Machine Design

Software can be deceptively simple to naive students — but simulations are "doomed to succeed"

Those of us old enough to remember computer punch cards often have an image of engineering simulation as a topic for grad students.

Those of us old enough to remember computer punch cards often have an image of engineering simulation as a topic for grad students.

That mind set was evident when software vendors first began to float the idea of design simulation on a PC in the 1980s.

Pioneers in this field were sometimes ridiculed and vilified. Algor Inc.'s Mike Bussler, for example, can recall competitors accusing him of "handing a loaded gun to children" because he had created the first full-featured FEA code that ran on 8088-based machines. "We couldn't go anywhere without people condemning us as the moral equivalent of baby killers," he says.

One of the arguments thrown at Bussler back then was that only specialists could be trusted to simulate engineering systems correctly. But that's not an opinion you'll likely hear today. Modern engineering students get introduced to simulation as sophomores. Educators teaching simulation to undergraduates also emphasize the need for verifying simulation results with actual data.

How best to get across such concepts was a hot topic at the recent National Instruments Corp. NI Week conference. Dr. Wayne Book, Husco/Ramirez chair of fluid power and motion control in Georgia Tech's school of mechanical engineering quipped that, "Simulations are doomed to succeed. I like to use that catchphrase because it sticks with people and reinforces the attitude that simulations must be verified. Their parameters can be adjusted arbitrarily and falsely. If you're not careful, they can predict pretty much whatever you want."

Don't think simulation for undergraduates is just a trivial exercise in pointing-and-clicking. Georgia Tech engineering students get introduced to simulation in a course on numerical methods. But simulations, of course, can encompass several disciplines such as dynamics, materials science, and control theory. "It's challenging as an instructor to tie all those together when you've only personally taught one or two of the courses yourself. And not every student ‘gets it,'" he says.

Moreover, graphically oriented simulation programs can be deceptively simple to naive students. "It is a struggle to make easy-to-use tools compatible with understanding what is really going on under the hood," says Book.

"That's even true of spread sheets," he says. "Some of the biggest mistakes I've personally made have been with spread sheets because they are so powerful. We tend to rely on them beyond our intuition and not think about the results. Students can get trapped this way and come out with crazy outcomes because they have no concept of the ball park magnitude of what they should be expecting."

Nevertheless, simulation helps students because it speeds the trial-and-error process. And future sophomores will likely experience simulation far differently than today. "Full simulation with out needing to know any numerical analysis is the ultimate end of all software development," says Algor's Mike Bussler. "Will you still have to know a lot about what you are designing? You bet you will. But simulation will help you learn faster and give you a better chance of getting it right the first time."



Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.