In addition to launching its new software initiatives, the ANSYS users conference here also featured a panel of industry luminaries discussing the past and future of simulation.

This panel of pioneers included:

Swanson began the discussion by calling for "robust" applications that store the results of past iterations of simulation tests. "You don't want to do something over and over again; you want to change things and see what happens."

Harry Schaeffer replied that he'd been studying the clash between the CAD and CAE industries since 1982, and noted several trends. In order, they were: the emergence of network computing; the disappearance of corporate culture; the rise of desktop computing; the dumbing-down of engineering education; the emergence of degrees-of-freedom greed (as models got bigger and bigger); and finally the design of high-speed processors (just in time to handle those large models).

A question-and-answer section followed.

Q: With more automation of analysis software, is there a danger that simulation problems are not thoroughly thought out?

Schaeffer: "People who are very good with PCs and mice tend to be bad at tests, so more mentoring is needed."

MacNeal: "I think most of us up here are old enough to know not to trust computers."

Swanson: "Whoever created the application knows its limitations and boundaries well enough to constrain a person with less knowledge; that's crucial because we all know that no one reads the manual. Also, you must inspect the actual geometry to know what you've just analyzed; today's visualization tools help to do that."

Wulf: "Garbage in, garbage out; it's always been true. The emphasis should be on education."

Q: As analysis software becomes more sophisticated, how do you see the balance changing between physical testing and simulation?

Swanson: "I was talking to an auto industry VP today who said that when there's a clash between results, he goes down to the testing floor to see what went wrong. That reliability is increasing slowly. But still, I don't want to ride in anything that hasn't been tested."

Q: How will current hardware innovations affect the software?

Wulf: "We see a trend toward Linux clusters in CFD. But it's a challenge for us to put meshing and solving into a distributed environment."

Swanson: "We can't live in a 32-bit environment anymore. The coming of 64 bits is long overdue."

MacNeal: "The nature of hardware is to get overwhelmed by software."

Q: There have been many mergers lately, like MSC and EDS buying up competitors. Is that good for customers?

Swanson: "It's OK as long as you don't get down to just one company. The key is AI*Workbench, because the technology is no longer in one place."

MacNeal: "I'd always assumed CAD companies would eventually acquire the CAE companies, but I heard last night it may be the other way around. There are parallels to the auto industry, which is down to three major ones now; the federal government's job is to keep an eye on the marketplace."