The Engineering Process (or Of Course We Know What We Are Doing! Except That, We Have No Clue About That…)

Oscar Gordon

A Navy Turbine Tech who learned to spin wrenches on old cars, Oscar has since been trained as an Engineer & Software Developer & now writes tools for other engineers. When not in his shop or at work, he can be found spending time with his family, gardening, hiking, kayaking, gaming, or whatever strikes his fancy & fits in the budget.

Related Post Roulette

54 Responses

  1. Francis says:

    Thanks for this. I don’t know a damn thing about engineering, but whenever someone says that there’s a simple fix that “they” are ignoring which will have some 5star effect, I’m suspicious. Yes, human progress can occur by individuals bucking the trend, but that’s rare. Most progress is incremental and consistent with the consensus.Report

    • Oscar Gordon in reply to Francis says:

      This one has been itching to be written for a while. The highlight of this is that all sorts of very non-obvious improvements have been made to commercial airline safety over the past few decades, such that air travel is impressively safe. But because people are bad at assessing risk, they still focus on the splashy event and look for an obvious way to assuage the risk, rather than understanding that the risk is already so very small.Report

      • Kim in reply to Oscar Gordon says:

        Airplane much less dangerous than surgery. Much much less dangerous than driving car.Report

      • El Muneco in reply to Oscar Gordon says:

        Since you’re more likely to know: just what proportion of incidents have happened (1) at cruising altitude/speed (or sufficiently early/late in landing/takeoff), (2) attempting emergency landing was not feasible (i.e. over open ocean), (3) did not involve a missile?

        I’m finding it hard to think of a mass-publicized incident that fits – planes just aren’t dropping out of the sky…

        The way I heard it is basically that if nothing has gone wrong in the first minute after you start accelerating for takeoff, you’re either safe until the last minute before landing, or you’ll hear about the problem with plenty of time to start worrying, so you might as well relax.Report

        • Oscar Gordon in reply to El Muneco says:

          The one that jumped into my brain was the Air France flight that was heading to Brazil. But even in that case, the pilots knew something was wrong, but hadn’t realized they were slowly falling out of the sky.

          Maybe the Malaysian Air flight that went missing, if the cabin crew could have initiated a cabin release (assuming that the pilots were incapacitated but the cabin crew was functional).

          But yeah, planes rarely fall out of the sky, it’s either a critical failure at take off, or a botched landing. Perhaps, if the pilots knew they were going to have to attempt a dangerous landing (gear failure or something), they could climb, drop the cabin, then try the landing, but even that would be a rare event, and probably be a hull loss since landing without the supporting structure of the cabin would be bad, so it’d have to be a real bad thing.Report

        • Kim in reply to El Muneco says:

          Not around here. Pittsburgh can get pretty dangerous with sudden, unexpected downdrafts… (which, granted, is still not much of a problem a mile up in the sky, but if you’re coming down for a landing, you can get hit on the approach).

    • Kim in reply to Francis says:

      Depends on how simple you want it to be.
      I’ve got probably ten or fifteen inventions that I could pull out of a hat that took less than twenty people and had a fivestar effect. Of course, if those 20 people are all world-leaders in their fields, the fix probably isnt’ all that simple, now is it?Report

      • Francis in reply to Kim says:

        Kim you are so delightfully unusual that I have no problem believing this to be true. So, if you wish to file your inventions with the USPTO I will take a leave of absence from my current employer and work with you to prosecute the applications in return for an equity share. After taking the patents international, we can license the claims to the global community and retire to alternating months between Manhattan, Paris and Bora-Bora.Report

        • Kim in reply to Francis says:

          *snort* The patents are already registered (often not stateside) and the devices are already in production.
          That is, if at least one of them hasn’t been sued into oblivion —
          Step 1: Make Company
          Step 2: Sell Product Like Hotcakes
          Step 3: Sell Company and run for cover.
          (Step 1.5 was sell idea to military to use for destructive purposes).

          … no, I’m not kidding.

          That one actually only took one guy with a decent idea…

          [Although if you want a patent on a “cool device”… apparently Archer has the real specs posted for a sonic generator that produces a brown note…].Report

  2. notme says:

    Nice picture of one the soviet ground effect vehicles. Or is that a hydrofoil?Report

    • Oscar Gordon in reply to notme says:

      Russian Ekranoplane. I’ve done lots of research over the years on those Wing in Ground Effect birds (even have plans of my own to build one).

      I picked it because the Russians built those beasts with massive tails (aerodynamically necessary, but massive all the same).Report

      • notme in reply to Oscar Gordon says:

        Yes they did. The MD-160/Lun class ekranoplane was probably the pinnacle of their designs. I always wonder why Boeing didn’t build their Pelican.Report

        • Oscar Gordon in reply to notme says:

          Because the DOD couldn’t accept the tradeoffs the design needed (Pelican was to be a hybrid WIG, capable of WIG & free flight, which means it would be inefficient at both unless the wings were somehow reconfigurable, which means added weight and complexity).Report

      • Joe Sal in reply to Oscar Gordon says:

        Are you planning to build a single seater?Report

        • Oscar Gordon in reply to Joe Sal says:

          First one will be a two seater.

          If my plans work well, I’ll explore a larger, 6-8 seater design.Report

          • Joe Sal in reply to Oscar Gordon says:

            able to sustain short flights? how many total pounds ya running? I designed mine as one seat and very light, results in models showed a light gust would lift it up well past the 12 feet ceiling, which required some considerations on how to glide back down to ground effect.

            Of course dynamic stability going in and out of ground effect was fun. I started with the ekranoplane also, but I sure did end some where else.Report

            • Oscar Gordon in reply to Joe Sal says:

              Maybe, I haven’t settled on the wing shape yet, so limited flight isn’t a given, and thus I don’t have a fixed mass target yet either (ball park around the displacement of a family weekend outboard runabout.

              A lot of it will depend on how much of a shop I can build/get access to.

              Popping out of ground effect can be exciting, and if you don’t have the right wing, settling back in can be even more so. Ever heard of the Flarecraft?

              If you don’t want to be leaving ground effect, you’ll need to adjust your tail size, or add weight. The pitching moment in GE can be pretty impressive.Report

              • Joe Sal in reply to Oscar Gordon says:

                I found the Ekranoplanes were terrible at pitching on model scales. Too much tendency to nose up and stall. At full scale with the mass they carry, it’s probably more optimum.

                Chose to not add weight which makes the dynamic stability a very tough goal. The cruise speed was low, I think 40-58 mph, so a 35-40 mph gust would affect the lift of the wing beyond a point the tail could counter the lift. Up she goes.

                Realized I needed to safely glide back down to ground effect. Researched a lot on dynamically stable low speed gliders.

                It will be interesting to see were your design leads. Flarecraft and Hoverwing were interesting. My work was about 4 years ago. I shelfed the design until retirement is closer.Report

  3. Jaybird says:

    I love these posts but they remind me so very much that I was a Humanities major.Report

    • Oscar Gordon in reply to Jaybird says:

      The best part about being an engineer is sitting down with my peers and thinking up all the ways X can go horribly wrong, then getting to work figuring out how to keep X from going horribly wrong.

      And sometimes, at the end of it all, we come to the consensus that X will just plain “Not End Well”, then we go drinking.Report

  4. Autolukos says:

    Remember the old saw: You can have it done fast, well, and cheap – pick any two.

    You real engineers and your high standards; in software, you’re lucky to get one.Report

    • El Muneco in reply to Autolukos says:

      Yeah. It’s more like “aim for two and pick which one is first to be sacrificed”.Report

    • Oscar Gordon in reply to Autolukos says:

      I swear it’s because of the fact that software isn’t seen as a physical product.

      It’s better with my employer, since I write software for computer aided engineering. You can’t be slapdash with it, since people are using it to create actual products, and if it fails to give reliable answers, customers will stop using it faster than they will a crap email app or web browser.Report

      • Autolukos in reply to Oscar Gordon says:

        I’m pretty happy that standards are higher for the software you write than what I write.

        For what I do, the incentives are pretty clear: failures are unlikely to kill anyone, and money from various sources is pretty easy to come by at the moment, so speed becomes the overwhelming priority.Report

        • Oscar Gordon in reply to Autolukos says:

          Speed is critical for us as well (we release every 4 months), but if a planned feature can’t hold up to testing, it doesn’t deploy, and we apologize to customers if necessary.Report

          • Autolukos in reply to Oscar Gordon says:

            4 months? Do you mail it to your customers on floppy discs, too? 🙂Report

            • Oscar Gordon in reply to Autolukos says:

              Most engineering software is lucky if it puts out a release annually. Some have gone years between releases absent a critical bugfix/patch.

              When your software is often cited for regulatory compliance, you need some pretty rigorous testing.Report

              • Morat20 in reply to Oscar Gordon says:

                We do major releases yearly (we’re in the same boat, regulations wise. The FAA uses our stuff. If it flies in air or space, engineers have used our code — either in design, testing, compliance…somewhere) ourselves.

                And it’s because of exactly that. It has to be done right — and while we have a list of things customers want that’s like 40 pages long, a lot of it is “Figure out how to DO this” too. It’s not just software — they have engineering problems to figure out first, along with lengthy testing so we can prove to our clients that our answer is good — and what we mean by good, exactly.Report

          • The State of Colorado has a single intake software system for public assistance. Modifying the intake software is often the most time-consuming part of changing eligibility requirements. Releases are quarterly, and major changes may not be rolled out for eight or 12 months after the law changes. Building new test cases, modifying existing ones, and full regression testing takes time, and the consequences for getting things wrong can be millions or tens of millions of dollars in fines by the federal government. There’s also priority conflicts — if the federal Department of Ag requires changes in their audit interface in three months, that’s going to get much higher priority than changing a state-mandated eligibility condition.Report

  5. Kazzy says:

    I’d venture to guess that a similar phenomenon exists in most, if not all, fields. “Why don’t you just…?”

    Hell, even within separate subfields. In education, teachers rarely understand the various pushes and pulls of admin and, despite most of them having been former teachers, most administrators forget the day-to-day lives of teachers.

    To your last point — about the response you garnered on Facebook — how many features DO exist precisely for that reason (i.e., peace of mind)?Report

    • Oscar Gordon in reply to Kazzy says:

      Not many, because on a plane, weight == money, so if it doesn’t actually do something significant, it better not weigh much.

      One of the few I can think of is the strange lock a pilot has to put their gun in if they are one of the few pilots the government has approved to carry. Somehow the government feels better if a pilot can not casually pull out his pistol, but instead has to go through a big hassle to get it. In response to the danger of that, I’ll point to a certain GermanWings flight…

      ETA: Yes, most fields have some form of this, where simple, obvious solutions aren’t, for whatever reasons.Report

    • Mike Schilling in reply to Kazzy says:

      If teachers could just corporally punish the kids that are acting out …Report

  6. Richard Hershberger says:

    This post is massive overkill. I admire that. Well played, sir. Well played.

    What I take away from this is that I should bring a personal jet pack with me when I fly.Report

  7. scott the mediocre says:

    This is some nice work, Oscar.

    I worked for Boeing (military/space, doing various C4I comms and image processing thingies) long ago (courtesy of their acquisition of McDonnell Douglas). I found parts of the company very impressive. I take it from your past tense that you no longer work for Boeing. Can you say what kind of software you write now, if not with/for whom?

    Since leaving aerospace (1998) I don’t work on things which directly cause large scale human/dollar losses if they fail (though some medical devices I helped design can cause high acuity medical conditions to be missed if the devices fail), which is OK with me. But what I have found working for smaller companies that still have to produce industrial/medical (versus consumer) level reliability and use lifetimes, while taking the wetware’s norms of reaction into account, is that the engineering development process is at its heart the process of figuring out what your spec actually was (~should have been) rather than the written down thing that you thought was the spec.

    Airframes bend quite a bit during flight; it’s why aircraft doors are next to impossible to open in the air.

    I thought it was because they are plug doors, but I suppose without any pressure differential, bending based binding would still make them stick. Doesn’t the door’s metal to metal mating surface have to take into account a fair amount of differential thermal expansion? Can it really be designed that the door edge is always under compression?

    However, I must take exception to:

    The Hyatt Regency case is one of these, where the initial design engineering was good, but an onsite modification that was not fully analyzed caused a disaster.

    Your own link includes the (editorial) finding:

    Even as originally designed, the walkways were barely capable of holding up the expected load, and would have failed to meet the requirements of the Kansas City Building Code.

    (Also the crucial change was made at/by the fabricator of the hanger rods and well before construction began, not onsite)

    • Oscar Gordon in reply to scott the mediocre says:


      Thanks. I worked for Boeing for 5 years (my wife still does). I’m a computational physics guy & Boeing had me doing other stuff (some of it fun, some of it scut work). I got an offer from a computational physics software company & i jumped at it. We just got acquired by a ginormous German company, which should give you enough information to figure out who I work for without saying it directly.

      The doors are plug doors and the Pdiff keeps them closed for the most part, but the door frame bends just enough to bind the latch & hinges during flight. They aren’t impossible to open, but the drunk in first class isn’t going to wrench it open unless he belongs in the pages of a comic book. They aren’t designed that way explicitly, but they aren’t designed to open in flight, either.

      You are right about the walkway! My bad, I was going off my 17 year old recollection of an undergrad discussion of the case, which was more focused on the problem with the change itself, rather than the initial design. I found the link after I wrote it & didn’t reread the report.

      Anyway, glad you enjoyed it.Report

  8. You can have it done fast, well, and cheap – pick any two.

    Then there was the replacement span for the Bay Bridge …Report

  9. DavidTC says:

    As someone who is not a mechanical engineer, and knows nothing about airplane designs, even I saw some serious nonsense in the proposal.

    My first thought was: Aren’t a lot of plane crashes during takeoff and landing? During which, duh, parachuting something out of the plane cannot work.

    Second thought: And aren’t a lot of the remaining crashes *sudden*? Like, they collide with something midair?

    Third thought: And of the remaining crashes, aren’t they usually in really bad weather? The sort of weather in which the airplane is already being thrown around, so ejecting from it is is very unsafe, and even if that works, parachutes do not work particularly well in those winds?

    Forget the actual engineering problems. Even if it was a *trivial* change that airplane manufacturers could do tomorrow without any effort…it would *maybe* help in like 10% of crashes.

    And those would be exactly the sort of crashes that pilots are supposed to make emergency landings in, but *fail* to pull it off, so at which point, exactly, are the pilots supposed to know to pull the rip cord? Bearing in mind that parachuting like this is probably *more* dangerous that managing a *controlled* forced landing. (Because it is, in fact, an *uncontrolled* forced landing.)

    ‘My plane has been so badly damaged that I cannot make it to an airport. I can try to find a highway or field or somewhere to land in, which works a majority of the time, but sometimes doesn’t work and kills everyone, or I can pull the ripcord and leave the fate of my passengers up to random chance, and hope that their parachute works right and they don’t land on an occupied building or in water or something.’

    These idiots seem to be assuming that planes crashes are caused by airplanes magically having their wings disappear or something in perfectly calm weather, and it’s an obvious decision to abandon the airplane at that point. (And if *that* was how it worked, a more reasonable suggestion would be to have *passenger* parachutes, which at least won’t crush people on the ground and can be controlled to some extent even by untrained people. But because that’s *not* how airplane crashes work, we don’t do that.)Report

    • Oscar Gordon in reply to DavidTC says:

      And yet, the video had many thousands of “likes” on FB.

      One of my pet peeves are people who fancy themselves “designers”, who cobble together neat looking animations of their designs and promote them online. “Designers” who usually have very little to no training or experience regarding the design of whatever their latest idea is.

      Sorry buddy, you aren’t a “designer”, you are barely a competent animator.

      They are an endless source of inane questions from my podunk family & friends.Report

      • Morat20 in reply to Oscar Gordon says:

        I’m sure it does have tons of likes. Everyone wants safer things.

        And not to put too fine a point on it, it’s not exactly unknown for ‘safety’ to get cut in favor of other priorities. Caveat emptor. Penny wise, pound foolish. A million cliches to cover the basic “we’re gonna think short-term” or “not invented here”.

        Which means good sounding but entirely impractical in reality ideas get lumped into “Not done for penny pinching reasons” ideas.Report

        • Oscar Gordon in reply to Morat20 says:

          It’s not bad that people want safer things, it’s bad that people see parachutes and think, “Oh, that’s safer.” It’s bad that an uncontrolled, if slower, descent is better in their mind than the pilots working to put the plane on the ground in one piece. It’s bad that people have such poor ideas as to what constitutes “safer”.

          Honestly, if we were going to try to put parachutes on a plane, we’d be better off building a system similar to the small aircraft system I linked to. It would require less structural modifications. Of course, a 747 would need like 5 massive parachutes (3 along the fuselage, one on each wing), and before deployment, the pilot would need to basically throttle back and essentially stall the plane to reduce the airspeed enough to safely deploy, and you’ll still give everyone a nasty case of whiplash. Provided, of course, that you aren’t in a storm, or only 1000′ off the ground, or in a half dozen other situations where the parachutes won’t do squat except cover the crash site with polyester or nylon.Report

          • DavidTC in reply to Oscar Gordon says:

            It’s not bad that people want safer things, it’s bad that people see parachutes and think, “Oh, that’s safer.” It’s bad that an uncontrolled, if slower, descent is better in their mind than the pilots working to put the plane on the ground in one piece. It’s bad that people have such poor ideas as to what constitutes “safer”.

            I’m just imagining one of those things ramming a building.

            Honestly, if we were going to try to put parachutes on a plane, we’d be better off building a system similar to the small aircraft system I linked to.

            Heh, I missed that link, I didn’t know we already had airplanes like that, but I just made the same point. They’re talking about parachuting a huge amount of plane, and by the time you’ve done that, you might as well parachute the entire thing.Report

          • Morat20 in reply to Oscar Gordon says:

            Oh I agree. That’s why I said “good sounding but impractical” (or impossible). Because I can assure you that what the average person thinks is “We have parachutes for when you jump out of planes for fun, and also for emergencies when you have to jump out of planes. Parachutes work! So yeah, attach a parachute to a plane sounds like a great idea! Plane breaks, we parachute down!”.

            And if you ask them why they don’t already have them, you’ll get mostly “Dunno, probably costs too much to add or something” more than “Dunno, must not work even though I just saw a video that said it did” because of the aforementioned trust issues.

            (A fun fact on parachutes — returning capsules that use them deploy more than one set, and a critical design element is “How to make sure that, no matter WHAT, parachutes we want to get rid of before deploying the next set GO AWAY”. Because not going away would foul the next set of chutes. The answer involves explosives. Oh, they call it a chute cutter — but it’s a wedge of steel with an explosive charge behind it. When that thing goes off – -from what I am sure are triply redundant igniters, a fairly sharp chunk of steel slams into the chute lines at several hundred miles an hour)Report

            • Oscar Gordon in reply to Morat20 says:

              “We have parachutes for when you jump out of planes for fun, and also for emergencies when you have to jump out of planes. Parachutes work!

              People watch too much TV.

              What pilots actually use parachutes to escape a doomed aircraft? Combat pilots, and they have complex ejection systems to do the job. Why do they have said systems? Is it because every human life is precious and invaluable? No, it’s because combat pilots are damned expensive to train, so being able to recover them is worth the cost. Also, having an ejection system has a psychological benefit to the pilot, in that they have a possible out (helps them get into the cockpit in the first place, even knowing they will get shot at).

              returning capsules that use them deploy more than one set

              Yep, first set (drogues) slows you down enough to pop the second, second set (more drogues) slows you down enough to pop the third, third set is the one that can actually bring you down safely.

              The explosive cutter is cool, though. Is that the default method, or the emergency method?Report

              • Morat20 in reply to Oscar Gordon says:

                Been around since Apollo at least. It’s basically a guillotine blade fired into a soft metal “anvil” using a small explosive charge.

                Few points of failure and lightweight.

                With the shuttle, there was this umbilical arm that was connected to the Shuttle until main engine ignition. It supplied power while it was on the pad (therefore never touching it’s internal stores). Problem was, when the engines went off you had this arm still attached. They didn’t want to remove it when crew loaded on (sometimes they’d sit there a good long while). They wanted it on as long as possible.

                On the other hand, they only retracted it when the engine ignited — they couldn’t have a “failed to retract fully” problem — it’d scrape the shuttle and you’d be out one shuttle at the end. And it had to retract fairly fast.

                Their solution? A heavy weight, a cable, and a trapdoor. When they wanted the umbilical retracted, they yanked back the latch on the trapdoor (which, frankly, I wouldn’t be surprised to learn had an explosive charge as a backup). The door fell open, the weight dropped, the cable went taut and pulled the umbilical arm in. Gravity always works. 🙂

                Same idea with the chutes. If the igniter(s) works, that cord IS getting cut.Report

      • DavidTC in reply to Oscar Gordon says:

        I especially like the fact that the animators seem to realize that parachutes alone are not really workable, and appear to have some sort of *retrorockets* during the landing.

        I also like the fact they don’t seem aware that putting the emergency exit at the *cockpit* end (Instead of the side) would be exceptionally stupid, as you’d need *another* emergency exit to use if the cockpit had not detached, and where the hell is that stairway fitting anyway? And also causing the back of the airplane to open up is nonsense, as no passenger airplane has a sloped back like that. The passengers obviously go almost all the way back, because, duh, empty space is wasteful. (Which you mentioned in the article, but even non-engineers should know that.) Airplanes *already have emergency exits*, you twits.

        Again, I’m not any sort of mechanical engineer…this is *basic* information about airplanes.

        Also, they don’t seem to know large passenger airplanes *are more than one floor*, that there is an entire area down below filled with luggage and whatnot. Then again, they apparently don’t know how wings attach, either.

        Here’s the incredibly baffling thing about this: If this *actually worked*, a much saner solution would be to *put parachutes on the airplane*, and have the entire thing float down. If you had the pilots come *with you* instead of them staying behind in the cockpit to die, and put smaller controllable parachutes in the *wings*, you could even control the landing a bit. And at least that way the pilots can level the plane and slow down and *then* open the parachutes, instead of just tossing the thing out to roll wildly until it’s far enough that it can open parachutes.

        Of course, that would not actually work, for pretty much the exact same reason this wouldn’t work, and even if it did work, the situations in which it did would be almost nothing…but at least it wouldn’t require building impossible break-away airplanes.Report

        • Oscar Gordon in reply to DavidTC says:

          Yeah, I like the retro rockets. We have problems with batteries catching fire on planes, let’s mount some solid rocket boosters in the cabin, that won’t cause any issues…Report

  10. Oscar Gordon says:

    Totally unrelated to the topic at hand, but this is a fun video:

    Warning, very salty language!Report