Principles is Principles

This isn’t really surprising:

The SR-71 Black Bird is truly a wonder of engineering. You can tell that’s the case just by looking at it even if you have no prior knowledge of aviation. The way it looks just screams “radical design.”

Designed in part by Clarence “Kelly” Johnson for Lockheed and its secret division called Skunk Works, this bird was years ahead in innovation.

That’s the point of this article. It had to be thought through. In the early 60s no computers existed that could render or even come close to computing how this aircraft should be built. It was all done on drawing boards with pencils, rulers and protractors (if you remember what any of these are.) That’s one incredible feat if you think about it.

What’s even more incredible is that in the 2000s this design was ran through a sophisticated computer program used to design planes. You’re talking about thousands of rivets, angles of the fuselage and about a million other factors that this computer checks for.

The end result? The computer wouldn’t do anything different. The design is as efficient as it could be. It was perfect.

I might have expected some minor redundant elements to be identified in the computer model, but not much else.  The purpose of computer-aided modelling is not to give you a different design as one worked out by hand, but to make that process of working out quicker and easier to visualise.  The underlying engineering principles – bending moments, second moments of area, allowable stresses, material properties, etc. – on which the manual calculations and drawings are based would have been programmed into the computer software: the computer isn’t making any decisions, it is merely performing hand calculations very quickly and in vastly greater quantities.  The SR-71 engineering team would have performed all necessary calculations by hand and scrutinised the whole design to eliminate any redundancy, seeing as the design was so close to the edge in what was feasible.  You’d be far more likely to find a bog-standard 1950s motorway bridge to differ from a computer-modelled version because there wouldn’t be as much pressure to optimise the design.  It’s a matter of resources and time/economic priorities, not capability.

People seem to think that computers, when executing calculations, can do what humans can’t.  Presumably they wouldn’t expect a result calculated by an Excel formula to differ from a hand calculation, so I’m not sure why we’d expect engineering calculations to change once the forumlae are calculated by machine instead of on a piece of paper.

(On the subject of the SR-71, this is an awesome little story.)

Share

21 thoughts on “Principles is Principles

  1. Having worked in/around IT for years, I always offer the advice to clients that a computer cannot do anything people can’t, it just makes the process affordable or sustainable in terms of effort.

    I go on to advise that they work out how to do something without a computer (ie the process) first, and then use IT to make it quicker/cheaper/easier afterwards (ie automate the process).

    There’s a very strong correlation between those that ignore this advice and those who fail.

  2. I highly recommend reading Clarence Kelly Johnson’s autobiography – he ran an engineering operation that just wouldn’t be possible in today’s bureaucratised corporate culture. Also good is Ben Rich’s book on skunkworks.

  3. I agree that computers can only do what humans tell them to do, so they are just a form of automation for information-processing tasks. But they do allow scientists and engineers to run huge calculation tasks that simply couldn’t be done without automation because it would be far too costly to employ enough human “computers” (the word was originally a job title) to carry them out. The speed at which they work also makes it possible to carry out very extensive calculations for time-sensitive tasks like weather forecasting. Computers don’t allow people to do anything that couldn’t be done manually, but they do significantly change what it is practical and economically viable to do. They create new possibilities in the sense of greatly expanding the range of information processing tasks that can be done on any given timescale and budget.

  4. When my son started in software engineering, he was asked by a company to design a program that would assist them in calculating the various factors in their tasks (they were in the central heating business). He did, but at the end of his demonstration these people said while they liked it, where was the “magic button?”

    My son asked what that was, and the company kindly explained it was the thing you pressed to “magically do everything all at once.” I believe my son is, after years of working in the IT business, still searching for the coding for magic.

  5. Up to a point. Some of my software (for a petrochemical company) has done calculations that had never been done by humans nor ever would be. The task was beyond human capacity in any useful time. Once you’ve got multivariable iterations going on, with iterative calculations nested inside them, and further inside them, and so on, and the whole shebang drawing on a thermodynamic package which itself undertakes massive calculations, humans would be stumped.

    Happily you can compare output to observations from existing plant or from the lab. So this is unlike the case of mathematical models being trusted in the absence of searching tests which, alas, happens all the time. The great temptation of mathematical modelling is to fall in love with your model, and to be unwilling to test it properly. People routinely succumb .

  6. AndrewZ: They create new possibilities in the sense of greatly expanding the range of information processing tasks that can be done on any given timescale and budget.

    True, which makes it even more awe-inspiring what engineers accomplished before the drudgery of the computational work was relieved by computers. One of my all-time favorites in that respect is the aqueduct that the Romans built to bring water to the city of Nimes and of which the Pont du Gard is one element. The entire aqueduct is 50 kilometers long, mostly underground (i.e., tunneled through rock), but with a bridge to span the valley of the Gardon, and over its entire length it descends in height by only 17 meters to allow for the constant slow flow of water from the source to the city. The bridge itself is almost 50 meters high and 456 meters long and the descent over that span is of only 2.5 cm. It may be more a matter of measurement than of computation, but those Roman engineers knew what they were doing with pencils, rulers and protractors. And compasses, too, I guess.

  7. Andrew Z,

    But they do allow scientists and engineers to run huge calculation tasks that simply couldn’t be done without automation because it would be far too costly to employ enough human “computers” (the word was originally a job title) to carry them out.

    Indeed, that’s what I meant by “it’s a matter of resources and time/economic priorities, not capability.”

  8. “those Roman engineers knew what they were doing with pencils, rulers and protractors. And compasses, too, I guess.” Not pencils; English, 16th century.

  9. “Roman engineers knew what they were doing”

    Yep they certainly did and the same rules still apply today. Although the concern now is that given our ever increasing reliance on computers there is now a low awareness of the practical application of first principles amongst engineering practitioners such that it may prevent troubleshooting or stifle creative ideas or innovation.

  10. So the fastest jet so far was designed around 1960 and has been retired since. The Concorde was designed in the mid-1960s and stopped flying in 2003. Sounds like a retreat on that front rather than progress.

  11. If speed were the overriding factor, you’d have a point. However, for xommerxial airccraft, efficiency is more highly sought after in most of the market, with luxury being desired at the topend. For mil jets, it’s more or less accepted the constraints imposed by the seat to stick interface mean you won’t be able to outrun the best missiles, so the emphasis is on not letting one lock onto you in Tue first place through stealth, maneuverability of some combination of the 2.

  12. To expand on what TomJ said, as far as passenger jets are concerned they figured out a combination of comfort, passenger numbers, and fuel economy was preferable to speed once a certain speed was attainable. People would rather fly for 7 hours in comfort than 4 hours with less comfort it seems, doubly so if the former is a lot cheaper.

  13. Hedgehog: Thank you for the information on Pont de Gard.
    Wow, I’ve made at least three visits to Pont de Gard, I didn’t know that about it. I’ve spent hours wandering around, and up & down it, admiring the construction, even going around the “man-proof” barrier on the top.

  14. I’ve just realised the Pont de Gard is only a few kilometres from Avignon. If I’d known that, I’d have visited it when I was there last July.

  15. Regards high speed aircraft: On one of my late night rambling engineering-for-the-layman internet surfs, I stumbled upon some stuff (a bit technical for me, but I got the drift) about the limiting factor in aircraft airspeed is that above a certain speed (I forget it now, but say mach 1.8) jet engines aren’t so much creating propulsion as colliding with the air which at that speed has become a solid mass, this causes the engine to disintegrate.

    Apparently this presented quite an engineering problem, and was eventually solved (after a fashion) by Lockheed. The soviets never did solve it.

    Don’t pick me apart on small points, but y’all got the general point there. I’ll have to go back to read (& print or bookmark) information on the matter, as it is something almost any hands-on type bloke, from farmers through to design engineers, loves to hear about.

  16. Steve,

    Similarly, I read a great article on helicopters that explained their top speed is restricted by the fact that one side of the rotor is travelling in the forward direction, and hence is providing less lift than the opposite side which is travelling rearward. If a helicopter moves forward too fast one side of the rotor is effectively stationary.

    This document is very heavy, but covers everything brilliantly.

Comments are closed.