Under Control


One reader writes: I was talking to my old buddy Rusty the other day. Actually, that's not his name, but he's a metallurgist with red hair, so of course that's what I call him. Rusty's all right, even though he likes to glorify the role of metallurgy in the history of civilization (he likes to call metallurgy the second oldest profession.) I asked what was new in the metals processing world.

"Chaos," he said.

"What about Chaos?" I reply. Rusty knows I go on the occasional diatribe about chaos theory and its all-encompassing uses. So I'm a little suspicious here that he's trying to give me the business.

"I've been thinking about hypersensitivity, unpredictable behavior, and complexity, which you say are the characteristics of chaotic systems," he said.

Now I begin to warm up to the conversation. I've been preaching this stuff to Rusty for some time and he'd never shown any real interest before. Perhaps there is hope for him yet, I'm thinking. "Go on, go on," I urge.

"Seems to me that if there are any words that describe most metals processing activities, they would be complexity, hypersensitivity and unpredictable behavior. I wonder if applying chaotic ideas would give us a handle on how to control these processes."

I have to suppress a chuckle at this point. Ever notice that people will accept new ideas more easily if they believe they thought them up themselves?

"Really?" I say, leading him on. "How are you going to predict processes that are unpredictable? And what are you talking about, specifically? If materials processes were unpredictable, how come we can make all the different things that we do?"

(This way I'll see if he's really caught on or if he's just making idle chatter.)

"To answer your last question first," Rusty began, "most manufacturing processes are developed from laboratory experiments into full-scale industrial, repeatable, controlled processes over long periods of time, with a lot of guesswork and human experience thrown in. Once we guess the right set of points, we program the PID to hold everything in place, and then make stuff. Works great if the stuff you're making is a bazillion widgets. But it doesn't address the problem of cutting down the cost of developing new processes and materials, or how to do it quickly and efficiently. Ever heard of 'agile manufacturing' or 'flexible manufacturing'?"

Rusty sometimes treats me as if I live on a farm in New Hampshire and never get to look at the latest issues of 'Manufacturing Vogue.' Okay, I figure, I'll play his silly game. "So chaos is going to help control unpredictable processes?" I ask. "How?"

"It goes something like this. We recognize that the refinement, solidification, deformation and joining processes are inherently hypersensitive to external conditions, which means just about everything in and around the process. So we model the process, not with linear finite-element models, but with independent agents, object-style modeling, and we use the model to determine where and how the processes behave in a chaotic or complex fashion."

"What does that buy you?" I ask. After all, Rusty is still only repeating just what I've been telling him for some time now.

"By itself, that gives you an understanding of the process's 'dynamic system behavior,' which is no small thing. But we want to do more than just map the state of the process; we want to move the process into states that we determine will produce the material, or formed product, that we want. So we link the simulation, in real time, to the sensing and control connections. Then we 'predict' the next state of the process based on the current state, which we can measure in real time, and the simulated new state. If the program doesn't like the predicted outcome, it makes changes to the controlling output, which it uses the simulation to check, and pushes the system toward that state. Simple."

"Sounds like you need a really fast machine to do all that simulating, sensing, and controlling," I comment. "I wonder if such a machine exists?"

Then Rusty smiles a smile just like the smile of that goofy Warner Brothers cartoon character who says, "Eh, could be."

Then Rusty says, "Eh, could be." - Alan Campagna, president, Theta Systems, Woburn, Mass

As appeared in Manufacturing Systems Magazine August 1993 Page 8

Manufacturing Systems Magazine Articles
References - Table of Contents

Send mail to rmi.info@barn.org for more information.
Please send mail to webmaster@barn.org regarding web site structure.
Copyright © 1996-2002 R.Morley Inc. All Rights Reserved

R. Morley Incorporated
614 Nashua Street, Suite 56
Milford, NH 03055-4992 USA
Tel: 603-878-4365 FAX: 603-878-4385