Posted Fri, 11 Jun 2010 13:02:02 GMT by manu-al
Hello out there,

I have a problem in modeling the expansion of saltwater... first to my modell (feflow5.2):

I have time controlled disposal wells, without any casing, so the screen is about 300 m through seven slices of my model. In the Flow-boundaries I put the injection-rate (via timecurve xy) in slice 1, all other slices have a well at that special node, but without any injection rate.... I do use well bore conditions, so the well is linked and the flow is functional in all slices...

so far

but now I want to update this model for [b]transport of saltwater[/b]... In the problem class I`ve choosen "flow and mass transport",  "single species", so my Species is NaCl.... 

I´ve entered a background or better an initial concentration of about 13000 mg/l (this area is outside europe, the values are save, believe me) for the whole model area, one border is a coastline, there i`ve put an average value of 45000 mg/l...

now for the disposal wells I have checked serveral different input methods and alternatives. first of all the water injected to the aquifer has a TDS of about 150000 mg/l, so the concentration should rise and rise and rise in the ambience of the wells but what happens: from slice 3 to 7 there is a negative concentration given after the run (-12000 mg/l for example)... in one slice (sl2) there is a little positive fringe, showing that some of the high concentration water is really injected, but in all layers below the concentration is < 0 mg/l !!??? ...  can somebody think of a soultion?? i cannot explain it to myself..... i would think the high-concentrated water should sink through the aquifer, shouldn´t it? by the way, this should be the aim of my model, showing where the flowpaths are going from this injection points... but density- and concentration-coupled...

or somebody knows a tutorial 'bout that? [move]gerne auch auf deutsch ;)[/move]

regards!
AL
Posted Mon, 28 Jun 2010 08:44:38 GMT by Denim Umeshkumar Anajwala
What you observe seem to be typical numerical oscillations due to a too coarse discretization. Please try to refine the model vertically and (if also necessary) horizontally to avoid these.

FEFLOW does never suppress calculated negative concentrations so that they can be used as an indicator for too coarse mesh geometry.
Posted Wed, 30 Jun 2010 09:12:11 GMT by manu-al
Thx a lot, yes this problem seems to get better, the more refined the mesh gets.... time to calculate is more than one week I guess, so better in 2d...

You must be signed in to post in this forum.