I Worked through the FEFLOW Demonstration Exercise so You Don't Have To

 Here’s my review of FEFLOW 6 based on the Demonstration Exercise Tutorial.  First of all, a disclaimer: before today, I had virtually no practial experience with finite element grids.

The demonstration exercise called for the use of FEFLOW Classic. There is also a FEFLOW Standard GUI, which is probably a bit more graphically sophisticated. During the demonstration exercise, you work through a simple groundwater modeling problem with two wells, three layers and a mass transport component. At first, FEFLOW Classic appears blocky and old-fashioned but I quickly found it to be uncluttered and easy to navigate. The lack of over-stylized graphic effects was actually a relief on the eyes. That said, my HP mini had issues with it. Several of the forms got cut off at the bottom and were completely unusable from the HP mini (I didn’t test out FEFLOW Standard…it may be free of this problem.) I also managed to crash the program a couple of times by hitting the wrong buttons.

The biggest difference between finite difference and finite element models is that finite difference modes are constructed on a rectangular grid where the node is located inside the grid cell, and finite element models are constructed on a mesh (in this case triangular) with the node located at mesh junctures. In FEFLOW, the construction of the mesh is automated. Option mesh refinement is available around model elements and the model border. The mesh can then be refined further around areas of interest. The basic groundwater modeling concepts are, of course, consistent across modeling platforms. The processes of assigning boundary conditions, interpolating slice surfaces, positioning wells, computing mass transport, and importing observation data will be familiar to those savvy with other software. Assigning aquifer properties is straightforward, with values either held constant or interpolated from point data.

There are some definite advantages over GUIs.  First of all, the problem editor menu has been purposefully structured such that the most efficient way to build a model is to start at the top of the menu list and work downward. I love the idea of the GUI helping you to move through the problem in the most logical manner. Also, in demo mode, you can save small models (max of 500 nodes per slice and max of 5 slices).  The three-dimensional model projections were clean and intuitive. On the whole, I think FEFLOW classic is very easy to work with and it seemed very easy to use overall.

The biggest disadvantage with finite element that I can think of after working through this tutorial is that it would make programming more difficult. The nice thing about MODFLOW is that cell properties can be displayed and easily edited as a grid of numbers. How easy would to be to look at the raw FEFLOW node database and understand what’s going on? My company, GroundwaterGo, is focused on the automation of model construction, which I think would probably be more difficult with finite element than it was with finite grid (I’ll try, if I can find an open source version). Working with this type of mesh would also be more challenging in GIS.