Denim Umeshkumar Anajwala

After discussing this matter at FEFLOW Services we have found the reason of the problem:
Additional to the [i]Power Function ID's [/i]the user has had defined [i]Time Constant Data[/i] too (in the <Assign> Database > Data regionalization dialog).
Therefore, as Time Constant Data were defined, FEFLOW uses these values for the regionalization, not as expected the Power Funtion ID's.

In a transient model, it should be possible to assign power fxns as BCs the same way as with const. values in a steady state model. The field you link would be the power fxn ID. You pbly have to make sure the power fxns get defined in FeFlow first, otherwise it might default to a const. value. I'm pretty sure my colleague did that recently .... Chris

Hey Zebra,

I use C++ myself. Isn't that "state of the art" as well? ;-)

Chris

Hey Zebra,

I didn't know that was possible yet. I thought, that was part of the new FeFlow 6 which won't be available for another little while from what I have been told.

What interface are you using to write IFMs in Python right now?

Chris

Hey Steve,

I just finished a great mesh building exercise myself, and here is an idea that I used to get refinements into my model without having to do it manually in FeFlow/GridBuilder.

Define the refinement in your GIS and import the already refined super-elements into FeFlow.

I know, that sounds easy in theory, and is a little tricky in practice. But it works.

Here is roughly how I did it.
1) Defined my model domain
2) Defined "cut-lines" which were later used to cut the domain into super-elements
3) Refined the cut-lines where necessary
4) Cut the model domain into super-elements (which preserved the refinement)
5) Imported those super-elements to FeFlow via the super-mesh import filter

In step 3) I used a script that iterates through selected lines and adds extra vertices along the lines to define the refinement. That's probably the trickiest part. I can provide the script for the GIS Manifold.

How does that sound?

Chris


PS: The screenshot shows part of the mesh. You can see how there are different refinements defined along the model border and the red vs. the blue lines.

I don't have any experience with TMesh, but could you add line add-ins to define those wells? Would that force TMesh to generate nodes and elements with the desired discretization? This is just as guess and might work, or totally blow up the mesh generation process.

Good luck, Chris

I agree with Zebra,

this is pretty well documented in the online help. Take a close look at that.

Make sure you got your folders set up properly: you need a sub-folder "femdata" and "results" within the folder with the batch file.[font=Courier]
    FeFlowBatchFolder
       + BATCHFILE.BAT
       |- femdata (with your FEM files)
       |- results (for your dar/dac files)[/font]

This is what a line in my batch files typically looks like:[font=Courier]
feflow -run             -work "C:\FeFlowBatchFolder"  -log "feflow.log"  -dar "model.dar"  "model.fem"
feflow -run -hide -exit -work "C:\FeFlowBatchFolder"  -log "feflow.err"  -dac "model.dac"  "model.fem"[/font]

For details on the different options, type "FeFlow -help" in a DOS window and you'll get a summary of all the options.

For this to work, the FeFlow folder has to be included in your PATH variable (which I think is the case by default).

Any folder or files names containing white-spaces (blanks) need to be enclosed in quotes! I do that for any file and folder name, so I don't forget by accident.

Good luck! Chris

[quote author=dstone link=topic=406.msg920#msg920 date=1209144551]
The FEFLOW online help indicates that a 3D (X-Y-Z-F) export option should be available for outputting nodal/elemental values as points in the Plotting Output Options of the Data View, Operation, and Export dialogue box for 3D models. However, this option does not appear for my 3D model; it only allows 2D output for the current slice. Is there some other switch that needs to be set to enable this feature? I encounter the same problem whether I am trying to export from the Problem Editor (e.g., Flow Materials, using the "Special" option) or from the Postprocessor (View Results).
[/quote]

Dan

If you select ASCII databases (XYZF) (*.dat) as the output type, you will be given another dialog where you can choose all slices (or layers) OR only the current slice (or layer).

I am not aware of any special switch or setting which controls this behavior. I tried setting up a mock 3D, 1 layer model, and it asked me if I wanted to export data for all layers. A 2D model only allows export of XYF data.

Dwaine Edington

Hey Pejman,

yes, you will need a C++ compiler (I don't think a C compiler will do). However, Microsoft started providing some of its software for free, as what they call their Express versions. You can download Visual C++ Express from

http://www.microsoft.com/express/download/

I haven't used it (we use a licensed version of Visual Studio .Net), but I tried SQL Server Express before and that worked fine.

The trickiest part will pbly be to get the compiler set up correctly. There is a section in the IFM documentation, that describes it, but you might have to adjust that for using Visual C++ Express and the newest version of FeFlow, which saves some shared files in different locations (see my post in "creating module with published source code" in the "IFM and Filter Programming" section). If you can't figure it out (with help from the support team), let me know and I'll see what I can do.

Good luck, Chris

Hey Pejman,

personally, I would stay away from messing around with the FEM file. Once you've read the file specs, you'll pbly understand why.

The "wise way" that I'd rather suggest is using the InferFace Manager (IFM) to create your own data importers (and exporters, if needed). This can be done, so that these importers are called automatically before every flow simulation (PreSimulation) for example. All you need to do then, is manipulate the default input files for those importers externally. This, however, can be done in a much safer way, since you define the format of these input (ASCII) files yourself, and hence you know exactly how to modify the data.

We've used this approach with SensAn, a Sensitivity Analysis tool, that's freely availbale as part of the PEST package. See attached flowchart for details.

Coming from MatLab, you pbly know enough C syntax already. That plus a C compiler is all you need really.

Chris