Tuesday, May 31, 2011

A Story...

The news has been covering the Air France flight 447.
Like all modern aircraft crashes the crash was the result of many causal events chained together.

Its clear that the root cause is a failure of both pitot static systems. Probably caused by super cooled freezing rain. One could reasonable argue that the root cause was the failure to replace these sensors, as that particular model had a history of freezing rain problems.
I believe that the static ports clogged and the pitot ports remained open.


We pick up our story shortly before this happened.

The experienced captain has retired to the crew rest area leaving two more junior crew to watch the airplane. Its four hours into a very long flight, its dark and cloudy flying between layers one cant really see the sky or the surface of the ocean. The cruise flight is usually uneventful and almost fully automated. The airbus flies as high as it can given the fuel load, it seeks thin low drag air for maximum efficiency. The margin between the cruise speed and stall speed where the aircraft stops flying is thin. They fly as high as they can given their weight. Maybe as little margin as 15%. In this warm cocoon with dim cockpit lights one struggles to stay alert.
Its hard to stay alert your human clock has no idea what time its it's dark, quiet peaceful.
Little has prepared you to be sharp and focused, you are about to loose the fight for your life.

The plane flies though a mist of super cooled water. This immediately makes a smooth solid sheet over the airframe. The various heaters and bleed air system make short work or removing this ice from most of the important surfaces.

The Pitot static system has two ports. One is used to measure the ambient "static" pressure. The other port is used to measure the pitot or impact pressure.
The absolute pressure on the static port is used to measure altitude.
the difference in pressure between the pitot and static pressures measures the indicated airspeed. IE dynamic air pressure. At these altitudes the airspeed might read 150mph even though the airplane is going 600mph. The air is thin here so the impact/airspeed pressure is
not adjusted for density as its real use is to tell what aerodynamic conditions the airplane sees, not the real speed over ground.

The static port heaters aren't quite up to the job. They seal shut under this icy glaze.
Since the system is basically flying in equilibrium there is no sudden change. After some time,
maybe quickly, maybe many tens of minutes the computer system on the aircraft determines that it can't tell what speed or altitude its flying at. The autopilot is officially confused. So it does what its supposed to do when confused, it shuts itself off and tells the pilots I have no clue you figure it out. The junior pilot is awakened from his day dreams with an alarm the basically says you've got it. So he is now been jarred into action to fight some flaw or failure of unknown origin and type. He can't really see outside he is now hand flying the airplane focused on the instruments. Maybe he sees the airspeed is a little bit high, so the natural reaction is to pull the nose up and slow down. When he pulls up the airspeed slows down as it should, as the plane climbs the pitot pressure decreases and the indicated airspeed slows down. So his first reaction seems to be correct, nose comes up speed slows down. Its kind of strange the altitude did not change. No he's confused. But his brain says nose up slow down it worked. Airspeed is ok.
Back to messing with the computers to see what error code caused this problem.
Yet maybe he keeps applying a little back pressure the plane climbs and slows down to the point it stalls. Now he has multiple different alarms, stall warnings, computer alarms and confused autopilot he must choose what problem to address. Scared and confused he goes back to basics, fly the dammed plane. At this time the aircraft is stalled and descending rapidly in the dark clouds. The outside air pressure is increasing so the pitot pressures is increasing showing a rising airspeed. Yet in the real world the airplane is slow REALLLY slow, falling from the sky.
Yet the pilot saw the airspeed respond to his first inputs when he took control, that must be working. soon the airspeed is really climbing, maybe even approaching redline as the pressure increases. The Altimeter shows lots of altitude, the view out the window is dark confusing and useless. The more the pilot tries to pull up to slow down the airplane the faster the airspeed shows.... he now fears that he's going to rip the wings off..... His brain screams your screwing up something is wrong, yet noting seems to behave correctly, trying to just do the basics and its not right. Its not clear that the pilot ever figured out what was wrong before the plane smacked the water killing everyone on board.



When I was a 100 to 150 hour pilot in Alaska I almost had the identical crash.
My college girlfriend had come up to visit at the end of the summer and I wanted to take her flying to show her where I grew up. The airplane I was used to flying, a super cub on floats, was down for maintenance. So we were flying my Dads Turbo 206. It was completly tricked out with all the slow speed stol slats, big engine amphibian floats (not regular floats like the cub)
I Probably had 6 to 10 hours in the 206. It was a bit much for me. We were going from Ketchikan to Craig Alaska, typical Alaskan day raining dreary 1500 ft ove cast. Getting through the pass across prince of wales island was tight but not unusually so. Getting near Craig there was patchy fog around and the water was glassy. Glassy water landings are a challenge in a sea plane because you can't judge height over a mirror surface. So to land you set up an attitude and rate of descent and just hold it till you hit the water. I set up fro a straight in landing heading directly at Craig, I'm all configured, wheels up, flaps down, cowl flaps, etc... I check the gear twice. All the check list things are done, but the plane is not flying right. The airspeed is way high and I just can't get it to slow down. I'm trying to hit the target glassy water speed to set the right approach angle. Its not working the nose is way up, the airspeed is climbing. With all the STOL gear on this plane the stall is mushy not sharp so I don't notice a buffet or stall break.
I'm fully stalled and descending to the water at a high rate of speed. My brain is screaming something is wrong fix it. I can't figure it out something is obviously wrong. If it had been the super cub I had a lot of experience in I would have realized what was wrong by the feel of the airplane. I did not have enough time in the 206 to realize this. I eventually realized that I was in trouble and decided to abort the landing, full power (this was the souped up plane so full power would climb at a 45 degree angle) Stop thinking about landing start thinking about going around.
I started looking outside exclusively as I did not want to fly into the town of Craig, I needed to go around. When I stopped focusing on the airspeed and started flying the airplane I lowered the nose. At that moment we hit the water. I'd just powered my way out of a stall into minimum controllable airspeed by pure power. We hit firmly but not hard enough to break anything.
(remember this is glassy water so there is no height clue) I immediately pulled the throttle back and we were bobbing on the water, not moving at all. I looked down and the airspeed said 130 knots. I reached down and flipped the alternate static port valve and altimeter and airspeed suddenly read the correct values.


Here I was flying by visual flight rules in a simple non automated airplane in the day time while being on high alert due to the newness of flying this plane and a frozen static port almost killed me. I can close my eyes and imagine the terror as the pilot struggled to figure out what was wrong, his brain screaming your screwing up something is wrong fix it, and being unable to figure it out. I've been there it was not fun. Its been 29 years since that flight and retellign the story still makes me feel terrified.



A side note on multiple causes:
With the STOL gear the plane stalls 10 or 15 mph slower than the factory stock version. The bush pilots that flew that plane on a regular basis found the continuous stall warning horn on approach annoying so they put a chunk of tree branch in the stall warning vane so it would not go off. If I'd heard a stall warning I might have figured it out sooner...the cub I usually flew did not have or need such things.

19 comments:

heroineworshipper said...

Greenspun called it right, 1st. Pitot tubes always came across as hacks & hobbyists on rcgroups seemed to have endless problems with them. Can't believe the most advanced aviation systems are still at the mercy of this tiny, unreliable, 200 year old, mechanized artifact.

Thom Vincent said...

Paul,

It seems your combined GPS/IMU system could be used in situations like this - on aircraft or rockets or spacecraft. Any progress? I know it's a very substantial undertaking.

Also, I left you a question on your last post, if/when you get a minute.

Cheers,
Thom

Andrew Mooney said...

Absolutely Awesome: A ghost story for any airman. Every warning light is screaming at you and yet you cannot work out why...

A thought that I have had about the information provided by pitot tubes is are the engines functioning? The engines can provide information about air pressure due to the power that they are generating. You would need some sort of force measurement device examining the moving propeller or fan assembly in relation to the airframe. As the propulsion system generates power, how much power it is generating is then capable of being measured by the force it is exerting to the airframe.

Pitot tubes just seem unreliable on their own. If one of them freezes up, a substitute will, and so a separate route to determining air pressure is important.

MTO said...

EFIS systems available to the experimental aviation market, such as the grand rapids units, will correlate their gyros, air pressures and GPS as an error check. It seems incredible to me that an airbus can't match an RV6 for redundancy on this front.

Anonymous said...

Only one question:

Was ONLY ONE PITOT on the aircraft,which, this being by far a not reliable instrunet, would be a criminal negligence??

And if there were several, do you believe that both type of faiure you descrbe happend at same time at each of them??

Anonymous said...

I correct my post 7.34 PM

Was ONLY ONE PITOT on the aircraft,which, this being by far a not reliable instrument, would be a criminal negligence??

And if there were several in different airflow circuits,do you believe that both type of faiure you descrbe happend at same time at each of them??

The bea REPORT, WITH REFERRING TO THE CONVERSATIONS INTHE COCKPIT AND CLEARLY INSINUETING A PLAINLY IDIOTOTIC COLLCTIVE IDIOCY OF THE PILOTS, IS A SCSM YO ALLOW SOME TIME TO THINK OF A WAY TO SAVE AIRBUS OF A DISCLOSURE OF the Failure of their already HEAVILY CRITICIZED ""FLY BY WIRE" SYSTEM

Paul Breed said...

It had dual redundant pitot static systems. They were identical, and the design was prone to failure under certain icing conditions.

So I believe that both static ports got the ice glaze treatment and both had the same design failure to clear them.

Realize that with rapid descent an ice plug would be driven deeper into the hole/tube as ambient pressure increased.

OOton Tisch (ex-anonimous!) said...

Thank PAUL BREED

What DUAL imply?? OOne the Tp of thta aircraft, one on the side of the other?? one on each side of the Aircraft??

As for the ice pressured into the Pitot ehen descending is correct, but it was nor descending when it happened!

Max said...

I too find it sad that even in this day and age, there are no backup ways to sense airspeed / altitude on a major airplane - like GPS. I'm not advocating wiring the autopilot fully over to a GPS, but I seem to recall this being not the first deadly incident caused by the Pitots (ex. taped over on AeroPeru 603?), and I fail to see how a backup GPS reading would anything but help in such cases, even if not normally used as a main instrument (but as a check).

In fact, the biggest problem I see with the tubes is no indication of failure if it happens. Perhaps - being such important instruments - they could be a bit better diagnosed (in-flight): I could imagine a simple setup that periodically tries to reverse the pressure in the tubes ("blow" in them), to verify they are not clogged up (and possibly even de-clog them along the way); I have no idea whether that was ever actually done though...

Anonymous said...

Agreed, seems like decent and pressure would only be a factor after the problem occurred. There must be some other MAJOR factors here. If the tubes froze at cruise it should be something a pilot could recognize or some other system would identify. Are we really saying that if both tubes freeze at the same time the end result is death? Doesn't make sense. Maybe the indicators can't handle that case? Maybe the pilot can't?

Paul Breed said...

The pilot is supposed to recognize this. I'm really technical.
I have my aircraft and power plant mechanics license, I was alert, it was day time, I was 100% awake and it almost killed me.

When the basics of airspeed control, drilled into you from the first student flight, aren't working you need to reset your expectations and figure out the puzzle. I barely missed crashing, the air France pilot did not.

The only real protection we have from double failures in strange places are the human brains we carry around on the airplane for just such events. This was a triple failure static source #1, static source #2 and junior pilot brain, had any one of the three systems worked as intended it would have been a non event.

heroineworshipper said...

You could probably have a table of valid pitot readings for each altitude & GPS velocity, that created a warning. Unfortunately, it probably wouldn't be within the 15% margin of an airliner.

The computer could directly compare GPS acceleration to pitot acceleration. The autopilot could command slight acceleration & decelerations to test the instruments, but by then, you've just made air travel uneconomical.

Makes you wonder if this space travel business is commercially feasible, since the margins required to make air travel economical are barely real.

Brian H said...

Gripping stuff!

But it's "lose the fight for your life", not "loose". Which rhymes with "goose".

Bruulstraat 2C, 3360 Lovenjoel said...

Shouldn't the air france crew have felt the descent? Shouldn't they have felt that they're falling?

When I'm flying as a passenger I feel every force on the plane that interacts with inertia. How come the crew didn't get this clue?

C. Scott Ananian said...

You only feel *acceleration* not *velocity*. They were descending rapidly, but probably not in free fall.

Beside, the pilot was getting conflicting information from several sources. His inner ear was just one more "instrument reading" at odds with all the others. Pilots are actually taught not to trust their inner ear too much, since you can actually flip your plane completely over on a dark cloudy night if you try to fly by feel alone.

peeFactor said...

Paul,

I don't know of too many incidences of blocked static ports on modern airliners (the A330 has static ports on both sides of the nose for the F/O and CAPT) - but I awoke today to the same theory.

The FDR does not record ADR2 data so we do not get a picture of what speed was presented to the F/O. Furthermore, I think ADR switching could have contributed to the confusion (this is my speculation). At one point the PNF had switched the AIR DATA switch to 'F/O on 3'. Later on after the captain is in the cockpit, they switched the AIR DATA to 'Capt on 3'. So now they were presented with ADR 3 information on the PNF side, which agreed with the ISIS, and ADR 2 info on the PF side. As I've seen before with ADR switching, the information does not immediately get shown but rather 'scrolls' to the indicated value. I could be wrong but this is what I've noticed with normal working ADRs, when using the ADR switch on the 320 family.

The cockpit conversation would explain the confusion then, when upon switching the ADR to 'Capt on 3', the pilots note the descent in PFD1 and the ISIS, however see the altimeter on PFD2 (from ADR2) scrolling UP to an erroneous (frozen?) original altitude of 35,000, despite the fact they are descending at 10,000fpm. Furthermore, the PF had indicated he had lost V/S indications on his PFD, which further hindered the situation.

Again, all we can do is speculate, but the F/O Static Port being blocked (however unlikely), seems to fit very well with the actions of the PF and the later cockpit conversation about 'you're climbing there, but there you're descending'.

www.muebles.pl said...

It can't succeed in fact, that's what I believe.

best custom writing services said...

This is one of your best posts, I think.

college paper writing service said...

This article is great! If just one had been to purchase an argument on the authentication of this post, I believe that you would have to throw of your energy because this article is ideal in addition to everyone