Difficult to say. I would suggest that you start with an elastic model, and then move your way down your suggested ladder until you meet a good match. However, ideally, your stress strain curve from lab results should be able to help in deciding your final selection.

Hi,

1) start with the empirical formulation about settlements

S=f(Volume loss at surface , z, k) [Attewell, New]

2) numerical models: with plaxis you can apply a volume loss on the excavation. Attention: you have to read the manual about the magnitute (doble the value provided, if I remember well), and remember that the volume loss at excavation is less than at surface

3) your question: the excavation causes an unload of the soil at bottom of excavation. This causes raising on the bottom/invert, and wrong results on settlements values but also for the settlements shape. Is common with very soft soils that Mohr-Coulomb lift the whole tunnel and settlement are upwards (non physical sense).

The hardening soils is a good solution: limitate data required, and you can perform easily sensitive analysis.

For the other const. models the question is: do you have a data for define all the parameters required? Have you confidence with these models?

My suggestion is starting from simple to complex. First empirical formulation, then numerical model with Hardening soil . Following insights are obviously welcome, but remember you have to adapt your models to reality, not the other way around.

Luca

Article Numerical study of the segmental tunnel lining behavior unde...

Article Ground Movements Prediction in Shield-Driven Tunnels using G...