Cornelius.K

Hello everyone!

I want to simulate the intrusion of freshwater via an infiltration pond in saline groundwater. Therefor I use the convective form of the transport equation (there had been some trouble with the dispersive form (
http://forum.mikebydhi.com/index.php/topic,2276.msg5172.html#msg5172)) with the Free-Surface approach.
The Infiltration pond is represented with a well-BC and a 1st kind concentration-BC.

In the most cases this works fine. But in scenarios with a low Infiltration rate or a high hydraulic conductivity, which come along with only a low rise of the hydraulic head in the area of the infiltration pond, the freshwater flows under the denser groundwater in the toplayer. (see picture)

[img width=200 height=50]http://i1383.photobucket.com/albums/ah315/Mieszko2/Schenkel_zps31fd5e6d.jpg[/img]
[sup]Cut through 3D-Model:  Freshwater: violet; Groundwater: Red; Lines: Isolines of Nodal Darcy-Flux[/sup]

The density-contrast might be quite low, but considering the long simulationtime (>2000d) there should be a mixing with the water with lower density in the lower slices. Furthermore the isolines of the Darcy-Velocity and the hydraulic head are orthogonal to the surface, so that the mass should be transported with the flow. Thus this seems to be unphysical behavior of the model.

I tried vertical as well as horizontal refinement, a different solver and much more, but nothing fixed the problem.

Maybe someone encountered a similar problem? What couses this behavior of the model?
AS I said, this only happens with low rises of the hydraulic head.

Thank you
Cornelius

Hello everyone,

I’m trying to model the Infiltration of water with a low concentration into the top layer of an unconfined aquifer (high concentration).
I tried following versions/scenarios:

1: 4th kind Flow BC (well) and 4th kind Mass transport BC (well) with the divergence formulation of the transport equation.
2: 4th kind Flow BC (well) and 1st kind Mass transport BC (Dirichlet) with the convective form.

In both cases I tried to model it with a free moveable top-slice. But there were some problems regarding the results:

1:
The Mass-BC-flux (Rate Budget) is correct at the Inflow-Area. Unfortunately mass is accumulating at the border of the zone in which the hydraulic head rises (picture1). This causes unphysical concentrations (higher than the ambient concentration (3000mg/l)). In picture 2 and 3 one can see that the concentration of the top slice rises and the one of the underlaying falls in Areas where the slices get uplifted. So it seems that this is also connected to some Interpolation caused by the moving of the slices.

[URL=http://s1383.photobucket.com/user/Mieszko2/media/FEFLOW-Snapshot2_zps59ddaa09.jpg.html][img width=200 height=169]http://i1383.photobucket.com/albums/ah315/Mieszko2/FEFLOW-Snapshot2_zps59ddaa09.jpg[/img][/URL]
[sub]Picture1[/sub]
[URL=http://s1383.photobucket.com/user/Mieszko2/media/crosss_ts10_35d_zps38bb7f6c.jpg.html][img width=200 height=183]http://i1383.photobucket.com/albums/ah315/Mieszko2/crosss_ts10_35d_zps38bb7f6c.jpg[/img][/URL][URL=http://s1383.photobucket.com/user/Mieszko2/media/crosss_ts11_409d_zpsafcb5189.jpg.html][img width=200 height=183]http://i1383.photobucket.com/albums/ah315/Mieszko2/crosss_ts11_409d_zpsafcb5189.jpg[/img][/URL]
[sub]Picture 2 and 3: Rise of the layers causes uplift and accumulation of mass[/sub]

2:
The Well-BC (Flow) causes a rise of the hydraulic head. The ‘Point Source/Sink BCs’ in the Mass Rate Budget shows exactly the amount, the BC should have. Unfortunately there is also a big negative (Outflow) Dirichlet-BC of Mass in the Rate Budget (but no Dirichlet in the Flow-Budget) (See picture 4).
The Dirichlet-Mass-Flux converges to 0 as the rise of the hydraulic head converges to its maximum. It looks like that the rise of the top layer causes a (dispersive?) flux over the Dirichlet-BC – Is that possible?

[URL=http://s1383.photobucket.com/user/Mieszko2/media/Unbenannt_zps2eb85a81.png.html][img width=140 height=200]http://i1383.photobucket.com/albums/ah315/Mieszko2/Unbenannt_zps2eb85a81.png[/img][/URL]
[sub]Picture4: Inflow boundary there is also an outflow of mass over the Dirichlet-BC (should be only Inflow)[/sub]

So my final question: Is there a better way to model the Infiltration of big amounts of water and/or is there a function to limit the moving of the layers to the top slice?
I also tried to model it with the phreatic approach but the simulation time seems to be much higher and the results more unstable.

Thank you,
Cornelius