Spook Story

This here is a post which you probably shouldn't read if you hate to fly. A pilot raising safety-related questions about airplanes just doesn't make for good bedtime reading if you're squeamish about these things.

I had to catch a flight up to Montreal the other day, so that I could spend the next day in a hotel before flying the airplane back to Kentucky. Then go home. (Now that's my idea of a back-breaking work week.) Anyway, the flight out of O'Hare up to Montreal was on a new Embraer 170, a new airplane design which I'd seen taxiing around but hadn't ridden on yet. It looks like a mini Airbus A-319, a generic modern twinjet. Inside, it has a three-row 2/1 first class section, and about 18 rows of 2/2 economy class for roughly 80 passengers. Nice airplane. I stopped by the cockpit on my way in to chat with the flight crew about their new toy.

I used to fly an Embraer E-120 "Brasilia" at my first airline job, and while not a very big airplane, it was my first plane with an airline cockpit, one which is operated like any other large airplane. (I remember the stories about the ordeal of going down to Brazil to pick up new airplanes from the factory as they were produced. A substantial amount of cash--a couple hundred grand per airplane--was needed for payoffs in order to get the airplane out of the country. I wonder if these stories were true and if the practice continues.) The Brasilia was not, to me, a very impressive airplane. It was fast, but we had lots of maintenance issues with it, and the propeller systems needed to be treated with respect, as prop mishaps at other companies had brought down, or nearly so, several airplanes (the propeller system of a turboprop is a hardy system, but mechanical mishaps with the system can be problematic; the lack of this system on a jet is one of the things that makes jets statistically safer). The new E-170 is a jet, so that problem needn't come into play.

But on our way out to the runway the pilots reported that they had an electronic glitch caused, they thought, by a strong blast of jet exhaust to one side of our airplane from a taxiing 767. This caused (I'm speculating now) the sensors on one side of the airplane--pitot tubes, static ports, angle of attack indicators, etc.--to register a "weather" phenomenon that the other side did not agree with. Ergo, error message. That's simple enough, but what's odd to me is that it's a non-clearable error. Their solution was to completely de-power the airplane and let it sit dark for seven minutes and then reboot things. Like rebooting your computer (exactly like that, actually), this was the generic system reset that would fix things, so they speculated. And they were right. They gave us the option of going back to the gate and going inside while they did it, or just agreeing to sit in darkness out by the runway. We all chose the latter, and after a few minutes' pause we were up and running and on our way to Montreal. End of story, I suppose.

Now, I've flown a couple electronics-intensive airplanes in my career, and I recall doing exactly this thing (come to think of it, at exactly the same spot on the O'Hare field in one case). But here's my issue: in the airplanes I flew, these electronic glitches might have caused my navigation display to show erroneously, or maybe one of my systems warning panels (air conditioning, hydraulics, etc.) to annunciate incorrectly. But this new Embraer is, like virtually all Airbuses and all Boeings from the 777 onward, a fly-by-wire airplane. This means that the pilots, by manipulating the flight controls, are actually operating computer input devices--joysticks, essentially--that tell the computer what they want to do, and the computer arbitrates whether that's a good idea or not. The computer is doing the flying, and the pilots are inputting to the computer. The safety built into this system is that the airplane is prohibited by programming from doing things like going too fast or too slow or rolling upside-down.

But the real motivation in using fly-by-wire technology in a civilian airplane is economic. I don't want to give a lesson on aerodynamics here, but the stability of a traditional airplane is a function of its weight distribution, and that built-in stability costs fuel in the form of additional drag. The main rationale for civilian fly-by-wire use is that the basic airframe can be designed to be neutrally stable, that is, without any inherent (and fuel-sucking) stability; and the stability and docile flying manners are then supplied by software. This setup saves gas, pure and simple. (Military aircraft also can be designed to be inherently UNstable for maneuverability benefits, and then made flyable by electronics; basically, the survival parameters stipulated by having an organic-matter pilot on board are programmed into a computer, keeping that pilot safe in an airplane whose capabilities vastly exceed his or her own.)

So my question is this: if it's actually a computer system which is in ultimate control of the airplane, what prevents an electronic glitch in THAT system in flight, a non-clearable error in the flight control system? De-powering an airplane is not an option in flight, and especially when electricity is the only connection between pilot and flight controls. The electronics-intensive airplanes I flew were actually regular, mechanical airplanes with some sophisticated electronic systems, so this issue did not apply. Likewise, my DC-8 is flown with cables and judicious use of hydraulics, and it has considerable backups in case of a malfunction in any of those mechanical systems. We rely on electronics for very little in the old steam-powered DC-8, so this issue just doesn't come up for us. But fly-by-wire airplanes are the wave of the future.

I can't say I'm worried about it, because I'm confident the issue is addressed in some way--otherwise (like claims that the aspartame in my beloved Diet Coke causes cancer) I think we'd see ample evidence of a problem at this point. But I'm curious, because I've not been shown how they get around this potential problem. I remember something about the space shuttle using five identical computers and giving each one a vote, and then going with consensus opinion. Is that the way? Do redundant computers reboot themselves, one at a time? The question may well be more interesting than the answer, but I'd love to know. (I have a couple friends who fly the Airbus A-320--the world's first commercial jetliner with a fly-by-wire control system--and maybe they'll come to my rescue here.)

Lastly, I can't help thinking that Burt Rutan and Boeing and Beechcraft and many others have developed airframes with canard wings which provide all lifting surfaces--that is, airplanes who achieve the same or better efficiency as fly-by-wire without resorting to this isolating electronic layer. Beech's airplane, a turboprop pusher, was called the "Starship," and it went into production but never amounted to much. Boeing's idea was called the Sonic Cruiser, and it looked beautiful, like a huge jet version of the Starship. But the public just couldn't be brought along when the airplane looked so different. And so the solution we're stuck with to keep bodies in the seats is one which seems much less satisfactory to me (though I honestly don't know whether the Sonic Cruiser was meant to be fly-by-wire or not). I know I've never met any pilot who thinks the basic idea that a computer does the flying and the pilot informs the computer is a good one. Maybe we're all overconfident control freaks, but giving the final say in things to a computer goes against all our training. And when there are predictable problems with giving computer control to any other system, you have to wonder...

OK. You can open your eyes now. Scary stuff all gone.