Sunday, December 6, 2015

Don't fly the computer, fly the aircraft!


An investigation into the crash last year of Indonesia AirAsia Flight 8501 has revealed some startling lapses by the flight crew - and revived a debate that's divided the flying community.  Flight Global comments:

The ill-advised decision by an Airbus A320 crew to reset flight computers, without appreciating the ramifications, and the pilots’ failure to cope with the subsequent in-flight upset ... is not the most dispiriting aspect of the crash. More bewildering is that it occurred just five years after Air France AF447 – an accident which had left the airline community stunned that highly-trained and experienced pilots could have failed to recognise one of the most rudimentary upset conditions in aeronautics.

Pulling a circuit-breaker demands specific detailed system knowledge to comprehend the risk. Stalling, in contrast, does not. To misunderstand stalling is effectively to misunderstand the basic concept of lift and the role of the wing. In short, to misunderstand the most fundamental reason why an aircraft is able to fly.

Handling an aircraft at rarefied cruise altitudes requires care and finesse, and stable flight is a delicate balancing act normally entrusted to the flight-management system and autopilot. Yet this environment in which pilots need a separate array of handling abilities is also the one in which they are least likely to gain hands-on experience.

The result is that pilots, used to operating the aircraft at low altitude, suddenly encounter different behaviour characteristics if they are forced to take over in cruise – when they are likely to be facing pressing matters as well as a possible degradation in flight-control laws and related envelope protections.

There's more at the link.

It's not just flying the aircraft at cruise altitude, either.  Consider the crash while landing of Asiana Airlines Flight 214 at San Francisco in 2013.  Wikipedia reports:

The NTSB determined that the flight crew mismanaged the initial approach and that the airplane was well above the desired glidepath. In response the captain selected an inappropriate autopilot mode, which without the captain's awareness, resulted in the autothrottle no longer controlling airspeed. The aircraft then descended below the desired glide path with the crew unaware of the decreasing airspeed. The attempted go-around was conducted below 100 ft by which time it was too late. Over-reliance on automation and lack of systems understanding by the pilots were cited as major factors contributing to the accident.

The NTSB further concluded that the pilot's faulty mental model of the airplane’s automation logic led to his inadvertent deactivation of automatic airspeed control. In addition, Asiana’s automation policy emphasized the full use of all automation and did not encourage manual flight during line operations. The flight crew’s mismanagement of the airplane’s vertical profile during the initial approach led to a period of increased workload that reduced the pilot monitoring’s awareness of the pilot flying’s actions around the time of the unintended deactivation of automatic airspeed control. Insufficient flight crew monitoring of airspeed indications during the approach likely resulted from expectancy, increased workload, fatigue, and automation reliance.

Again, more at the link.  Bold, underlined text is my emphasis.

Note how many references there are to automated or computerized systems, and the pilots' lack of understanding of the aerodynamic realities underlying those systems.  It seems that pilots are relying more and more on automated systems to manage almost every aspect of a flight.  The aircraft can even be landed automatically at some airports.  If the pilot can get it off the ground, he can engage automated systems within seconds after becoming airborne, and never need to touch the controls again (except to issue updates to the automated systems, which is computer data entry rather than actually flying the plane) until arrival at his destination.

As one who's been "up close and personal" with bush flying in Africa, this horrifies me.  The thought of a pilot who's not completely familiar with his aircraft, so that he could almost fly it blindfolded if necessary, is so weird that it boggles my mind . . . yet apparently that's not what many airlines expect of their pilots these days.  They expect them to control, or "fly", the computer systems rather than the aircraft.  If those computer systems are given the wrong inputs, or their response is not properly understood (as in the crashes of Air France 447 or AirAsia 8501), the pilots are seemingly incapable of responding correctly to the information they're receiving. 

This makes me glad I don't have to fly commercially very often . . .




Peter

20 comments:

Well Seasoned Fool said...

The concept of autopilots doing everything is foreign to me having no experience beyond the wing leveling function in a Mooney. My first instructor started every lesson with one question, "What is the most important thing you will do today"? The answer, Fly the airplane, first, last, always".

Roy said...

"This makes me glad I don't have to fly commercially very often . . ."

Some perspective is in order. I agree with you and I also avoid flying commercial but for a different reason. Travelers generally want three things - Safety, comfort, and reliability. Airlines today are not delivering comfort and reliability, but they are delivering safety. Indeed, commercial aviation has never been safer than it is right now, while at the same time, the total number of flight hours has never been less.

And while I will also agree with you that *over reliance* on automation is definitely a negative trend, keep in mind that routine automation also tends to eliminate the primary cause of accidents - human error.

Better training is the answer.

Roy said...

(WSF)

My flight instructors always stressed that the three things you do in order were - Aviate, Navigate, and then communicate. In other words, in an emergency, the very first thing you did was *FLY THE AIRPLANE*. There have been quite a few accidents where the proximal cause was the pilots failure to do that first step.

Old NFO said...

One other contributing factor in ALL the Airbus crashes, is there is NO stick feedback. It's a joystick, just like the one you play games with. You don't know if you're actually doing anything... sigh

Chuck Pergiel said...

" the pilot monitoring’s awareness of the pilot flying’s actions"
Am I being thick or does that not make any sense?

J G Pelham said...

To continue the creed. Aviate, Navigate, Communicate, Manage Systems

Chuck the pilot monitoring is the pilot not flying in a dual controls setup. The idea is they are supposed to have some situational awareness of what the pilot doing the flying is doing and why. Dependant on personal relationship, background, habits, mindset, and training this can vary widely in how difficult it is to do.

This is something within aviation an cybernetics which we have been very much aware of but which is incredibly difficult to study. If you want to read the basis of the field you can go back to the work by Fitts in the 50's which everyone agrees is wrong but at least is wrong in useful ways. That was a decomposition approach where either the pilot or the computer did something but not both at the same time. The AGARD organisation did some excellent work into so called cognitive cockpits in the 80's and a PACT(Pilot Authority control of tasks) framework was developed for specifying the levels of authority between pilot and computer.

The trouble is that these different approaches are more like philosophies than proven theories. Almost all the experimental work done used highly trained pilots drawn from the same background and organisation and who were already proficient and able to work together.

I myself as a systems guy and distrustful of pilots would dearly love to have a gold plated single true way to design a system that a pilot couldn't misunderstand in an emergency while also being complex and abstracted enough to be useful for everyday use without affecting pilot proficiency by taking away their practice of good flying habits.
However for the moment there is no such way. Only different flawed ways which all either distract the pilot, don't draw the pilots attention when needed or hide the complexities of flying from the pilot.

Anonymous said...

Airspeed, Altitude, Attitude.

In the Air Asiana crash, the plane could not land itself using its CAT III system as the ILS for that runway was offline. The pilots had to manually fly the approach using the PAPI lights, which was an unusual procedure for them. All three pilots had their eyes on the lights and none of them checked the airspeed indicator strip on the PFD. The pilots I work for are trained in CRM, Cockpit Resource Management, if one pilot is flying a visual approach the other monitors the instruments (PFD in a glass cockpit) and calls out airspeed and altitude as needed. The Air Asiana copilot or the relief pilot sitting in the jumpseat should have been tasked by the pilot with monitoring the airspeed and altitude or assumed that duty on their own initiative.

As an apprentice, I flew a few times in the right seat of a floatplane (Beavers and Otters with the original round engines) and many times in the jumpseat of HS748s and a Boeing 727 with pilots who earned their stripes landing Twin Otters on ice floes and gravel beaches in the Arctic. These were pilots who refused to use ILS if they could see the runway, it wasn't real flying to them and they were very proud of their ability to grease it onto the runway. Give me a pilot who has paid his dues in the bush or the Arctic making thousands of landings on lakes or gravel strips with no aids whatsoever and I'm a happy passenger. When I fly commercial I have no idea of what caliber of pilots are up front and I am an anxious passenger.

Al_in_Ottawa

Chuck Pergiel said...

Thank you, JGP. I was afraid it was something like that, impenetrable jargon for anyone who isn't totally in the know. That might be part of the problem right there.

Anonymous said...

I've always used "Pilot Flying" and "pilot not flying", to clarify who had their hands on the controls when things got interesting. The Pilot In Command might not have been flying, so we used PF and PNF. The PIC was still the one with final responsibility.

LittleRed1

Rolf said...

And yet... there are people who think that self-driving vehicles are a panacea that will solve all road-related injury and fatality problems.

*sigh*

Jacquejet said...

First, I'm not sure what you mean by " 'up close and personal' with bush flying in Africa." Does that mean you were a pilot in Africa or you were a passenger? I'm not sure how being a passenger in a light airplane gives anyone insights into the operation of a commercial transport. A light airplane pilot, maybe a little.

The Wikipedia description of the malfunction in the A320 was confusing. There are 2 rudder travel limiters on the airbus. One is required for dispatch but the aircraft flies just fine with out either. Only that the pilots need to be aware when using rudders that at high speeds, there are no limits on rudder travel and one can easily damage or over control the aircraft.
I have no idea why any one would use CB's to unpower the FAC's (flight augmentation computers). The on/off switches are on the overhead panel, one on the left side above the captain and one above the F/O. If you lose a travel limiter, a reset can be attempted by cycling the fac's off and back on, one at a time. Did the crew do this but turned them both off at the same time?
If they did, the airbus degrades into "alternate law" , a mode in which the the plane can be stalled.
To me, it looks like the pilots tried to climb above some big T Storms, at or near their service ceiling, got distracted by a minor problem that presented no real complications to their flight, messed up a simple reset (that wasn't needed at the time) put the a/c in a degraded flight law, and stalled the aircraft.
Based on the comments above, maybe the Asian airlines ought to make their pilots be bush pilots for three or four years learning real flying skills before allowing them to sit in the cockpit of modern jet.
For that matter, having flown the A320 for 25 years, I didn't realize my lack of Bush Pilot time made me so inexperienced.

Will said...

Jacquejet,

essentially, what we have been seeing for some years now, is how the "integration" of computer systems into passenger jets has added an additional complex layer of new and confusing means for the pilots to lose control of the basic function that they are charged with: flying the damn plane!

"For that matter, having flown the A320 for 25 years, I didn't realize my lack of Bush Pilot time made me so inexperienced."
This may be a truer statement than you think. Do you fly anything besides your airliner? Own your own plane? Do aerobatics? Sailplanes?
A statement of time in cockpit doesn't necessarily mean you have a lot of experience. Just like I can claim I have been operating computers for 35+ years, that doesn't mean I am an expert with them. In earlier days, it meant that you had encountered lots of problems, and survived or overcame them. Not so much these days, with things so reliable.

As has been pointed out by you and others, airline pilots don't really fly anymore, they manage flight systems, especially the Airbus planes. It seems that the only common problems that pilots encounter are the ones programmed into the simulators. So, to a greater extent these days, simulators are where "real" flying is done. It would appear that the non-US airlines do a bad job of this sort of training.

Anonymous said...

Thanks Jacquejet. The quality of comments is a good indicator of the quality of a blog, and your comments are always high quality.

Jacquejet said...

Will...I'm not going to list my flying vitae. Suffice that a bunch of different C130's, 727's and DC10 and 25 years in the 320. I readily admit that there are pilots with better skills and more experience. That said, I am pretty comfortable and pretty confident in the 320. I comment on some of these articles only because this is my world and others might not understand how aircraft systems work. The 320family is different than most other commercial jets because of the fly-by-wire. It is even different than Boeings fbw (I know nothing of how Boeing does things in the 777, 787).

Believe it or not, airline pilots in every jet flying manage more than fly. Most pilots engage the autopilot soon after takeoff and disengage it just before landing. Otherwise only one pilot is looking outside, doing chores, talking on the radio, you name it while the other guy flies.
The airbus has a few modes of flight where one has to be especially aware of certain inputs. You can get into a real hole if you aren't careful, but then how is that different from any other type. I could name a couple if you wish!

Simulators are the place where all these gotcha's are practiced, besides the approaches you never fly on the line. Single engine ILS to minimums with a single engine missed approach. A barrel of fun, but never done in the real world (I hope). Smoke in the cockpit with a non precision approach in the weather. ILS PRM breakout maneuver. Eight hours of hard work every 9 months at a large airline based in Atlanta.

You are right about Asian airlines and the problems with experience or the lack of it. Many airlines over there hire guys off the street with zero flight time and train them to proficiency. Look up how many airlines use Ab Initio hiring practices.such pilots have no experience other than what they got at whatever airline they fly for. No bush pilot, no civilian, no military, no nothing. Yikes!!!

Sorry to be so long winded. I'll be happy to chat about some of the 320 quirks if you wish.

Jjet said...

If it ain't Boeing, I ain't going.

Take that to the bank.


USAF Fighter Pilot and Juvat(Ret)
AA Captain (Ret)

Jacquejet said...

Jet... How many 320's does AA have now?

Stretch said...

If it ain't Boeing I ain't going.

Will said...

Jacquejet:

I wasn't really asking for your details. Just wanted to make the point about being well-rounded as a pilot. Common in the US, not so much elsewhere, apparently.

Something I overlooked was that the crash five years ago was an Air France flight. One would expect that the government's own airline, and where the plane is built, would have properly trained pilots. That cockpit crew came off looking like a Three Stooges skit.

Why isn't there a master direction control switch that locks out one of the side sticks, so there is no question as to who is in control? Since they are not physically connected, or even by feedback servos, they should never be active at the same time. It's not like they need two people to attempt to pull the control wheels back when the old style systems lose the boosters.

What it reminds me of is two people with R/C controllers set to the same freq, and the poor toy is heading for a major smash-up. Same result.

0007 said...

The thing I remember about Airbus is the video of one of their first models flying into a gradually rising hill as it perform a landing(IIRC) in front of a large group of important people.

Tal Hartsfeld said...

No wonder people worship technology the way they do.

Computers are so intelligent. The way they can handle everything by themselves so immaculately. You don't ever need to do anything yourself---just let automation do all the work for you. You don't need to know or know about anything anymore---your computers have all the necessary information and knowledge.

Technology is always perfect.
For example: Think about how much the social climate worldwide has improved with the proliferation of cell phones, the iPhone, the iPad ...