Multilayer well for unconfined aquifer in free mode (movable top slice)

Hello,

I was wondering if there is a way to simulate a multilayer well using the free strategy (i.e. movable top slice) without the nodes of the deformed mesh going down, beyond the bottom of the well (I have noticed that the initial number of nodes assigned as a Multilayer Well BC remains the same after the model run but they have changed their positions adjusting to the deformed mesh). Or else if it is possible to "turn off" the well nodes when they go down beyond the bottom of the well (due to the moving mesh solution method). The problem is that the wells seem to keep pumping even when the water level has descended past the bottom of the well (i.e. the condition sometimes referred to as "dry well").

Thank you in advance,

Jorge

Correct, the perforation lengths inevitably change with changing mesh geometry. Moreover, in Free&Moveable settings the Multilayer Well (MLW) is always within the saturated zone. If you want to avoid changes in the perforation lengths I suggest to switch either to the Phreatic approach or solving the Richards equation.

Thank you for your reply.

I'm using the [i]Phreatic approach[/i] now but convergence seems harder to achieve, I think it is probably because I have a variable thickness in my layers and the value for the [i]residual water depth[/i] is constant. I've tried with an average value for the residual water depth so it is adequate both for thin and thick elements; the simulation does converge but the solution seems to be inaccurate. The [i]Free strategy[/i] allowed the model to converge with the solution appearing to be more accurate, so I think it will be great to use it with [i]Multilayer Wells[/i]. Definitely, solving the Richards equation will give a more accurate solution but I have no information about some of the parameters required.

A quick suggestion: Did you plot the water table (0 kPa-isoline) on several cross sections? If the shape of the water table does not look reasonable (e.g. zigzag shape) you may check if a finer vertical layer discretization helps.

Thanks again, Björn

A finer vertical discretization helped indeed. The accuracy of the solution was not seriously affected, it just showed one "hanging" water body of small dimensions above the main phreatic level, which I was able to identify plotting the water table in 3D view (zero isosurface).

In addition, I've used another strategy to run the simulation using the Free approach: I set Discrete features (Hagen-Poiseuille) in the same join edges as the original Multilayer Wells then I set the the node in the bottom of each well as a Hydraulic head B.C. (instead of a flux B.C., which I believe is the approach used in Multilayer Wells) with minimum flow constraint equal to the time series of pumping in each well (through time series ID) and maximum flow constraint equal to 0 m3/s. In this way, the wells will stop pumping when the water level goes down below the bottom the perforation length.