Thanks to the work of David Learmount at Flight Global, and that of the Wood's Hole Oceanographic Institution and the Bureau d’Enquêtes et d’Analyses, enough is now known about this accident to start looking for useful lessons and to analyze the data along with the BEA. Flight safety and the future of the piloting profession depend on this becoming a wide and serious conversation.

Pilots obsess about accidents for good reason. There is always so much to learn. The AF447 tragedy is an epochal example.

There is a mind-boggling number of lessons to be learned here, in a host of areas and disciplines: Pilot Training, Standard Operating Procedures, Instrument Flight, and Aircraft Design are but a few of them.

I will commit today to joining the conversation. I begin with a consideration of Angle of Attack.

Angle of Attack

Wolfgang Langewiesche (father of William) emphasized Angle of Attack in his excellent Stick and Rudder, published in 1944. Advocacy of AoA was an uphill battle then and it still is today. Instead of talking about AoA, we prefer to use airspeed and explain why certain speeds we use change with aircraft weight and G loading. Many or even most aircraft flying today have no Angle of Attack indication. The accident aircraft had two AoA sensors. The flight recorders had access to the signals from these sensors but the pilots did not, at least not when they needed it most.

Lift is produced when the air flowing over the top of a wing has a longer distance to travel than the air flowing underneath. The air “stretching out” over the top produces a lower pressure, allowing the higher pressure underneath to push the wing up. There is a caveat, however. The airflow must remain attached to the upper surface of the wing.

Imagine a cross-section of wing, with a line drawn from the middle of the rounded leading edge to the pointed trailing edge. This is the chord line. Now imagine an arrow pointing at the leading edge. This is the airflow.

If the arrow meets a (symmetrical) wing head-on there will be no lift. But let the wing meet the air at a slight angle and the airflow around the wing will no longer be symmetrical: it will meet the rounded leading edge at an angle and it will divide lower on the curve of the leading edge. The air flowing over the wing will have a longer distance to travel. Lift will be produced.

The angle at which the airflow meets the chord line is called the Angle of Attack. Up to a point, increasing the Angle of Attack will increase lift. But beyond a certain point – usually about 16° – lift will instead decrease because the airflow is beginning to separate from the upper surface of the wing. This is called the aerodynamic stall, and it always happens at the same Angle of Attack.

Angle of Attack is controlled by the elevators, the control surfaces on the trailing edge of the horizontal tail. When the pilot pulls back on the stick, the elevators lift, causing a down-force on the tail and forcing the wing to meet the air at a higher Angle of Attack. Trim tabs (small surfaces at the trailing edge of the elevators) can be moved to change the neutral position of the stick. (Another way to think of it is the trim tabs change the Angle of Attack at which there is zero stick force.)

In a modern jet transport the entire horizontal tail is usually moveable. This is because of the very wide speed range of the jet and because flaps and leading edge slats also change the “trim.” The other side of the coin is that this horizontal tail, or stabilizer, is very powerful in modern jet transports. A runaway stabilizer is a true emergency. Traditionally there has been a STAB IN MOTION aural warning, and an emergency cutout switch close to hand. Most cases where the stabilizer ran all the way up or down in flight have resulted in the loss of all on board.

In most aircraft the pilot is used to trimming as he flies. A change of speed or configuration, be it in a Beech Bonanza or a DC-9, will require a trim change. With some experience on type the pilot knows (for example on a DC-9) that extending the leading edge slats will require two beeps (of the STAB IN MOTION aural warning) of nose-up trim. He can use the thumb switches on the yoke to move the stabilizer as the slats are extending and thus remain stick-neutral during the configuration change. This is part of anticipation, or staying ahead of the aircraft.

Airbus aircraft, from the A320 onwards, are different. They are fly by wire, where computers are interposed between the pilots' sidestick inputs and the control surfaces. This arrangement allows some elegant additions to aircraft design, such as envelope protection (which among other things makes it impossible for the pilot to stall the aircraft) and, relevant to our discussion today, stick force per G and autotrim.

In Normal Law, which is where the Airbus is most (and the pilot hopes, all) of the time, configuration changes can be made hands off, even flying by hand. Of course the pilot has the tips of his fingers on the sidestick, but he can make a configuration change with no pitch input because the control system, in Normal Law, will maintain 1G flight. When he calls for FLAP 1 and the leading edge slats extend, the nose-down pitch is sensed and countered by the system, maintaining 1G flight. (1G is what you experience sitting in a chair at home or in an aircraft at cruise in smooth air). In effect, the airplane is doing the anticipation for the pilot.

Like the transition in the late 1950's from props to jets, fly-by-wire has been a major change for pilots. In general we welcome it for the many advantages it offers.

Experience has shown that to do his job, which is to ensure the safe arrival of his aircraft, the pilot must fully understand a much more complex airplane. Chesley Sullenberger reached up and started the APU (the Auxiliary Power Unit, a small turbine in the tail which can supply electrical and hydraulic power on the ground or in flight) as soon as his engines lost power. Why? Because he knew his airplane and he knew he wanted to keep it in Normal Law until touchdown.

The transition from props to jets was all about speed range, speed brakes and spoilers, high Mach number, coffin corner, Dutch Roll and super-stall, but in everyday life it was more about high drag on approach, no propwash, slow spool-up times, and operating on the back side of the power curve. This change took some adjustment on the part of pilots: the more experienced pilots had more adjustments to make. The same is true with the transition to fly-by-wire.

In a traditional airplane the pilot controls Angle of Attack with the elevator and the trim tabs or stabilizer. (More often he will be thinking of Airspeed, which is the constant-weight, 1G manifestation of AofA). He is used to feel, which is essentially the change in elevator neutral point with AofA. Should the aircraft slow on approach, the nose will get “heavy”, prompting him to pull back or trim nose-up.

That feel is totally absent in Airbus aircraft. (Boeing, in the B777, have added artificial feel to their fly-by-wire system). The Airbus pilot points and shoots, so to speak. Flying by hand he can take the bird, turning on a symbol (like a bird or an aircraft seen from behind) on his Primary Flight Display. The bird shows where his velocity vector is pointed; in other words, where is airplane will be so many seconds from now if he makes no further adjustments. On approach he can pin the bird on his flare point on the runway and either let the autothrust take care of the speed or adjust the thrust levers manually. If he does the latter, he must remember that there is no feel or feedback in the sidestick.

Obviously there are quite different assumptions operating during an approach in a Bonanza, one one hand, and an Airbus, on the other. This is not necessarily a bad thing. Take for example driving a car versus riding a motorcycle. In a car you steer with the steering wheel. In a motorcycle you counter-steer, putting pressure on the inside foot-peg and forward pressure on the inside bar, in effect trying to steer the front wheel the opposite way.

But you know you're on a motorcycle and not in a car. You have learned how to ride a motorcycle.

Consider, however, flying an Airbus if something goes wrong with a sensor or a computer and you wind up in Alternate Law or Direct Law. You are in the same vehicle but suddenly the rules have changed; the assumptions have changed. It is, in effect, no longer the same machine. This is a recipe which messes with a pilot's head.

Unfortunately, experience has shown that Direct Law, where control displacement is proportional to stick force and the airplane handles like a wet fish, is actually the more benign of the two degraded modes. There is a big message in red on the ECAM saying USE MAN PITCH TRIM. The pilot moves the THS (Trimmable Horizontal Stabilizer) by moving a wheel almost a foot in diameter. This is old-style, normal airplane flying, commanding AofA with stick force and trim. There is still no feel in the sidestick, but the procedure is familiar.

Alas, in Alternate Law there is no such familiarity. It is still point-and-shoot, sort of, but autotrim is still working. As long as there is back pressure on the stick the THS trims nose-up, and vice-versa. There is NO Stabilizer in Motion warning except the movement of the trim wheels. That would seem to be an easy thing to detect, but I can testify from personal experience that it is not. On every landing (in Normal Law) the flight control computers memorize the attitude at 50 feet Radio Altitude and at 30 feet start rolling in nose-down trim, in effect trying to mimic the feel of a normal aircraft slowing in the flare. In almost a decade of flying as Captain and Training Captain, whether as Pilot Flying or Pilot Not Flying, I cannot remember ever seeing the trim wheels move.

In two recent accidents an Airbus has hit the ocean with the THS wound to full nose up. In both cases the aircraft was in Alternate Law.

I am not an engineer. There are likely many ramifications that have not crossed my mind. But sitting here this afternoon my personal recommendation would be as follows:

Disable Autotrim in Alternate Law

Thanks to the work of David Learmount at Flight Global, and that of the Wood's Hole Oceanographic Institution and the Bureau d’Enquêtes et d’Analyses, enough is now known about this accident to start looking for useful lessons and to analyze the data along with the BEA. Flight safety and the future of the piloting profession depend on this becoming a wide and serious conversation.

Pilots obsess about accidents for good reason. There is always so much to learn. The AF447 tragedy is an epochal example.

There is a mind-boggling number of lessons to be learned here, in a host of areas and disciplines: Pilot Training, Standard Operating Procedures, Instrument Flight, and Aircraft Design are but a few of them.

I will commit today to joining the conversation. I begin with a consideration of Angle of Attack.

Angle of Attack

Wolfgang Langewiesche (father of William) emphasized Angle of Attack in his excellent Stick and Rudder, published in 1944. Advocacy of AoA was an uphill battle then and it still is today. Instead of talking about AoA, we prefer to use airspeed and explain why certain speeds we use change with aircraft weight and G loading. Many or even most aircraft flying today have no Angle of Attack indication. The accident aircraft had two AoA sensors. The flight recorders had access to the signals from these sensors but the pilots did not, at least not when they needed it most.

Lift is produced when the air flowing over the top of a wing has a longer distance to travel than the air flowing underneath. The air “stretching out” over the top produces a lower pressure, allowing the higher pressure underneath to push the wing up. There is a caveat, however. The airflow must remain attached to the upper surface of the wing.

Imagine a cross-section of wing, with a line drawn from the middle of the rounded leading edge to the pointed trailing edge. This is the chord line. Now imagine an arrow pointing at the leading edge. This is the airflow.

If the arrow meets a (symmetrical) wing head-on there will be no lift. But let the wing meet the air at a slight angle and the airflow around the wing will no longer be symmetrical: it will meet the rounded leading edge at an angle and it will divide lower on the curve of the leading edge. The air flowing over the wing will have a longer distance to travel. Lift will be produced.

The angle at which the airflow meets the chord line is called the Angle of Attack. Up to a point, increasing the Angle of Attack will increase lift. But beyond a certain point – usually about 16° – lift will instead decrease because the airflow is beginning to separate from the upper surface of the wing. This is called the aerodynamic stall, and it always happens at the same Angle of Attack.

Angle of Attack is controlled by the elevators, the control surfaces on the trailing edge of the horizontal tail. When the pilot pulls back on the stick, the elevators lift, causing a down-force on the tail and forcing the wing to meet the air at a higher Angle of Attack. Trim tabs (small surfaces at the trailing edge of the elevators) can be moved to change the neutral position of the stick. (Another way to think of it is the trim tabs change the Angle of Attack at which there is zero stick force.)

In a modern jet transport the entire horizontal tail is usually moveable. This is because of the very wide speed range of the jet and because flaps and leading edge slats also change the “trim.” The other side of the coin is that this horizontal tail, or stabilizer, is very powerful in modern jet transports. A runaway stabilizer is a true emergency. Traditionally there has been a STAB IN MOTION aural warning, and an emergency cutout switch close to hand. Most cases where the stabilizer ran all the way up or down in flight have resulted in the loss of all on board.

In most aircraft the pilot is used to trimming as he flies. A change of speed or configuration, be it in a Beech Bonanza or a DC-9, will require a trim change. With some experience on type the pilot knows (for example on a DC-9) that extending the leading edge slats will require two beeps (of the STAB IN MOTION aural warning) of nose-up trim. He can use the thumb switches on the yoke to move the stabilizer as the slats are extending and thus remain stick-neutral during the configuration change. This is part of anticipation, or staying ahead of the aircraft.

Airbus aircraft, from the A320 onwards, are different. They are fly by wire, where computers are interposed between the pilots' sidestick inputs and the control surfaces. This arrangement allows some elegant additions to aircraft design, such as envelope protection (which among other things makes it impossible for the pilot to stall the aircraft) and, relevant to our discussion today, stick force per G and autotrim.

In Normal Law, which is where the Airbus is most (and the pilot hopes, all) of the time, configuration changes can be made hands off, even flying by hand. Of course the pilot has the tips of his fingers on the sidestick, but he can make a configuration change with no pitch input because the control system, in Normal Law, will maintain 1G flight. When he calls for FLAP 1 and the leading edge slats extend, the nose-down pitch is sensed and countered by the system, maintaining 1G flight. (1G is what you experience sitting in a chair at home or in an aircraft at cruise in smooth air). In effect, the airplane is doing the anticipation for the pilot.

Like the transition in the late 1950's from props to jets, fly-by-wire has been a major change for pilots. In general we welcome it for the many advantages it offers.

Experience has shown that to do his job, which is to ensure the safe arrival of his aircraft, the pilot must fully understand a much more complex airplane. Chesley Sullenberger reached up and started the APU (the Auxiliary Power Unit, a small turbine in the tail which can supply electrical and hydraulic power on the ground or in flight) as soon as his engines lost power. Why? Because he knew his airplane and he knew he wanted to keep it in Normal Law until touchdown.

The transition from props to jets was all about speed range, speed brakes and spoilers, high Mach number, coffin corner, Dutch Roll and super-stall, but in everyday life it was more about high drag on approach, no propwash, slow spool-up times, and operating on the back side of the power curve. This change took some adjustment on the part of pilots: the more experienced pilots had more adjustments to make. The same is true with the transition to fly-by-wire.

In a traditional airplane the pilot controls Angle of Attack with the elevator and the trim tabs or stabilizer. (More often he will be thinking of Airspeed, which is the constant-weight, 1G manifestation of AofA). He is used to feel, which is essentially the change in elevator neutral point with AofA. Should the aircraft slow on approach, the nose will get “heavy”, prompting him to pull back or trim nose-up.

That feel is totally absent in Airbus aircraft. (Boeing, in the B777, have added artificial feel to their fly-by-wire system). The Airbus pilot points and shoots, so to speak. Flying by hand he can take the bird, turning on a symbol (like a bird or an aircraft seen from behind) on his Primary Flight Display. The bird shows where his velocity vector is pointed; in other words, where is airplane will be so many seconds from now if he makes no further adjustments. On approach he can pin the bird on his flare point on the runway and either let the autothrust take care of the speed or adjust the thrust levers manually. If he does the latter, he must remember that there is no feel or feedback in the sidestick.

Obviously there are quite different assumptions operating during an approach in a Bonanza, one one hand, and an Airbus, on the other. This is not necessarily a bad thing. Take for example driving a car versus riding a motorcycle. In a car you steer with the steering wheel. In a motorcycle you counter-steer, putting pressure on the inside foot-peg and forward pressure on the inside bar, in effect trying to steer the front wheel the opposite way.

But you know you're on a motorcycle and not in a car. You have learned how to ride a motorcycle.

Consider, however, flying an Airbus if something goes wrong with a sensor or a computer and you wind up in Alternate Law or Direct Law. You are in the same vehicle but suddenly the rules have changed; the assumptions have changed. It is, in effect, no longer the same machine. This is a recipe which messes with a pilot's head.

Unfortunately, experience has shown that Direct Law, where control displacement is proportional to stick force and the airplane handles like a wet fish, is actually the more benign of the two degraded modes. There is a big message in red on the ECAM saying USE MAN PITCH TRIM. The pilot moves the THS (Trimmable Horizontal Stabilizer) by moving a wheel almost a foot in diameter. This is old-style, normal airplane flying, commanding AofA with stick force and trim. There is still no feel in the sidestick, but the procedure is familiar.

Alas, in Alternate Law there is no such familiarity. It is still point-and-shoot, sort of, but autotrim is still working. As long as there is back pressure on the stick the THS trims nose-up, and vice-versa. There is NO Stabilizer in Motion warning except the movement of the trim wheels. That would seem to be an easy thing to detect, but I can testify from personal experience that it is not. On every landing (in Normal Law) the flight control computers memorize the attitude at 50 feet Radio Altitude and at 30 feet start rolling in nose-down trim, in effect trying to mimic the feel of a normal aircraft slowing in the flare. In almost a decade of flying as Captain and Training Captain, whether as Pilot Flying or Pilot Not Flying, I cannot remember ever seeing the trim wheels move.

In two recent accidents an Airbus has hit the ocean with the THS wound to full nose up. In both cases the aircraft was in Alternate Law.

I am not an engineer. There are likely many ramifications that have not crossed my mind. But sitting here this afternoon my personal recommendation would be as follows:

Disable Autotrim in Alternate Law