The Java Virtual Wind Tunnel: Discussion

Version 1.0B1

The way the Java Virtual Wind Tunnel helps students learn is by providing a new way to interact with and explore a flow. At the undergraduate level, the applet helps students build an intuitive understanding of basic flow phenomena. At the graduate level, it helps them understand the power and limitations of CFD.

All of the examples included here are of the "Ni's Bump" geometry. Version 1.0Beta1 will accept arbritary grid geometries, so the simulation could just as well be of a rocket nozzle or an engine intake. With some modification, this program could also simulate an airplane wing. The flow geometry and initial conditions are stored in a data file on the Web server and downloaded by the applet on startup. Since all the data sits on the server, the geometry can be easily changed from week to week, or even day to day. One week, it might be a simulation of a rocket nozzle. The next week, it might be a wing in flight. From the student's point of view, it would just be a matter of typing in a URL and seeing what was up for that week.

When I entered the Java Applet Contest, I came up with a few examples of problems which might be explored using the Virtual Wind Tunnel.

And a more amusing problem:
Again, this program is a fairly simple prototype. But it opens up new possibilities for teachers and students alike.

Finally, there's one more question worth addressing: Why Java? Why not C, or FORTRAN, or any number of well established programming languages?

Fundamentally, it's a question of cost. In the late 1980's, the MIT department of Aeronautics and Astronautics developed a package of specialized educational programs for teaching things like fluid mechanics. The package was named Todor, and ran on the workstations supported by what was then MIT's Project Athena. Today, that package is obsolete. The workstations it was originally designed for are gone. The BLOX graphics package it was based on has diappeared, and no money is available to upgrade Todor to work with modern computers.

You see, the major problem with educational software is that it is very expensive to maintain specialized software in a distributed, multi-platform environment. Every time the university buys a new platform or even upgrades the system, the software has to be reinstalled and rebuilt. The Virtual Wind Tunnel, on the other hand, can be run on a variety of platforms simply using a Java capable browser like Netscape. This makes it much simpler to develop and maintain in a multi-platform environment. And, as this applet shows, Java is a flexible enough and fast enough to build numerical applications. Since hardware will only continue to grow faster with time, more complex and realistic simulations can be built for the next generation of chips and processors.

I am interested in encouraging the development of educational software for colleges and universities, and I can provide you with a copy of the source code for the Java Virtual Wind Tunnel if you would like it. Just send me email at

June 1996
David Y. Oh / Computational Aerospace Sciences Laboratory, MIT /