dewatering - strange behaviour

[matevz]

Hi all,

i am having some problem with modeling dewatering of excavation pit. I created a simple 4 layer model (with flat borders between layers). Hk for eac layers are as follows, 1e-3 m/s, 1e-6 m/s, 5e-3 m/s and 5e-3 m/s. Layers 3 and 4 are the same i just divided them, so that i coul model the depth to which sheet wall is designed.

I modelled dewatering with several (around 30) pumping wells. Here comes the problem. When setting pumping rate to -0.007 m^3/s(unit) i get really small drawdown. But if i set pumping rate just a little bit higher, like -0.008 (unit) i get all DRY CELLS in layer 1 BUT AT THE SAME TIME i get all FLOODED cells in layer two?! Attached is the file with taht solution (with -0.008 m^3/s pumping rate for each well., coverage ss). Does anyone have any idea why this ahppens?

Regards, m

[matevz]

I am sorry but i can't attach any file biger than 29 kb...

[Sean Czarniecki]

Well, I can answer part of your question. When attempting to arrive at a solution, the water level fluctuates such that it drops below the bottom of cell(s) in layer 1 during an iteration and dries out the cell(s), therby basically turning those cells off permanently (unless you use cell rewetting). Now, when it fluctuates back to a head above the bottom of layer 1 and converges to a solution at that elevation, the layer 2 cells show up as flooded, since these cells are now considered the top cells in the model. Note that once a cell dries out, it often has a ripple effect on neighboring cells. You want to do your best to minimize dry cells from getting created, so if you are able to adjust the elevation of your layers in some way to limit that from happening, that is what you should do.

[matevz]

Hi Sean,

thanks for explanation. Coudn't i simply use Cell rewetting? What do you mean by adjusting the elevation of layers?

Thanks, m

[Sean Czarniecki]

You can definitely use cell rewetting, and as your model sounds fairly simple, it may work just fine for you. Cell rewetting can often also result in non-convergence as cells get turned on and off.

Regarding adjusting the elevation of layers - it sounds like the bottom of your layer 1 is right around where your water table wants to be. If you don't have strong data showing you the elevation of the bottom of layer 1, lower it a bit so that the cells stay wet.

Another thing to consider is your starting heads. The closer you can get them to your final solution, the less problems you will have. For example, if you put your starting heads at the ground surface and they are expected to end up around 10 m below that, the first iteration will often swing the water elevation well below the 10 m, then it will come back up above the 10 m....and iterate until it converges on the final elevation. That is how the cells can dry up, so if you instead make the starting heads a certain thickness above the bottom of layer 1, that may be a better starting head array to use and could limit your cells drying out.

[matevz]

Hi Sean,

thanks fot the help. It works now. Do you have any idea how storage coefficient and specific yield is calculated in GMS?

thanks, m

[Sean Czarniecki]

Are you asking if I know how they are used? ...because you enter them into arrays for transient runs.

[matevz]

Hi, no i mean i can't find the actual equations on which these paramateres are calculated in GMS...

[GMS User]

I suggest you to read cap. 6.5.1 Geotechnical Problems: inflow of water into an excavation pit, from Processing Modflow PDF (just google it, pm5.pdf).

[matevz]

Thanks GMS user...i'll check it out.

m