Wells in a moving mesh model

I have used a well boundry condition in a moving mesh 3D saturated groundwater flow model and the
boundary condition is only applied across the layer that is considered the aquifer.  If I now apply a pumping rate that results in a groundwater level that is below the base of the aquifer (i.e. solves an groundwater level that is below the original base of the well boundary condition), will the  simulation continue.  If it does I am guessing that the slices that were originally within the aquifer layer will now be in a layer belwo the aquifer. 
I assume if this is the case, to stop this from happneing I could constrain the use of the well boundary condition such that the level can not go below a predefined base of aquifer elevation.  Is this correct? 

In a 'free&movable' model, the slices move with the water table elevation. As the boundary conditions are located on slices, they will move together with them. So a well might even move deeper with higher drawdown. It is not automatically turned off when the water level falls below the (initial) bottom of the well.

To limit well extraction by a minimum level, there are two possibilities:
1. You can use a 'minimum head' constraint on the well. That will turn the well bc into a fixed head bc when the pumping would lead to a water level below the minimum head (well bottom). This option works well as long as the water level does only drop caused by pumping in the well. If external effects can also lead to a water level below the well location, you should prefer the second option. Using a 'minimum head' constraint in this case could lead to infiltration at the well due to the changed boundary condition.
2. Instead of the well bc, you can set a head bc with the minimum head (well bottom) as the value. The boundary flow at this head bc then has to be constrained by a 'min flux' of 0 and a 'max flux' of the pumping rate. In this case, the original pumping rate will be applied (well bc) as long as the water table is much above the well bottom. At a water level slightly above the bottom, a head bc will be applied, leading to less pumping. At a water table below the bottom, the boundary condition will be turned off (well bc with value 0).

Thanks Peter. Is this the correct forum to ask why the well bc does not simply place a zero pumping rate if the head is at or below the constrained minimum head, regardless of how this happens?

'Why'-questions are always difficult to answer  ;)
The reason here is probably the general concept of the constraints. Setting a zero pumping rate if the water table is below the minimum head would mean a specific functionality for a well boundary condition. The constraints, however, are used in exactly the same way for all boundary conditions in FEFLOW. So here's a conflict of objective between flexibilty and user-friendlyness that has been decided in favor of the flexibility.

Yes I undersand, and thanks again for the answer.  One last question to finish this one for me.  I am guessing with the free surface moving slice option, all boundary conditions will move if the the slice that the bc is on moves.  That is, the bc is slice related and not aligned with the geological layering defined in the 3D slice elevation menu.

Exactly.

I have a similar problem to swilkes but with wells completed over several layers/slices in a quarry dewatering model that uses a ring of wells.  I can find no combination of wells or head boundaries and constraints or discrete elements that will allow the layers to dewater smoothly.  The best so far is to set the layer elevation and flow constraint at each slice where the wells are screened with the flow constraint set according to the permeability of the layer (the top two layers have the highest permeabilities).  The model dewaters smoothly to the base of the top layer and almost to the base of the second layer before going unstable.  Eventually the modelled water level stabilizes at the base of that layer but the boundary is still discharging at a high rate.  Curiously a layer three below this one stops discharging.  We covered the subject briefly on the Feflow course in Berlin last year and I have tried everything in my notes!

I wonder therefore: has anyone sucessfully modelled mutli layered wells around a quarry or excavation in which several slices are successively dewatered?  If not, can anyone suggest a way that it might be done?

I want to clarify a point regarding boundary conditions versus constraints.  In the FEFLOW user manual (p. 57) it is stated that "Outfluxes are defined as positive for boundary conditions.  However, for constraint conditions, the definition is opposite, here influxes are positive".  Therefore, if you are using first type of boundary conditions to represent wells  (i.e. constant head is set to the well screen bottom elevation) then the constraint to limit flow out of the model to the well rate should be Qmin = well rate NOT Qmax = well rate.  Similarly, to prevent discharge into the model when the aquifer head drops below the boundary condition elevation Qmax should be set to zero.  I have tried the WASY recommended method of setting constraints of Qmax = well rate, Qmin = 0 and monitored flows using observation point groups.  This clearly indicates that no flow occurs out of the model at the drain locations under this scenario.

Can you confirm I am correct on this point since there are several posts by WASY where this incorrect advice is given?

You are right, maxQ shall be 0, minQ shall be set to - pumping rate. I've found one location in the forum where this was stated wrong (in the modeling FAQ). I've corrected this. If you have seen another one, it would be great if you could let me know the location.

Hi Peter,

Thank-you for clarifying this.  I did a quick review of the forum posts and I do not see any other reference to first type of boundary condition constraints other than the one you mention.

Regards,

Stephen