Original subtitles

In the ocean near Los Angeles...

on a lonely mountain in Japan...

off the East coast of Canada...

a plane crash can reduce

an enormous jet plane to mangled pieces.

The cause is buried somewhere in the wreckage.

In the business, we refer to,

often, "finding the golden nugget."

A modern jet plane is made up of hundreds

of thousands of parts.

A failure in any one of them can lead to disaster.

A missing screw can jeopardize the safety of flight.

Constant checks keep planes flying,

and passengers safe.

A single oversight can end in tragedy.

Mayday!

Something exploded.

Oh, my God!

Oh, my God!

Help me, help me hold it. Help me hold it!

Mayday, mayday.

A hot summer night in Phoenix, Arizona.

It's eleven o'clock,

but the maintenance workers at Southwest Airlines

are just getting started.

Tonight, they're going to open up

a state-of-the-art Boeing 737-700.

Almost 40 inspectors and mechanics are going

to spend the night making sure the plane is fit to fly.

Without proper maintenance, airplanes don't fly.

Pilots are usually the focus

for the operation of the airplane.

But maintenance has an equally high priority role

in the safe operation of any aircraft.

To keep airplanes in peak condition,

they get more health checks than most passengers.

It's a very intricately weaved web

between the operation of the airplane

and the maintenance of the airplane

and the management of the airplane.

Passenger planes are examined

every time they come to a stop.

This is the A-check.

A brief walk-around inspection turns up

the most obvious problems.

The more intensive work is done at set intervals.

These are the B and C checks.

Tonight, workers are performing a C-check.

From start to finish, it can require

hundreds of man-hours of work.

It all has to get finished tonight,

so the plane is back flying in the morning.

It's a massive challenge

because modern jets are made

of hundreds of thousands of individual pieces.

In 1903,

when the Wright Brothers took their historic first flight

near Kitty Hawk, North Carolina,

their plane had some 1,500 parts.

A 737 has more than 360,000.

You have to ensure that every one of those components

is doing its respective job.

It doesn't matter how big the part is.

A missing screw can jeopardize the safety of flight.

It's a lesson the aviation industry has

learned the hard way.

January 31st, 2000.

On board Alaska Airlines flight 261,

the situation is desperate.

Operating a damaged plane,

the captain is trying to land at Los Angeles airport.

But the aircraft is not responding to controls.

The MD-83 is plunging towards the Pacific Ocean.

Holy!

Other pilots flying nearby report

the nightmare scene back to L.A. Air Traffic Control.

Push the blue sign up!

Here we go!

Flight 261 crashed off the coast of California

at over 400 kilometers an hour.

All 88 passengers and crew are killed.

Investigators from the National Transportation Safety Board

begin their work quickly.

The cockpit voice recordings provide some of the earliest clues.

We have a jammed stabilizer

and we're maintaining altitude with difficulty.

We immediately suspected

some problem in the tail of the airplane,

which is where the controls are.

Something was wrong back there.

Investigators examine

the MD-83's horizontal stabilizer.

The stabilizer controls the plane's pitch,

its ability to tilt up and down.

As the stabilizer moves up,

the plane's nose tilts down.

As the stabilizer moves down,

the nose moves up.

In the MD-83,

a motorized jackscrew on the tail moves

the stabilizer up and down.

When investigators recover the tail from the crash site,

they make a puzzling discovery.

The jackscrew wasn't mated

with the nut that it screws into.

It was just by itself.

And the nut was found in another piece of structure

a few feet away from where the jackscrew was.

To have a screw separate itself from a nut

with very thick threads surprised us.

Without the jackscrew,

the stabilizer was beyond control.

Without the stabilizer, the plane was doomed.

The investigators very quickly figured out

how the accident happened.

Now they want to know why.

The answer is tragically simple.

There was no lubrication or visible grease

on the working area of the screw.

That was surprising and strange.

The Federal Aviation Administration orders

an immediate check on all MD-80s in the USA.

At Alaska Airlines,

the jackscrews on six of its fleet

of 34 MD-80s fail inspection.

Investigators discover even more alarming evidence

as they go through the carrier's maintenance records.

Mechanics at Alaska Airlines report that they are

under tremendous pressure

to cut corners to keep the planes flying.

We interviewed all the mechanics

who had worked on these airplanes.

We knew that they had been falsifying records

or not doing the work they had indicated.

To survive an economic recession

in the 1990s,

Alaska Airlines slashed their maintenance regime.

With air carriers,

especially those that may be economically strapped,

they're going to stretch inspection cycles

to the maximum.

The FARs, the Federal Aviation Regulations,

set a minimum level of safety.

Now if you're going to operate on a shoestring,

you're only going to meet that minimum level of safety.

If I'm a good carrier, or I want to be a good carrier

and I want to show that we're going to operate

at the highest levels of safety,

I'm going to typically exceed the minimums.

It's going to cost more, but I'm going to exceed it.

A lot of companies that say,

"Wait, the regulations only say

"I only have to go to here. That's what I'm going to do."

Jackscrews in the company's fleet had been

inspected every 500 to 700 flight hours.

But in 1996, to cut costs,

Alaska Airlines began checking the jackscrews

every 2,500 hours.

At the same time, they doubled

the average daily use of their fleet.

If you had 600 hours between inspection points

and greasing points,

we have no chance of ever having

a metal-to-metal contact situation.

But if you put that up to 2,000 hours or 2,500 hours,

now what you do is eat into

some of these protective stages,

these barriers that we have towards catastrophic failure.

Proper maintenance becomes even more critical

when there is no backup to a component.

On the MD-83, there was no alternative

if the jackscrew failed.

So proper maintenance was a matter of life and death.

But in the aviation industry,

it's also a matter of dollars and cents.

There's a lot of pressure in the airline industry,

when you look at it, whether you're hauling boxes

or hauling people.

The fact of the matter is, is that competition is stiff

and how do you get the competitive advantage

against the next guy?

How am I gonna get more for less?

And a lot of times, it's labor.

The other times, it's maintenance.

If I can stretch the inspection

to 500 hours instead of 400 hours,

that saves me a lot of money.

To stay afloat financially,

Alaska Airlines put countless lives at risk.

But disaster can erupt even when an airline doesn't

cut back on its maintenance regime.

We're going to hit the mountain!

Keep trying!

It's past midnight in Phoenix, Arizona.

A maintenance crew works through a 737-700.

They're performing a so-called C-check,

one of the most detailed inspections

any plane can go through.

We work overnight because that's when nobody flies.

It's better for the airline to keep the airplanes on ground overnight

to fix them up.

Tonight, 339 individual inspections are

set to be made.

Each one of these is tracked by computer.

Anything that comes up yellow is an unscheduled procedure,

a problem that's just been spotted.

Unscheduled maintenance are those kinds of things

typically that people experience with their car

where they're driving down the highway

and, all of a sudden, the air conditioner doesn't work.

Well, the same with an airplane.

Tonight, the inspectors discover

a tire on one of the main landing gears is worn out.

They add it the list of unscheduled maintenance items.

It has to be replaced

before the plane goes back into service.

Obviously, the stakes are extremely high.

Every night, we come to work

and try to do our best job possible,

make sure everything's in working order

so people get to where they need to go.

But sometimes, despite all the maintenance,

the worst-case scenario comes true.

A simple repair can unexpectedly lead to disaster.

August 13th, 1985. Mount Osutaka, Japan.

This is the wreckage from the deadliest single plane disaster

in aviation history.

JAL Flight 123 crashed the night before,

killing 520 passengers and crew.

Only four people survived.

Because the 747 jet was built in the United States,

the National Transportation Safety Board joins

the investigation.

When I arrived in Tokyo,

the atmosphere in Japan was extremely stressful.

The news media were everywhere.

There was a tremendous amount of anger.

Soon after the crash,

experts get a helping hand from an amateur photographer.

He managed to take a picture

of the 747 minutes before it crashed.

The picture reveals that JAL Flight 123 was

flying without its massive tail fin.

The tail fin houses critical control surfaces

like the rudder,

as well as tubes that carry the hydraulic fluids.

What force could be strong enough

to tear off the tail fin?

Digging through the 747's maintenance history,

investigators discover that, seven years earlier,

the jet had landed with its nose too high.

The tail hit the ground and scraped along the runway.

The rear part of the plane had to be repaired,

including the pressure bulkhead.

Japan Airlines called in Boeing technicians

to help repair the cracked bulkhead.

After this unscheduled maintenance,

the 747 was given a clean bill of health

and flew for another seven years.

But this bulkhead becomes

a prime suspect for the investigators.

We had an idea that we wanted to find

the rear pressure bulkhead,

because we had a flight attendant

who had been interviewed, that described an explosion

in the back of the airplane and she could see out.

So we wanted to focus on the bulkhead.

During his investigation,

Schleede finds a piece of the panel

that had been spliced into the bulkhead

seven years before.

The mystery of Flight 123 is solved.

The 747 went down because of a faulty repair.

The repair had, in fact, not been done correctly.

There was only one row of rivets

holding that joint together,

uh, where there should have been two rows

of rivets holding the joint together.

With only one row of rivets straining to hold

the repaired panel in place,

this was a disaster waiting to happen,

especially because this was such a busy jet.

This particular airplane was

used in Japan on a domestic operation.

So, it made multiple takeoffs and landings

on domestic operations,

unlike most 747s that make long-range hauls.

So, this was considered a high cycle airplane.

Investigators calculate

that, with the repair job, the bulkhead would survive

approximately 10,000 flights or cycles.

But on the day of the crash,

the 747 had already racked up over 12,000 cycles.

On 747 jets, the cabin is pressurized, but not the tail.

During flight, the pressurized cabin air presses against

the repaired bulkhead.

After some 12,000 cycles,

this pressure stretched the faulty repair

to breaking point.

The highly-pressurized air

blasted into the hollow tail fin

and blew it off.

Flap up! Flap up!

Losing part of the tail crippled

the plane's hydraulic systems.

The Boeing 747 had

four independent hydraulic systems

to power its systems.

So it had quadruple redundancy.

Unfortunately, these four lines came together

in the lower part of the spar,

and when it separated, it sheared those four lines.

All four hydraulic systems were depleted.

Both hands!

For some 30 minutes,

the crew tried to fly their 747 using only thrust.

This is like trying to drive

a car using only the accelerator,

no steering wheel, no brakes.

Despite their heroic efforts,

it was a losing battle.

All this death and destruction boils down

to a missing row of rivets.

Why had the growing metal fatigue

in the bulkhead remained undetected

through seven years of scheduled maintenance

and inspections?

The primary inspection method for the bulkhead area

and the seams was a visual inspection.

And at heavy maintenance periods when they...

they would take the insulation out,

uh, off the walls and everything and off the bulkhead,

they would do a detailed visual inspection.

And during subsequent maintenance checks,

the faulty repair was never found.

Two decades after JAL Flight 123,

airlines are constantly looking for hidden flaws

that aren't visible from the outside.

Back at the Southwest maintenance hangar,

inspectors are using a borescope,

a tiny flexible camera, to inspect the engines.

Engines are the heart of passenger planes.

If they stop working, pilots don't have the option

of pulling over to the side of the road.

Yeah, there we go.

In this area, we're looking for cracks.

Looking at the, uh, blades, the rotor blades,

and we're looking for missing material off of them.

You know, any hot spots

that have worn through the metal cracks,

radial and axial cracks.

Any kind of crack or trace of metal fatigue

in any of the fan blades could spell disaster.

- Take-off, check. Below the line. - I'll clear your lights.

August 21st, 1995.

Atlantic Southeast Airlines Flight 529,

an Embraer Brasilia, is about to take off

with 29 people on board.

It's bound for Gulfport, Mississippi.

It was, at the time,

the fastest, sleekest turbo-prop around.

Before the plane even reaches

its cruising altitude,

something seems to explode outside.

Autopilot, engine control...

The sound of that was tremendous.

It was as if someone had taken a baseball bat

and hit an aluminum garbage can as hard as they could.

It was just a... gigantic crashing sound.

The airplane immediately lurched to the left.

No matter what the flight crew tries to do,

the plane pulls violently to the left.

Autopilot. Engine control.

Help me hold it. Help me hold it!

Over there!

The captain and co-pilot are pushed

to the brink of their experience.

Help me. Help me!

Help me hold it. Help me hold it!

Atlantic Southeast Flight 529 crashes

near the small farming community

of Carrollton, Georgia.

Emergency. Yes,

we have a plane crashed in our backyard.

A plane crash?

All 29 people survive the violent landing.

But ten passengers eventually die from their injuries.

Called into action, the NTSB creates teams

to examine various parts of the plane.

Jim Hookey, an aerospace engineer,

is in charge of the Propeller Maintenance Group.

We came along a lot of pieces of the wing,

came along the, um...

the propeller assembly

that was missing one part of the blade.

The blade broke in a very specific fashion,

leaving behind all the telltale signs

of a fatigue fracture.

A fatigue fracture tends to be a very flat fracture.

It also has what we call beach marks

radiating out from the origin.

So you see these radiating concentric rings

coming from the origin of the crack.

Hookey had good reason

to focus on the broken propeller blade.

Seventeen months before ASA 529,

identical propeller blades broke

on separate flights over Canada and Brazil.

Fortunately, in both cases,

the aircraft managed to land safely.

The manufacturer of the propeller was Hamilton Standard.

Hookey and his team start combing

through Hamilton Standard's maintenance records.

They're looking for anything out of the ordinary.

It's whatever's abnormal.

You really don't know what you're looking for until you find it.

But you just go through,

and there's a lot of routine maintenances done,

regular inspections, A-checks, B-checks, C-checks.

And there's the non-routine maintenance that's occurred.

If something is broken, or a truck hits the airplane

or they have a bird strike or something like that.

And it's those that you look for.

The maintenance records reveal

that the broken propeller blade had earlier problems.

We found out that that propeller blade had

actually been removed from service once already,

uh, for a crack indication.

And that became the first clue about there may be a problem

with that propeller blade and those inspections.

Deep inside the hollow propeller,

investigators find what they're looking for.

In the hollow interior, or taper bore,

weights are inserted to balance the prop.

They're kept in place by a cork.

This simple cork was the trigger

in a deadly chain of events.

About 95% of the cork that's produced in the world

is used by the medical industry.

And, for aesthetic purposes and for sterilization,

they like to have the light color.

So, the cork is bleached with chlorine.

The NTSB discovers

that moisture inside the propeller caused

the chlorine in the cork to leach out

and corrode the propeller's aluminum alloy.

They also notice something else on the broken blade.

On the inner surface,

extending about four centimeters

from the fracture, there are a series of sanding marks.

Going through the blade's repair records,

Hookey notices the initials "CSB".

Christopher Scott Bender.

This technician worked

at a Hamilton Standard repair facility.

When Christopher Bender watches news of the accident,

he learns that the investigators

are examining the Hamilton propeller.

As soon as I heard that,

my heart just sank. I was like...

you know, I think I might have even cried a little bit

'cause I was just, you know, just emotionally overwhelmed

that, you know, something I had put my hands on,

a procedure somebody trusted me to do, failed.

Uh, and because of it, somebody had died.

After discovering that it was Bender

who last worked on the deadly propeller blade,

the NTSB now has to find out how the blade had

passed inspection at Hamilton Standard.

Investigators ask Bender to perform

his standard maintenance technique

on the propeller.

He demonstrated how he would go down into the barrel

of the taper bore

with a fiberoptic borescope and look for cracks.

And therein lied one of the primary problems.

The borescope that he was using

had a bright white light

that would put a lot of glare

back into the inspector's eyes.

It really did not lend itself

to the inspection that was required.

And investigators also find a gap

in Bender's training.

He had never been shown what a crack would look like.

He was just told to find a crack,

and he would look for a crack.

When he was examining the propeller blade,

Bender had been unable to detect

any evidence of corrosion.

He then did what he had been told to do,

polish the inside of the blade.

He was given a directive to use a repair

to blend out the inside of the taper bore.

He blended it out, he did an inspection.

And the blending that he had done

had roughened the surface,

so it actually masked the indication of the crack

in the subsequent inspection.

And the blade was returned to service,

where the crack continued to propagate

until it ultimately reached critical length and separated.

The draft accident report we present to you today

involves Atlantic Southeast Airlines Flight 529.

According to the NTSB,

by polishing the blade, Hamilton Standard had

unwittingly removed all traces of the crack.

Even a later, more thorough ultrasound examination

could not detect it.

The company that manufactured Flight 529's propeller is

now renamed Hamilton Sundstrand.

Its inspection and repair process was made

more stringent, in some cases exceeding FAA requirements.

Flight 529 was the last time

one of its propellers failed in flight.

You know, I wish this had never happened.

I wish I could go back in time and fix it

and take care of it and that it didn't happen.

Out of the thousands of parts on board

an Embraer Brasilia, a simple small cork was

the key to a horrific accident.

ASA Flight 529 underlines

the critical need for proper maintenance.

But sometimes maintenance can

create the potential for disaster

when a new component is

installed into an older airplane.

Moncton center, Swiss Air 111 heavy is declaring

Pan Pan Pan. We have smoke in the cockpit.

It's early morning in Phoenix, Arizona.

Southwest Airlines engineers are continuing

their scheduled maintenance of a 737.

Southwest is unique among larger airlines.

It flies just one kind of plane, the 737.

Tonight, engineers are working on a 700 model,

one of the newest 737s.

But the company's very first 300 model,

bought in the mid-1980s, is still flying.

You can still operate an old airplane,

as long as you have inspection protocols.

When you look at some of the cargo carriers,

they're operating airplanes that are 30 and 40

and even 50 years old.

They're still reliable airplanes.

They've been maintained,

they've been retrofitted with modern day equipment.

Updating older planes is

a standard part of maintenance.

But sometimes, installing a new component

in an older plane can lead to tragedy.

Inside this makeshift lab are the shattered remains

of Swiss Air Flight 111.

On September 2nd, 1998,

the passenger jet crashed off the coast

of Nova Scotia, Canada,

killing everyone on board.

Recovered from the seabed, the debris is overwhelming.

There is almost 250 kilometers of wiring alone.

In Swiss Air, we'd had about two million pieces of airplane.

And we pretty much almost had to look at them all.

In the business, we refer

to often "finding the golden nugget."

That's saying, "A-ha, there's the cause of the accident."

Somewhere in this wreckage,

investigators hope to find that golden nugget,

the one piece that will reveal the reason

why Swiss Air 111 crashed into the Atlantic ocean.

The cockpit voice recorder gives

investigators their first critical clues.

Do you smell something?

Yeah. What is that?

Go have a look. I'll take the controls.

Roger. You have control.

The First Officer checks

the area around the air conditioning vent.

Nothing seems wrong.

Don't see anything there.

And there's nothing up there now.

Captain Zimmerman is troubled

by the smell of smoke.

There it is again.

He starts to divert the plane

to the nearest airport.

Find the closest place to land, Stefan.

He radios air traffic control

in Moncton, New Brunswick.

Moncton Center, Swiss Air 111 heavy is

declaring Pan Pan Pan. We have smoke in the cockpit.

"Pan Pan Pan" is an international term used

to notify air traffic control of an urgent situation.

...to fly to the threshold.

It's one step below declaring Mayday.

Uh, I guess Boston, we need...

Swiss Air 111 is directed to Halifax

and starts its descent.

Can I vector you to set up for runway zero-six at Halifax?

The pilots appear calm and in control.

Halifax is just 20 minutes away.

They want us to turn to the south.

At that point, everything was normal.

I gave the pilot an initial descent

and he requested to level off

at an intermediate altitude

to get the cabin in order for the landing,

which took to mean that they needed

to pack away dinner trays and things like that.

Cabin bus off.

Cabin bus off. Roger.

But the seemingly controlled situation

on board Swiss Air Flight 111 escalates

into a full scale emergency.

Autopilot, disconnect.

We are declaring emergency now!

Swiss Air 111 at time 0-1-2-4!

All my screens are down!

I'm planning on steering by instruments.

Maintaining 300!

Shortly after declaring an emergency,

the plane goes silent.

It was probably one of the most helpless feelings

that any individual can have,

not being able to do anything

but just sit and watch the target

and hope that it would turn back toward the airport.

And of course, it didn't.

At 10:31 p.m. Atlantic time,

residents of Peggy's Cove hear a devastating explosion.

From the cockpit voice recorder,

investigators know they are dealing with a fire

and not a plane that was malfunctioning.

We found no anomalies

or no problems in any of that flight data

that suggested there was a problem with the aircraft.

Investigators work their way through the hangar

of wreckage recovered from the Atlantic Ocean.

Finally, they find scorch marks,

which reveal that the source of the fire was

in the back of the cockpit,

directly behind the First Officer.

Following this trail leads the team to an unlikely suspect,

the entertainment system in First Class.

The Swiss Air MD-11s provided First Class

with one of the world's most sophisticated entertainment systems.

Passengers in first class

could choose their own movies,

access the Internet, and even gamble.

This entertainment system was not part

of the original MD-11 design.

The system had some major deficiencies.

It was getting very hot,

it drew a lot of power.

And, uh, thereby, for example,

raising the cabin temperature

considerably, because it was always running.

They did not install a simple "off" switch,

nor did they install appropriate cooling systems.

Any time you have an electrical system

where you're putting an after-market install

into an airplane,

you run the risk of compromising

the integrity of the aircraft itself

as it was originally designed.

When informed about the flaw in the wiring,

Swiss Air immediately disabled the entertainment system

on the rest of its fleet.

But investigators discover that another piece

of the jet had helped the fire spread

with alarming speed.

And in this instance, we did discover a wire

that arced in that way.

And right next to it was some very flammable material

called metalized polyethylene terephthalate covering material

that covers the insulation blankets.

This polyethylene insulate was

common on commercial airlines around the world.

It had somehow passed

the industry's flammability tests,

which require materials to self-extinguish

after a reasonable period of time.

This thermo-acoustical material

that was in this aircraft was very flammable.

Even though it passed a test,

it does sustain and it does propagate flame.

The fire spread quickly from the cockpit

back into the First Class galleys.

Some metals showed heat damage

from temperatures as high as 1,100 degrees Fahrenheit.

Less than 12 minutes after the crew declared

a Pan Pan, the fire disabled

all electronics in the cockpit.

In the aftermath, Swissair removed

the flammable insulate from its entire MD-11 fleet.

The rest of the industry was required to follow suit.

In Phoenix, Arizona, flight engineers continue

their C-check on the 737.

It includes testing the plane's rudder.

The rudder is one of the jet's most vital control surfaces.

It allows a plane to turn left and right.

Okay, C-Bass, rudder should turn.

All right, go ahead.

A problem here could have terrifying consequences.

Uh, you're clear on the left.

In fact, despite years of proper maintenance,

a problem with a tiny component

with the 737 rudder killed more than 100 people.

Not even the most diligent maintenance workers could have spotted it.

March 3rd , 1991.

United Flight 585 begins

its final approach into Colorado Springs.

Another ten-knot gain.

Thirty flaps.

Oh, God, flip!

- Fifteen flaps. - Fifteen!

Oh, my God!

Oh, my God!

- Oh, God! - Oh, my God!

All 20 passengers and five crew are killed.

Investigators from the National Transportation Safety Board descend

on Colorado Springs.

My first sense that it was going to take some time

to investigate the accident was

the damage that we saw on the parts.

An aerospace engineer by training,

Greg Phillips is in charge

of investigating United 585's flight control systems.

We focused in after eliminating

other flight control surfaces

that we thought could contribute to the roll.

We started looking at the rudder.

But investigators face a critical obstacle.

Most of the plane's parts are too crushed

or burned for testing.

Luckily, one vital component is still reasonably intact.

This is the power control unit or PCU.

Used constantly during flight,

especially during landings,

the PCU performs like a car's power steering.

When the pilot pushes on a rudder pedal,

the PCU uses hydraulic fluid

to convert the gentle movements

of a pilot's foot into the pressure needed to move

the 737's enormous rudder.

The heart of the PCU is the dual servo valve.

This valve is roughly the size of a soda can.

It contains two extremely thin slides

that glide past one another.

These slides direct the flow of hydraulic fluid

which moves the rudder.

When a technician opens up the power control unit,

it seems to be in working order.

We didn't have any absolute indication

or information that we could point to that said

the rudder power control unit, the servo valve or any part

of that flight control system caused that accident.

It's a pass.

For only the fourth time in its history,

the NTSB releases a report

that does not reach a conclusion.

We had put a lot of time and effort in

into the investigation

and we just weren't sure what had happened.

Less than two years later,

Greg Phillips and the NTSB would be brought back

to the mysterious disaster

with the crash of another 737.

Hold on. Hold on.

- Hold on. - Shoot!

What the hell is this?

Oh, God, no!

September 8th, 1994.

US Air Flight 427 has crashed near Pittsburgh,

killing everyone on board.

When we first arrived at the crash site, first of all,

there was no aircraft there.

There were only bits and pieces of the airplane.

It wasn't really recognizable as an airplane.

Investigators begin to see that this crash is

a mirror image of United Flight 585.

On final approach, United 585 rolled right,

while US Air 427 rolled left.

Both crews are caught by surprise.

After a terrifying struggle, both crash with no survivors.

Once again, investigators test the dual servo valve.

But they come up empty-handed.

That unit passed all its operational tests.

There wasn't any indication that it had failed.

We were going up against an aircraft that had an incredible safety history.

It was really, everything you could see for 30 years,

this had been a great airplane.

We were trying to prove that there was something wrong

with the straight A student.

The team reaches another dead-end.

But almost two years later,

they get a third chance to solve this deadly mystery.

In Phoenix, Arizona,

a long night of maintenance is winding down.

Over 30 mechanics

and some half-a-dozen inspectors

have combed through a Southwest Airlines 737.

The team has made more than 400 separate checks.

Parts have been replaced.

Southwest 737 N427WN is almost ready to fly again.

Hopefully it'll push back out of the hangar.

We'll do some leak checks, make sure everything's working,

nothing's leaking.

Then it'll go to the gate for departure time.

But it's a truth in maintenance

that engineers can only fix what they know is broken.

For several years, every 737 that flew had

a hidden danger that not even the most careful technician could have spotted.

In the early 1990s,

two 737s crashed in mysterious accidents.

In both cases, the jets spiraled out of control.

In 1996, the same malfunction strikes again.

It's June 9th.

Captain Brian Bishop prepares to land in Richmond, Virginia.

Then, just like United 585 and US AIR 427,

his plane rolls out of control.

I turned the yoke the opposite direction

and stood on the opposite rudder pedal.

The pedal didn't move for me.

We didn't know to what extent,

but we knew we had a problem with the rudder.

For over 30 seconds,

Bishop struggles to control his renegade plane.

And then, just as suddenly,

the 737 calms down and goes back to horizontal.

We had started the checklist.

And almost before I could finish the sentence,

all of a sudden there was just a 'wham!'

The 737 is once again out of control.

Then, out of the blue, Eastwind 517 is back on track.

Wasting no time, Captain Bishop gets it onto the tarmac.

Taxiing in is when I realized my legs were shaking.

We marched to the scene.

The airplane literally didn't move.

Stayed at its location at the airport

until we got down there.

Suddenly they had a 737 that had had a rudder incident,

that was intact,

and they had a pilot who was alive and who could talk about it.

I think they were much happier to have the airplane than me.

Investigators zero in

on the 737's rudder controls.

The power control unit is tested again and again.

But it performs perfectly.

Refusing to give up, investigator Tom Haueter decides

to try a different test.

One fellow mentioned a test they had done in the military

of a thermal shock.

The power control unit is soaked in dry ice

and blasted with nitrogen gas at minus 40 degrees Celsius.

Then it's injected with super-heated hydraulic fluid.

It's then given a command to start working.

As we were standing there listening

to the actuator move left and right, left and right,

it stopped, and it was not commanded to stop.

It just jammed.

Stopped working completely.

The team has discovered

that a small hydraulic valve that controls the rudder

of the world's most popular jetliner can jam

in the right circumstances.

And the valve can jam without leaving behind any traces.

When investigators double-check their results,

they discover another major flaw.

In careful analysis of the data,

a couple of the engineers recognized

that it not only stopped working,

but actually left became right and right became left.

There was actually a movement

of fluid in... into places that it shouldn't have gone.

And the reversal is like driving your car.

You turn to the right, it goes left.

You're not gonna figure out this failure mode

until you go off the road.

And in these cases, that's what the pilots were faced with,

something so unusual

that they didn't understand what was happening.

What the hell is this?

That would explain why the First Officer Chuck Emmett would keep his foot

on the rudder pedal,

because he's thinking,

"Why isn't the plane going right?",

and he's feeling the plane go to the left.

In the aftermath of these disasters,

pilots received better training

on how to deal with sudden rudder problems.

Boeing spent hundreds of millions of dollars

redesigning and replacing the rudder's dual servo valve

on thousands of 737s around the world.

One thing we don't like at the Safety Board is to have

an undetermined accident,

because then we can't make a change to improve safety.

So out of US Air 427,

United 585,

we have a much safer 737 fleet.

It's 7 a.m.

After an eight-hour shift, the maintenance is finished

on this Southwest Airlines 737.

According to their maintenance reports,

the team has conducted 78 unscheduled procedures

and 339 scheduled inspections.

I think each one of us out here has

a sense of pride in themselves,

that we do the best job we possibly can, day in and day out.

Obviously, the stakes are high. Every life is important.

Despite the horror of airplane disasters,

they are still extremely rare,

especially given how often passenger planes take off and land.

Sometimes we obscure the fact

that we fly millions and millions

and millions of people day in and day out,

without putting a scratch on even the airplane, let alone the people.

This is the most amazing system.

This system depends on the dedicated team

of professionals committed to taking care

of these 21st century masterpieces,

planes so well built that they could fly

almost as long as we're willing to take care of them.

We've learned now how to inspect and maintain these things

and even rebuild them to where they should have an indefinite life.

They're built tough

and they should be able to last forever,

if they're maintained properly.