Thirty Years Since Air France Flight 296
Thirty years ago, on the 26th of June 1988, a brand new Airbus A320 crashed during a demonstration of its fly-by-wire capabilities, killing three passengers and demolishing the plane as airshow guests watched helplessly. The tragic accident was filmed from start to finish, one of the few commercial crashes captured on video.
But it wasn’t until I was asked to take part in a documentary that I took a hard look at the events to understand how on earth a fully functional Airbus could crash in front of thousands of spectators. I know many of you already know and remember this accident but I hope you will be as interested as I was in the detail.
I’m afraid this is going to be another two-part post, though, because to make sense of this accident means understanding a number of modern aircraft systems and aviation jargon and I think that needs to be discussed and understood before we can look at what happened, especially because there are a lot of articles that use this accident as evidence that automation in aircraft is fundamentally flawed, without actually understanding the mechanics of what happened.
So today’s post will focus on the aircraft systems and jargon which relate to this incident. As always, please feel free to leave a comment to correct or even just to improve my explanations. Next week we’ll look at the accident itself in detail.
The term fly-by-wire reflects the fact that there are no mechanical connections between the sidestick in the cockpit and the control surfaces (eg the elevators) of the aircraft. In a conventional aircraft, the controls are connected to the yoke: if you push the yoke forward, there are cables and rods which directly connect the yoke to the elevator. In a large plane, where it would be too difficult to move the control column with human strength alone, there’s a hydraulic system in place.
In a fly-by-wire system, the movement of the controls sends a signal to the system, which uses electronic circuits to move the control surfaces. If you push the control column forward, a signal is sent to the computer and the computer commands the elevators to move. The result is the same, the nose pitches down, just the way that the information is transmitted is different.
Air France flight 296 at Habsheim was the Airbus 320’s first passenger flight and the first public demonstration of the first civilian digital* fly-by-wire aircraft. The demonstration was focused on showing the exciting new technology of alpha protection and alpha floor.
*See comments; I missed the difference between analog and digital fly-by-wire.
Alpha protection gives the pilot the best lift while preventing the aircraft from stalling. The speed brakes retract automatically, the aircraft maintains the current AOA (angle of attack). If the pilot continues to apply backpressure to the stick, the engines will kick in (Alpha Floor activates) providing TO/GA thrust (the power setting used for take-off and go arounds) and the aircraft will climb at a slow airspeed.
Bear in mind I have never flown a fly-by-wire plane so you know, check the comments for corrections, but here’s my explanation by way of an example.
You, the pilot, are cruising along quite happily when all of a sudden you see something in front of you and you need to climb to avoid it, for example a mountain or an oncoming plane. Obviously, you would pull the stick back, which moves the elevators, causing the aircraft’s nose to pitch up, which will make the aircraft climb. However, at the same time you have to think about whether the aircraft has enough energy (airspeed or thrust) to climb.
Instinctively (if you were trained as a pilot) as you pull that stick back, you will also increase your engine power, in order to ensure that you have enough energy to allow the aircraft to climb. You also will probably not pull the stick back as far as you can, because the steeper the climb, the more energy you need, and it takes some time for those engines to spool up. You want to make sure you climb over your obstacle, but you also want to climb as efficiently as possible. Above all, you don’t want to run out of energy, because the aircraft not only won’t climb, you’ll lose the lift under your wings and you won’t be flying at all but falling out of the sky. This is a stall, which means the critical angle of attack has been exceeded and the aircraft has lost lift.
So don’t pull back too far. But then, if your climb isn’t steep enough, you might not clear that obstacle, and you will still crash. So you need to pull back enough but not too much while getting that power on and (possibly) reconfiguring the aircraft. It’s a lot to think about on the spur of the moment and I haven’t even started on contributing issues like aircraft type, altitude, weather and temperature, all of which can also have an effect.
This is where alpha protection comes in. The idea is that the pilot shouldn’t have to think about exactly how far back he should pull the stick to clear the object in front of him without risking a stall. In an Airbus fly-by-wire, you demand ‘up’ on the sidestick and the computer works out exactly what the maximum possible climb is in your current configuration without risking stalling the plane. It can take you right up to the critical angle of attack and keep you there.
This means that you don’t have to worry about whether you are applying enough back pressure but not too much, instead you simply pull all the way back, which is a command in itself: max backpressure. Having pulled all the way back, you’ve commanded the aircraft to achieve the best lift while preventing the aircraft from stalling. The speed brakes will retract and engines will increase to TO/GA thrust. As the airspeed increases, so will the pitch angle, throughout offering the maximum climb while protecting the aircraft from stalling.
In modern aircraft, the flight management computer will determine the power needed by the engines to take off, based on a number of factors such as runway length, wind speed, temperature, and most importantly the weight of the aircraft. In older aircraft these calculations were performed by the pilots before a takeoff. The advantage of having such a system is the ability to reduce wear and tear on the engines by only using as much power as is actually required to ensure the aircraft reaches a safe take off speed.
When taking off, the pilots take the aircraft to 40-60% RPM on Boeing, 50% on Airbuses, then increase the thrust levers to TO/GA. The aicraft increases to the computed take-off power.
During landing the TO/GA switch allows for quick modification of the autopilot mode so that the aircraft is no longer following the ILS glide slope to the runway and overrides the current authothrottle mode. Using the TO/GA switch is the quickest way of increasing thrust to abort a landing.
There’s been at least one Airbus 300 crash caused by the first officer accidently pressing the TO/GA switch on landing. There have been a few other incidents that I know of, where the pilots hit the TO/GA switch to go around and then changed their minds and attempted to land. And there is at least one that I’ve been working on for a while where they attempted to go around without hitting the TO/GA switch and without realising that the autopilot was still trying to follow the glideslope.
On Airbus planes, the TO/GA switch is effectively activated if you push the throttles to TO/GA power, that is, if you increase thrust to TO/GA while on final approach, the flight management system understands that you want to abort the landing and it reconfigures the aircraft for this.
If you pitch the nose up without enough airspeed, an aircraft stalls. Every pilot is taught to respond immediately to an impending stall by increasing the airspeed: add power and, if the aircraft is climbing, pitch the nose down back to level flight, waiting for the airspeed to increase before pitching up again to climb away. A stall close to terrain can be very difficult to recover from because you can’t afford to lose any height. If there’s an obstacle in your way, you have a real problem.
The Airbus alpha protection recognises an impending stall – the critical angle of attack has been exceeded – and reacts the same as a pilot: the aircraft pitches the nose down and adds thrust – TO/GA power. The fly-by-wire system is able to do a massive amount of computation in milliseconds in order to ensure that the least possible amount of height is lost; however, there is still a risk of loss of height and it is certainly not possible to climb until the airspeed has increased.
The Habsheim Airshow
The Airbus A320 was the first civilian fly-by-wire aircraft and, as you can imagine, there was a huge amount of excitement at the launch of the aircraft. An event was organised at Habsheim in France to show off the capabilities of the A320. This was the first public demonstration of any civilian fly-by-wire aircraft and thousands of people attended to see the plane.
The concrete runway 02/20 at Habsheim was 1,000 metres (3,280 feet), which is not enough runway for the Airbus A320. Instead, the passengers, journalists and the winners of a raffle, would be picked up at Basel-Mulhouse, ten nautical miles (16 km) away. Then the Airbus A320 would climb to 1,000 feet and turn right to fly to Habsheim. Once they had the runway in sight, they would take the flaps to 3 (20°), extend the landing gear, and descend to 100 feet to fly over runway 02/20.
The spectators (and the passengers) would be wowed by the low approach in landing configuration. Slowly flying the aircraft at 100 feet above the ground would show how the fly-by-wire system could hold the A320 exactly at critical angle of attack in a way that a human pilot could not hope to achieve. Then the aircraft would climb away, circle around and do a second fly over at 100 feet, this time at high speed, before taking the passengers to Mont Blanc (France’s highest mountain) for a sightseeing flight around the peak.
An ambitious plan but certainly doable. However, within the first few minutes of the flight, things had already started to go wrong. And as we all know, several very small issues can quickly build up into tragedy.
This is already very long and there’s a lot to cover, so I will step through the actual flight and what went wrong in a separate post. In the meantime, I hope that at least some of you have found the background information useful.