Posted Tue, 19 Feb 2008 14:57:53 GMT by Thomas O.
Hello all,

I try to run a model of an aquifer system located in a valley surrounded by two hills. This system has two different aquifers. These two aquifers are separated by a confining layer (see a cross section of the problem in the attached file):

1- The upper aquifer is unconfined and found only in the centre part of the valley;
2- The lower aquifer is semi-confined and is found both (a) in the centre of the valley [confined] and (b) on the sides of the neighbouring hills [unconfined]. Looking at the attached file, you will notice that the lower aquifer is made of two different materials.

The confining layer is only found in the centre of the valley.

I try to run this model in steady state for flow only, unconfined. However, the model will only reach convergence if I use the option confined aquifer. According to me, this is not how the system should be modelled, since the upper aquifer and the lower aquifer (on hill sides) have free surfaces...

When I tried to run the model unconfined, I assigned the following slices properties:

Slice 1: Free & Movable
Slices 2-5: Unspecified
Slices 6-11: Fixed (the water level in  the upper aquifer will not get lower than the top of the confining layer )

Any ideas on why the model will not reach convergence when run unconfined? Is there any changes you think I should do?


Posted Tue, 19 Feb 2008 20:55:33 GMT by Michael

I had this problem many times. For these cases I always run my model considering confined aquifer. Anyway, your hydraulic properties will ''guide'' the model.

Take care if you use Free and movable because surface layer will change of shape after a run. You will agree than the topography cannot move in the real world.

I suggest you to try several run with different settings with a transient setup. It generally help to reach the convergence.

Good luck,

Posted Wed, 20 Feb 2008 17:47:56 GMT by Thomas O.
Thanks for your answer, Michael.

I was wondering if setting the top slice as phreatic in the unconfined mode was something you sometimes consider?

I think that if you simply run the model confined, the saturated thickness of an aquifer becomes equal to its  total thickness, thus exagerating the transmissivity.

What do you think

Merci bien pour ton temps! ;)


Posted Thu, 21 Feb 2008 16:00:49 GMT by Michael

It surely depend if you run the model fully saturated. I don't know about the basic equation for case of a confined aquifer. Anyone know the difference between unconfined and confined in the mathematical solution of Feflow.


PS: C'est si évident que je suis francophone  :-\

Posted Thu, 21 Feb 2008 21:52:25 GMT by Thomas O.

pour le français... c'est seulement ton adresse email qui t'a trahi


Posted Fri, 22 Feb 2008 15:31:00 GMT by Alexander Renz
Hi Michael and Thomas,

The difference between the confined model and unconfined ("phreatic"-option) model is quite simple:

In a confined simulation, FEFLOW calculates only the Darcy-equation.

The phreatic model is quite similar, just that if the cells pressure (not hydraulic head) falls below zero - meaning that it becomes dry - its conductivity is reduced linearly with its saturation. To avoid that the cell becomes absolutely impermeable, the water depth in the cell is not allowed to fall below a certain residual water depth (this can be adjusted in specific option settings).

The "free+movable" approach is essentially a confined approach, whereas the top slice is constantly shifted to match the water table. It can yields excellent results for homogenous aquifers, but also has a lot of pit-falls and should be handeled with care, especially if having heterogenous materials, 2nd, 3rd or 4th kind BC (which are shifted as well) or if the aquifer is falling partially dry.

In these cases, "phreatic" is more convenient and robust. It also gives the possibilities to simulate several free water table on top of each other.

The mathematics are described in detail in the reference manual chapter 2 and Whitepaper Vol I, chapter 2

give it a try!

Posted Tue, 01 Apr 2008 23:44:32 GMT by mnovotny

Am I correct in my understanding that in the phreatic mode the overall conductivity, including that in the z-direction, is scaled by saturation or, at a minimum, the residual water depth?

I have had the experience that recharge (applied as inflow) seems to become trapped at the top of the aquifer if the layers below are dry, resulting in unrealistically high (10e8!) heads, and that this would occur if Kz were similarly scaled.  The same model runs fine in confined mode.


You must be signed in to post in this forum.