Mayday (2013) s09e07 Episode Script
The Final Blow
NARRATOR: Just minutes from Strasbourg Airport Merde! .
.
an Airbus A320 slams into a mountaintop.
(SCREAMING) There are survivors .
.
but what they don't realise is that no-one knows where they are.
(SIRENS BLARE) Before investigators can begin searching for what caused the crash of Air Inter Flight 148, they must first find the plane.
January 20, 1992.
Air Inter Flight 148 has departed from Lyons, France.
124.
905, thank you.
Captain Christian Hecquet and First Officer Joel Cherubin are experienced pilots with over 12,000 hours of flying time between them.
The flight is a short hop between Lyons, in central France, and the city of Strasbourg, in the mountainous Alsace region.
The French airline Air Inter caters mostly to business travellers.
Crews are motivated to avoid delays, as former Air Inter pilot Gerard Arnoux explains.
We were famous for our very short turnaround .
.
and the faster we flew, the better wages we got.
Have we been flying for 35 minutes yet? 41 minutes.
The crew is flying an Airbus A320, one of the most technologically advanced commercial airplanes in the world.
Even before take-off, the pilots program the autopilot to land on a specific runway in Strasbourg.
The cockpit of the A320 is also very different from other planes.
Instead of analogue gauges, the pilots look mostly at digital displays.
AIRPORT RECORDING: Strasbourg.
Good evening.
Runway in use - 05.
Transition level, 50.
Wind, 040 at 1-8 knots.
Visibility - 10km.
A recording from Strasbourg Airport informs the crew of a change in plan.
Due to high winds and poor winter weather, they'll have to land on an alternate runway 05 in service.
.
.
not the one programmed into the autopilot.
05? What sort of wind are they giving us? 18 knots.
18 knots.
Captain Hecquet doesn't like the idea of changing runways.
No chance.
He was hoping to use runway 23, an approach that provides the autopilot with a precise navigational fix.
The new runway, runway 05, is surrounded by mountainous terrain that can interrupt radio signals sent to the autopilot.
No, if we go with the runway 05 procedure, we No, no.
(CLEARS THROAT) Captain Hecquet suggests a compromise.
I am putting back runway 23 - otherwise, I couldn't make the ILS interception.
He'll program the autopilot to fly towards runway 23 .
.
but near the airport, the captain will take over the controls and make a visual landing on runway 0-5.
We are taking 23, then? (SHOUTS) Yes! Agreed.
Roger 854.
Proceed to GTQ air level 140 Strasbourg.
Uh, yes.
We intend to proceed to do an ILS on runway 23, then an indirect for runway 05 after that.
The Strasbourg controller considers the captain's plan.
He warns that there will likely be a delay due to heavy traffic.
Given that we're going to have three take-offs on 05, you risk waiting in a stack at 5,000 feet.
We're not going to mess about like that descending at full speed.
If they had warned us in advance! Cripes! (SIGHS) Inter France, Strasbourg.
I hear you.
Aware of the captain's frustration, the controller offers assistance.
If you want, I can take you with the radar to lead you to ANDLO at 5,000.
ANDLO is a navigational point on the approach to runway 05.
It helps pilots align the plane for landing.
Yeah.
Yeah, that's good.
- Oh, yeah.
- OK, then.
Turn left to heading 230 degrees.
Since runway 05 doesn't allow for a full autopilot approach, the captain must calculate the angle of descent on his own.
That makes 3.
3 degrees.
(SPEAKS FRENCH) TRANSLATION: 3.
3 degrees is a normal flight angle that provides a good slope for landing.
Turn left, steer 90.
090 degrees, Delta Alpha.
The controller talks Flight 148 through the last turn to align the plane with the runway, now 25km away.
Then, First Officer Cherubin notices the plane is slightly off-course.
We are headed inside.
You're inside there.
You should have started with 070.
Yeah.
- At least that much.
- (CLEARS THROAT) The controller also notices that the plane is off-course.
Air Inter Delta Alpha? It has missed aligning itself with ANDLO, the runway's electronic guidepost.
Delta Alpha? You're passing to the right of ANDLO.
Nevertheless, he authorises the landing.
Authorised for final approach 05.
Delta Alpha.
The captain initiates the landing sequence.
Gear down.
Hecquet notices that the plane is travelling too fast, so he extends the speed brakes.
They disrupt airflow over the wing, which helps create more drag to slow the plane.
We have to watch our descent.
The approach axis.
The First Officer is more concerned with their horizontal position in relation to the runway.
It was 60.
Check it out.
But before the crew can adjust their course Merde! (SCREAMING) Delta Alpha, your position? Air Inter Delta Alpha, Strasbourg.
The crash is catastrophic.
The A320 has flown into the side of a mountain.
Delta Alpha, your position? Flight 148 is no longer on radar .
.
nor responding to radio contact.
An emergency is declared at Strasbourg Airport.
This is the last hit we got.
They were flying about 20km away from the airport.
Officials need to pinpoint the crash site, but it's not as easy as it might seem.
The airport's radar is not recorded, there has been no signal from the plane's emergency beacon and, surprisingly, no-one has reported seeing a plane go down.
It could be anywhere in here.
The proposed search area covers more than 20 square kilometres of dense forest just outside Strasbourg.
Nicolas Skourias survives the crash of Air Inter Flight 148 with only minor injuries.
(COUGHS) I saw a fire in front of me and I panicked because I say to myself "I'm going to burn.
" I went to the back of the plane - what remained of the plane (COUGHS) I found some other survivors.
Come on! It's gonna blow! With the smell of leaking jet fuel in the air, the survivors move away from the burning plane.
We stay together, waiting for the first aid.
But the wait will be longer than anyone might expect.
The first reaction that we have after the crash was, "OK, in half an hour, one hour most, "and the worst, OK, the rescue team will be here.
" And it wasn't here.
One hour after the crash, rescuers still have no idea where the wreckage lies.
(GROANS) Skourias and the others now face a new ordeal .
.
surviving sub zero temperatures in a dark and isolated forest.
2.
5 hours after Flight 148 disappeared from radar near Strasbourg Airport (SIRENS WAIL) .
.
the missing plane has still not been located.
Amidst growing tension, the French aviation bureau, the BEA, sends in its lead investigator, Jean Paries.
I immediately called my two main investigators and we organised the go-team .
.
and we got prepared to rush to the site as soon as the site was located somewhere.
The delay feels like an eternity.
A little bit surprisingly long.
We can expect this in the jungle or the rainforest but not exactly in a highly dense populated area like the Strasbourg area.
With no help in site, Skourias returns to the wreckage to look for more survivors.
I think that some people that died could have survived if the first aid come sooner.
Nearly 1,000 people search for the missing plane, but three hours after the crash, there's still no sign of it.
Frustrated, Skourias goes looking for help.
He stumbles into a TV crew trying to find the crash but with no wreckage in sight, they react with scepticism.
They didn't expect survivors from an airplane crash.
Hey! Hey, you've got to believe me! They didn't believe that there was one of survivor but believe me, I was, because my face was black due to the smoke, the kerosene and so on.
Come on! The journalists follow him back to the crash site .
.
where they discover eight other survivors.
- Finally, the first rescuers arrive.
- (SIRENS WAIL) The crash site is located near the top of the 2,500ft Mont Sainte-Odile .
.
19km from the runway.
They found us after 4 hours and 30 minutes, so it was a mess.
I was very, very disappointed that 20km from Strasbourg and they couldn't find us.
A total of 87 passengers and crew have died, including the pilot and co-pilot.
The survivors begin to tell their stories but no-one reports anything that might explain why the plane crashed.
(SPEAKS FRENCH) TRANSLATION: I don't know what happened.
We were landing.
I lost all consciousness.
We must have hit the trees.
Bob Macintosh, an American NTSB investigator, arrives at the crash site.
The BEA of France recognised the international attention would be on this accident, even though it was a domestic accident.
He invited a group of international accident investigators to come and participate.
Welcome to the team.
The first priority for investigators is to retrieve the plane's black boxes.
We are very anxious about the state of the tape inside.
Will it be possible to use it? Will we get the critical information we need? In France, aviation accidents are also investigated by the justice system.
Paries and his team are not allowed access to the site until judicial officials secure the black boxes.
In a previous crash, the crash of Air France Flight 296 in 1988, investigators waited 10 days before turning the black boxes over to police.
Rumours persisted that these boxes had been tampered with.
This time, police are keeping investigators at bay.
I can recall seeing the glowing embers and seeing the flight recorder sitting there .
.
and not being able to intervene and say "Get that thing cooled down as soon as you can.
" After midnight, the boxes are retrieved from the plane and sent for analysis.
Investigators can only hope it's not too late.
PARIES: They were extremely hot.
They looked damaged and they looked burned.
(FEMALE REPORTER SPEAKS FRENCH) In the light of day, investigators get some of their first clues from the crash site itself.
They discover why the plane's emergency locator beacon didn't send a signal.
It was actually destroyed by the impact.
The beacon is located inside the cockpit and is designed to start working after a crash.
Its failure suggests an unusually forceful impact with the ground.
We had this first feeling the descent was abnormally steep.
Investigators examine the engines to see if they may have stalled before impact.
If you find the blades curved and a lot of wood sucked inside the engines, then you understand that the engines weren't working properly.
And that's exactly what they find.
The plane clearly had power, yet it ploughed steeply into a mountainside without ever sending out a distress signal.
Investigators are puzzled.
They hope that the box which recorded the plane's flight data will help them solve the mystery.
The black box is designed to survive temperatures up to 1,100 degrees Celsius for half an hour.
The tape recorder inside is protected by a capsule filled with water.
MAN: When the recorder heats up, the water turns to steam, absorbing the energy and actually vents out through a little hole in the crash enclosure.
But when the flight data recorder is opened, investigators make a troubling discovery.
The FDR was totally damaged.
Impossible to read anything from it.
There's now only one hope for recovering the plane's flight data - a device called a Quick Access Recorder, or QAR.
Maintenance workers use the QAR to access the plane's computers, but it also records some flight data.
Unlike the black boxes, the QAR is stored near the cockpit.
Quick Access Recorders are not protected at all.
They're up in the front end of the aircraft, typically in the electronics bay.
They're generally destroyed just from the impact damage.
Investigators are encouraged to discover that in this case, the QAR has survived.
But on closer examination, their optimism turns to frustration.
The last 20cm of the tape were burned and stretched and were damaged to the point that we couldn't use them into a machine - we couldn't read it.
Investigators are desperate to retrieve the data, so they take a chance on an experimental technique.
Known as the 'garnet technique', a light is shone through a mineral lens made of garnet.
The special lens helps the technicians differentiate between the positive and negative magnetic pulses which translate as binary digits, or 'bits'.
There are 768 bits per second, so that's a lot of ones and zeros.
You have to be very precise in moving the tape under the lens, or the garnet, to make sure you don't miss a bit or read the same bit twice, you know? So it's difficult.
It took about a day to read a second of recording.
Any additional seconds recovered could reveal something that would make a difference.
The effort to retrieve all the QAR data could take a month or more.
In the meantime, the focus of the investigation shifts to the cockpit voice recorder.
The cockpit voice recorder, which was just inches away but outside of the ashes, had air passing over it, survived.
HECQUET: (ON RECORDING) Runway 23.
The recording reveals the captain's anxiety early on in the flight.
- You are taking 23, then? - Yes! Investigators know that landing on runway 05 requires what's called a 'non-precision approach'.
That means pilots receive electronic guidance only on their horizontal position - left and right.
They get no guidance when it comes to altitude.
05.
What sort of wind are they giving us? - 18 knots.
- 18 knots.
The non-precision approach increases the demands on the pilots.
Investigators can also hear that the captain had concerns about landing on runway 05.
RECORDING: 48 Delta Alpha, you are number one with runway 05.
10 nautical.
That won't work.
It's a lot of distress over non-precision approach.
Wondering what can cause such distress, investigators research pilot training at Air Inter.
They find that most pilots did not have extensive training making non-precision landings in the new A320.
I think we should have had double the training compared to an older plane.
Investigators ask the airline for detailed records on the pilot's history of runway approaches.
They're intrigued by what they discover.
Captain Hecquet had landed at Strasbourg countless times but he had never landed an A320 there using a non-precision approach.
Not going to mess around like that descending at full speed.
Clearly, the captain was uneasy about having to execute a landing he had never made before.
I think the captain was worried about making it in in a minimum amount of time and the minimum amount of delay.
Have we been flying for 35 minutes yet? And the co-pilot was worried about not getting in trouble by offending the captain.
At least that much.
More research into the pilot's work history offers yet another revelation.
While the two pilots had flown more than 12,000 hours between them, they were both still relatively new to the highly advanced A320.
05 in service.
The aviation community misunderstood the magnitude of changes brought by the new Airbus A320.
The captain had only 162 hours in the A320 .
.
and the co-pilot, even less - just 61 hours.
PARIES: Behind this accident scenario, there is an issue of confidence of the crew - in themselves, in the aircraft.
18 knots.
No chance.
They were not prepared, really, to fly in this kind of condition.
If they had warned us in advance.
Cripes! Investigators conclude that the crew's training was insufficient, but that alone does not explain the crash.
Merde! (SCREAMING) Investigators search for other factors in the crash of Flight 148.
They review the conversations between the crew and air traffic controllers.
If you want, I can give you radar headings and take you to ANDLO at 5,000.
Yeah.
Yeah, that's good.
TRANSLATOR: The radar vector makes flying easier.
The captain was happy because it was reducing his workload.
Turn left, steer 90.
With the controller's assistance, this landing should have been very simple.
But when investigators reconstruct the plane's trajectory using radar information from various stations around the airport, they discover a shocking error.
The 090 heading started here.
090 degrees, Delta Alpha.
But it won't take them to ANDLO.
Last radar vector the controller gave was incorrect.
It sent them Thank you.
.
.
closer to the mountain.
PARIES: They were off-course because of following the heading they got from the radar vectoring.
They found themselves in this undershoot situation.
You're inside there.
- You should have started with 070.
- Yeah.
Investigators are also troubled by the controller's choice of words when he warned the pilots, incorrectly, that they were headed to the right.
Delta Alpha.
You're passing to the right of ANDLO.
From the pilot's perspective, the plane was on the left side of the runway, not the right.
It could only add to their confusion.
TRANSLATION: It was very poor guidance because he didn't employ the usual terminology.
(SPEAKS FRENCH) Investigators recommend that controllers use only compass points when giving directions - never the words 'right' and 'left'.
The controller's mistakes clearly brought the plane closer to the mountain.
Turn left, steer 90.
But once again, investigators feel they don't have the whole story.
It's not something totally abnormal to start a descent from this situation.
Flaps towards 2.
Flaps towards 2.
It's not what you expect to do every day, but it's not outside the tolerance of the concept of this approach.
Gear down.
When investigators study the plane's reconstructed flight path, they discover something more alarming than the plane's horizontal misdirection.
As it circled the mountain, the plane inexplicably entered a dangerously steep and rapid descent.
Perhaps 2.
5 times the normal rate of descent.
It's lethal at that altitude.
Without the steep descent, they would have cleared the mountain.
If the vertical trajectory had been correct, they would have no problem at all.
Finding the cause of that sudden descent is now key to understanding why 87 people died in one of the most advanced passenger planes on earth.
Authorised for final approach 05.
Descent was initiated at 1800 hours, 19 minutes and 38 seconds.
That Delta Alpha.
.
.
is the point of no return.
By studying Flight 148's trajectory, investigators determine that the rapid descent began 60 seconds before the crash.
Delta Alpha.
There is no indication on tape that the descent was deliberate.
How it happened and why the crew didn't notice is a mystery.
It should be a no-brainer, keeping track of the altitude.
The cockpit altimeter gives pilots a constant readout of their altitude.
Altimeter - that's a very precise instrument.
They've become very reliable.
They're accurate between 5 or 10 feet.
Ignoring it would be a major error in flying protocol.
Flaps towards 2.
The recording reveals just one single remark from the crew about their descent.
We have to watch our descent.
(CLEARS THROAT) It occurred 16 seconds before the crash.
The captain had just extended the speed brakes.
The aircraft was accelerating abnormally.
The captain started to realise there was something wrong with the descent rate.
But the first officer changed the subject.
The approach axis - we're hitting the axis a half-point off.
There.
- It was 60.
Check it out.
- Yeah.
(CLEARS THROAT) He refocused the captain's attention on the lateral situation rather than the vertical situation, which was the main problem, of course, and they both failed to recognise the situation.
I think they were planning they were going to break out of the clouds and they'd be able to see the runway and they wouldn't need to do the full instrument approach.
It was 60.
But the plane never left the clouds.
There's an old adage in aviation - rocks have been known to hide out in those clouds.
Merde! It now seems clear that the crew was not monitoring their altitude closely enough.
- But a bigger mystery remains - We can only guess .
.
what caused that deadly descent? After months of work, investigators may finally have the answer.
All the available flight data from the damaged Quick Access Recorder has been recovered.
The data confirms that just before the crash, the plane was speeding towards the ground at an extremely high rate - 3,300 feet per minute.
It also confirms that the angle of descent was dangerously steep .
.
much greater than the 3.
3 degrees selected by the captain.
3.
3 degrees.
That's quite a difference.
Investigators now wonder - did the autopilot malfunction? Did it somehow fail to obey the captain's safe descent angle and send the plane into a deadly nosedive? What state was it in before the accident? Unfortunately, the flight control unit which houses the autopilot is too badly damaged to provide any definitive answers.
We could never demonstrate that this FCU on this aircraft during this flight functioned properly or not.
But then when he returns to studying the flight data, Paries discovers something that may finally reveal the cause of the crash.
He notices a similarity between two key numbers - the plane's vertical speed - 3,300 feet per minute - and the intended flight path angle - 3.
3 degrees.
Coincidence? Paries uses a flight simulator to test a new theory.
Can you show me a descent of 3,300 feet per minute? He believes that the similarity is no mere coincidence.
On the autopilot, there are two descent modes - flight path angle and vertical speed - but they are both displayed on the same window .
.
so 3,300 is abbreviated to 33.
Now, show me a flight angle of -3.
3 degrees.
The problem on this aircraft was that the two values were visible on the same window and controlled by the same knob.
3.
3 degrees.
-3.
3 degrees.
Paries strongly suspects that the confusing display tripped up Captain Hecquet.
The confusion is quite easy between the two modes if you don't do it carefully.
If the captain failed to push the mode selector knob, then entering '33' would not have initiated a safe 3.
3 degree angle of descent.
Instead, it would have put the plane into a deadly rate of descent of 3,300 feet per minute.
(SCREAMING) Two months after the crash, another Air Inter plane enters a dangerously steep descent for the same reason.
The crew only discovered the problem when they broke out of the clouds.
Those pilots also confused the plane's flight path angle with its vertical speed.
They were lucky enough to have a much higher cloud base so they could correct the problem.
Further research reveals an industry-wide problem with the A320.
Many people confuse these modes - especially during training - and many of them fell into the trap even after the training.
Eager to test his new theory, Jean Paries programs a simulator with all the known data from Flight 148.
He then inputs the same rate of descent he believes the Air Inter pilots selected.
If Paries is correct, the simulation will end with the plane hitting the mountain.
- But it doesn't.
- We are missing something.
Strangely, this didn't lead to a crash.
Every approach would overfly this obstacle by a significant margin.
Have we factored in the wind? We started to work on other alternate hypotheses Let's try again, but initiate the turn sooner.
.
.
but nothing was really credible.
No matter how hard he tries, Paries cannot simulate the crash.
Unable to explain why, he turns to the plane's manufacturer for help.
Thanks for bringing this to my attention.
After much research, an Airbus designer comes to Paries with an explanation about a little-known element of the autopilot's design.
In emergency situations where the A320 needs to change direction quickly, the autopilot is programmed to reverse the plane's direction at twice the normal rate.
The reaction of the autopilot would be much faster in these cases where typically, when you are descending and asking the autopilot to climb, or climbing and asking the autopilot to descend.
We immediately went back to the data at the very second at which the descent was commanded by the crew.
Gear down.
Paries discovers a tragic coincidence.
Sadly, we found at this very second, there was turbulence.
There was an ascent.
It's very slight, but there it is.
The momentary turbulence caused the plane to climb slightly.
And this led to a positive 600 feet per minute vertical speed for maybe half a second.
It was during that same half-second that the crew commanded the plane to descend.
It was 60.
Check it out.
But the autopilot read this as an emergency, requiring a blazingly fast descent.
That could be it.
Investigators now contemplate a terrible thought - could a random gust of wind, hitting at exactly the wrong split second, have been the difference between life and death? Here it comes.
(MACHINE BEEPS) And we got a crash.
Paries' theory explains every aspect of the crash.
It's a fascinating lesson about the random dimension of accidents.
Half a second before, half a second later, they wouldn't have the accident.
The discovery of a confusing cockpit display .
.
has enormous implications for the entire industry.
The flight instrumentation of aircraft like the DC-10, MD-11s, the 74s and so on - all the Boeing products and all the commuter products that were using that avionics suite had this vulnerability about it.
Investigators now face a daunting question affecting aircraft safety around the world.
If the design of the autopilot interface isn't changed, how many more people could die? There's mounting evidence that the design of the autopilot interface on Airbus A320s led the Air Inter pilots to accidentally dial in a dangerous rate of descent.
HECQUET: 3.
3 degrees.
We felt a need to start the industry to work on this.
The plane's manufacturer, Airbus, responds immediately.
The main change, which was very quickly made, was to change the display window.
With the new design, if a pilot selects a vertical speed of 3,300, the entire 4-digit number is displayed.
So the confusion between an angle and the vertical speed was no longer possible.
For investigators, only one mystery remains.
All Airbus A320 jets are designed to be equipped with a safety device known as a ground proximity warning system, or GPWS.
Which is a downward-looking, single-purpose radar that tells you how high you are above the ground directly beneath the airplane and if it gets to be too low, it'll set off a warning.
- (ALARM SOUNDS) - WARNING SYSTEM: Pull up! Terrain! - (ALARM SOUNDS) - Pull up! But Captain Hecquet We have to watch our descent.
.
.
never received a warning for one very simple reason - his A320 didn't have that alarm.
Merde! (SCREAMING) The first question, of course, was why the aircraft was not equipped.
So it's not part of the minimum equipment list? The Air Inter management had decided they did not like the false warnings that had been produced by GPWS equipment.
TRANSLATOR: Normally, most planes fly slower than 250 knots when under 10,000 feet but we flew at 350 knots until the final approach.
So, at those speeds, GPWS was always giving off false alarms.
This decision, while legal, prevented the pilots from having one last line of defence before crashing into the mountain.
It's impossible to imagine that the pilot wouldn't have pulled up if he'd heard the alarm.
We should have a GPWS on commercial flights in any case, yes.
That's an obvious conclusion.
The report will list these causes - flight deck ergonomics Investigators conclude that there was no single cause for the crash of Flight 148.
The tragedy involved an ill-fated combination of many different weaknesses in the airline industry.
We made 35 or so recommendations, including pilot training, about the ground proximity warning system, and so on.
The recommendations lead to sweeping changes.
Pilots must now have more A320 training before getting behind the controls.
One of the two pilots now need to have at least 300 hours on the plane.
They estimated that 300 hours were enough.
Another change - the design of a more heat-resistant black box.
The FAA did a test, did some studies, what the thermal characteristics of post-crash fires were and came up with a value of 260 degrees C for 10 hours.
Delta Alpha, your position? Air Inter Delta Alpha, Strasbourg.
As a result of the Strasbourg crash, the A320 is now a safer plane.
You can only get this change if there is what people perceive to be a good reason.
And sadly, a good reason is still an accident.
But improved aviation technology is still no substitute for well trained, well prepared pilots.
There's an old axiom in aviation that you're taught early on - never let an airplane take you somewhere that your brain hasn't visited at least five minutes ahead of time.
This is an excellent example of a flight crew that didn't follow that particular axiom.
.
an Airbus A320 slams into a mountaintop.
(SCREAMING) There are survivors .
.
but what they don't realise is that no-one knows where they are.
(SIRENS BLARE) Before investigators can begin searching for what caused the crash of Air Inter Flight 148, they must first find the plane.
January 20, 1992.
Air Inter Flight 148 has departed from Lyons, France.
124.
905, thank you.
Captain Christian Hecquet and First Officer Joel Cherubin are experienced pilots with over 12,000 hours of flying time between them.
The flight is a short hop between Lyons, in central France, and the city of Strasbourg, in the mountainous Alsace region.
The French airline Air Inter caters mostly to business travellers.
Crews are motivated to avoid delays, as former Air Inter pilot Gerard Arnoux explains.
We were famous for our very short turnaround .
.
and the faster we flew, the better wages we got.
Have we been flying for 35 minutes yet? 41 minutes.
The crew is flying an Airbus A320, one of the most technologically advanced commercial airplanes in the world.
Even before take-off, the pilots program the autopilot to land on a specific runway in Strasbourg.
The cockpit of the A320 is also very different from other planes.
Instead of analogue gauges, the pilots look mostly at digital displays.
AIRPORT RECORDING: Strasbourg.
Good evening.
Runway in use - 05.
Transition level, 50.
Wind, 040 at 1-8 knots.
Visibility - 10km.
A recording from Strasbourg Airport informs the crew of a change in plan.
Due to high winds and poor winter weather, they'll have to land on an alternate runway 05 in service.
.
.
not the one programmed into the autopilot.
05? What sort of wind are they giving us? 18 knots.
18 knots.
Captain Hecquet doesn't like the idea of changing runways.
No chance.
He was hoping to use runway 23, an approach that provides the autopilot with a precise navigational fix.
The new runway, runway 05, is surrounded by mountainous terrain that can interrupt radio signals sent to the autopilot.
No, if we go with the runway 05 procedure, we No, no.
(CLEARS THROAT) Captain Hecquet suggests a compromise.
I am putting back runway 23 - otherwise, I couldn't make the ILS interception.
He'll program the autopilot to fly towards runway 23 .
.
but near the airport, the captain will take over the controls and make a visual landing on runway 0-5.
We are taking 23, then? (SHOUTS) Yes! Agreed.
Roger 854.
Proceed to GTQ air level 140 Strasbourg.
Uh, yes.
We intend to proceed to do an ILS on runway 23, then an indirect for runway 05 after that.
The Strasbourg controller considers the captain's plan.
He warns that there will likely be a delay due to heavy traffic.
Given that we're going to have three take-offs on 05, you risk waiting in a stack at 5,000 feet.
We're not going to mess about like that descending at full speed.
If they had warned us in advance! Cripes! (SIGHS) Inter France, Strasbourg.
I hear you.
Aware of the captain's frustration, the controller offers assistance.
If you want, I can take you with the radar to lead you to ANDLO at 5,000.
ANDLO is a navigational point on the approach to runway 05.
It helps pilots align the plane for landing.
Yeah.
Yeah, that's good.
- Oh, yeah.
- OK, then.
Turn left to heading 230 degrees.
Since runway 05 doesn't allow for a full autopilot approach, the captain must calculate the angle of descent on his own.
That makes 3.
3 degrees.
(SPEAKS FRENCH) TRANSLATION: 3.
3 degrees is a normal flight angle that provides a good slope for landing.
Turn left, steer 90.
090 degrees, Delta Alpha.
The controller talks Flight 148 through the last turn to align the plane with the runway, now 25km away.
Then, First Officer Cherubin notices the plane is slightly off-course.
We are headed inside.
You're inside there.
You should have started with 070.
Yeah.
- At least that much.
- (CLEARS THROAT) The controller also notices that the plane is off-course.
Air Inter Delta Alpha? It has missed aligning itself with ANDLO, the runway's electronic guidepost.
Delta Alpha? You're passing to the right of ANDLO.
Nevertheless, he authorises the landing.
Authorised for final approach 05.
Delta Alpha.
The captain initiates the landing sequence.
Gear down.
Hecquet notices that the plane is travelling too fast, so he extends the speed brakes.
They disrupt airflow over the wing, which helps create more drag to slow the plane.
We have to watch our descent.
The approach axis.
The First Officer is more concerned with their horizontal position in relation to the runway.
It was 60.
Check it out.
But before the crew can adjust their course Merde! (SCREAMING) Delta Alpha, your position? Air Inter Delta Alpha, Strasbourg.
The crash is catastrophic.
The A320 has flown into the side of a mountain.
Delta Alpha, your position? Flight 148 is no longer on radar .
.
nor responding to radio contact.
An emergency is declared at Strasbourg Airport.
This is the last hit we got.
They were flying about 20km away from the airport.
Officials need to pinpoint the crash site, but it's not as easy as it might seem.
The airport's radar is not recorded, there has been no signal from the plane's emergency beacon and, surprisingly, no-one has reported seeing a plane go down.
It could be anywhere in here.
The proposed search area covers more than 20 square kilometres of dense forest just outside Strasbourg.
Nicolas Skourias survives the crash of Air Inter Flight 148 with only minor injuries.
(COUGHS) I saw a fire in front of me and I panicked because I say to myself "I'm going to burn.
" I went to the back of the plane - what remained of the plane (COUGHS) I found some other survivors.
Come on! It's gonna blow! With the smell of leaking jet fuel in the air, the survivors move away from the burning plane.
We stay together, waiting for the first aid.
But the wait will be longer than anyone might expect.
The first reaction that we have after the crash was, "OK, in half an hour, one hour most, "and the worst, OK, the rescue team will be here.
" And it wasn't here.
One hour after the crash, rescuers still have no idea where the wreckage lies.
(GROANS) Skourias and the others now face a new ordeal .
.
surviving sub zero temperatures in a dark and isolated forest.
2.
5 hours after Flight 148 disappeared from radar near Strasbourg Airport (SIRENS WAIL) .
.
the missing plane has still not been located.
Amidst growing tension, the French aviation bureau, the BEA, sends in its lead investigator, Jean Paries.
I immediately called my two main investigators and we organised the go-team .
.
and we got prepared to rush to the site as soon as the site was located somewhere.
The delay feels like an eternity.
A little bit surprisingly long.
We can expect this in the jungle or the rainforest but not exactly in a highly dense populated area like the Strasbourg area.
With no help in site, Skourias returns to the wreckage to look for more survivors.
I think that some people that died could have survived if the first aid come sooner.
Nearly 1,000 people search for the missing plane, but three hours after the crash, there's still no sign of it.
Frustrated, Skourias goes looking for help.
He stumbles into a TV crew trying to find the crash but with no wreckage in sight, they react with scepticism.
They didn't expect survivors from an airplane crash.
Hey! Hey, you've got to believe me! They didn't believe that there was one of survivor but believe me, I was, because my face was black due to the smoke, the kerosene and so on.
Come on! The journalists follow him back to the crash site .
.
where they discover eight other survivors.
- Finally, the first rescuers arrive.
- (SIRENS WAIL) The crash site is located near the top of the 2,500ft Mont Sainte-Odile .
.
19km from the runway.
They found us after 4 hours and 30 minutes, so it was a mess.
I was very, very disappointed that 20km from Strasbourg and they couldn't find us.
A total of 87 passengers and crew have died, including the pilot and co-pilot.
The survivors begin to tell their stories but no-one reports anything that might explain why the plane crashed.
(SPEAKS FRENCH) TRANSLATION: I don't know what happened.
We were landing.
I lost all consciousness.
We must have hit the trees.
Bob Macintosh, an American NTSB investigator, arrives at the crash site.
The BEA of France recognised the international attention would be on this accident, even though it was a domestic accident.
He invited a group of international accident investigators to come and participate.
Welcome to the team.
The first priority for investigators is to retrieve the plane's black boxes.
We are very anxious about the state of the tape inside.
Will it be possible to use it? Will we get the critical information we need? In France, aviation accidents are also investigated by the justice system.
Paries and his team are not allowed access to the site until judicial officials secure the black boxes.
In a previous crash, the crash of Air France Flight 296 in 1988, investigators waited 10 days before turning the black boxes over to police.
Rumours persisted that these boxes had been tampered with.
This time, police are keeping investigators at bay.
I can recall seeing the glowing embers and seeing the flight recorder sitting there .
.
and not being able to intervene and say "Get that thing cooled down as soon as you can.
" After midnight, the boxes are retrieved from the plane and sent for analysis.
Investigators can only hope it's not too late.
PARIES: They were extremely hot.
They looked damaged and they looked burned.
(FEMALE REPORTER SPEAKS FRENCH) In the light of day, investigators get some of their first clues from the crash site itself.
They discover why the plane's emergency locator beacon didn't send a signal.
It was actually destroyed by the impact.
The beacon is located inside the cockpit and is designed to start working after a crash.
Its failure suggests an unusually forceful impact with the ground.
We had this first feeling the descent was abnormally steep.
Investigators examine the engines to see if they may have stalled before impact.
If you find the blades curved and a lot of wood sucked inside the engines, then you understand that the engines weren't working properly.
And that's exactly what they find.
The plane clearly had power, yet it ploughed steeply into a mountainside without ever sending out a distress signal.
Investigators are puzzled.
They hope that the box which recorded the plane's flight data will help them solve the mystery.
The black box is designed to survive temperatures up to 1,100 degrees Celsius for half an hour.
The tape recorder inside is protected by a capsule filled with water.
MAN: When the recorder heats up, the water turns to steam, absorbing the energy and actually vents out through a little hole in the crash enclosure.
But when the flight data recorder is opened, investigators make a troubling discovery.
The FDR was totally damaged.
Impossible to read anything from it.
There's now only one hope for recovering the plane's flight data - a device called a Quick Access Recorder, or QAR.
Maintenance workers use the QAR to access the plane's computers, but it also records some flight data.
Unlike the black boxes, the QAR is stored near the cockpit.
Quick Access Recorders are not protected at all.
They're up in the front end of the aircraft, typically in the electronics bay.
They're generally destroyed just from the impact damage.
Investigators are encouraged to discover that in this case, the QAR has survived.
But on closer examination, their optimism turns to frustration.
The last 20cm of the tape were burned and stretched and were damaged to the point that we couldn't use them into a machine - we couldn't read it.
Investigators are desperate to retrieve the data, so they take a chance on an experimental technique.
Known as the 'garnet technique', a light is shone through a mineral lens made of garnet.
The special lens helps the technicians differentiate between the positive and negative magnetic pulses which translate as binary digits, or 'bits'.
There are 768 bits per second, so that's a lot of ones and zeros.
You have to be very precise in moving the tape under the lens, or the garnet, to make sure you don't miss a bit or read the same bit twice, you know? So it's difficult.
It took about a day to read a second of recording.
Any additional seconds recovered could reveal something that would make a difference.
The effort to retrieve all the QAR data could take a month or more.
In the meantime, the focus of the investigation shifts to the cockpit voice recorder.
The cockpit voice recorder, which was just inches away but outside of the ashes, had air passing over it, survived.
HECQUET: (ON RECORDING) Runway 23.
The recording reveals the captain's anxiety early on in the flight.
- You are taking 23, then? - Yes! Investigators know that landing on runway 05 requires what's called a 'non-precision approach'.
That means pilots receive electronic guidance only on their horizontal position - left and right.
They get no guidance when it comes to altitude.
05.
What sort of wind are they giving us? - 18 knots.
- 18 knots.
The non-precision approach increases the demands on the pilots.
Investigators can also hear that the captain had concerns about landing on runway 05.
RECORDING: 48 Delta Alpha, you are number one with runway 05.
10 nautical.
That won't work.
It's a lot of distress over non-precision approach.
Wondering what can cause such distress, investigators research pilot training at Air Inter.
They find that most pilots did not have extensive training making non-precision landings in the new A320.
I think we should have had double the training compared to an older plane.
Investigators ask the airline for detailed records on the pilot's history of runway approaches.
They're intrigued by what they discover.
Captain Hecquet had landed at Strasbourg countless times but he had never landed an A320 there using a non-precision approach.
Not going to mess around like that descending at full speed.
Clearly, the captain was uneasy about having to execute a landing he had never made before.
I think the captain was worried about making it in in a minimum amount of time and the minimum amount of delay.
Have we been flying for 35 minutes yet? And the co-pilot was worried about not getting in trouble by offending the captain.
At least that much.
More research into the pilot's work history offers yet another revelation.
While the two pilots had flown more than 12,000 hours between them, they were both still relatively new to the highly advanced A320.
05 in service.
The aviation community misunderstood the magnitude of changes brought by the new Airbus A320.
The captain had only 162 hours in the A320 .
.
and the co-pilot, even less - just 61 hours.
PARIES: Behind this accident scenario, there is an issue of confidence of the crew - in themselves, in the aircraft.
18 knots.
No chance.
They were not prepared, really, to fly in this kind of condition.
If they had warned us in advance.
Cripes! Investigators conclude that the crew's training was insufficient, but that alone does not explain the crash.
Merde! (SCREAMING) Investigators search for other factors in the crash of Flight 148.
They review the conversations between the crew and air traffic controllers.
If you want, I can give you radar headings and take you to ANDLO at 5,000.
Yeah.
Yeah, that's good.
TRANSLATOR: The radar vector makes flying easier.
The captain was happy because it was reducing his workload.
Turn left, steer 90.
With the controller's assistance, this landing should have been very simple.
But when investigators reconstruct the plane's trajectory using radar information from various stations around the airport, they discover a shocking error.
The 090 heading started here.
090 degrees, Delta Alpha.
But it won't take them to ANDLO.
Last radar vector the controller gave was incorrect.
It sent them Thank you.
.
.
closer to the mountain.
PARIES: They were off-course because of following the heading they got from the radar vectoring.
They found themselves in this undershoot situation.
You're inside there.
- You should have started with 070.
- Yeah.
Investigators are also troubled by the controller's choice of words when he warned the pilots, incorrectly, that they were headed to the right.
Delta Alpha.
You're passing to the right of ANDLO.
From the pilot's perspective, the plane was on the left side of the runway, not the right.
It could only add to their confusion.
TRANSLATION: It was very poor guidance because he didn't employ the usual terminology.
(SPEAKS FRENCH) Investigators recommend that controllers use only compass points when giving directions - never the words 'right' and 'left'.
The controller's mistakes clearly brought the plane closer to the mountain.
Turn left, steer 90.
But once again, investigators feel they don't have the whole story.
It's not something totally abnormal to start a descent from this situation.
Flaps towards 2.
Flaps towards 2.
It's not what you expect to do every day, but it's not outside the tolerance of the concept of this approach.
Gear down.
When investigators study the plane's reconstructed flight path, they discover something more alarming than the plane's horizontal misdirection.
As it circled the mountain, the plane inexplicably entered a dangerously steep and rapid descent.
Perhaps 2.
5 times the normal rate of descent.
It's lethal at that altitude.
Without the steep descent, they would have cleared the mountain.
If the vertical trajectory had been correct, they would have no problem at all.
Finding the cause of that sudden descent is now key to understanding why 87 people died in one of the most advanced passenger planes on earth.
Authorised for final approach 05.
Descent was initiated at 1800 hours, 19 minutes and 38 seconds.
That Delta Alpha.
.
.
is the point of no return.
By studying Flight 148's trajectory, investigators determine that the rapid descent began 60 seconds before the crash.
Delta Alpha.
There is no indication on tape that the descent was deliberate.
How it happened and why the crew didn't notice is a mystery.
It should be a no-brainer, keeping track of the altitude.
The cockpit altimeter gives pilots a constant readout of their altitude.
Altimeter - that's a very precise instrument.
They've become very reliable.
They're accurate between 5 or 10 feet.
Ignoring it would be a major error in flying protocol.
Flaps towards 2.
The recording reveals just one single remark from the crew about their descent.
We have to watch our descent.
(CLEARS THROAT) It occurred 16 seconds before the crash.
The captain had just extended the speed brakes.
The aircraft was accelerating abnormally.
The captain started to realise there was something wrong with the descent rate.
But the first officer changed the subject.
The approach axis - we're hitting the axis a half-point off.
There.
- It was 60.
Check it out.
- Yeah.
(CLEARS THROAT) He refocused the captain's attention on the lateral situation rather than the vertical situation, which was the main problem, of course, and they both failed to recognise the situation.
I think they were planning they were going to break out of the clouds and they'd be able to see the runway and they wouldn't need to do the full instrument approach.
It was 60.
But the plane never left the clouds.
There's an old adage in aviation - rocks have been known to hide out in those clouds.
Merde! It now seems clear that the crew was not monitoring their altitude closely enough.
- But a bigger mystery remains - We can only guess .
.
what caused that deadly descent? After months of work, investigators may finally have the answer.
All the available flight data from the damaged Quick Access Recorder has been recovered.
The data confirms that just before the crash, the plane was speeding towards the ground at an extremely high rate - 3,300 feet per minute.
It also confirms that the angle of descent was dangerously steep .
.
much greater than the 3.
3 degrees selected by the captain.
3.
3 degrees.
That's quite a difference.
Investigators now wonder - did the autopilot malfunction? Did it somehow fail to obey the captain's safe descent angle and send the plane into a deadly nosedive? What state was it in before the accident? Unfortunately, the flight control unit which houses the autopilot is too badly damaged to provide any definitive answers.
We could never demonstrate that this FCU on this aircraft during this flight functioned properly or not.
But then when he returns to studying the flight data, Paries discovers something that may finally reveal the cause of the crash.
He notices a similarity between two key numbers - the plane's vertical speed - 3,300 feet per minute - and the intended flight path angle - 3.
3 degrees.
Coincidence? Paries uses a flight simulator to test a new theory.
Can you show me a descent of 3,300 feet per minute? He believes that the similarity is no mere coincidence.
On the autopilot, there are two descent modes - flight path angle and vertical speed - but they are both displayed on the same window .
.
so 3,300 is abbreviated to 33.
Now, show me a flight angle of -3.
3 degrees.
The problem on this aircraft was that the two values were visible on the same window and controlled by the same knob.
3.
3 degrees.
-3.
3 degrees.
Paries strongly suspects that the confusing display tripped up Captain Hecquet.
The confusion is quite easy between the two modes if you don't do it carefully.
If the captain failed to push the mode selector knob, then entering '33' would not have initiated a safe 3.
3 degree angle of descent.
Instead, it would have put the plane into a deadly rate of descent of 3,300 feet per minute.
(SCREAMING) Two months after the crash, another Air Inter plane enters a dangerously steep descent for the same reason.
The crew only discovered the problem when they broke out of the clouds.
Those pilots also confused the plane's flight path angle with its vertical speed.
They were lucky enough to have a much higher cloud base so they could correct the problem.
Further research reveals an industry-wide problem with the A320.
Many people confuse these modes - especially during training - and many of them fell into the trap even after the training.
Eager to test his new theory, Jean Paries programs a simulator with all the known data from Flight 148.
He then inputs the same rate of descent he believes the Air Inter pilots selected.
If Paries is correct, the simulation will end with the plane hitting the mountain.
- But it doesn't.
- We are missing something.
Strangely, this didn't lead to a crash.
Every approach would overfly this obstacle by a significant margin.
Have we factored in the wind? We started to work on other alternate hypotheses Let's try again, but initiate the turn sooner.
.
.
but nothing was really credible.
No matter how hard he tries, Paries cannot simulate the crash.
Unable to explain why, he turns to the plane's manufacturer for help.
Thanks for bringing this to my attention.
After much research, an Airbus designer comes to Paries with an explanation about a little-known element of the autopilot's design.
In emergency situations where the A320 needs to change direction quickly, the autopilot is programmed to reverse the plane's direction at twice the normal rate.
The reaction of the autopilot would be much faster in these cases where typically, when you are descending and asking the autopilot to climb, or climbing and asking the autopilot to descend.
We immediately went back to the data at the very second at which the descent was commanded by the crew.
Gear down.
Paries discovers a tragic coincidence.
Sadly, we found at this very second, there was turbulence.
There was an ascent.
It's very slight, but there it is.
The momentary turbulence caused the plane to climb slightly.
And this led to a positive 600 feet per minute vertical speed for maybe half a second.
It was during that same half-second that the crew commanded the plane to descend.
It was 60.
Check it out.
But the autopilot read this as an emergency, requiring a blazingly fast descent.
That could be it.
Investigators now contemplate a terrible thought - could a random gust of wind, hitting at exactly the wrong split second, have been the difference between life and death? Here it comes.
(MACHINE BEEPS) And we got a crash.
Paries' theory explains every aspect of the crash.
It's a fascinating lesson about the random dimension of accidents.
Half a second before, half a second later, they wouldn't have the accident.
The discovery of a confusing cockpit display .
.
has enormous implications for the entire industry.
The flight instrumentation of aircraft like the DC-10, MD-11s, the 74s and so on - all the Boeing products and all the commuter products that were using that avionics suite had this vulnerability about it.
Investigators now face a daunting question affecting aircraft safety around the world.
If the design of the autopilot interface isn't changed, how many more people could die? There's mounting evidence that the design of the autopilot interface on Airbus A320s led the Air Inter pilots to accidentally dial in a dangerous rate of descent.
HECQUET: 3.
3 degrees.
We felt a need to start the industry to work on this.
The plane's manufacturer, Airbus, responds immediately.
The main change, which was very quickly made, was to change the display window.
With the new design, if a pilot selects a vertical speed of 3,300, the entire 4-digit number is displayed.
So the confusion between an angle and the vertical speed was no longer possible.
For investigators, only one mystery remains.
All Airbus A320 jets are designed to be equipped with a safety device known as a ground proximity warning system, or GPWS.
Which is a downward-looking, single-purpose radar that tells you how high you are above the ground directly beneath the airplane and if it gets to be too low, it'll set off a warning.
- (ALARM SOUNDS) - WARNING SYSTEM: Pull up! Terrain! - (ALARM SOUNDS) - Pull up! But Captain Hecquet We have to watch our descent.
.
.
never received a warning for one very simple reason - his A320 didn't have that alarm.
Merde! (SCREAMING) The first question, of course, was why the aircraft was not equipped.
So it's not part of the minimum equipment list? The Air Inter management had decided they did not like the false warnings that had been produced by GPWS equipment.
TRANSLATOR: Normally, most planes fly slower than 250 knots when under 10,000 feet but we flew at 350 knots until the final approach.
So, at those speeds, GPWS was always giving off false alarms.
This decision, while legal, prevented the pilots from having one last line of defence before crashing into the mountain.
It's impossible to imagine that the pilot wouldn't have pulled up if he'd heard the alarm.
We should have a GPWS on commercial flights in any case, yes.
That's an obvious conclusion.
The report will list these causes - flight deck ergonomics Investigators conclude that there was no single cause for the crash of Flight 148.
The tragedy involved an ill-fated combination of many different weaknesses in the airline industry.
We made 35 or so recommendations, including pilot training, about the ground proximity warning system, and so on.
The recommendations lead to sweeping changes.
Pilots must now have more A320 training before getting behind the controls.
One of the two pilots now need to have at least 300 hours on the plane.
They estimated that 300 hours were enough.
Another change - the design of a more heat-resistant black box.
The FAA did a test, did some studies, what the thermal characteristics of post-crash fires were and came up with a value of 260 degrees C for 10 hours.
Delta Alpha, your position? Air Inter Delta Alpha, Strasbourg.
As a result of the Strasbourg crash, the A320 is now a safer plane.
You can only get this change if there is what people perceive to be a good reason.
And sadly, a good reason is still an accident.
But improved aviation technology is still no substitute for well trained, well prepared pilots.
There's an old axiom in aviation that you're taught early on - never let an airplane take you somewhere that your brain hasn't visited at least five minutes ahead of time.
This is an excellent example of a flight crew that didn't follow that particular axiom.