Saltwater Intrusion numerical dispersion

Hello everyone

I'm simulating a saltwater intrusion on a simple model which is set to be bounded by a hydraulic head BC by the western side and a saltwater head by the sea boundary (to the East). As far as regards mass transport boundaries, I have set a constant mass concentration BC by the sea boundary (35000 mg/l) and I am setting a saltwater inflow in the western boundary, only in the last two model slices (as my saltwater intrusion should cross all my model in the last two slices). However, when  starting simulating the mass transport I can't avoid oscillations and negatives values in the concentrations at the edge of the intrusion. Moreover, even if I set a saltwater inflow only in the last two slices, salt spreads all over the system by the end of the simulations. The mesh was set up to maintain a Peclet number of less than 1 given the dispersivity and I have also ensured that it had <5% obtuse triangles.  I've tried different upwinding techniques, but none of those remove the problem entirely either.  Also, I have tried a more gradual introduction of saltwater by varying the western concentration, but as soon as any saltwater enters the system negative values appear...

Any suggestion that can help to avoid these problems?
Thanks!!!

You will most probably not be able to completely avoid negative values (even if the Péclet criterion is fulfilled), as FEFLOW will show them as they occur in the solution. The 'art' is to find the balance between stability and possible mesh resolution. From your description, the spreading over the entire system does not seem to be caused by numerical oscillations, but by the boundary conditions you have chosen. However, this is hard to judge from only a few lines of description.

It's the information I'm looking for as well.

[quote author=CP link=topic=21705.msg28594#msg28594 date=1572877389]
Hello everyone

I'm simulating a saltwater intrusion on a simple model which is set to be bounded by a [b]hydraulic head BC by the western side and a saltwater head by the sea boundary (to the East)[/b]. As far as regards mass transport boundaries, I have set a [b]constant mass concentration BC by the sea boundary (35000 mg/l) and I am setting a saltwater inflow in the western boundary[/b], only in the last two model slices (as my saltwater intrusion should cross all my model in the last two slices). However, when  starting simulating the mass transport I can't avoid oscillations and negatives values in the concentrations at the edge of the intrusion. Moreover, even if I set a saltwater inflow only in the last two slices, salt spreads all over the system by the end of the simulations. The mesh was set up to maintain a Peclet number of less than 1 given the dispersivity and I have also ensured that it had <5% obtuse triangles.  I've tried different upwinding techniques, but none of those remove the problem entirely either.  Also, I have tried a more gradual introduction of saltwater by varying the western concentration, but as soon as any saltwater enters the system negative values appear...

Any suggestion that can help to avoid these problems?
Thanks!!!
[/quote]

Hi CP,
I'm not sure what exactly the problem might be, but from the two sentences that I have bolded in your message, it seems that you're assigning a salt mass concentration on [u]both sides[/u] (why do you have saltwater inflow in the western boundary if your sea is to the east?). If that indeed is the case, then salt spreading across your entire domain is the expected result.

can i get any reference on how to assign longitudinal dispersivity?
it depends on scale, which means distance from source?