The placebo and the engineer

An engineer, updating one of his machines, replaced the old analog controls with a plc. In his haste to deliver the machine to the customer, he specified the old control panel faceplate. Rather than leave an unsightly hole where a knob had been, the shop foreman installed the old knob and potentiometer. Neatly taping the unattached wires, he pointed out his service, and told the engineer where the freshest doughnuts could be purchased.

Having delivered and tuned the machine, a technician locked all the settings, inviting the customer, if unsatisfied with the machine’s performance, to call for service. Weeks later the customer’s old, experienced operator told the technician the machine worked well, except the operator needed to adjust the knob once or twice during the shift. Wisely holding his tongue, the technician nodded and made a note. The manufacturing company’s old experienced engineer told the younger engineer to be content, and leave the switch on all models, as a placebo. And so that’s what they did, illustrating the difference between engineers, and smart engineers.

A fable of quality

A manufacturer, eager to save money, bought parts from a developing nation. Finding many unusable, he hired temporary workers to examine them and discard those not meeting his standards. Hearing his customers speak of the recently poor quality of his product, the manufacturer asked a consultant what was wrong. The consultant split his fee with a professor at State U to produce something the consultant could give the manufacturer. The professor purchased a paper from a service that outsources writing to the third world. The manufacturer paid the fee, ignored the report, and prospered, because his product was cheaper than his competitors. The moral is, with globalism everybody wins.

—

My little fable is fictional, and a caricature at that. Mark Graban looks at a related real-world phenomenon in GM to try to Inspect Quality In After Letting Experience Walk Out the Door:

As with many business decisions, such as outsourcing or offshoring, the “savings” or “benefit” from such a move is easy to calculate. It’s easy to calculate the savings from paying workers $14/hour instead of $28/hour, even considering the buyouts and “go away” payments to departing workers. But the COSTS are much less easily quantified.

The Detroit News (quoted in Graban’s post above) shows why executives should let the PR guy talk to the press:

The biggest challenge for GM may be accomplishing the massive undertaking without compromising the quality of its cars and trucks. Having begun to win new respectability on the quality front, the automaker can’t afford costly and reputation-marring mistakes on the factory floor, which is a risk when there is significant turnover.

“We are very intensely focused on making sure our quality isn’t compromised,” said Joe Mazzeo, GM’s executive director of manufacturing quality. “Our customers don’t know this is going on, and they don’t care.” — GM preps for new hires after buyouts

To avoid “reputation-marring mistakes” in front of the microphone it would be better keep an experienced PR guy on staff. I bet that’s one function GM executives think is worth top dollar. Well Mister Mazzeo, some of your customers do know this is going on. Whether they care or not we’ll see.

Bridge collapse

See Minnesota Bridge Collapse, part One, with good links in “UPDATES”. Somebody said politicians would rather build new bridges than fix old ones, which is certainly true. That’s because voters do not reward politicians for fixing things. (We’ll see if they punish them for failing to.) Things are expected to work. It’s kind of like fixing up your house. It may help a little to point out new shingles, but houses are expected to have roofs. To increase the sales price, ignore the thirty-year-old roofing and add a skylight. Beyond that, I have nothing I have not already said in Pharoh’s engineer. I look forward to DOF’s part two.

Calculating

DOF comments on the May 2006 Scientific American article by Cliff Stoll, and explains why he still uses a slide rule himself.

I have several slide rules, but I use them rarely, and then only for fun. For regular calculations I use an HP-11C. For teaching, I have to demonstrate with a TI-83/TI-86, though I don’t much like it. It is modern and capable, and everyone uses it, but I find it an awkward nuisance (Another blogger has a less charitable opinion of graphing calculators.) The algebraic entry system is inferior to RPN. The graphing features make the calculator too complex, but are not powerful enough to take the place of Excel or Mathematica. The size and shape make the TI difficult to operate. The HP’s horizontal layout makes it easy to hold while I push buttons with both thumbs, and the keys on the HP have a nice solid feel. The HP fits in my pocket. If I need graphing I’ll use a workstation.

The arcane practice of nomography

Sounds like a third-century heresy

Roger Pollack has a blog of Cold-war calculators: “A collection of calculators created during the Cold War for the use of civil defense planners and military commanders.”

I suppose you could say that a nomograph is a graphical representation of a multi-variable formula. They can be difficult to make, especially for advanced applications, and there is really little pay-off to putting in the time to learn how. Like learning to sew well, it is a skill that cannot be expected to repay the cost of learning it.

I have always liked Nomograms. I learned a little about them in high-school drafting class, and a bit of the theory in college, but it was becoming archaic. Still, even today there are alignment charts of different kinds in use. I occasionally see (or use) one in a small engineering shop or at a company that makes specialized machinery. Someone forty years ago plotted out a nomograph to select the bearings for a particular application, and it is still in use, laminated and chained to a filing cabinet. Most have been replaced with spreadsheets, or suppliers’ proprietary computer programs.

They do have their advantages. A nomograph is relatively easy to use, and hard to misuse. It only works inside a defined range, and can only be used within that range. With a spreadsheet, any goof can change the formulas or copy another row to extrapolate outside of its legitimate range of use. With a (correctly made) nomograph, you can only use it for what it has been made to do. For example, if it is set up choose bearings to operate from 60 to 200 rpm, then the speed scale only runs from 60 to 200 rpm. It does not exist where it does not apply.

Supplement, nutritional, holiday; 1 each

Food baskets from the engineering department

A few years ago I was working in an engineering office as the holidays approached. One day we got a memo that said, in paraphrase,

At this festive season we are again distributing food baskets to the needy. Rather than bringing in whatever food you want to donate, we ask that you donate the money you would have spent. We will use this to purchase a nutritionally optimal selection of food that will then be distributed equitably.

So yes, engineers really are like that. But after all, what is the alternative? Do you want a sub-optimal selection of food? Is it about feeding the hungry, or feeling good?

The Gleaners

The Biblical injunction was to leave some for the gleaners. The modern approach would be to pick the field clean and give 8.6% to an umbrella organization which would distribute the wheat to the poor. The chairman of the umbrella organization would make a six-figure salary. Does this just lead (after hundreds of years of misery) to fat poor people? Here I would insert a picture of obese people in back-braces waddling through a wheat field, stuffing grain in their mouths. Fortunately for us all, I have no skill with a paintbrush.

Now clearly I am being kind of snarky, and taking some pot-shots at organized charity and big government. To be fair, it is this organizational impulse that allows us to feed so many people so well that world-wide more people are fat than hungry. Not that we gain nothing from the structure of modern society; We gain a lot. My question is, “What do we loose?”

Bustin’ Rod

My Foray into Psycholinguistics

For a while I was kind of a temporary assistant to the engineer on a bridge construction project. The bridge was being built with reinforced concrete. Concrete is cheap and very strong in compression, but weak in tension. Steel is strong in tension, but expensive. Before the concrete is poured, long steel rods are placed in a framework where the concrete will be. These rods, sometimes called ‘re-bars,’ take the tension while the concrete takes the compression. It’s an ingenious system.

At lunch I got to talking with one of the men building the bridge. It turned out we had a few acquaintances in common. I asked him what his job was. He said, “I bust rod.”

My friend was a union ironworker, and his particular job was cutting and bending three-quarter-inch steel rods to the prescribed shape, and then putting them in place to form a kind of cage that the concrete would be poured around.

Busting rod is hard work for good pay. It demands intelligence, strength, and stamina. Heavy construction is dangerous, and working high up on a bridge over a river, next to highway traffic, makes it more dangerous yet.

“Hey, boss, the chop saw’s busted again.”

But it is construction. Nobody’s busting anything, except by accident. The rods are cut with a special saw, they’re bent with a bender, and they’re placed and wired together by a skilled worker. The man was building a bridge. If I were building a house, I wouldn’t say I was “bustin’ board.” So what’s up with this?

Part of it is just a way of speaking. Like Pushin’ broom or Pushing Tin. But why ‘Bust?’ Is it unworthy of a man if it’s not violent? Do computer programmers bust code? [Insert lame Microsoft joke]

When I was a boy I busted stuff all the time. Usually old stuff that was already busted: Clocks, toasters, model airplanes, Army men. As I grew older I busted bigger stuff, sometimes by accident. In high school, I almost busted half the physics lab. Once in a while I got busted.

When I was a young man I had a job busting stuff for Uncle Sam. I busted a lot of stuff violently with explosives. I blew up buildings, cars, all kinds of things. Instead of getting in trouble I got a good salary with full medical and dental. If people asked, I said I was ‘in the Army.’ If pressed for detail I used the equivalent civilian job title and said I was an ordnance engineer. I never described my job as ‘Bomb buster;’ I and my fellow soldiers would have regarded that as embarrassingly lame.

“I’m a tenth-level Paladin”

Joe Carter writes about how, for effete intellectuals, ideology fulfills a fantasy role. I wonder if what leads an intellectual to satisfy his fantasy in ideology isn’t a broader human need. That is, it’s not that they’re inadequate Che Guevara wannabes; It’s just that they’re men. We all need to feel significant. The difference between the activist professor and my friend the ironworker is in how this need manifests itself.

The guy who glamorizes heavy construction into dragon-slaying is far less pernicious than the intellectual with a big idea. The ironworker who imagines he’s slaying a dragon builds a real bridge. The ideologue who sets out to build a metaphoric bridge ends up building a labor camp.

Don’t it always seem to go…

That you don’t know how it works till it breaks

Some thoughts on machine design

You start out with a good, solid design; It’s the result of ignorance, caution and hubris:

“How thick should the side-plates be?”
“I don’t know, we’ve never made anything like this. Two inches!”

Really it’s over-designed. It’s a 250,000 mile transmission is a 100,000 mile car. And a good thing too, because for the next iteration you take the established design and start pushing it. You make a machine just like the old one but wider; longer; faster. This process works fine until it fails.

As a mechanical engineer, I’ve spent a chunk of my career designing stuff. This has been more-or-less customized industrial machinery. Suppose that my fictional^* employer, TinyCo, has a product line. One of our standard models of BigMachine is tweaked, sometimes out of all recognition, to meet the needs of the customer. Starting with what the salesman faithfully promised to deliver in three months, I’ll pull out the plans for the most-recent/most-similar machine that we’ve successfully built. Then I’ll make the whole thing four inches wider, or twenty percent faster. I’ll make the rollers static resistant, or Teflon-coated, or chilled. I’ll update the design to use the gear box from the new supplier. Maybe I’ll put a black box on the side to do something special. The electrical engineer will do the same kinds of things with the parts for which he’s responsible. Remarkably enough, this works pretty well almost every time.

There’s not a lot of science to it. Suppose space is tight this time:

“Why are the side plates 1.25″ thick?”
“Because that’s the way we did it on the C2.”
“How thick do they have to be?”

I could set up a testing program to figure it out; That would cost a fortune. I could do some theoretical calculations of forces and deflections; That’ll take a week I don’t have, and nobody will believe me unless it’s what they want to hear anyway. Or I could just make it 1.125″ thick. That’ll probably be fine. One of the guys in the shop saw a competitor’s machine that had 0.75″ side plates, and that one had some problems when it ran at full speed, which he thinks was 600 feet per minute. So I design the C3 with inch-and-one-eighth side-plates, and it works fine.

Fast forward two years

Another engineer, a new guy, gets the specifications from sales. The customer needs, and sales promised, 800 feet per minute. Okay, that’s very similar to this C3 from a couple of years ago, and that worked fine. It’s going to be a tight fit. He thinks, “We’ll just shave a bit off the side plates.” And that works fine; That design process gets used, and works fine, right up until it doesn’t. Someone pushes the envelope too far; The envelope pushes back and there’s a disaster. If the problem is caught before delivery the disaster just involves expensive re-work. Worst case, someone gets hurt.

Finally in a slack period someone sits down and does some analysis, examining every machine the company’s ever sold. He finds an empirical relationship between machine width, machine speed, side-plate thickness, and, surprise! machine height. He puts it all together in a spreadsheet, and we finally have something solid to go on.

Besides the body of machine drawings on paper and in Autocad, there may be half a dozen spread sheets like this, and one tattered old nomograph in a drawer somewhere. These together make up the “book” we use when we do things by the book. At any given time the written parts are about two-thirds of the total knowledge base. The rest is habit, guess work, and lore.

^*All of the details are fictional.