United Airlines flight 232: how hard to control?

The crash of United Airlines Flight 232 came up in a recent thread. (TL,DR: in 1989, uncontrolled engine failure disabled all three hydraulic systems at cruising altitude, pilots could only control plane by varying left/right engine thrust, nursed it down to a crash-landing that allowed 60% of occupants to survive.)

ISTR reading/hearing at that at some point after the crash, this scenario (i.e. a complete loss of hydraulics during cruise) was programmed into a flight simulator, and it was found that no one else was able to successfully control the plane (via differential engine thrust) the way Capt. Haynes and his colleagues did.

Is that an urban legend, or is it true? If true, did the simulator scenario include having the pilots aware from the start that they were supposed to try to control the plane with diff thrust, or did it involve totally uninformed pilots facing a surprise loss of all hydraulics (which would leave them having to figure out for themselves in the space of a few seconds, like Haynes et al., that they needed to try using diff thrust)?

Seems they really did try simulator testing according to this report by the Chicago Tribune. And it really was uncontrollable uness they simulated landing on a vast level open space.

Since then there’s been some practicing this stuff. And other outrageous mishaps like Sully’s.

Keeping the airplane right side up isn’t too hard on a nice day if . Steering left/right within a mile or so is doable. Much tighter is damn tough.

Controlling touchdown point fore/aft is also tough within a couple miles. The big thing is your descent angle is non-standard. Once you turn final you’re sorta stuck with where that descent angle intersects the surface. Given enough time and fuel you could make multiple passes until you learned how to judge that.

But every moment you’re staying in the air is another moment wherein something upsets that apple cart and you lose control. Like walking a tightrope, as long as you’re well-centered you’re OK. Once you start wobbling, it’s easy have that amplify until you lose it.

This has happened more than once.

In 1985, a Japanese flight lost all of its hydraulics. They managed to keep the plane in the air for a while, but had difficulty steering it and maintaining control of it. They eventually crashed into a mountain.

In 2003, a DHL plane was hit by a missile over Iraq. They managed to land the plane safely. I believe that this was the first time that a plane suffered a total loss of hydraulics and they managed to get the plane down onto the ground in one piece. The plane veered off the runway and plowed through a bunch of sand, so it wasn’t exactly a pretty landing, but the plane didn’t roll over and cartwheel like United 232 either.

In both of these cases, the only control they had was differential thrust, like United 232.

Also in both cases, the pilots figured out pretty quickly that all they had was differential thrust, and they then set about figuring out exactly how to control the plane doing exactly that.

I’ve seen Capt Haynes speak a few times at unit training days - he is a very unassuming guy, and really talks about the team effort it took to get the aircraft on the ground (makes sense, as his crash is often taught in crew resource management (CRM) training for just that). A confluence of events enabled the plane to crash land, rather than just become a smoking hole (the instructor on board), and allowed the folks injured on board to survive (the airport had crash training in the recent past and the hospital had an oncoming and off-going shifts their at the same time), but you can’t forget that he did a great bit of leadership in getting that plane down.

For sure. The fact that since then folks have considered these kinds of failures and rehearsed them is nice and all. But being the first to encounter a situation that was supposedly impossible and then having the cool & calm to not lose control in the first 30 seconds is the real game-changer here.

Had they augured in from altitude nobody would have thought the crew was a failure; the impact would have been declared inevitable if they were able to determine the total hydraulic loss from what little wreckage would have survived.

The infamous “Gimli Glider” incident, where due to a metric-to-imperial conversion error, a 767 ran out of fuel and essentially glided down to an abandoned air force base halfway across Canada - I recall reading that the hydraulics died when the engine died (or a bit later when the auxiliary power units ran out) and to maneuver the plane, the pilots were essentially standing on the pedals to make them move.

I assume this is different from the other incidents because the hydraulic lines were intact, it was just the power assist pumps than stopped working.

The Gimli Glider still had some hydraulic pressure from the Ram Air turbine. Not as much as the engines would provide, and they had some trouble getting the landing gear down and locked, but enough for the plane to be flyable.

Yup. 767 & 757 have a https://en.wikipedia.org/wiki/Ram_air_turbine which provides normal hydraulic pressure, but vastly reduced quantity even if both engines are seized and you have no electricity.

Absent that the airplane is totally uncontrollable. But with it you can steer OK-ish. The big difference is much larger cockpit control displacements produce much smaller & slower movements of the control surfaces. It’s not mechanically like a car without power steering, but the net effect is the same: Pull like heck & wait a long time for the undersized result to appear.

The Gimli Glider guys did the right thing for an engine out approach: aim to land long and then use a forward slip to burn off excess energy once they’re sure they’re not going to land short. What almost cooked their goose was the RAT had barely the capacity to hold the constant aileron and rudder displacement they needed. Which fact was unknowable by them before they tried it.

The Japanese flight never really had much control. It’s altitude continued in phugoid waves and heading-wise the crew couldn’t even keep it over the ocean to avoid populated areas. They merely kept it in the air as long as possible because, well, what else could they do? It eventually hit the side of a mountain at enough of a glancing angle that four people survived. Also unlike Sioux City it was a 747.

The ram air turbine, which all airliners have for emergency use, provided bare minimum hydraulic power for the Gimli, enough to control the plane. The landing gear were lowered using a backup ‘gravity-drop’ method, however because the nose gear had to swing ‘forward’ into the wind it never fully deployed or locked (the airliner skidded on its nose). Many things fell into place for that case as well. One of the pilots was an experienced glider pilot, one of them knew about the old Air Force Base at Gimli (the only strip close enough), the nose gear collapsing allowed the plane to not overshoot the end of the runway (and crash into the crowds of people or their cars) etc. The crew did such a good job of landing that the plane suffered only minor damage. It was repaired on-site enough to be flown out of Gimli, fully repaired and returned to service.

If I recall correctly, the pilot in this event had unpowered glider experience to apply to the problem as well.

Well, it was Iowa. :slight_smile:

Only four were rescued alive. A hell of a lot more survived the crash. Japanese authorities inexplicably turned down the assistance of the US military, who would have been on the scene within an hour. The Japanese only sent out rescue teams the next morning, some ten hours after the crash.

Well the Japanese had very poor ability at control, they pitched and banked, perhaps because they were trying to fight bank with differential thrust.

The Iraq plane was an airbus, and as it was in take off, it may have been in neutral attitude for rudder and bank… with pitch stable. (not extreme… )
Much better than having two or all three in a bad setting…

The OP ( United crash at Sioux) plane had some control, seemingly they had pitch and bank left at a fairly neutral setting, but they had rudder and they couldn’t get differential pitch to keep it straight, they could slow their turn but not stop it.

Its understandable that no one in simulator could do better, the simulator is quite accurate in terms of the physics of the flight, but the feed back to the pilot isn’t quite the same , and the pilot isn’t in a panic situation…they’d feel exhausted and drowsy… In the real situation, the pilot fills up with adrenaline… The pilot filled with adrenaline is not as good as casual relaxed pilot with nothing to worry about, but better than being drowsy from exhaustion… So the sim test isn’t really the same… its just too much to think about without adrelanin, so the question about whether a computer can do it remains…

The computer may have to be programmed to learn the turning capability of the damaged aircraft, and then use that to predict the touch down to coincide on a runway…

On the other hand, the computer won’t think of an out of the box solution, such as a wide open disused carpark that is easier to come down on than a runway surrounded by corn fields ???

The plane just happened to be trying to land at Sioux, which unfortunately had soft ground beside the runway…

I’m going from memory here so I hope this is accurate, but I believe all three planes ended up oscillating in phugoid waves. The Iraq and Sioux City plane pilots managed to get control of the oscillations using thrust. I believe the Japanese pilots got some control over their plane as well.

It seems to me that one of the key differences is that when the Japanese pilots lowered the landing gear, it made the plane less stable, where the Iraq plane became more stable.

I don’t recall reading anything one way or the other about the effect the landing gear had on the Sioux City flight.

Landing gear extended will tend to reduce yaw stability and produce a pitch-down moment. Whether that’s enough to upset the apple cart is mostly a matter of luck.

Landing gear extension will also produce small transients in all 3 axes as everything doesn’t move exactly in sync. This is even more true for backup gear extension methods. Which methods also leave various doors hanging open that are normally retracted after the gear is fully extended. Which extra doors further reduce yaw stability and increase pitch-down moment.


What impressed me was the news photo - they landed in Gimli so precisely that the nose of the plane came to rest centered right on the barrier running down the middle of the runway that had been converted to a drag strip.