fixed versus unspecified

Gday Feflowers

I am fairly new to Feflow with a background in MODFLOW, and I need some advice, please.

I am creating a flow and transport model in FEFLOW for a glacial Quaternary aquifer system. I have excellent drillhole coverage (over 200 holes in a small area) and have been able to produce a fairly accurate geological model which generally consists of sand and till sequences that are individually 10-20m thick, so that the whole sequence is between 50 and 70m. The sequence is sand, clay, sand, clay, basal sand, basal clay. 

I have defined the problem as unconfined (phreatic) aquifers. When i select the clay layers as "fixed" layers, I get a good convergence for steady state, however, when I set the clay layers as "unspecified", the convergence becomes a serious problem, and the calibration is not good. Can someone give me some advice on when to set fixed versus unspecified for saturated and thick clay sequences in unconsolidated aquifers... the help option in the program gives a brief explanation but does not give specific examples of when to utilise each.

Thankyou

Tim

Hi mate,

the answer hides in your question.
whenever you wish to "fix" your layers you should use "fixed" option.
as i understand it, the moveable layers option was created for the upper layers of phreatic aquifers, in which groundwater table can shift up and down.  so, this option is actualy 1) raise your resolution and 2) allow you to change groundwater table without driyng the uppermost cells.
i don't think that there is a "rule" for your question, rather, you should (and did)used your common sense.

Elad

Indeed, there is no general rule. However, 'fixed' means that the layer below the 'fixed' slice will always be treated as fully saturated. So you should use 'fixed' only for layers which are always fully saturated. The most important application of 'fixed' is for 'free&movable' top slices, where the 'fixed' setting can avoid slice movement below the groundwater table.
If your simulation converges with 'fixed' settings while it does not converge with 'unspecified', probably the fixed layers are not fully saturated in the model. So 'fixed' would be wrong. For your kind of layering, the first question is if you expect only one or several (perched) water tables. In the first case, 'phreatic' and 'unspecificied' would be the primary choice (not talking about free&movable now). To stabilize, you can increase the residual saturation for the dry or partially dry cells on top (options - specific option settings). That will lead to higher conductivities in dry cells, less conductivity contrast and a more stable solution.
If you expect several perched water tables, the top of each layer with an expected groundwater surface should be set to 'phreatic'.

Thanks guys!

It appears that I have three distinct aquifers here. The two bottom sand aquifers are confined, however, the degree of communication between those layers is currently unkown as I do not know if the thick clay layers pinch out in any areas. We have recently installed bores and piezometers into all layers and will be test pumping and sampling soon to gather more information on the interconnection (or lack of) of the sand horizons.

Assuming that these two basal sand layers were fully saturated, then a fixed layer in the clays above and below them may be a potential solution. I am currently running PEST (SVDA in native format) for both the fixed and unspecified options to compare the results. The perched aquifer in the upper sand unit is limited in extent and ceratinly appears to dry up on occasion. I do not want slice movement to connect the water tables in the upper sand and the middle sand units and convert the middle sand aquifer into an "unconfined aquifer".

To follow this on. Hypothetically, if I wanted to depressurise the basal confined sand (which has a fixed clay layer above and below) with a gallery of pumping wells, would this impact on the stability of the model considering the aquifer may not remain fully saturated at some point in the future?

cheers
tim