adacovsk

I think it's broken. I can't compile anything, even my previous plug-ins.

I found this but it's not very useful...  :'( http://feflow.info/html/bugls_70.pdf

No, they aren't. Values are only changed once you assign the expression. See this thread: http://forum.mikepoweredbydhi.com/index.php/topic,2610.0.html

A good question to ask yourself is how you derived your transient flux. The problem could be that the average of those transient values over the entire transient simulation is different than the steady-state derived flux values. The averaged transient values and the steady-state should be similar if you want to have reasonable heads (typically). This is especially important is smaller-scale models.

Another problem might be that the top of the model is acting like a capillary barrier in unsaturated-saturated flow problems: dry season may lower your hydraulic conductivity so much that you get lots of surface head once you have a wet season.

Another problem might be that the larger fluxes over your model may be too much if your model is a relatively small volume, such that it cannot store all the additional water.

Hope this helps.

Adam

To build on this thread: I've tried exporting data from the Subdomain Boundary Period Budget ... See attachment. Problem is that there's no easy way to work with the delimiting in this file...

I started a thread for suggestions on projecting free XYZ points to nearest slice:

http://forum.mikepoweredbydhi.com/index.php/topic,2596.0.html

Ahh, I understand the problem now. I was not aware that it wouldn't update every timestep. If this difference is not preserved, you should consider writing a plug-in through the FEFLOW API.

Here's what I'd do:

Pre-timestep:

IfmSetBcHeatValueAtCurrentTime(pDoc, your_node_plus_5000, IfmGetBCHeatValue(pDoc, your_node) + 5000)

To note: the node index starts at 0 instead of 1, so add 1 to your node to reference the right node.

If you know the temperature of the other node throughout the period via a boundary condition you could set up two time series, one for each of the nodes.

I think this should work?:

Current_new = Current(No_n(111)) + 5000

Where No_n is the node number and Current is the property/BC.

The initial condition remains the same throughout: in unsat-sat flow conditions you'll notice this the most: you get parameters that are optimzed on your current head distribution. If you run a steady-state with a different head distribution with your calibrated parameters, you're going to get a worse calibration.

[quote author=S Heermann link=topic=2597.msg6142#msg6142 date=1444149169]
I recently used Python to extract parameters and create tables summarizing parameters for a range of "zones"-- it works great.  I would like to incorporate it into the parameter estimation process, but I haven't quite figured out how yet.  Any thoughts?
[/quote]

You would need to do is create element selections to define your parameter zones. Not sure if there's an easy way to do that. If you're trying to adjust parameters outside what FePEST has, here's a thread courtesy of Ludwig that has some groundwork on it: http://forum.mikepoweredbydhi.com/index.php/topic,2047.0.html
Not sure if this answers your question.

Hope this helps!

Adam

You can try using Python instead :)

I think MS Visual Studio is still available, although not express... Should work the same? I still use the 2010 version...https://www.visualstudio.com/en-us/products/visual-studio-express-vs.aspx