Oh boy, composites! That brings me back. Time for another story, from the lands of not-necessarily-big-B-but-could-very-well-be, told by someone who has no idea whatsoever what he's doing.
New product, new structures. Composites! Leadership says that Flight Science needs to sim every single top level assembly permutation - that's a hundred some different airframes, in a few dozen flight regimes each. Now, if you paid money for a decent CAD/CAM package, those big packages have prefabs for all sorts of things when it comes to composites - you don't need to sim from nothing. This being the beloved military industrial complex[1], no one wants to pay money for anything, so sim from nothing it was. Poor bastards in Flight Science spend seven hundred hours simming every single flyable configuration.
Why did I care? I was keeping a thumb in this pie because I planned to integrate with their sim runs, keeping my team from needing to drawn graphs manually every time they updated. Hook up the data to a graphing doohickey, key in the right aircraft / configurations, and never need to draw graphs again. Anyway.
A week goes by, I send an email. Nothing. A month goes by, I drop by their hallway with sandwiches to get the lowdown. After the "SIM EVERYTHING" directive, a dozen or so major changes and additional flyable configurations got railroaded into engineering release. So they ran out of time to sim anything before the heat death of the universe. Now, hold on a second, I said. The military needs operational limits and maintenance schedules - how do you know how many flight cycles, say, this widget here can take before it breaks? Whelp. The Overlords got on the horn with the military procurement dudes and they basically ballparked a number - say, 20 flight hours, for this one carbon component - that it would need to be inspected and serviced. They did this for everything, dreaming up a LORA out of thin air. But these components had only been through, at best, a partial model run, and since everything was changing twice weekly, some of those part configurations never saw one single pixel side of a stress-strain analysis. The military procurement guy waived absolutely everything; I want to emphasize this, because this sort of thing, it's much harder to get away with in civilian, pax-carrying aircraft.
Something to be emphasized here is that composites don't deform, they don't absorb energy as they fail. They fail, everywhere, all at once - the nonlocality of it can be a little scary. What this means is that the margins are razor thin. Remember that little part with the 20 hours? Weeeeellllllll turns out that got halved by saltwater/spray, then halved again for extreme cold, and halved because extreme 3-axis vibration . . und so weiter, und so weiter. Our little part didn't make it through one single climbout. Even more fun, since everything else was so close to its limits, when that little part went, everything it was attached to got whomped, and 20% of those other parts failed. To draw a big picture: imagine an aircraft that takes off, then before it gets a few thousand feet it . . sort of self-disassembles. The ones that do manage to make it back - keeping enough lifting surface and controls to wing around - are totalled - the airframe's thrashed, with so many loose fasteners inside it sounded like a pair of maracas. It's sort of impressive that it's even possible to build something so fragile it springs itself into bits. It's like a prank plane.
And I never did get that flight sim data to make my team's life any easier. The sim runs were never allowed to get to completion, and no sim, no data, no graph widget. So the graphs got drawn just like they always did; that is, from the imagination of one of the senior flight ops guys. "Just draw this curve . . up . . up . . and ooooooover . . . and then it comes back down again. Science!"
Alright, what's the lesson here? What's the takeaway?
Well, aside from the obvious (don't hire fast food franchise executives to run your airplane company), composites are very different, and old school engineers used to steel and aluminum really need to think when they're doing stress and risk analysis. Sure, they're strong as hell, but in the same vein, when they fail they break. They're done. I think the CAD systems have a good background for handling this, but the LORA and the risk methods haven't caught up - they're still stuck thinking about metal.
[1] And also a place where none of the Overlords had any aerospace background; in fact, most of them don't even have STEM backgrounds. One VP Engineer thought FEM (Finite Element Method) modeling was a euphemism for something dirty.
You weren't working on the F-35 by any chance, were you? Because that could be the start of an explanation as to why the development program went sideways. :-)
New product, new structures. Composites! Leadership says that Flight Science needs to sim every single top level assembly permutation - that's a hundred some different airframes, in a few dozen flight regimes each. Now, if you paid money for a decent CAD/CAM package, those big packages have prefabs for all sorts of things when it comes to composites - you don't need to sim from nothing. This being the beloved military industrial complex[1], no one wants to pay money for anything, so sim from nothing it was. Poor bastards in Flight Science spend seven hundred hours simming every single flyable configuration.
Why did I care? I was keeping a thumb in this pie because I planned to integrate with their sim runs, keeping my team from needing to drawn graphs manually every time they updated. Hook up the data to a graphing doohickey, key in the right aircraft / configurations, and never need to draw graphs again. Anyway.
A week goes by, I send an email. Nothing. A month goes by, I drop by their hallway with sandwiches to get the lowdown. After the "SIM EVERYTHING" directive, a dozen or so major changes and additional flyable configurations got railroaded into engineering release. So they ran out of time to sim anything before the heat death of the universe. Now, hold on a second, I said. The military needs operational limits and maintenance schedules - how do you know how many flight cycles, say, this widget here can take before it breaks? Whelp. The Overlords got on the horn with the military procurement dudes and they basically ballparked a number - say, 20 flight hours, for this one carbon component - that it would need to be inspected and serviced. They did this for everything, dreaming up a LORA out of thin air. But these components had only been through, at best, a partial model run, and since everything was changing twice weekly, some of those part configurations never saw one single pixel side of a stress-strain analysis. The military procurement guy waived absolutely everything; I want to emphasize this, because this sort of thing, it's much harder to get away with in civilian, pax-carrying aircraft.
Something to be emphasized here is that composites don't deform, they don't absorb energy as they fail. They fail, everywhere, all at once - the nonlocality of it can be a little scary. What this means is that the margins are razor thin. Remember that little part with the 20 hours? Weeeeellllllll turns out that got halved by saltwater/spray, then halved again for extreme cold, and halved because extreme 3-axis vibration . . und so weiter, und so weiter. Our little part didn't make it through one single climbout. Even more fun, since everything else was so close to its limits, when that little part went, everything it was attached to got whomped, and 20% of those other parts failed. To draw a big picture: imagine an aircraft that takes off, then before it gets a few thousand feet it . . sort of self-disassembles. The ones that do manage to make it back - keeping enough lifting surface and controls to wing around - are totalled - the airframe's thrashed, with so many loose fasteners inside it sounded like a pair of maracas. It's sort of impressive that it's even possible to build something so fragile it springs itself into bits. It's like a prank plane.
And I never did get that flight sim data to make my team's life any easier. The sim runs were never allowed to get to completion, and no sim, no data, no graph widget. So the graphs got drawn just like they always did; that is, from the imagination of one of the senior flight ops guys. "Just draw this curve . . up . . up . . and ooooooover . . . and then it comes back down again. Science!"
Alright, what's the lesson here? What's the takeaway?
Well, aside from the obvious (don't hire fast food franchise executives to run your airplane company), composites are very different, and old school engineers used to steel and aluminum really need to think when they're doing stress and risk analysis. Sure, they're strong as hell, but in the same vein, when they fail they break. They're done. I think the CAD systems have a good background for handling this, but the LORA and the risk methods haven't caught up - they're still stuck thinking about metal.
[1] And also a place where none of the Overlords had any aerospace background; in fact, most of them don't even have STEM backgrounds. One VP Engineer thought FEM (Finite Element Method) modeling was a euphemism for something dirty.