Excavation query

I am trying to figure out the flow rates into an excavation. (I have a multi-layered 3D model, with the layers having different hydraulic properties; recharge is specified at the surface; and top slice is specified as free and movable.) To dewater the excavation, I have specified a seepage face BC at the base and slopes of the excavation. I have run the model in steady state and it converges fine. However, the excavation does not seem to dewater all the way to the base – I checked this by looking at a cross-section of the pressure field. Any ideas as to what might be the problem?

Any thoughts would be appreciated. Thanks.

I suppose your seepage nodes are set up in slice 1. Is it possible that the piezometric water level in slice 2 is higher than the bottom of the excavation ? In this case you may have to put BC in slice 2 too (?) to achieve a complete dewatering ...

Thanks for the reply. I have set up the model as a saturated-type problem. The initial condition is some saturation at the base of the model, such that rainfall at the surface is expected to recharge the model. So, I am thinking that even if the model saturates completely, the BCs (seepage face) specified at the base (call it slice X – a slice near the base of the model) and side of the excavation should keep the excavation dry? You are probably suggesting that I assign a seepage face BC to the nodes within the perimeter of the excavation (not just the perimeter) at the slices above slice X?

I sometimes had problems with the constraint condition of BCs (for seepage face nodes FEFLOW does set a 1. kind BC with the head equals the z-coorinate of the node and applies a constraint that will no infiltration will be allowed). If your head is below the z-coordinate, the 1. kind BC will be converted to a 4. kind BC with flux=0, if the heads gets above and the constraint is chacked again the BC is reconverted to a 1. kind BC. In steady state models it could happen, that there is a convertion of the whole problem before all of the BCs have there finally state.

What you can do is, that you rerun the model sometimes and check whether something is changing. Or you can run the model in a quasi-steady-state (transient run with steady BCs) here the constraints are at least checked every timestep.