Please wait...
×

Error

Posted Fri, 03 Aug 2012 02:12:35 GMT by tkershaw
Hello,
To follow up on a previous post regarding a vertical model vs. horizontal model for modeling seawater intrusion (thank you Pete and Peter for your input): I am pursuing a simplified horizontal model but having trouble making it behave predictably. I have a simple 4 layer model with a sloped seawater boundary represented by the top first slice and with the slices below following the upper sea-slope with a minimum distance below in order to abide by the FEFLOW rule of having each layer extend through the entire model boundary. My problem is that even when I leave [i]all layers with default material property parameters[/i], their landward saltwater migration seems to be quite segmented for each layer as shown in the image attached. Also there is not much of a seawater intrusion toe with more migration on the lower elevations. The run seems to take quite a long time (12hrs for a 6.5 day run) so maybe this migration stepping (and toe behavior) will even out after a longer runtime but I am suspicious.

My questions are as follows:

1. Are there certain material properties that need to be set for these very thin stacked layer sections that would promote a more uniform mass migration? I will eventually want migration stepping behavior when I add aquitards but is is not intuitive when the layers have the same material properties.
2. Should I only put seawater boundary conditions (saltwater head: 0m, mass concentration: 35000mg/L, min mass flow constraint: 0g/d) on the first slice's nodes or on the slices below where the thin layer sections are very close to offshore?
3. My model is very simple structurally yet seems to have a very long runtime. Is this most likely from the thinly layered sections?

In case there is anyone interested in helping I have attached the model file and would love if anyone had the time to look at it to see if there are any egregious oversights.

Thank you very much!
Posted Fri, 03 Aug 2012 09:33:36 GMT by Giovanni Formentin
If the vertical scale of your model is the same of the pictures you uploaded in your previous message (approx. 400 meters), one thing I'd suggest is to dramatically increase the number of layers, even I can't tell if that alone would help recovering those very negative concentrations you now find in your run (I haven't opened the .fem yet).
Bye,

Giovanni
Posted Sun, 05 Aug 2012 23:35:09 GMT by tkershaw
Thanks for the reply Giovanni. The images shown in the original post are with a 10X vertical exaggeration. The vertical scale is 316m and the horizontal scale 12000m X 300m. I would agree that more layers would create more resolution and possibly clean up the concentrations, but i first wanted to get a simplified model to exhibit predictable behavior and it is already quite computationally intensive with only 4 layers. Attached is a 1X vertical exaggeration for comparison.
Posted Wed, 08 Aug 2012 07:18:45 GMT by Blair Thornburrow Groundwater Modeller
I would suggest taking the following approach:

1. Remove the sloping layer elevations and use layers of equal thickness - I think the thin layers are causing problems
2. Add more layers (try maybe 10 m thick layers?)
3. Refine the mesh globally, and locally near the coastline (you want refinement where the salinity gradient/saline interface will reside)
4. Smooth the mesh
5. Assign a zone of BCs (to differing extents in each layer) representing the model elements that exist above the sea floor
6. Change the upwinding option to shock capturing (not sure if this will make much difference through)
7. Possibly limit the maximum time step if you get instability issues

Good luck

You must be signed in to post in this forum.