30

I've seen several times on shows on television, and even talking to older pilots, where they will tell me that old planes were "over-engineered" to add some margin for errors on the designed engineering calculations. Of course, no specifics are given, and so I was thinking in this current age of super-fast computers, better wind tunnels, new software tools, and decades of actual experience, when people design new planes, how much of a margin to they design into them now?

It seems like confidence in computer modeling seems to be pretty high, and "over-engineering" seems to have typically meant making things stronger than they "need" to be, more reinforced elements, all things which add to cost and weight of an airplane which are factors that are at odds of profitability and performance.

Canuk
  • 8,442
  • 7
  • 47
  • 89
  • What do you mean under "old"? You can't say planes before 1920 were over-engineered, as they ware barely being able to fly short distances before their engine stalled or their wood and canvas wings snapped. – vsz Dec 26 '15 at 23:22
  • 1
    @vsz I think it would be fair to consider the large numver of Cessnas, Pipers, and similar still being flown that were built in the 1940s-1970s “old”, and everything after that (which has been able to take advantage of computers, wind tunnels, etc.) as mid-aged or new. – Slipp D. Thompson Dec 27 '15 at 02:39
  • 1
    @vsz I don't think that pre-1920 airplanes were, as a class, as incapable as that: Many of the aircraft of WW1 (1914-18) were capable of acrobatics and had ranges of hundreds of miles. – Wayne Conrad Dec 27 '15 at 16:16
  • 2
    Consider the amount of engines used on a plane - older planes tended to have more than necessary (4) simply because they wanted the plane to keep flying even if two engines failed, vs modern planes that will generally keep flying with only one engine out (with two engines normally) – user2813274 Dec 27 '15 at 16:30
  • 2
    @WayneConrad : if you look at the statistics, you'll see that almost half of the aircrew casualties were from accidents instead of enemy fire. – vsz Dec 27 '15 at 23:55
  • 2
    @SlippD.Thompson I guess I was a little vague. I was thinking the 1930-1970s. What prompted the question was Smithsonian's series on "Planes that Changed the World" where they featured the DC-3 and they talked about how it was "over-engineered" but just threw that term around without definition. – Canuk Dec 28 '15 at 20:45

5 Answers5

33

You are right that older designs used (and needed!) higher safety margins. I would, however, not call this over engineering. But that is besides the point.

The biggest advances have been made in materials! Not just new alloys, but much better quality control in manufacturing. The aluminium sheets and slabs you would get from a factory 50 or 80 years ago had much greater variations in local strength, and in strength between different production batches. The same goes for fasteners, forgings, whatever. Computer-controlled manufacturing and the relentless effort to improve quality and consistency have made this possible. Only wood is still the same as it was then, but Basil Bourque is right: Processing the wood has also improved by leaps and bounds.

But now consider what we do when we stress a wing: We use the maximum static lift coefficient, add aileron deflection and use that bending moment at v$_A$ to size our wing spar. We do this by applying a safety factor of 1.5 (§25.303). For parts with a more error-prone production process additional factors need to be added (see §25.619).

And now consider what happens in the real world: The dynamic maximum lift coefficient, which does not obey the rules and regulations imposed by the FAA, will easily be 1.3 times bigger than the static lift coefficient used for stressing the wing spar. In tests up to 150% have been achieved. Makes you think how sensible that safety factor really is. While we understand the static loads very well (because they are easy to study), dynamic and fatigue loads still happen to surprise the engineers from time to time. A little extra margin sometimes can be quite useful.

On the Eurofighter the structural safety factor was indeed reduced to 1.4 with the reasoning that more precise modeling and simulation would allow to use a smaller margin.

Community
  • 1
Peter Kämpf
  • 231,832
  • 17
  • 588
  • 929
  • 2
    If you want to add a real life example of exceeding the 1.5 safety factor in a dynamic situation, you could refer to American Airlines 587 (pdf). Page 61 shows the limit and ultimate design loads, and the loads created by the crew inputs. – DeltaLima Dec 27 '15 at 00:55
  • 5
    @DeltaLima: This is a good example of dynamic loads which would have been almost trivial to calculate, but were not covered by regulations. The safety factor is fine, what was missing was the consequence of repeated inputs at the eigenfrequency of the yawing motion. Airbus complied with the letter, but not the spirit of the (flawed!) law in this case. A good engineer could and would anticipate the higher loads. – Peter Kämpf Dec 27 '15 at 07:38
  • Is that really something that could've been anticipated? Of course you can assume that a pilot will do just about anything to the aircraft, but how realistic is it given that pilots are supposed to be highly trained and not put these kinds of loads on the airframe? Plus modern Airbuses (what's the plural of Airbus?!) flying in normal law won't allow these kinds of input to reach to the control surface anyway. – JustSid Dec 27 '15 at 11:06
  • I'm just wondering, obviously pilots are humans and make errors, but is the rate of that so high that the safety margins need to be increased so that pilots don't accidentally tear their planes apart in mid flight? It doesn't seem to happen very often from what I can tell. – JustSid Dec 27 '15 at 11:08
  • 4
    @JustSid: Yes, most pilots know what they are doing. On smaller aircraft without power boost they feel from the control forces what they do to their aircraft. But some are overconfident and do not realize how close they operate to the limits. Like the British certification pilot for the Do-228 who allowed the trim to run all the way forward and then could not pull out from the resulting dive. You are supposed to react within 1.5 s from the start of the trim malfunction. If you wait too long, that's it. His last words were "help me on the stick!". This is a general characteristic of complex ... – Peter Kämpf Dec 27 '15 at 12:45
  • 3
    @JustSid: ... technical systems that they are more complex under the hood than most operators assume. If they don't stick to the rules, they can die, be it in a nuclear power station (Chernobyl is a classic case) or in aircraft. Only very few people have the detailed knowledge to define the rules, but they don't operate the equipment. – Peter Kämpf Dec 27 '15 at 12:48
  • @PeterKämpf I totally understand where you are coming from, and I understand over confidence, but is this something that should be fixed by making the airframe stronger? Obviously I don't want to suggest to overburden the pilots with things they have to keep in mind, but to me as an absolute layperson it seems like flying the aircraft within its constraints shouldn't be too hard, especially since most aircrafts these days make it back to the ground in a single piece. – JustSid Dec 27 '15 at 12:53
  • 1
    As you can tell, I'm a programmer, I would rather the user gets fixed than the engineer having to do it :p – JustSid Dec 27 '15 at 12:54
  • A famous example of poorer manufacturing tolerances in the WWII era was a batch of defective engines made by the Wright Aeronautical Corporation: https://en.wikipedia.org/wiki/Curtiss-Wright#Defective_parts_sold_to_U.S._military_in_World_War_II . This was the basis for the Arthur Miller play All My Sons. –  Dec 27 '15 at 14:36
  • @JustSid: you can always beef something up, but you can also install discipline into the operators and inspectors to operate optimized equipment safely. Where do you draw the line? Would you make the wing and flaps strong enough to extend flaps at top speed? The beancounters at the airlines would not buy the product. We arrived where we are by trial and error and have a situation that is 99.99% safe already. – Peter Kämpf Dec 27 '15 at 16:34
  • 3
    Only wood is still the same as it was then. As for wood, hasn’t even that improved dramatically in recent decades as well with modern fabrication of plywoods? Better process control, better shaping, better glues and resins? (I am no expert) – Basil Bourque Dec 28 '15 at 02:26
  • @PeterKämpf -- of course, making it so that misoperation is benign is the best solution -- for your flaps case, instead of having something break if you exceed Vfe and try to extend the flaps, making it so the flap actuators stall/are pushed back by the airloads (i.e. airload-driven flap load relief) would be a better solution, no? – UnrecognizedFallingObject Dec 28 '15 at 05:30
  • 2
    @Basil Bourque LVL wood also improved in size/length ; There were also huge improvements over the quality of massive woods : computer automated imaging, automated x-rays controls etc.. – JB. Dec 28 '15 at 09:15
  • Thanks for the great answer. I guess that's why I used the term over-engineering in scare quotes, I knew there was a better word or phrase for what I was trying to ask. When I think about it, I was probably asking about what you called higher safety margins and @aeroalias called more conservative load estimates, but you definitely answered my question. – Canuk Dec 28 '15 at 20:52
  • @BasilBourque You are right, processing has improved tremendously. I was thinking of the raw materials. – Peter Kämpf Dec 29 '15 at 00:14
15

The legal "over engineering" factor has not been changed since 1970:

FAR Part 25, section 303:

Unless otherwise specified, a factor of safety of 1.5 must be applied to the prescribed limit load which are considered external loads on the structure. When a loading condition is prescribed in terms of ultimate loads, a factor of safety need not be applied unless otherwise specified.

[Amdt. 25-23, 35 FR 5672, Apr. 8, 1970]

Community
  • 1
DeltaLima
  • 83,202
  • 11
  • 272
  • 366
12

Over-engineering is a concept in value engineering, not aircraft design. In general, over-engineering means making the design more robust or unnecessarily complicated than necessary. In that sense, I would say that aircrafts are/were rarely over-engineered. Older aircraft were more conservative in determining the loads.

In structural design of any aircraft, the designer tries to make the aircraft safe in a number of steps:

  • Safety factor (which for transport aircraft, is 1.5)

  • Conservative material properties

  • Conservative loads.

The process followed now is pretty much the same; however, the engineers are able to determine the loads acting on the structures to a greater precision compared to the earlier times and are able to optimize the structure accordingly. The safety factor and taking of conservative material properties have not changed much.

Though one can say that, for example, making the wing spar thicker than necessary to account for unforeseen loads is good from structural point of view, it is bad from performance/weight/fuel consumption views. The problem is that if the designer knowingly pads the safety factor more than necessary, he's actually reducing the aircraft performance. As a result, over engineering (more than design or regulatory requirements) is plain bad design and is not recommended.

In short, the process haven't changed over time- it is simply that the tools have improved and more data is available for use.

There are other issues here- aircraft like F-35 Lightning II are repeatedly called over-engineered, but the point is that they were designed for the given specifications. For example, F-35 was required by design to perform the tasks of a number of aircraft and as such is having a significantly difficult engineering period.

aeroalias
  • 100,255
  • 5
  • 278
  • 429
  • 2
    +1 for pointing out the difference between "over-engineering" and safety margin. All mechanical and civil engineers dealing with safety-critical systems use safety margins in their calculations. The aerospace industry is no different. "Over-engineering" normally means making a system more complicated than it really needs to be to get the same job done. This usually tends to result in less robust systems, since more complex systems have more things that can go wrong. – reirab Dec 27 '15 at 20:13
3

On a different note from the other answers, there's been a few articles indicating that newer 'bone' like designs are being considered and implemented by companies like Airbus.

Airbus has partnered with Autodesk to rethink the design of those lowly partitions. Its new partition debuted today at the Autodesk University conference in Las Vegas and, thanks to 3-D printing and some wild new algorithms based on slime mold and bone growth, it weighs in at just 66 pounds. Airbus’s current partitions weigh 143 pounds apiece. “Our goal was to reduce the weight by 30 percent, and we altogether achieved weight reduction by 55 percent,” says Bastian Schaefer, innovation manager at Airbus. “And we’re right at the beginning.” http://www.wired.com/2015/12/airbuss-newest-design-is-based-on-slime-mold-and-bones/

There are even some pretty impressive futuristic designs:

enter image description here

http://www.airbus.com/innovation/future-by-airbus/the-concept-plane/

http://www.smithsonianmag.com/arts-culture/aircraft-design-inspired-by-nature-and-enabled-by-tech-25222971/?no-ist

NPSF3000
  • 157
  • 3
  • 2
    Rest assured, this concept plane pictured is as likely to become reality as those prognoses about the year 2000 from the Fifties. Writing this from my underwater habitat which can be reached by flying car only. Yeah, right! – Peter Kämpf Dec 27 '15 at 08:09
  • @PeterKämpf sure which is why I lead with an article showing how these techniques are being considered (at small scale) today. The point isn't that this is THE FUTURE but rather that there is room to improve current designs, in part using new and improved CAD techniques. I figure the OP might be interested in such developments. – NPSF3000 Dec 27 '15 at 08:11
  • 5
    There are many small parts in today's aircraft which can benefit mightily from getting a little more attention. Most are left-over compromises which worked well enough not to be looked at in detail later, so they were never analyzed or optimized properly. – Peter Kämpf Dec 27 '15 at 08:49
  • Is that structure really futuristic, or is it revisiting Barnes Wallis' WW2-era geodesics? – user_1818839 Dec 27 '15 at 16:22
  • @PeterKämpf Care to elaborate on those details a bit? Sounds much more interesting than the picture above! – Pavel Dec 28 '15 at 13:31
  • 1
    @PavelPetrman check this example out: http://www.wired.com/2015/12/airbuss-newest-design-is-based-on-slime-mold-and-bones/ – NPSF3000 Dec 28 '15 at 23:21
2

I think calculating stresses and loads are no better now than 50 years ago. Engineers understood forces and physics no differently then than now. They just used a slide rule instead of software, but they got the same results. Instead of modeling things, they build real life ones and tested them. They got the same answers. They built a wing, bent it till it broke and discovered its strength and were able to build one to any load specification needed without overbuilding it. Modern software just gets results cheaper and faster.

The big differences between then and now are materials and system evolution. Not calculating forces and stresses.

Peter Mortensen
  • 524
  • 3
  • 8
Steve H
  • 1,961
  • 12
  • 16
  • 4
    You cannot calculate stresses/strains/displacements in an arbitrary mechanical system, at least not in an analytical way (closed form solution) and not very accurately. But you can get very accurate results with a known margin of error with numerical mechanics, e.g. with the Finite Element Method (FEM). In short, I strongly disagree with your calculations statement. On a sidenote, I do think that experiments are still useful and still used in order to validate your mechanical models. However, I do agree with your materials and construction evolution statement. –  Dec 26 '15 at 16:41
  • 2
    Please note that there are still a lot of systems, which cannot be modelled and simulated satisfactory - not even with the Finite Element Method. Examples include contact mechanics, fluid-structure-thermo-interactions, damage modelling (e.g. wear, fatigue) and many more! –  Dec 27 '15 at 16:13