If you have access to OneMine as part of a member benefit, log in through your member association website for a seamless user experience.
|INTRODUCTION The majority of ground water flow models have been developed for large computer systems. Most of these operate under a batch mode of operation, that is, the data are input with a deck of punched cards. The advent of micro- and minicomputers over the last few years opens the possibility of adapting many of these large flow programs to the smaller computers. In general, this adaptation will require more computer time to run the program, but in many cases if the small computer is in-house, the overall cost may be less than that required for a large computer. There are also a number of peripheral devices that can be used to make the data input much easier. As an example, the use of a digitizer with a finite element or finite difference mesh reduces the data input time considerably; the use of a graphics terminal for display of the mesh immediately after it is developed shows immediately whether there are errors in the location of node points. DEVELOPMENT OF AN INTERACTIVE SYSTEM The finite element program, UNSAT2, has been used over the past several years by the authors for several ground water flow problems (Bloomsburg, 1977; Bloomsburg and Wells, 1978; Zahl and Bloomsburg, 1980). In the past these programs have been run on an IBM 370/ 145 computer under the batch mode of operation. One difficulty that has been encountered has been in determining the length of time step that must be specified for operation of the program. The problem is that if the time steps are too large, the program becomes un- stable and operation ceases. If this happens in the middle of the problem, the only alternative is to reduce the length of time step and repeat the entire run. This problem has been alleviated in other pro- grams by using a subroutine that determines the time step automatically. If the solution becomes unstable, the time step is reduced automatically and that portion of the solution is repeated. With an interactive system this can be done by storing the output pressures and restarting the program with the new output pressures. Under the batch mode of operation, there is an option in UNSAT2 that allows restarting of the program from the last point of solution, but this must be specified when originally starting the program. The interactive computer system that we use currently consists of a PDP 1 1 /23 with hard disk drive, CRT terminal, plotter, graphics terminal, digitizer, and printer. The computer has 256 K bytes of core storage and the hard disk drive will accept a 5.2 M byte disk. To run a large program on a computer such as this, some overlaying must occur; that is, subroutines are stored on disk and put into core storage only when they are actually needed for running the program. UNSAT2 required very few changes to run on this computer but some changes were necessary to take advantage of the interactive capabilities of the system. One of these changes consists of the development of a program called DBUILD to build the data file for running the program. This program calls for each piece of data by prompt statement and gives instructions for using the digitizer. The data are then entered in completely unformated form. The node points must be entered in order and the coordinates are determined automatically by the digitizer. The pressure readings or initial conditions on each node point are entered through the terminal. All element information also is entered through the terminal. After DBUILD is run the mesh may be displayed immediately on the graphics scope whereupon any error in node point|