Recently a couple of pilots found themselves in a situation that was foreign and perplexing to them; a scenario the designers of the airplane hadn’t fully expected. They fought their way for 3 minutes and 30 seconds while trying to understand what was happening after a failure of one of the pitot static systems on their Airbus A330. At times the flying pilot’s inputs exacerbated the problem when he assumed they were flying too fast rather than too slow.
Because they hadn’t seen anything like this in the simulator, and the airplane was giving conflicting information, the recovery would have been all the more difficult.
Pilots are taught that an erroneous airspeed indicator can be countered by paying close attention to their pitch and power. It sounds so simple that many pilots wonder aloud, just how anyone in the situation could mess it up.
In the early morning hours of June 1st, 2009, the pilots of Air France flight 447 were working their way around thunderstorms while flying from Rio de Janeiro to Paris in the widebody Airbus A330.
A faulty pitot tube created a situation where any changes in pressure resulted in fluctuations in the airspeed indicator. To understand how difficult it is to recognize this problem and then correct for it, let me use the following analogy:
Imagine you’re driving a car at night. You come down a hill and you feel the cruise control back off on the gas to prevent the car from going too fast. Just as you look down at your speed noticing that it is, in fact increasing, a siren and lights go off behind you. A police car has woken you up from your late night drive.
Instinctively you kick off the cruise control and apply the brakes. The speedometer indicates you’re still accelerating, so you press harder on the brakes. Your car has now decided that because you’re trying to slow so quickly, it will shut off the anti-skid braking system and allow you to use manual brakes. You then skid off the road and into a ditch.
Based on the released information about one of the most mysterious accidents in recent history, it appears the pilots of Air France 447 faced a set of circumstances similar to our driving example.When flying in turbulence, it’s important to watch your airspeed. Flying too fast will result in a situation called mach tuck, where the nose can slowly pitch over and the controls lose their effectiveness.
Flying too slow can result in a stall. Not an engine ‘stall’ as might be incorrectly reported by the press, but an aerodynamic stall where the wings aren’t developing enough lift, and an immediate increase in the airspeed is needed to recover. Here’s a tip for reporters. In aviation, the term ‘stall’ will never be used to describe engines that fail. Ever.
Up at altitude, the difference between flying too slow and too fast can be as little as 20 knots. It’s called the ‘coffin corner,’ a morbid term used to describe narrow band of airspeed that we need to maintain.
In this 767 example above, the airplane has a very safe margin between
too slow and too fast, as shown on the airspeed indicator on the left.
While working their way around clusters of cumulonimbus clouds in the inter-tropical convergence zone that night, our two first officers (the captain was in the back on his planned rest break) did their best to stay away from the weather.
A side note: whoever takes the second break is usually the pilot who made the takeoff and who will also make the landing. So the relief pilot (who’s a type-rated copilot) took over the flying related duties while the captain slept.
Back to the flight: During turbulence, maintaining that speed can be more difficult, much in the way it’s tough to hold the speed in a car going over hills. You may look down after studying the weather only to notice that the auto throttles aren’t holding the .80 mach speed you have selected and the airplane is now at mach .83 and accelerating. In a moment, the clacker goes off, indicating you’re now exceeding the normal cruise speed of the airplane, which certainly gets your attention, much like the sirens of the police car in our example.
In the case of Air France 447, the autopilot kicked off in response to the overspeed, and was followed by a warning Airbus calls a ‘cavalry charge’ sound which is designed to get your attention quickly. About 30 seconds later the auto throttles were turned off manually and the throttles were pulled back, but it takes an eternity to slow down such a slippery airplane, and it may have seemed to the flying pilot that he was still accelerating anyway. So he pulled the nose up, an effective way to slow down in a critical situation like this. (See last week’s post on the eight ways to slow a jet.)
Amazingly, as the airplane climbed from 35,000 feet to 38,000 feet the airspeed continued to increase, at least that’s what it looked like on the flying pilot’s side of the airplane. He must have been surprised then to hear the stall warning activate moments later, indicating that they were flying too slow.
The other pilot likely noticed the airspeed on his side was decreasing, and perhaps because he saw the difference between both airspeed indicators, he’s heard to say on the recording that “we’ve lost the speeds.”
They had slowed from 275 knots indicated to 60 knots, at which point the airplane went into a mode called ‘alternate law’ which meant the automatic protections that kept the airplane from stalling were removed.
To make matters worse, the stabilizer trim moved from 3 degrees to 13 degrees nose up, which meant the airplane may have needed almost full nose down inputs on the stick just to fly level.
And to further confuse and confound the pilots, it’s recently been reported that as the airplane slowed further, the stall warning stopped. When max power was applied and the nose was lowered at one point, the stall warning came back. This is opposite of what the pilots were looking for in a recovery.
The airplane ‘mushed’ in a 15 degree nose up attitude all the way to the water, at a rate of 11,000 feet per minute.
We occasionally train for unreliable airspeed indications, but it isn’t covered during every recurrent training period. Stall training is often limited to the low altitude variety, which is far less critical than one occurring at 35,000 feet. I’m certain training departments all over the world will soon be required to train for high altitude stall recoveries.
Since this will take some time to become a requirement, on a recent simulator session, I asked my instructor to give me a loss of airspeed scenario at altitude. I told him I’d prefer to have the failure at any random point during our four hours of simulator time that day.
When he eventually failed it, causing the airspeed to slowly increase, I immediately pulled the throttles back and raised the nose a bit. The non-flying pilot simply said ‘airspeed’ which I thought was obvious, as it appeared to me that the airplane was accelerating rapidly and I was doing my best to get it back under control.
But on his side, the airspeed was dropping rapidly. When he said “airspeed” he actually meant that the airspeed was slowing and that I needed to do something about it. I finally looked over at his side, and saw that his speed was actually decreasing while mine increased. This all occurred within ten to twenty seconds.
I immediately lowered the nose and told him that I suspected my airspeed indicator had malfunctioned. Since my indicator was useless, I offered the airplane to him.
It’s easy for pilots to harp that “pitch and power equals performance” but it’s not easy to ignore the instruments you’ve trusted for thousands of hours. For the pilots of Air France 447, the incorrect airspeed indications and confusing stall warning sounds that were caused by a failure of the pitot static system proved to be too much to handle.
Furthermore, the Airbus design reinforced ideas that counter everything a pilot is taught, Specifically, these pilots learned that pulling the stick full aft would not result in a stall when the airplane was operating under a condition known as “normal law.” Much of their careers had been flown in airplanes with this feature. That night, after the initial climb, they were operating under “alternate law” which allowed far greater changes to the flight envelope, and removed that protection.
I wouldn’t have wanted to be in their shoes.
Much of the focus of the accident in the press has been to blame the pilots for clearly stalling the plane. One strange headline read “Baby pilot at the controls of AF 447.“
The ‘baby pilot’ was actually 32 years-old and had previously flown an A320 for 4 years and the A330/340 for just over a year. He had 2,936 hours of flight time with 807 hours in the A330/340.
The other copilot, at age 37 had 6,547 hours with 4,479 of them in the A330/340. Sadly, his wife was also on board the aircraft as a passenger.
And the 58 year-old captain, who came to the cockpit from his break halfway through the event had 11,000 hours of which 1,747 were in the A330/340.
I’ve said it before; in the eyes of the media, pilots are either heroes or villains depending on the outcome of the flight. These pilots faced challenges few of us have ever come across. Given the mechanical failures that started the chain of events, there’s certainly plenty of blame to go around. Events like these have a profound impact on our training and help prevent future accidents. And at least that is something we can be thankful for.
In case you’re interested in even more details, AvHerald has an excellent summary of the BEA preliminary Air France 447 accident report.
Cockpit Chronicles takes you along on some of Kent’s trips as an international co-pilot on the Boeing 757 and 767 based in New York. Have any questions for Kent? Check out the Cockpit Chronicles Facebook page or follow Kent on Twitter @veryjr.