Alexander Renz

Hi Gemma,

The issue you have is related to the way how the mass flux BC are defined and about the choice between the two formulations of the transport equation (divergence form, convective form).

Please have a look at the attached document, it should explain the theory behind it.

If you have problems with the lateral flow, using a finer mesh should increase the accuracy of your model (Because less water is flowing laterally through your cells, where the 1st kind BC is prescribed).

best regards,
Alex.

Hi csena,

(1) you should use your choice (iii), the 2nd kind (Flux) boundary condition

To apply the recharge value in the way you described, several steps are necessary:

(1) first, you need the upper topographic boundary as a line (shp, ascii)
(2) second, you need a pow-file where you define a power function (time-varying function for the reacharge) for every node.
(3) You have to create a database-file where you specify the x,y-coordinates of these nodes and the respective number of the pow-function (syntax per line: x,y,pow-ID)
As a further prerequisite, the nodes of your mesh have to be identical with the nodes you described in these file.

You can then use the JOIN-tool with DATA-option to assign the Pow-ID's to the nodes. Choose the option "1D-Interpolation" in the upcoming dialog and set and set a reasonable high "Distance for non-interpolated data points". Load the pow-file in "Time-vaying function ID's" and load the database file with "Import powerfunction IDs". Finally click appply to start the assignment.

Annotations:

- Since the 2nd kind BC is mathematically applied to the edges of the cells and not to the nodes, there will be a minor "smearing" of your assignment between the nodes. Usually it should not matter.

- This procedure can be easier if you do not strictly define a pow-function to every single node. Alternatively, you can simply define the recharge for a small number of points on the border, and let FEFLOW the missing pow-functions by interpolation.

Hi Faby,

I looked up the FEFLOW 5.2 help, but could not find the discrepancy you described, yet. Could you please specify (Equation-Numers in WP, File and Equ. Numbers in FEFLOW Help Files) ?
Thanks a lot,
Alex.

when uising the "-iconic"-option, FEFLOW starts in interactive mode (with the GUI), but minimized. Usually, you prefer "-hidden" mode (since it yields the better performance).

if the head falls below the model domain, the cells around the well will surely be dry (negative pressure). Dry cells become less permeable, together with a constant pumping rate of the well this will further decrease the pressure/head.

If you trust the numerical result, you might want to switch of the wells if they fall dry; you can use boundary complemtary constraints (BCC) for this.

Using the line add-ins should work to force a maximum length of the edges at the well (equal to the distance of the fix-points). However, T-Mesh might decide to set extra nodes between the fix-points, causing even smaller element sizes.

You can also use the Triangle mesh generator, which offers a function as you ahve described.

Hi,

When using an unsaturated (Richards' equation) model, FEFLOW handles both the saturated and unsaturated zone in the same model. You can adjust the temporal parameters of the model in "Temporal&Control Data" Menu. You can also define time-vaying (power-) functions and use them for transient modelling of BC, material data, etc.

Dear Pejman,

when saving your results, you may choose the FEM-Interchange (ASCII) format instead of the default binary format. Afterwards, you can open the fem-file with an editor or do an automatic manipulation using external software.

The file format of the fem is described in the "file format library" in the FEFLOW helpm system (press F1 in master menu).

Note that the ASCII-fem-format is not as fast as the binary-fem-format; also the variables are stored with single-precision only (double-precision in binary).

Interior Flux-BC (as well as 2D-DFE's) are not suitable to model an impervious section inside the model domain (these are only additive terms in the flow equation, setting them to zero would be the same as applying nothing at all).

You can create a gap inside your superelement mesh to prevent flow across it, or apply a very low conductivity (~factor 1000) to this zones (at least two cells wide, otherwise the averaging of the material props can "smear away" these low conductive zones).

Choose an unconfined model in this case and set the respective constraints (top-/bottom-aquifer elevations). As long as the water table touches the top, the model will automatically handeled as confined. If not, the model switches to unconfined.