisaillement variant en fonction des contraintes déviatrices et de la contrainte intergranulaire moyenne, et le module de compression variant en fonction de la contrainte hydrostatique et de la contrainte intergranulaire moyenne (d) anisotrope, non linéaire, élastique, parfaitement plastique, en utilisant le modèle cidessus (c) (e) isotrope, non linéaire, élastique, parfaitement plastique, la rigidité au cisaillement et la rigidité à la compression variant en fonction des contraintes déviatrices, de la contrainte intergranulaire moyenne et des contraintes subies récemment. Les analyses modélisent la géométrie des tunnels doubles du prolongement de la ligne de métro Jubilee Line sous St James's Park, et présentent des données relevées sur le terrain à titre de comparaison (Standing et al., 1996). En examinant le tassement de surface prédit, l'étude montre l'importance de la modélisation de l'élasticité non linéaire, et l'effet d'un module de cisaillement tendre indépendant. L'article fait ressortir les différences de déplacements souterrains des modèles isotropes et anisotropes. La modélisation subséquente de l'excavation d'un tunnel adjacent met en évidence des aspects plus détaillés de tous les modèles. L'article conclut (a) que les paramètres anisotropes propres à l'argile londonnienne n'améliorent pas les pré- visions des mouvements de sol sous l'effet de déformations planes dans la modélisation des déformations non linéaires avant la rupture; (b) qu'un module de cisaillement anisotrope tendre mène à de meilleures prévisions dans le cas de nouveaux chantiers, mais pas celui de tunnels doubles; et (c) que l'historique des contraintes récentes compte peu dans l'analyse de ce problème (tunnels doubles de St James's Park).

The influence of previous stress history and stress path direction on the surface settlement trough induced by tunnelling

“…Simpson et al, 1979). More specifically, in the case of tunnels, Addenbrooke et al (1997) demonstrated that the introduction of non-linear pre-failure stiffness improved the predicted surface settlement profile, as compared with linear elastic models.…”

“…This model has already been used by Addenbrooke et al (1997) for the analysis of the twin tunnels beneath St James's Park, and was called model J4. The expressions that describe the variation of the tangent shear and bulk moduli, in the non-linear range, as implemented into ICFEP, are given in Appendix 2.…”

“…For the model J4 the use of the parameters adopted by Addenbrooke et al (1997), for the non-linear elastic part of the model, resulted in a softer response when compared with the available test data from the Canons Park site. For this reason a new set of parameters was derived; the predicted response can be seen in Fig.…”

“…The predictions of the M3-SKH model are also compared with the predictions of an elasto-plastic model with a nonlinear pre-yield behaviour, which has already been used in the analyses of Addenbrooke et al (1997). Further comparisons are made with field measurements.…”

Numerical investigation of the effects of tunnelling on existing tunnels

“…The pre-yield model (J4) has been described and calibrated by various researchers (e.g. Addenbrooke et al, 1997). The parameters for the non-linear elastic model J4 and for the Mohr-Coulomb model for all the soil layers are tabulated in Tables 2 and 3 respectively.…”