Introduction

In this issue we turn the Spotlight on TWODAN 5.0 by Fitts GeoSolutions. TWODAN is a two-dimensional analytical element ground water flow model that runs on Windows95 (Win95) or Windows NT. Hardware requirements for TWODAN are minimal: 4 MB of hard disk space and 8 MB of RAM. Fitts GeoSolutions sells TWODAN for $495 ($255 for qualifying educational institutions).

How we tested

We tested TWODAN on three Win95 machines: a Micron Millenia XRU 233 Mhz Pentium II with 128 MB of RAM; a Packard Bell R500 200 Mhz Pentium with 42 MB of RAM; and a Dell Latitude XPI laptop with a 133 Mhz Pentium and 32 MB of RAM.

We implemented and ran a variety of test problems. One two-dimensional test problem of our own design involves a dolomite aquifer overlain by glacial tills in the eastern region of the model domain and glacial outwash in the central region. A full description of that test problem is available on the World Wide Web at http://www.ems.psu.edu/Hydrogeologist/softspot.htm. Another test problem was based on the hydrogeology of the Rocky Mountain Arsenal in Colorado. We appreciate the assistance of Dr. L. Konikow of the U.S. Geological Survey in designing the specifications for that problem. Reviewers also used flow problems that they were personally familiar with to evaluate TWODAN. One reviewer had extensive prior experience with a DOS version of TWODAN; the others had no prior TWODAN experience. The reviewers collectively spent about 32 hours evaluating TWODAN.

What we found

TWODAN is a flexible and useful ground water model for both teachers and professionals, no matter what their modeling background may be. TWODAN can simulate one or two hydraulically connected, confined or unconfined layers. Each layer may have multiple isotropic heterogeneities, input as closed polygons. These may be nested inside of one another. Impermeable or resistant boundaries such as slurry walls may be modeled using open or closed line segments. Steady-state and transient wells are supported. The author of TWODAN states that it is also capable of optimizing discharges of steady wells based on specified head and aquifer discharge conditions, as referenced in Ground Water 32, no. 4 (1994), although we did not test that functionality. Spatially uniform or variable infiltration and leakage can be simulated. Constant head and constant discharge linesinks are supported. A uniform flow across the aquifer can be input at a user-specified rate and angle.

TWODAN€™s Windows interface facilitates easy and rapid user input of model domains, such as heterogeneities and boundaries, through on-screen digitizing. On-screen display shows those areas, AutoCAD drawing exchange files (DXF) you may have imported to use as base maps, as well as contours of calculated head, potential, or stream function. Pathlines from user-defined lines or circles extend either upstream or downstream, with arrowheads plotted at user-specified time intervals. Output to a wide variety of windows-supported devices, the windows clipboard, ASCII Surfer grid files (GRD) and DXF files is supported. TWODAN also will generate ASCII lists of contour positions and pathline data.

What we liked

TWODAN€™s strengths lie in the ease and speed of model construction, and the ease of model alteration. Using TWODAN, one can arrive at a solution to even relatively complicated two-dimensional ground water flow problems in an afternoon, compared to the days or weeks commonly required for more sophisticated finite element or finite difference numerical models. The hydrogeologist can put together a preliminary model in a few minutes, and test specific hypotheses with nearly instant feedback. Even a die-hard MODFLOW user will find TWODAN to be a valuable tool for preliminary two-dimensional model development, fine-tuning the conceptual model for a specific system, and trying different recharge rates or boundary conditions, for example.

This flexibility and rapid feedback also make TWODAN an excellent teaching tool for introductory hydrogeology courses. Students can more readily grasp concepts such as the influence of heterogeneity or boundary conditions on hydraulic potential, because they see the cause-and-effect relationship drawn for them on the screen.

TWODAN€™s implementation of the Windows Graphical User Interface (GUI) is robust and gives access to all of the model functions, eliminating the need for the user to formulate input data files. The main TWODAN menu screen contains controls or buttons for the critical operations related to model development and execution, eliminating the need for the impatient user to hunt around. The main screen also includes a plot window where the components of the model are displayed. The user determines which of these components are displayed. Elements can be added very quickly, using the mouse to digitize locations. The user is reliably prompted to fill in required input, and the necessary cells are highlighted. Input data descriptions of model elements, aquifer geometry, and hydrogeologic properties are entered in spreadsheet-like cells. When the cursor passes over the data cell, the user is supplied with context-sensitive information about the appropriate data and any restrictions to the range of values that might be relevant for the data type. For example, the user is informed that the minimum thickness and hydraulic conductivity for the required lower layer are greater than zero. Fourteen supplied example problems illustrate the capabilities of TWODAN.

The cost of ownership of TWODAN is moderate, and the cost of learning TWODAN, in terms of hours spent to acquire mastery of the fundamentals, is also moderate, but variable. The variability arises to a large degree from the modeler€™s initial conceptualization of the flow problem. We believe that additional time spent using TWODAN will yield progressively greater rewards as the user€™s TWODAN skill and familiarity levels increase.

The 34-page hard-copy documentation uses a tutorial approach, with some background material describing the mathematics of the analytic element method (AEM), but is not a reference source. The online help file is fairly extensive. Major relevant AEM books and journal articles are cited. Although only one user called Fitts GeoSolutions for tech support, he found that support to be outstanding.

What we didn€™t like

There are still some difficulties with the ergonomics of the modeling interface. Two separate main menu items are used for input of model data files and plot setting files. There is no familiar File menu item. Consequently, the Exit menu item does not appear underneath the nonexistent File menu, but rather at the end of the main menu bar. The nonstandard paradigm (if such a thing is possible) can lead to some misunderstandings. For example, the model data and plot window settings data are stored in different files. This gives the user greater flexibility in generating multiple plots showing different aspects of the same data, but, because there is no default plot view, loading and solving a data set without loading a plot data set can generate an error condition, causing termination of the program. Of course, once the user gets used to this, there is no additional problem. There is no windows dialog for altering printer selections and settings. TWODAN will use the printer set as default in Windows.

We experienced some minor difficulty with output file generation. Printing to a Tektronix Phaser 550 yielded only a two-color plot which did not use the on-screen colors. There were some difficulties with the output of Surfer grid files of the Rocky Mountain problem. Some of the data values were large and nonphysical. (We suspect that they were associated with the rock outcrop areas.) These data caused the zmin, zmax part of the file header to run together and confounded Surfer, although it still read the file without difficulty. Once we fixed the line in the file header using an ASCII editor, Surfer correctly contoured the data. Although TWODAN produces plots that are very informative to the analyst, we will most likely use Surfer or some other graphical package for post-processing of output for reports.

The method TWODAN uses to build a third dimension of varying thickness is cleverly done, but was not described well enough for one reviewer to master without technical support.

All reviewers reported difficulty with scaling of imported DXF files, even though these files were read by other applications without difficulty. We suggest that some upgrade to the import scaling mechanism, perhaps with user controls added, might be in order. We would also be delighted to see support for imported bitmaps as basemaps. Again, some sort of user control for scaling from pixels to model domain would be required. We experienced some difficulty with panning and zooming during digitizing mode, causing run-time errors that shut down the application with data loss and work interruption. We would discourage the developer from using yellow in displays, as these are often difficult to see.

Overall

One reviewer reported "TWODAN was fun to use, and this may be its greatest strength." We agree that there is substantial gratification provided by the instant feedback to changes in model design and addition of model elements. All of our complaints about TWODAN are of a relatively minor nature and easily remedied with minor-level revisions. TWODAN requires only a moderate investment of money and time to buy and learn, yet it has considerable power and flexibility, resulting in rapid solutions for two-dimensional flow problems. Used either for problem solving or teaching, TWODAN is a valued and handy tool in our hydrogeological software toolbox.

Ratings

Reviewers were asked to rate TWODAN on a scale of 1 to 5 in the following categories. The results reported are the arithmetic means of the three reviewer€™s responses. The tech support rating is based on only one reviewer€™s experience.