I am modeling a quadrant of a braced excavation enclosed by diaphragm walls modeled with liners, following the procedure in ‘Dewatered Construction of a Braced Excavation’:

The bracing consists of alternating layers of struts and slabs every approximately 3.5 m (total depth is about 30 m).

The procedure to create liners in the example follows the following logic:

- Configure Zones with CM and fluid density
- Create Liner with zero density and real stiffness
- Assign gravity, water density, water table, K0 stresses and init springs
- Initialize materials, and apply BCs.
- Come to equilibrium with fluid off and mech on

In both my model and the example, the liner is initially activated with a tensile coupling normal effective stress equal to the pore pressure. Right after initializing the coupling:

Upon initial solve, this is corrected:

and the model proceeds. However, in my case, the lining is initialized with the coupling normal eff stress = pore pressure as in the example:

But upon initial solve, the tensile coupling normal effective stresses remain high:

and the stress field is unphysical.

If I allow the simulation to continue, other unphysical behavior is observed, like the interface slipping under minimal shear stress, and strange looking moment distributions. I’m not sure if the initial coupling stress being off is responsible, but I would like to fix this anyway. Any advice on how to diagnose this problem would be greatly appreciated.

Thank you in advance,