Mayday (2013) s17e02 Episode Script
Deadly Myth
We're in a dive.
We're in a dive! - A commuter flight to Detroit - Come on! - Get out of it, for god's sake! - .
.
ends in a harrowing spiral.
No! Tremendous destruction of the aircraft and it was a very severe impact.
Whatever happened happened in a heartbeat.
Close analysis of flight data - Ice protection.
- Standard seven.
.
.
leads to a stunning realisation.
It's Roselawn all over again.
What doomed flight 3272 is a danger the industry has known about for years.
- Why would you wait? - In aviation, myths die hard.
To fulfil their mission, investigators confront some of aviation's biggest players.
But it was time to break that myth.
26 passengers are on their way to Detroit, Michigan, aboard Comair flight 3272.
It's a short afternoon flight, operated by one of America's largest regional airlines.
The workhorse of the Comair fleet is the Embraer 120, a Brazilian-made turboprop.
The aeroplane handled quite nicely just about every regime that you could think of that would be appropriate for the type of operation that we were doing.
Darinda Ogden has been a Comair flight attendant for five years.
Here's your drink, Ken.
Thank you very much for the ice.
First Officer Kenneth Reece and Captain Dann Carlsen have been trading piloting duties all day.
This is the day's third leg, and First Officer Reece is at the controls.
I flew with Kenneth two or three times.
Stewart Lauer is a former Comair pilot.
I found him to be a very capable pilot.
He knew what he was doing.
Out of Detroit, we're looking at 239 radio.
Flight 3272 left from an airport near Cincinnati, Ohio, at 3:08pm.
It's expected to land in Detroit just after 4:00 in the afternoon.
What we want to do is just fly from point A to point B, get the passengers there safely and not scare them to death.
Heading 030 to join the arrival, Comair 3272.
The crew has the plane on autopilot as they descend from 19,000ft.
Closing in on their destination Just some bumps.
.
.
they hit some unexpected turbulence.
Despite the rough ride, the flight is still running on schedule.
At Detroit approach control, Steven Cochran guides flight 3272 on its path to the runway.
Comair 3272, Detroit approach, heading 050, vector to runway 3-right.
It's another busy Thursday afternoon for the veteran controller.
Detroit, Cactus 50 level at 13,000.
The crew of an Airbus radios in.
Cactus 50, Detroit roger.
Expect vectors for 3-right.
The controller wants the larger Airbus to land before Comair 3272.
The jet was much faster than the Comair flight, so the air traffic controller made the decision to give the jet priority over the turboprop and gave them a slower speed.
Comair, maintain one-niner-zero knots.
If unable, advise.
Roger.
One-niner-zero knots, Comair 3272.
First Officer Reece reduces his speed to 190 knots to let the incoming Airbus land first.
The fact that 3272 was placed behind Cactus 50 added just a little bit to the manoeuvring that was required.
Ladies and gentlemen, we're on our final approach to Detroit.
Please fasten your seatbelts and stow away your tray tables.
The turboprop should be on the runway in less than 15 minutes.
Let's run the descent check.
The approach phase is the busiest time, from a workload standpoint, for the pilot, and that's the time that he would like to have the assistance of the autopilot.
Ice protection.
- Windshield, props, standard seven.
- Pressurisation? As they descend towards 8,000ft, the pilots perform a series of standard checks.
Auto.
As you come in to land, we go into what's called a sterile cockpit.
- On.
- There's no chitchat.
It's just strictly business.
Comair 3272, reduce speed 150.
The controller now wants the turboprop to reduce its airspeed to 150 knots.
Comair 3272, reduce speed 150.
Uh, speed 150, Comair 3272.
They always gotta tell us everything twice.
He's got a short-term memory disorder.
Is that what that is? Yeah, that's what it is.
Comair 3272, now turn left heading 090.
Heading zero-niner-zero, Comair 3272.
The aeroplane was responding properly to all the air traffic control inputs and not indicating any problem.
The plane banks left as the pilots make their final turn.
First Officer Reece expects the autopilot to roll the plane back to level.
- Check your low speed indicator.
- Yeah, I see it.
Power.
The captain advances the throttles to gain speed.
But it doesn't work.
Instead, the plane's autopilot disconnects.
When the autopilot disconnected, a very sudden left roll began.
Reece struggles to level the plane by hand.
The amount of effort it takes to move those control surfaces, you've gotta muscle it, until you can get it where you want it to go.
He wrenches the control column to the right, trying to correct the steep left bank.
Nothing really happened.
The aircraft continued rolling over to the left.
We're in a dive.
We're in a dive! - No! - Oh, god! Stop it.
You never give up.
You do what you gotta do to try to make it.
Come on! Get out of there, for god's sake! - No! Ahh! No! - God.
But there seems to be nothing the pilots can do to save their spiralling aircraft.
There was virtually no controllability.
Nothing can be done.
There's just insufficient altitude to recover.
No! It was a very severe impact, a tremendous destruction of the aircraft.
Comair flight 3272, en route from Cincinnati to Detroit, has crashed into a field 22 miles southwest of the runway.
First responders and news crews arrive at the scene.
They soon discover that all 26 passengers and three crew members are dead.
We'd like to offer our condolences to the families.
Our hearts go out to those that have lost their loved ones today.
The sudden loss of 29 lives leaves families and friends in anguish.
How could a commercial flight on approach to a major US airport go so horribly wrong? The morning after the crash, a blackened scar on a frozen field marks the fatal impact zone of flight 3272.
Families and friends attend a memorial service for the crash victims.
A team from the National Transportation Safety Board is already at work recovering pieces of wreckage for analysis.
God, it's cold out there.
NTSB systems specialist John DeLisi, faces one of the biggest challenges of his career.
We knew the aeroplane was on approach into Detroit and something dramatic happened.
Something went wrong suddenly.
What a mess.
The NTSB's Richard Rodriguez leads the investigation.
Our mission is to find the cause of an accident, make recommendations that will prevent it from ever happening again.
Their first task is to find the aeroplane's black boxes and get them to a lab for analysis.
We spent about eight hours that first day pulling the wreckage apart, but by late afternoon, we were able to find the cockpit voice recorder and flight data recorder and got those on their way back to Washington, DC.
Let's get those both packed up.
While they wait for news on the black boxes, investigators head to Detroit Metropolitan Airport.
They want to hear from the last person to speak with the pilots, the approach controller.
It was coming on rush hour.
We had winter weather.
From air traffic control, we learned that the weather was bad and aeroplanes were starting to pile up on their approach into Detroit.
That's when American West Airlines flight 50 made contact, right after 3272.
Detroit, Cactus 50 level at 13,000.
So I told Comair pilots to slow up.
Comair, maintain one-niner-zero knots.
If unable, advise.
I made sure there was plenty of distance between them.
There were other aeroplanes on approach to Detroit, so sometimes, when they're in close proximity, there can be an issue with wake vortex coming off one aeroplane that may affect another.
A wake vortex is a horizontal tornado that trails behind an aircraft.
If one plane flies too close to the wake of another, it can encounter sudden and extreme turbulence.
So, here's the Airbus, American West Airlines flight 50.
The Airbus is almost twice the size of the Embraer 120.
And here's our Comair flight 3272.
Did the big jet's wake vortex knock flight 3272 out of the sky? We needed to see if the accident aeroplane flew underneath that one and crossing its wake.
The Embraer is miles back and 1,500ft below the Airbus.
It could have got caught in the wake.
Get the data off to NASA.
Let's see what they think.
A wake vortex incident seems possible.
Oh, god.
But investigators won't know for certain until experts at NASA analyse the radar data.
Meanwhile, Rodriguez explores other possibilities.
What happened to the propellers? He's very familiar with this type of plane, having investigated previous accidents involving the Embraer 120.
The Embraer 120 propeller blade had separated due to a fatigue crack over in an accident earlier in Georgia.
The crew was able to land it, but it was very difficult to control.
In two previous cases, part of a propeller blade broke off in mid-flight.
Could it have happened again? If they were barely able to control the aircraft, this could have been a problem for 3272.
Rodriguez reviews details of where the two propellers were found at the Detroit crash site.
He makes a worrying discovery.
Part of one blade ended up 75ft away from the main impact crater.
He can't help but wonder, "Is this the third time he's seen an Embraer 120 propeller fail?" We need to take a look at those propeller blades.
We wanted to get those blades of the propeller to ensure they were intact at impact.
Rodriguez studies the blade fragment, looking for evidence that they failed.
If the propeller blades were all attached and spinning at impact, they would have all hit the ground with tremendous force, resulting in distinctive damage on every blade.
This is impact damage.
They were spinning right till the end.
The analysis leaves no doubt.
Though the propellers fragmented when they hit the ground, they did not fail in flight.
We were able to tell that nothing broke off the aeroplane during the flight path.
Just as investigators rule out one possible cause, they uncover another intriguing clue.
Richard.
What'd you find? It's the engine fire lever.
It looks like it's been pulled.
- Well, what do you know? - This is a lever that a crew member would grab and pull and twist if one of the engines was on fire.
It's a discovery that has frightening implications.
So, are we looking at an engine fire here? Yeah.
Put that one here.
NTSB investigators explore a disturbing theory as they try to figure out why Comair flight 3272 plunged from the sky.
It looks like one of the pilots pulled an engine fire shut-off lever.
The crew may have been fighting an in-flight fire.
No doubt there was a fire.
But when? While the engines are clearly fire-damaged, investigators need to determine if the fire started before or after the plane hit the ground.
In-flight fire would probably be at a higher temperature than ground fire.
So you would have some, perhaps, disintegration or disappearance of the metal in the engine cowling and so forth.
If the fire started while the plane was moving forward, Rodriguez should find soot marks in a horizontal pattern.
If it started on the ground, the flames would have gone straight up.
In this case, it was a relatively low temperature.
All the soot patterns were vertical.
There was no airstream affecting it to show in-flight fire.
I'm seeing post-impact fire.
There was no fire in flight.
No way.
But upon further examination, we realised that the extended position of that handle was just the result of the hard crush damage that occurred.
With engine fire ruled out, investigators turn to NASA analysis of airport radar data.
Perhaps there's evidence that the small turboprop flew too close to a big jet.
And that settles that.
- But it's another dead end.
- It wasn't a wake vortex.
The analysis reveals that the wake from American Airlines flight 50 could not have dropped to the altitude of Comair 3272.
Investigators are back where they started.
Why 29 people died just minutes before landing, is still a mystery.
OK.
Are we all ready? They finally have access to the information stored on the plane's flight data recorder.
Left engine, please.
It should give them important details on how the aircraft - was performing - Now the right engine.
.
.
right up to the moment of impact.
Here, it's starting.
The engines are guzzling fuel.
The speed stays the same.
The data reveals a puzzling detail.
Just before the plane's sudden upset, the engines were working hard, but to little effect.
The flight data recorder indicated that the aircraft had received significant degradation in the performance.
Look what's happening here.
Autopilot is on.
Turn is done.
It should level off here, but it doesn't! It keeps rolling.
What really jumped out at us was the aeroplane continued to roll even though the autopilot was commanding it to go back to 'wings level'.
Something's holding the engines back.
As Rodriguez studies the data, he thinks he knows what may be causing the loss of performance.
4,000ft and it falls from the sky.
It has all the hallmarks of something he's seen before.
It's Roselawn all over again.
In October 1994, the pilots of American Eagle flight 4184 suddenly lost control of their twin turboprop.
It plummeted from the sky and crashed near the town of Roselawn, Indiana.
All 68 people aboard were killed.
The cause was a phenomenon known as 'icing', where a build-up of ice on the wings leads to an aerodynamic stall.
When that happens, you're not producing the lift that you're supposed to.
The loss of lift is what the word 'stall' is pertaining to.
When we stall a wing, we lose lift.
Following the Roselawn accident, the FAA was very concerned about the performance of turboprop aircraft in icing conditions.
The Federal Aviation Administration was so concerned that it conducted full-scale tests, flying a turboprop in super-cooled precipitation sprayed from a water tanker.
They found that the aircraft did accumulate ice on the top of the wing to generate significant drag.
The flight test's most dramatic finding was that a layer of ice as thin as a sheet of sandpaper could cause major problems for pilots.
And we were very surprised to see that a trace amount of ice was enough to affect the aerodynamics and possibly cause a wing to lose lift.
In March 2001, a turboprop suffered severe icing and plummeted 8,000ft before the captain managed to regain control.
That captain was Stewart Lauer.
I had the shakes something fierce.
I mean, the adrenalin was flowing.
Just like Stewart Lauer's flight - PILOT: Check the low speed indicator.
- Yeah, I see it.
.
.
Comair 3272 dropped from the sky in an instant.
It's very complex aerodynamically, especially when there's ice contamination on the aeroplane.
At that point, it's very hard to recover unless you gain airspeed, and the only way to do that is point the nose to the ground, not something that a pilot wants to do while trying to recover an aeroplane.
Come on! Get out of it, for god's sake! We're in a dive.
We're in a dive! - No! - God.
I was lucky.
I had plenty of altitude.
It took me over 7,000ft to regain control of the aircraft.
They didn't have it.
They only had 4,000ft.
OK.
Let's see what they flew through.
Investigators need to learn all they can about the weather conditions the Comair pilots faced as they approached Detroit.
OK.
Let's pull up the weather data.
They know that for ice to build up on the wings of a plane Now let's see the flight path.
.
.
temperature, precipitation and aircraft speed must all fall within a very specific range.
The weather data began to paint a picture of an aeroplane that was on approach, flying in what were considered light icing conditions.
Cactus 50, be advised there are slick runways and poor visibility.
How is it up there? According to Detroit controllers, other aircraft, including the flight directly in front of Comair 3272, reported icing conditions.
Yeah, it's 237 at 32 up here, moderate icing with a possibility of freezing drizzle.
Roger that.
Comair 3272, how is it up there? Strangely, the Comair pilots never mentioned icing at all.
Comair 3272, uh, just a little turbulence but that's all.
It definitely flew through icing conditions.
Investigators know that even a thin layer of ice can be dangerous.
Perhaps the Comair pilots didn't see ice on their wings.
The Embraer is probably a good 20ft from the cockpit window back to the top of the wing where the ice would be forming, so it's very difficult to see in poor lighting conditions as 3272 was operating, in cloud and late in the afternoon.
But there's something that doesn't add up.
If the Comair pilots lost control due to ice build-up, why were they the only ones affected by the weather? None of the other aeroplanes that were on approach to Detroit experienced any problems.
So we needed to look hard about what was unique about this aeroplane.
NTSB investigators are now certain that Comair flight 3272 flew through icing conditions.
What they don't know is if those conditions were severe enough to bring down the plane.
We wanted to understand what role icing played in the slow and steady left roll that the aeroplane seemed to be experiencing.
In search of answers, they visit a NASA test facility.
Here, they can see with their own eyes how and when ice would have formed on the wings of flight 3272.
We ready to go? We were helping the NTSB identify potential ice shapes and what those ice shapes could do to the flight characteristics.
Tom Ratvasky is a NASA researcher.
A section of a wing is installed in a wind tunnel.
We were able to get our hands on an actual EMB-120 wing section.
Let's initiate.
Temperature and wind speed set? The temperature is cooled down to match the flight condition.
The fan is turned on to get the speed to match the flight condition.
OK.
Let's add the precipitation.
There's also a spray bar system, which introduces the liquid water into that air stream.
The test simulates the exact conditions faced by flight 3272.
Slowly, almost imperceptibly, ice begins to accumulate on the wing.
Even in a lighted wind tunnel, it was very difficult to see.
The type of ice that was accreted was relatively clear.
It was very rough.
Ice reshapes the wing surface so these aerodynamic surfaces aren't doing what they're supposed to do.
Can we get some measurements, please? The test leaves little doubt.
A thin but dangerous layer of ice almost certainly formed on the wings of flight 3272.
NASA gave us ironclad knowledge that the accumulation of thin amounts of ice was very dangerous and troublesome to the pilot and to the aircraft.
OK, thanks for your help with this.
Thanks.
But there's still something about the Comair crash that the NASA test doesn't explain.
The Embraer 120, like most commercial planes, has anti-icing and de-icing systems for winter weather flying.
Investigators need to know if the pilots activated those systems.
You guys ready? Let's play it.
They turn to the cockpit voice recorder.
PILOT: Let's run the descent check.
Ice protection.
Wind shield, props, standard seven.
The pilots are required to go through an approach checklist.
One of the items on that approach checklist is the standard seven, which is the various anti-icing facilities that they have on the aircraft.
The anti-icing system electrically heats the windshield and the propellers.
But for the wings, there's a different system, one that relies on mechanical devices called de-icing boots.
The boots are made of rubber and are part of the leading edge of the wings.
When activated, they inflate and expand to crack off any accumulated ice.
The ice is just hard enough that it cracks and the airstream blows it away.
PILOT: Pressurisation? CO-PILOT: It's reset for our landing in Detroit.
Rodriguez listens, waiting for the pilots to activate the de-icing boots.
Landing lights.
Landing lights on.
Cross-feed? Cross-feed's off.
That completes that.
They cover everything but activating the boots.
What gives? There was no discussion to indicate that they were concerned about icing on the aircraft.
Do you have the pre-flight documents? Investigators want to know why the pilots didn't activate an on-board system that could have saved their plane.
They check dispatch records.
Maybe the crew was never briefed on the weather conditions they'd face.
The flight had received their papers for dispatch from Cincinnati, which included warnings about icing in the Detroit area.
There could be some ice in our path.
What a shock.
Ice in January.
They knew they were gonna hit ice but they never popped the boots.
The de-ice boots did not appear to have been activated.
So we really wanted to take a good, hard look at when that system should be used and why it wasn't used on this approach.
NTSB investigators know flight 3272 flew through dangerous icing conditions.
They also know that the pilots never activated a critical de-icing system, but they don't know why.
They should have been aware there was icing taking place.
They certainly had to suspect it.
The pilots' work records reveal that Captain Carlsen was a seven-year veteran with Comair.
He even helped write pilot training procedures for the airline.
The captain was known as a by-the-book type of pilot.
First Officer Kenneth Reece was an equally proficient pilot, who was also a certified flight instructor.
These pilots were well-trained and had plenty of experience flying in the Midwest corridor where there's a lot of icing conditions.
So, if ice is just beginning to form on your wing, what would you do? Investigators contact other Comair pilots for insight into how they would handle a similar scenario.
What would you do? You'd wait? You'd wait? You'd wait.
How much ice? We were getting feedback that they'd wait until 0.
25 to 0.
5-inch of ice had accumulated before they'd activate the de-ice boots.
That was a surprise for us to see.
The revelation suggests that at Comair, outdated ideas about icing are putting pilots and passengers at risk.
Why would you wait? Bridging.
When de-ice boots were first invented, they inflated at a very slow pressure and stayed inflated for a while and pilots became concerned that ice might form over the inflated shape of the boot, a phenomenon known as 'bridging'.
Ice bridging was a concern with older de-icing systems that inflated and deflated slowly.
There was a risk that if pilots activated the boots too soon, ice might build up and form a shell, or bridge, beyond the limit of the inflated boot.
That could render the de-icing system useless.
Investigators know that modern de-icing boots have evolved.
Bridging is no longer a problem for turboprop aircraft.
The de-ice boots on this aeroplane inflate rapidly in less than a second, would break off any amount of ice that was accumulated and deflate again right away with vacuum suction.
So de-ice boot bridging was not gonna happen on this aeroplane.
Alright.
Let's take a look at the manuals these guys were following.
We were interested in what guidance had the crew received, what instructions do they have, what are they trained to do when they're experiencing icing? Both Comair and the plane's manufacturer, Embraer, published flight manuals for the 120 turboprop.
When investigators compare the two, - they make a surprising discovery.
- What? What jumped out at us was that Comair was instructing their pilots to wait until a significant amount of ice had accumulated before activating the de-ice boots.
That was surprising to us.
The Comair manual tells them to wait.
The Comair manual advises pilots to wait until half-an-inch of ice forms before de-icing, while the Embraer manual says something completely different.
Embraer says, "Activate boots at the first sign of icing.
" The material with respect to the icing conditions were not clear and they were not consistent for the pilots to know what they should be doing.
Let's see what we can find.
The Comair accident is the first fatal crash of an Embraer 120 caused by icing.
But how many close calls have there been? Six in less than eight years.
- Man, oh, man.
- There were six previous loss of control accidents involving icing with the Embraer, which is pretty significant.
So what did the FAA do about this? It's a disturbing record of accidents.
Investigators wonder what steps the FAA has taken to fix the problem.
The Federal Aviation Administration is responsible for setting safety standards throughout the industry.
Rodriguez finds disturbing evidence in FAA files.
Icing.
Icing.
Icing.
Well, you sure created a lot of reports.
Comair pilots may not be the only ones who held out-of-date views about de-icing.
The federal regulators themselves may be part of the problem.
I think we began to realise that we were working not against knowledge that wasn't out there.
We were working against a long-thought trend in aviation that de-ice boot bridging is a concern.
It was De-ice boot bridging was a myth.
In aviation, myths die hard.
It was time to break that myth.
Investigators are trying to understand why Comair and Embraer provided conflicting instructions to pilots regarding icing.
Digging through FAA files, they find some intriguing documents.
"Develop means to prevent ice formation.
" An FAA memo makes it clear that the regulator was concerned about icing.
There's only one way to do that - activate the boots.
And a revision to the Embraer 120 flight manual shows that the manufacturer had clearly spelled out the need to de-ice quickly.
"Activation of de-icing boots at the first sign of ice formation.
" Embraer knew the answer.
So why were the Comair pilots still in the dark? - There could be some ice in our path.
- Oh, what a shock.
Ice in January.
Investigators learn that the FAA approved the Embraer revision.
But then a critical step was missed.
The regulator did not require airlines to adopt the change.
Many, including Comair, did not.
Why not make it mandatory? It's a no-brainer.
While the FAA had a great deal of information, they never did officially sanction a standard that should be adhered to.
Investigators finally understand what caused the crash of Comair flight 3272.
Just before landing Ice protection.
.
.
the pilots enter icing conditions.
But the captain, following airline procedures, does not activate the de-icing boots.
It's reset for landing in Detroit.
It looks good.
- Landing lights.
- Landing lights.
The crew seemed to be aware that they were in light icing conditions, but per the instructions in the flight manual, they were waiting until a significant amount of ice had accumulated.
CONTROLLER: Comair 3272, reduce speed, 150.
The controller then instructs them to reduce speed.
That's getting pretty slow and it was a concern to us, why would they have accepted that without question? But in their manuals, there was no specific information that would tell them, "Do not go below a certain airspeed in the icing condition.
" Speed 150, Comair 3272.
If you have icing on there and you're slowing down, who knows where that stall point is? Comair 3272, now turn left heading 090.
Heading zero-niner-zero.
To make matters worse, the plane is on autopilot.
Since the autopilot was engaged, they were unaware that this trace amount of ice was starting to affect their aerodynamics.
When the autopilot is on, the pilot can't feel what's happening until he starts to see something on the instrument gauges.
It looks like your low speed indicator.
The captain was the first to notice the anomaly.
- Yeah, I see it.
Power.
- Thanks.
But it's too late.
The left wing has already stalled.
The aircraft rolled rapidly to the left.
They went from about 45 degrees angle of bank to the left to 140 degrees, which is basically inverted.
At that point, the aeroplane just - it's gonna do whatever it's gonna do and God help them.
We're in a dive.
We're in a dive! When that nose pitched straight into the ground, their fate was sealed.
No! No! No! I'm sorry.
I wish it hadn't happened.
The NTSB lays blame for the accident on the Federal Aviation Administration.
The agency failed to establish adequate standards for flights in icing conditions.
Ultimately, we didn't believe that the FAA put forth the right leadership to help dispel the myth of de-ice boot bridging.
They recommend that all pilots be informed of the need to activate the boots the moment they encounter icing conditions.
In this case, we had very good pilots, a very good aircraft.
It's very difficult to reconcile that, when you're talking with next of kin.
Hundreds of lives are affected.
We now have information that could prevent that.
Captioned by Ai-Media ai-media.
tv
We're in a dive! - A commuter flight to Detroit - Come on! - Get out of it, for god's sake! - .
.
ends in a harrowing spiral.
No! Tremendous destruction of the aircraft and it was a very severe impact.
Whatever happened happened in a heartbeat.
Close analysis of flight data - Ice protection.
- Standard seven.
.
.
leads to a stunning realisation.
It's Roselawn all over again.
What doomed flight 3272 is a danger the industry has known about for years.
- Why would you wait? - In aviation, myths die hard.
To fulfil their mission, investigators confront some of aviation's biggest players.
But it was time to break that myth.
26 passengers are on their way to Detroit, Michigan, aboard Comair flight 3272.
It's a short afternoon flight, operated by one of America's largest regional airlines.
The workhorse of the Comair fleet is the Embraer 120, a Brazilian-made turboprop.
The aeroplane handled quite nicely just about every regime that you could think of that would be appropriate for the type of operation that we were doing.
Darinda Ogden has been a Comair flight attendant for five years.
Here's your drink, Ken.
Thank you very much for the ice.
First Officer Kenneth Reece and Captain Dann Carlsen have been trading piloting duties all day.
This is the day's third leg, and First Officer Reece is at the controls.
I flew with Kenneth two or three times.
Stewart Lauer is a former Comair pilot.
I found him to be a very capable pilot.
He knew what he was doing.
Out of Detroit, we're looking at 239 radio.
Flight 3272 left from an airport near Cincinnati, Ohio, at 3:08pm.
It's expected to land in Detroit just after 4:00 in the afternoon.
What we want to do is just fly from point A to point B, get the passengers there safely and not scare them to death.
Heading 030 to join the arrival, Comair 3272.
The crew has the plane on autopilot as they descend from 19,000ft.
Closing in on their destination Just some bumps.
.
.
they hit some unexpected turbulence.
Despite the rough ride, the flight is still running on schedule.
At Detroit approach control, Steven Cochran guides flight 3272 on its path to the runway.
Comair 3272, Detroit approach, heading 050, vector to runway 3-right.
It's another busy Thursday afternoon for the veteran controller.
Detroit, Cactus 50 level at 13,000.
The crew of an Airbus radios in.
Cactus 50, Detroit roger.
Expect vectors for 3-right.
The controller wants the larger Airbus to land before Comair 3272.
The jet was much faster than the Comair flight, so the air traffic controller made the decision to give the jet priority over the turboprop and gave them a slower speed.
Comair, maintain one-niner-zero knots.
If unable, advise.
Roger.
One-niner-zero knots, Comair 3272.
First Officer Reece reduces his speed to 190 knots to let the incoming Airbus land first.
The fact that 3272 was placed behind Cactus 50 added just a little bit to the manoeuvring that was required.
Ladies and gentlemen, we're on our final approach to Detroit.
Please fasten your seatbelts and stow away your tray tables.
The turboprop should be on the runway in less than 15 minutes.
Let's run the descent check.
The approach phase is the busiest time, from a workload standpoint, for the pilot, and that's the time that he would like to have the assistance of the autopilot.
Ice protection.
- Windshield, props, standard seven.
- Pressurisation? As they descend towards 8,000ft, the pilots perform a series of standard checks.
Auto.
As you come in to land, we go into what's called a sterile cockpit.
- On.
- There's no chitchat.
It's just strictly business.
Comair 3272, reduce speed 150.
The controller now wants the turboprop to reduce its airspeed to 150 knots.
Comair 3272, reduce speed 150.
Uh, speed 150, Comair 3272.
They always gotta tell us everything twice.
He's got a short-term memory disorder.
Is that what that is? Yeah, that's what it is.
Comair 3272, now turn left heading 090.
Heading zero-niner-zero, Comair 3272.
The aeroplane was responding properly to all the air traffic control inputs and not indicating any problem.
The plane banks left as the pilots make their final turn.
First Officer Reece expects the autopilot to roll the plane back to level.
- Check your low speed indicator.
- Yeah, I see it.
Power.
The captain advances the throttles to gain speed.
But it doesn't work.
Instead, the plane's autopilot disconnects.
When the autopilot disconnected, a very sudden left roll began.
Reece struggles to level the plane by hand.
The amount of effort it takes to move those control surfaces, you've gotta muscle it, until you can get it where you want it to go.
He wrenches the control column to the right, trying to correct the steep left bank.
Nothing really happened.
The aircraft continued rolling over to the left.
We're in a dive.
We're in a dive! - No! - Oh, god! Stop it.
You never give up.
You do what you gotta do to try to make it.
Come on! Get out of there, for god's sake! - No! Ahh! No! - God.
But there seems to be nothing the pilots can do to save their spiralling aircraft.
There was virtually no controllability.
Nothing can be done.
There's just insufficient altitude to recover.
No! It was a very severe impact, a tremendous destruction of the aircraft.
Comair flight 3272, en route from Cincinnati to Detroit, has crashed into a field 22 miles southwest of the runway.
First responders and news crews arrive at the scene.
They soon discover that all 26 passengers and three crew members are dead.
We'd like to offer our condolences to the families.
Our hearts go out to those that have lost their loved ones today.
The sudden loss of 29 lives leaves families and friends in anguish.
How could a commercial flight on approach to a major US airport go so horribly wrong? The morning after the crash, a blackened scar on a frozen field marks the fatal impact zone of flight 3272.
Families and friends attend a memorial service for the crash victims.
A team from the National Transportation Safety Board is already at work recovering pieces of wreckage for analysis.
God, it's cold out there.
NTSB systems specialist John DeLisi, faces one of the biggest challenges of his career.
We knew the aeroplane was on approach into Detroit and something dramatic happened.
Something went wrong suddenly.
What a mess.
The NTSB's Richard Rodriguez leads the investigation.
Our mission is to find the cause of an accident, make recommendations that will prevent it from ever happening again.
Their first task is to find the aeroplane's black boxes and get them to a lab for analysis.
We spent about eight hours that first day pulling the wreckage apart, but by late afternoon, we were able to find the cockpit voice recorder and flight data recorder and got those on their way back to Washington, DC.
Let's get those both packed up.
While they wait for news on the black boxes, investigators head to Detroit Metropolitan Airport.
They want to hear from the last person to speak with the pilots, the approach controller.
It was coming on rush hour.
We had winter weather.
From air traffic control, we learned that the weather was bad and aeroplanes were starting to pile up on their approach into Detroit.
That's when American West Airlines flight 50 made contact, right after 3272.
Detroit, Cactus 50 level at 13,000.
So I told Comair pilots to slow up.
Comair, maintain one-niner-zero knots.
If unable, advise.
I made sure there was plenty of distance between them.
There were other aeroplanes on approach to Detroit, so sometimes, when they're in close proximity, there can be an issue with wake vortex coming off one aeroplane that may affect another.
A wake vortex is a horizontal tornado that trails behind an aircraft.
If one plane flies too close to the wake of another, it can encounter sudden and extreme turbulence.
So, here's the Airbus, American West Airlines flight 50.
The Airbus is almost twice the size of the Embraer 120.
And here's our Comair flight 3272.
Did the big jet's wake vortex knock flight 3272 out of the sky? We needed to see if the accident aeroplane flew underneath that one and crossing its wake.
The Embraer is miles back and 1,500ft below the Airbus.
It could have got caught in the wake.
Get the data off to NASA.
Let's see what they think.
A wake vortex incident seems possible.
Oh, god.
But investigators won't know for certain until experts at NASA analyse the radar data.
Meanwhile, Rodriguez explores other possibilities.
What happened to the propellers? He's very familiar with this type of plane, having investigated previous accidents involving the Embraer 120.
The Embraer 120 propeller blade had separated due to a fatigue crack over in an accident earlier in Georgia.
The crew was able to land it, but it was very difficult to control.
In two previous cases, part of a propeller blade broke off in mid-flight.
Could it have happened again? If they were barely able to control the aircraft, this could have been a problem for 3272.
Rodriguez reviews details of where the two propellers were found at the Detroit crash site.
He makes a worrying discovery.
Part of one blade ended up 75ft away from the main impact crater.
He can't help but wonder, "Is this the third time he's seen an Embraer 120 propeller fail?" We need to take a look at those propeller blades.
We wanted to get those blades of the propeller to ensure they were intact at impact.
Rodriguez studies the blade fragment, looking for evidence that they failed.
If the propeller blades were all attached and spinning at impact, they would have all hit the ground with tremendous force, resulting in distinctive damage on every blade.
This is impact damage.
They were spinning right till the end.
The analysis leaves no doubt.
Though the propellers fragmented when they hit the ground, they did not fail in flight.
We were able to tell that nothing broke off the aeroplane during the flight path.
Just as investigators rule out one possible cause, they uncover another intriguing clue.
Richard.
What'd you find? It's the engine fire lever.
It looks like it's been pulled.
- Well, what do you know? - This is a lever that a crew member would grab and pull and twist if one of the engines was on fire.
It's a discovery that has frightening implications.
So, are we looking at an engine fire here? Yeah.
Put that one here.
NTSB investigators explore a disturbing theory as they try to figure out why Comair flight 3272 plunged from the sky.
It looks like one of the pilots pulled an engine fire shut-off lever.
The crew may have been fighting an in-flight fire.
No doubt there was a fire.
But when? While the engines are clearly fire-damaged, investigators need to determine if the fire started before or after the plane hit the ground.
In-flight fire would probably be at a higher temperature than ground fire.
So you would have some, perhaps, disintegration or disappearance of the metal in the engine cowling and so forth.
If the fire started while the plane was moving forward, Rodriguez should find soot marks in a horizontal pattern.
If it started on the ground, the flames would have gone straight up.
In this case, it was a relatively low temperature.
All the soot patterns were vertical.
There was no airstream affecting it to show in-flight fire.
I'm seeing post-impact fire.
There was no fire in flight.
No way.
But upon further examination, we realised that the extended position of that handle was just the result of the hard crush damage that occurred.
With engine fire ruled out, investigators turn to NASA analysis of airport radar data.
Perhaps there's evidence that the small turboprop flew too close to a big jet.
And that settles that.
- But it's another dead end.
- It wasn't a wake vortex.
The analysis reveals that the wake from American Airlines flight 50 could not have dropped to the altitude of Comair 3272.
Investigators are back where they started.
Why 29 people died just minutes before landing, is still a mystery.
OK.
Are we all ready? They finally have access to the information stored on the plane's flight data recorder.
Left engine, please.
It should give them important details on how the aircraft - was performing - Now the right engine.
.
.
right up to the moment of impact.
Here, it's starting.
The engines are guzzling fuel.
The speed stays the same.
The data reveals a puzzling detail.
Just before the plane's sudden upset, the engines were working hard, but to little effect.
The flight data recorder indicated that the aircraft had received significant degradation in the performance.
Look what's happening here.
Autopilot is on.
Turn is done.
It should level off here, but it doesn't! It keeps rolling.
What really jumped out at us was the aeroplane continued to roll even though the autopilot was commanding it to go back to 'wings level'.
Something's holding the engines back.
As Rodriguez studies the data, he thinks he knows what may be causing the loss of performance.
4,000ft and it falls from the sky.
It has all the hallmarks of something he's seen before.
It's Roselawn all over again.
In October 1994, the pilots of American Eagle flight 4184 suddenly lost control of their twin turboprop.
It plummeted from the sky and crashed near the town of Roselawn, Indiana.
All 68 people aboard were killed.
The cause was a phenomenon known as 'icing', where a build-up of ice on the wings leads to an aerodynamic stall.
When that happens, you're not producing the lift that you're supposed to.
The loss of lift is what the word 'stall' is pertaining to.
When we stall a wing, we lose lift.
Following the Roselawn accident, the FAA was very concerned about the performance of turboprop aircraft in icing conditions.
The Federal Aviation Administration was so concerned that it conducted full-scale tests, flying a turboprop in super-cooled precipitation sprayed from a water tanker.
They found that the aircraft did accumulate ice on the top of the wing to generate significant drag.
The flight test's most dramatic finding was that a layer of ice as thin as a sheet of sandpaper could cause major problems for pilots.
And we were very surprised to see that a trace amount of ice was enough to affect the aerodynamics and possibly cause a wing to lose lift.
In March 2001, a turboprop suffered severe icing and plummeted 8,000ft before the captain managed to regain control.
That captain was Stewart Lauer.
I had the shakes something fierce.
I mean, the adrenalin was flowing.
Just like Stewart Lauer's flight - PILOT: Check the low speed indicator.
- Yeah, I see it.
.
.
Comair 3272 dropped from the sky in an instant.
It's very complex aerodynamically, especially when there's ice contamination on the aeroplane.
At that point, it's very hard to recover unless you gain airspeed, and the only way to do that is point the nose to the ground, not something that a pilot wants to do while trying to recover an aeroplane.
Come on! Get out of it, for god's sake! We're in a dive.
We're in a dive! - No! - God.
I was lucky.
I had plenty of altitude.
It took me over 7,000ft to regain control of the aircraft.
They didn't have it.
They only had 4,000ft.
OK.
Let's see what they flew through.
Investigators need to learn all they can about the weather conditions the Comair pilots faced as they approached Detroit.
OK.
Let's pull up the weather data.
They know that for ice to build up on the wings of a plane Now let's see the flight path.
.
.
temperature, precipitation and aircraft speed must all fall within a very specific range.
The weather data began to paint a picture of an aeroplane that was on approach, flying in what were considered light icing conditions.
Cactus 50, be advised there are slick runways and poor visibility.
How is it up there? According to Detroit controllers, other aircraft, including the flight directly in front of Comair 3272, reported icing conditions.
Yeah, it's 237 at 32 up here, moderate icing with a possibility of freezing drizzle.
Roger that.
Comair 3272, how is it up there? Strangely, the Comair pilots never mentioned icing at all.
Comair 3272, uh, just a little turbulence but that's all.
It definitely flew through icing conditions.
Investigators know that even a thin layer of ice can be dangerous.
Perhaps the Comair pilots didn't see ice on their wings.
The Embraer is probably a good 20ft from the cockpit window back to the top of the wing where the ice would be forming, so it's very difficult to see in poor lighting conditions as 3272 was operating, in cloud and late in the afternoon.
But there's something that doesn't add up.
If the Comair pilots lost control due to ice build-up, why were they the only ones affected by the weather? None of the other aeroplanes that were on approach to Detroit experienced any problems.
So we needed to look hard about what was unique about this aeroplane.
NTSB investigators are now certain that Comair flight 3272 flew through icing conditions.
What they don't know is if those conditions were severe enough to bring down the plane.
We wanted to understand what role icing played in the slow and steady left roll that the aeroplane seemed to be experiencing.
In search of answers, they visit a NASA test facility.
Here, they can see with their own eyes how and when ice would have formed on the wings of flight 3272.
We ready to go? We were helping the NTSB identify potential ice shapes and what those ice shapes could do to the flight characteristics.
Tom Ratvasky is a NASA researcher.
A section of a wing is installed in a wind tunnel.
We were able to get our hands on an actual EMB-120 wing section.
Let's initiate.
Temperature and wind speed set? The temperature is cooled down to match the flight condition.
The fan is turned on to get the speed to match the flight condition.
OK.
Let's add the precipitation.
There's also a spray bar system, which introduces the liquid water into that air stream.
The test simulates the exact conditions faced by flight 3272.
Slowly, almost imperceptibly, ice begins to accumulate on the wing.
Even in a lighted wind tunnel, it was very difficult to see.
The type of ice that was accreted was relatively clear.
It was very rough.
Ice reshapes the wing surface so these aerodynamic surfaces aren't doing what they're supposed to do.
Can we get some measurements, please? The test leaves little doubt.
A thin but dangerous layer of ice almost certainly formed on the wings of flight 3272.
NASA gave us ironclad knowledge that the accumulation of thin amounts of ice was very dangerous and troublesome to the pilot and to the aircraft.
OK, thanks for your help with this.
Thanks.
But there's still something about the Comair crash that the NASA test doesn't explain.
The Embraer 120, like most commercial planes, has anti-icing and de-icing systems for winter weather flying.
Investigators need to know if the pilots activated those systems.
You guys ready? Let's play it.
They turn to the cockpit voice recorder.
PILOT: Let's run the descent check.
Ice protection.
Wind shield, props, standard seven.
The pilots are required to go through an approach checklist.
One of the items on that approach checklist is the standard seven, which is the various anti-icing facilities that they have on the aircraft.
The anti-icing system electrically heats the windshield and the propellers.
But for the wings, there's a different system, one that relies on mechanical devices called de-icing boots.
The boots are made of rubber and are part of the leading edge of the wings.
When activated, they inflate and expand to crack off any accumulated ice.
The ice is just hard enough that it cracks and the airstream blows it away.
PILOT: Pressurisation? CO-PILOT: It's reset for our landing in Detroit.
Rodriguez listens, waiting for the pilots to activate the de-icing boots.
Landing lights.
Landing lights on.
Cross-feed? Cross-feed's off.
That completes that.
They cover everything but activating the boots.
What gives? There was no discussion to indicate that they were concerned about icing on the aircraft.
Do you have the pre-flight documents? Investigators want to know why the pilots didn't activate an on-board system that could have saved their plane.
They check dispatch records.
Maybe the crew was never briefed on the weather conditions they'd face.
The flight had received their papers for dispatch from Cincinnati, which included warnings about icing in the Detroit area.
There could be some ice in our path.
What a shock.
Ice in January.
They knew they were gonna hit ice but they never popped the boots.
The de-ice boots did not appear to have been activated.
So we really wanted to take a good, hard look at when that system should be used and why it wasn't used on this approach.
NTSB investigators know flight 3272 flew through dangerous icing conditions.
They also know that the pilots never activated a critical de-icing system, but they don't know why.
They should have been aware there was icing taking place.
They certainly had to suspect it.
The pilots' work records reveal that Captain Carlsen was a seven-year veteran with Comair.
He even helped write pilot training procedures for the airline.
The captain was known as a by-the-book type of pilot.
First Officer Kenneth Reece was an equally proficient pilot, who was also a certified flight instructor.
These pilots were well-trained and had plenty of experience flying in the Midwest corridor where there's a lot of icing conditions.
So, if ice is just beginning to form on your wing, what would you do? Investigators contact other Comair pilots for insight into how they would handle a similar scenario.
What would you do? You'd wait? You'd wait? You'd wait.
How much ice? We were getting feedback that they'd wait until 0.
25 to 0.
5-inch of ice had accumulated before they'd activate the de-ice boots.
That was a surprise for us to see.
The revelation suggests that at Comair, outdated ideas about icing are putting pilots and passengers at risk.
Why would you wait? Bridging.
When de-ice boots were first invented, they inflated at a very slow pressure and stayed inflated for a while and pilots became concerned that ice might form over the inflated shape of the boot, a phenomenon known as 'bridging'.
Ice bridging was a concern with older de-icing systems that inflated and deflated slowly.
There was a risk that if pilots activated the boots too soon, ice might build up and form a shell, or bridge, beyond the limit of the inflated boot.
That could render the de-icing system useless.
Investigators know that modern de-icing boots have evolved.
Bridging is no longer a problem for turboprop aircraft.
The de-ice boots on this aeroplane inflate rapidly in less than a second, would break off any amount of ice that was accumulated and deflate again right away with vacuum suction.
So de-ice boot bridging was not gonna happen on this aeroplane.
Alright.
Let's take a look at the manuals these guys were following.
We were interested in what guidance had the crew received, what instructions do they have, what are they trained to do when they're experiencing icing? Both Comair and the plane's manufacturer, Embraer, published flight manuals for the 120 turboprop.
When investigators compare the two, - they make a surprising discovery.
- What? What jumped out at us was that Comair was instructing their pilots to wait until a significant amount of ice had accumulated before activating the de-ice boots.
That was surprising to us.
The Comair manual tells them to wait.
The Comair manual advises pilots to wait until half-an-inch of ice forms before de-icing, while the Embraer manual says something completely different.
Embraer says, "Activate boots at the first sign of icing.
" The material with respect to the icing conditions were not clear and they were not consistent for the pilots to know what they should be doing.
Let's see what we can find.
The Comair accident is the first fatal crash of an Embraer 120 caused by icing.
But how many close calls have there been? Six in less than eight years.
- Man, oh, man.
- There were six previous loss of control accidents involving icing with the Embraer, which is pretty significant.
So what did the FAA do about this? It's a disturbing record of accidents.
Investigators wonder what steps the FAA has taken to fix the problem.
The Federal Aviation Administration is responsible for setting safety standards throughout the industry.
Rodriguez finds disturbing evidence in FAA files.
Icing.
Icing.
Icing.
Well, you sure created a lot of reports.
Comair pilots may not be the only ones who held out-of-date views about de-icing.
The federal regulators themselves may be part of the problem.
I think we began to realise that we were working not against knowledge that wasn't out there.
We were working against a long-thought trend in aviation that de-ice boot bridging is a concern.
It was De-ice boot bridging was a myth.
In aviation, myths die hard.
It was time to break that myth.
Investigators are trying to understand why Comair and Embraer provided conflicting instructions to pilots regarding icing.
Digging through FAA files, they find some intriguing documents.
"Develop means to prevent ice formation.
" An FAA memo makes it clear that the regulator was concerned about icing.
There's only one way to do that - activate the boots.
And a revision to the Embraer 120 flight manual shows that the manufacturer had clearly spelled out the need to de-ice quickly.
"Activation of de-icing boots at the first sign of ice formation.
" Embraer knew the answer.
So why were the Comair pilots still in the dark? - There could be some ice in our path.
- Oh, what a shock.
Ice in January.
Investigators learn that the FAA approved the Embraer revision.
But then a critical step was missed.
The regulator did not require airlines to adopt the change.
Many, including Comair, did not.
Why not make it mandatory? It's a no-brainer.
While the FAA had a great deal of information, they never did officially sanction a standard that should be adhered to.
Investigators finally understand what caused the crash of Comair flight 3272.
Just before landing Ice protection.
.
.
the pilots enter icing conditions.
But the captain, following airline procedures, does not activate the de-icing boots.
It's reset for landing in Detroit.
It looks good.
- Landing lights.
- Landing lights.
The crew seemed to be aware that they were in light icing conditions, but per the instructions in the flight manual, they were waiting until a significant amount of ice had accumulated.
CONTROLLER: Comair 3272, reduce speed, 150.
The controller then instructs them to reduce speed.
That's getting pretty slow and it was a concern to us, why would they have accepted that without question? But in their manuals, there was no specific information that would tell them, "Do not go below a certain airspeed in the icing condition.
" Speed 150, Comair 3272.
If you have icing on there and you're slowing down, who knows where that stall point is? Comair 3272, now turn left heading 090.
Heading zero-niner-zero.
To make matters worse, the plane is on autopilot.
Since the autopilot was engaged, they were unaware that this trace amount of ice was starting to affect their aerodynamics.
When the autopilot is on, the pilot can't feel what's happening until he starts to see something on the instrument gauges.
It looks like your low speed indicator.
The captain was the first to notice the anomaly.
- Yeah, I see it.
Power.
- Thanks.
But it's too late.
The left wing has already stalled.
The aircraft rolled rapidly to the left.
They went from about 45 degrees angle of bank to the left to 140 degrees, which is basically inverted.
At that point, the aeroplane just - it's gonna do whatever it's gonna do and God help them.
We're in a dive.
We're in a dive! When that nose pitched straight into the ground, their fate was sealed.
No! No! No! I'm sorry.
I wish it hadn't happened.
The NTSB lays blame for the accident on the Federal Aviation Administration.
The agency failed to establish adequate standards for flights in icing conditions.
Ultimately, we didn't believe that the FAA put forth the right leadership to help dispel the myth of de-ice boot bridging.
They recommend that all pilots be informed of the need to activate the boots the moment they encounter icing conditions.
In this case, we had very good pilots, a very good aircraft.
It's very difficult to reconcile that, when you're talking with next of kin.
Hundreds of lives are affected.
We now have information that could prevent that.
Captioned by Ai-Media ai-media.
tv