uqjzhang

Is there a way to quantify the numerical error in a trainsient simulation?

I know in steady state run, you can actually see the error and convergence chart.

And in MODFLOW, people like to use the discrepancy error curve to justify the accuray of a model run?

The topic is a while ago, anyway I share my walkaround here. It's not perfect but it's super easy.

1. select the element;
2. assign a big value (e.g. 1000) to an unused 'Material properties' (e.g. Out-transfer rate); assign to User Data if nessecery;
3. switch to node selection mode
4. use Select Nodes Expression Editor and put 'TRAFOUT > 999' after 'where' and apply

This approach will probably miss some nodes at the element boundary so some manual work will be needed.

Cheers
Ji

Hyped :)

Dear FEFLOW users,

What is the most convinient way to extract water table data out of Head and pressure and visualize it?
I know H = P/rho*g +Z, but what next?


Jim

This happens to me all the time. It behaves quite random, sometimes freeze, sometimes no problem.

I am guessing its related to to the size of the problem and frequency of you result saving interval. I feel like if you run a big transient problems (millions of nodes/elements) for like 2 years and you have your time steps size for 7 days and save every time step results, you will result in a huge fat dac file. And each time you make actions on the charts, the program has to retrieve data from this huge fat dac file, and it freezes.

I suggest you try to limit the output intervals when you run your model. For example, if you do calibration, you only need to output those dates that have monitoring data. If you already have a huge fat dac file, you can open it and "save as", then only select a few important dates and save it in a reduced healthy dac file.

I hope this helps.

Thanks Blair, I tried the constraints way, ant it worked! For those who are also interested on this, it is really simple, here is a picture I made for the definition of this trick.

Say that I need these nodes on fixed head (1st kind) for 1 year and cut them loose after then, How may I define that in ONE model run?

Thanks

[quote author=Julia Mayer link=topic=1344.msg3313#msg3313 date=1355385549]
Hello Ji,

this functionality is already available in FEFLOW 6.1. The smoothing operation will only shift the currently selected nodes and honor the original position of nodes which are not part of the current selection.
An effective workflow is to select all nodes along polygon borders/lines etc. which you would like to keep at their original position and then invert the selection before you start the smoothing operation.

Julia
[/quote]

Thanks Julia. I just read about the new features of v6.1 and found something called "selective mesh smoothing", should be the new function I am hoping for.

This raised another issue:

I just download v6.1 and installed on my computer, now I have a license issue. We have two network licenses (unlimited) purchased in 2010. When I connect the server and the licenses are still there, but I could not apply. The license version is still v6.0x. I am pretty sure our licenses are unlimited and should not be limited to a certain FEFLOW version.

Could you please advise me how to solve this problem?

Thanks

Ji

Dear FEFLOW developers

Currently if I smooth the mesh, I will have trouble later on when using selection functions such as "select by map polygon" or "select by map line". It is natural (often important) to keep these geometries information in our final meshes. Smoothing the mesh is important to get a numerically-reliable model, but often times I have to skip this step just to keep a good presentation of my model output, e.g. to keep the river straight, to avoid the zigzap of the zone boudaries.

So it is really important for me to have a function that keeps nodes on polygon boundaries and/or lines UNMOVED when I smooth the mesh.

Cheers

Ji

Suppose we have a high-permeable unit (A) sitting on a low-permeable unit (B). Landslide is happening on Unit A, possibly due to the rainfall recharge cannot penetrate Unit B and accumulate at the bottom of A.

We are design several horizontal drainage to help with the dewatering during rainfall events. see the picture attached.

Please give me some advice on how to set up the boundary conditions on the horizontal drainages.

Thanks~~