The Idea Factory

Home > Other > The Idea Factory > Page 14
The Idea Factory Page 14

by Pepper White


  Several friends were gone, though, and I missed them. Matt Armstrong had finished his master's degree in record time1 year-and was off to a job for $35K as senior research scientist for Owens Coming. Michael Picardi had stayed in TPP, had earned A's in all his classes, but had never found funding; he transferred to Princeton, where they gave him full financial support. Ike Thomas received two C's the second term, and, like Mary's lab partner, was given the boot. And Amrit, the squash player who thought that energy was easy, had landed a job modeling world oil demand for OPEC in Vienna.

  MIT is unlike law, business, or medical school. People come and go on their own schedules, so there is not, for example, a "Class of 1984 Mechanical Engineering Master's Degree Students." There's no cohesion; there's nothing but your own ability, or lack thereof, to meet people and make friends.

  My only summer class was Two-oh-two-three-System Dynamics and Control. It's an undergraduate class, but the summer version was populated mainly by what Rohsenow affectionately referred to as "the Navy Guys." Like the ones I'd met in Rohsenow's heat transfer class, they were generally older than the other graduate students and were working toward master's degrees.

  The smart navy guys had been to the academy at Annapolis, or Purdue, or maybe Georgia Tech. There was a good percentage of dumb ones, too, and I thought they would help the curve on the tests. However, the navy teaches teamwork, organization, and discipline: the navy guys help one another on the problem sets, their files of previous exams are impeccable, they work hard, and they concentrate well. They're also used to sleep deprivation. These qualities can make up for stupidity, so I would have to work for my grade in this class, too.

  "Feedback control systems are what makes anything that works work," David Miller said to the class in Building 13. "Everything is inherently unstable, and the feedback loop takes an unstable system and changes the system equation to make it into a stable system. For example, walking is unstable; just look at a baby trying to figure it out. He or she hasn't yet figured out or programmed himself to know the right gains to apply to his servomotors-his muscles. So he hits the stops a lot, or rather the carpet. But gradually we do get our gains tuned, and we can walk."

  David paused a second, then continued, But before you can control a system, you need to be able to model the performance of a system. We've worked out a number of techniques for doing this. That's the first part of the course. Then we'll go on to figure out how to find the gains we need to apply to the open loop systems we've put together in the first part of the course. That's basically what controls are-modeling open-loop systems and finding closed-loop gains, where by gain I mean how hard and fast you push the system to do what you want."

  Model model model. Eventually I'll be able to use that word without blushing. David certainly did. He looked as if he were pushing thirty, and had a curly mop of hair that he said his son's little league teammates pulled on, saying "Is this real?" He was exuberantly bright, talented, enthusiastic. He hadn't yet finished his Ph.D., so he was close enough to our level to have some compassion for our slowness. Plus, he really seemed to love to teach.

  His Ph.D. would put him among the MIT-cubed ranks that Mary had told me about. These are the talented few whose undergraduate grades are good enough to get them into MIT grad school, who then pass the doctoral qualifying exams, and successfully defend their doctoral theses. Good enough undergraduate grades means almost straight A's, because if it meant any less there wouldn't be room for graduate students from anywhere else-like me.

  David's Ph.D. thesis was to make a little underwater robot work. Once he was done, it would be able to crawl around deeply buried shipwrecks, the Titanic, for example; pick up debris; and send video camera messages back up to the control room of the ship on the surface.

  He continued his lecture. "So basically once you know how to model things, you can model anything. It doesn't matter whether it's a mechanical, fluid, thermal, chemical, electrical, or biological system. The concepts of modeling are the same. Do any of you have any examples from what you're doing you'd like me to go over?"

  Holy kill two birds with one stone, Batman. I raised my hand. "I've been working on this one problem, how to model a balloon being filled up with air." The word model rolled off my tongue naturally.

  Without missing a beat, David drew the picture that had taken me ten days to intuit in March. He quickly set up the equations I'd been trying to program on the Apple and expressed the problem in the parlance of two-oh-two-three. The only difference from my model was that he gave mass to the skin of the balloon and added something he called a dashpot.

  "And so your balloon problem will be a matter of accounting for the energy flow from the pressure of the fluid to the mass of the skin of the balloon, the elasticity of the skin of the balloon (the spring), and the internal friction of the balloon (the dashpot). The dashpot always slows down the motion of the mass, whether the balloon is expanding or contracting," he summarized. "If you put a feedback loop on the system, to, say, maintain a constant pressure in the balloon as the pressure in the tank varies, you'd have the same problem designers of airplane pressurization systems face."

  So there's a reason to learn to blow up a balloon.

  June 22

  Time for another dog and pony show-a research sponsors' meeting. The men in brown and blue had flown in from Peoria, Davenport, and Columbus to see whether we'd made any progress. Actually, it was a good idea for them to come every four months because in the last three weeks of the four months we made a lot of progress.

  "We gotta get some results for the meeting," Chet had said just after Professor Heywood's final. It was always something. So we'd cranked to replace what we could of the components of the RCM. The new driving air pressure tank was in, and we had mounted it in place. The new shaft was on order, and we knew there was no hope of further trials of the whole machine until after the meeting. The target then became a matter of completing as much other hardware in the test cell as possible.

  Scott and I had split up the machine into the front half and the back half; he would design the pieces of the new starting mechanism on the back, and I would design the fuel delivery system on the front of the machine.

  And then there was the electronics to sequence the whole thing. This involved assembling black plastic boxes and electronic components. I put them together on the basis of Chet's designsI drilled holes into metal plates, spray-painted the plates battleship gray, screwed the electronic components onto the plates, and connected them with wires and solder. The process made me understand how uneducated people can work on assembly lines and produce very complex equipment without having any idea what they are doing.

  On the front end of the machine, one immediate goal was to construct a support to fix the cylinder to the metal slots on the test bed. It was a two-day job; Nick and I cut pipes and plates, set up one of the plates on the machine, and welded the four angle-cut pipes between that plate and the plate on the test bed. Then we spray-painted the whole thing silver. It amazed me how good it looked when we were done-like something in Popular Mechanics.

  My talk was just after the coffee and cream cheese Danish break. I summarized the leaks we'd detected and the hardware we'd assembled.

  The guy from Caterpillar said, "So what you're telling us is the thing doesn't work yet. When is it going to work? I have to go back and report to my vice-president, and he's getting a little antsy. You know, what with the strike and everything, money's been getting a little scarce and we might have to cut the"

  Professor Heywood stepped in to the rescue. "Well, as you know, these things do take time. We expect to see a lot of progress during the summer when the pressure of classwork is less for all of us, and I'm sure that by the next meeting we will have all the components operational and have a test firing or two."

  Thank you, Professor.

  During the lab tour that afternoon the guy from Caterpillar said he liked the front support Nick and I had built. It could be interesting to w
ork for him; if only they weren't located in Peoria.

  That night after midnight, An and I were the only ones left in our offices.

  "How's it going?" I asked him.

  "Oh, not too bad," he said. "They could be better, though. My doctor said I have to stop how do you say 'burning the candle from both ends,' but I don't have much choice. Every day it's stay here till 1:00 A.M., then go home to my wife and children, then up at 7:00 to get the children ready for school and back here at the lab by 8:00. Seven days a week I am here for two years so far. "

  "Well, after you work your way up to general and on to defense minister and on to premier, you'll know it was all worthwhile," I said.

  "Maybe if I'm lucky I'll be a general. But the rest is politics, and you know, my friend, we are engineers, we do not lie, and politics is best suited for people who lie. Look what you people did to Carter. He was a good man, maybe even a good engineer, an intelligent man, an honest man, and your country spat on him," he answered.

  He offered me half of his bologna sandwich and I sat down in the old green Naugahyde office armchair across from his steelgray desk.

  "How can you get by on so little sleep?" I asked him.

  He answered, "Typically when I am on duty in Israel we live on four hours sleep. It's all a matter of what you're used to. There we had four hours so it would not be much of a decrease to zero hours. You know your Thomas Edison never slept; he just took fifteen-minute catnaps every four or six hours. But even in the army with the few hours of sleep, even during the wars, nothing there is as hard as this place. The Six Day War was hard, but at least it was only six days. Sometimes I wonder whether I'm going to make it out of here alive."

  "Hang in there," I said. "I'm sure you'll do fine. At least you've got the talent and the ability to work. I often wonder whether I have either."

  "Talent is a gift, my friend. And the ability to work, if you can stay here long enough, that is something you acquire. You become like Nehemiah, with a sword in one hand and a shovel in the other hand, and nothing is able to bring you down from your work."

  July 10

  Chet, West, and I drove to United Technologies Research and Development Center in Hartford to pick up the fuel injector. In diesel engines, the piston compresses the air in the cylinder, and it heats up the air, just as a bicycle pump does. At the end of the compression, the fuel injector squirts fuel into the cylinder; the fuel then vaporizes and spontaneously ignites.

  At the R & D center, the guys with Ph.D.'s had their own little offices with cinderblock walls and steel-gray desks like the one I had at the institute. Unlike my office, though, they had windows, which were a nice touch even if you couldn't open them. Their offices encircled the pigpens, where all the underlings-the newcomers or guys with just master's degrees-worked at computer terminals and drafting tables arranged in neat rows. If American industry wants better engineers, it occurred to me, why don't they give them better working conditions? Then the brightest ones might decide against going to business school or law school to get a private office and a decent salary.

  West was nearly totally out of the picture now; his company was rolling along with a grant from the Department of Energy for a report that would sit on somebody's shelf. But he had a buddy at UTC who had a spare EFIS fuel injector. That was what we would use for our experiments.

  EFIS stands for electronic fuel injection system, which was developed by UTC to improve tank performance. The only problem was that, like many other things developed for the army, it worked great in the research lab but was too complicated to work well in the field when the low-tech-but-reliable Russian tanks stormed toward Brittany.

  It would suit our purposes well, though, because in the RCM, the fuel injector only had to fire once per experiment. And the EFIS could do this better than any standard fuel injector.

  Chet, West, and I went to the workshop in the back of the building. It reminded me of the Sloan Lab. We went to the quiet, glass-enclosed office on the side and met the foreman. He picked up the telephone and said over the PA system, "George Brent to the office. George Brent to the office." The foreman gave no further information, and for all George Brent knew from the tone of voice on the PA, it was pink slip time.

  "George, these gentlemen here are from MIT, and we need to set them up with an EFIS injector for some research they're doing. I'd like you to help them out," the foreman said.

  "I'd be happy to, sir," George said. George was a friendly, middle-aged mechanic, with graying hair greased back, black shoes, and white socks; he reminded me of Riley in the "Life of Riley" TV show. He was like Nick, but younger.

  George took us to the EFIS injector he'd set up on the lab bench. "See, it's got a solenoid here; that's just a valve that operates with an electromagnet. So what you do is cut the power to the magnet and then the valve opens and the fuel goes out of the tiny holes in the tip of the injector."

  He flipped a switch and a small white cloud that smelled like diesel fuel burst out from the tip of the injector.

  George continued his explanation, "You know, it's interesting how we build these things. We make all the pieces, mostly rods and cylinders that the rods fit in, and then we give some of the ladies in the plant a box of rods and a box of cylinders. They take the rods and test them in the cylinders; they can tell by the feel in their fingers whether or not the dimensions are correct. It's a lot cheaper than machining the parts to a ten-thousandth of an inch tolerance."

  High technology.

  Anybody can have a good idea, even a stupid person. My good idea was to ask George whether he had any spare pieces of the EFIS, together with the complete unit that we'd come to pick up. I didn't have any real reason to ask for the junk pieces; it just seemed that I might be able to figure out how the thing worked if I had them handy. I could play show-and-tell. By myself.

  On the drive back from western Mass., I sat in the backseat and slept some while Chet and West talked in the front.

  "Do you consult?" West asked Chet bluntly.

  "Yeah, a little bit. I do some work on the side for the people who funded my thesis," Chet answered. Chet didn't seem to care about money. He was more into the science and the theory of it; he would follow in Heywood's footsteps.

  "What's your billing rate?"

  "Fifty dollars an hour."

  "They must be laughing at you, pal," West chuckled. "If I were you I'd go for at least a hundred. They'll pay it; they've got the bucks."

  At the tollbooth at Route 128 a flatbed truck pulled up beside us. Its cargo was wrecked cars with the exterior sheet metal stripped off; the joints and marrow were bare. There was a spring on an axle, and a dashpot-the shock absorber. The chassis looked like a mass, just like in David Miller's class.

  July 15

  At the laundromat I pondered how much the spin cycle should run its motor to wring the moisture from my clothes, and how the spin cycle energy compares with the dryer's heat energy. What combination of spin time and dryer time would result in the least total energy use?

  I picked up the aquamarine paperback that somebody'd left on the next dryer. The back cover was a picture of a man and a boy standing next to a motorcycle.

  Two sentences stood out. The first said, "The steps in the scientific method are: (1) statement of the problem; (2) hypotheses as to the cause of the problem; (3) experiments designed to test each hypothesis; (4) predicted results of the experiments; (5) observed results of the experiments; and (6) conclusions from the result of the experiments." The second was like unto it: "An experiment is never a failure solely because it fails to achieve predicted results. An experiment is a failure only when it also fails adequately to test the hypothesis in question, when the data it produces don't prove anything one way or another."

  The statements explicitly summed up what Chet, West, and Professor Heywood had implicitly tried to tell me. We had tested the machine to see whether it could take the higher pressure. The answer was no, and the conclusion was to rebuild parts of the ma
chine. Scott and his computer modeling would predict results of the experiments. We would do the experiments, I would analyze the data, and he would conclude what changes were required in his computer model.

  The book was Zen and the Art of Motorcycle Maintenance.

  August 12

  I got an A. David gave me an A.

  I wondered whether he gave everybody A's, because, who knows, one of the navy guys might review a grant application of his some day.

  There were three weeks left before my Incomplete with Greene would become an F and I would be driving a cab.

  The fact that August was cool and not its typical humid self, which makes it pleasant to be in an air-conditioned room with a computer terminal, didn't make things any easier. If it hadn't been for the demands of the lab and David's class I probably could have knocked the project off in a week or two at the beginning of the summer.

  I remembered Greene's assignment: program the incompressible balloon-filling problem with realistic parameters; for example, a typical balloon volume might be 100 cubic inches, pressure on the inside of the balloon might be 2 or 3 psi, and a typical stretchiness (elasticity) might be close to that of a skinny rubber band. I looked at my set-up of the equations-my model.

  The killer is the square term. If it were an ordinary differential equation, marching through time with the solution would be a piece of cake. How to deal with the square term?

  Solution. Go to the library again. Somebody somewhere sometime must have bashed his head up against this. Maybe even Bernoulli himself tried to solve this type of equation as a followon grant from the Medici or Sforza Foundation.

  I could ask Chet to give me a hint, but that would be cheating. But going to the library is not cheating. If by chance someone has already done this I'll be a free man. I went via a hall on the fourth floor of Building 4 with photographs of bats and gymnasts and a bullet going through a jack of diamonds, and on up to Barker Engineering Library. Before going to the card catalog I read the funnies and the sports, took a nap in one of the soft, cushy, leather chairs, and checked the price of the stock I'd been following since seventh grade. I didn't own any of the stock but it had done very well.

 

‹ Prev