Automation Bogie raises it's head yet again
 

Europe Regulators To Curb Autopilots


By ANDY PASZTOR


(WSJ) European air-safety regulators, stepping up warnings about excessive dependence on cockpit automation, have told pilots of two widely used Airbus jetliner models to avoid routinely using their autopilots in certain emergencies.


The directive issued last week by the European Aviation Safety Agency, or EASA, specifically covers emergency procedures for more than 1,000 Airbus A330 and A340 jets in the event of major airspeed-sensor malfunctions. More than two-thirds of the two-engine, wide-body planes are flown by European or Asian airlines.


More broadly, the move reflects growing concerns by international safety experts about the hazards of undue reliance by pilots on on-board automated systems, following a spate of incidents pointing to pilot mistakes and confusion stemming from improper use of automated safety aids.


Safety officials believe many pilots flying a wide range of jets-across the U.S. as well as other regions-may need additional training to cope effectively with emergencies when autopilots or automated thrust-control systems are unavailable or can't be trusted.


"Given the increasing dependence on automation" in the latest generation of jetliners, according to Deborah Hersman, chairman of the U.S. National Transportation Safety Board, "you've got to have pilots who are prepared for all scenarios, especially those in which they have a lack of information" from typical flight-management sources.


If two of the three external-airspeed indicators on the Airbus models singled out by last week's directive provide unreliable readings, autopilots and automated engine settings are designed to instantly disconnect. But instead of routinely re-engaging both at the first sign of trouble to try to control the aircraft, EASA now requires pilots to wait at least 30 seconds to determine if reliable information is flowing into the automated systems.EASA's directive also follows the recent release of a long-awaited U.S. air-safety study emphasizing that commercial pilots tend to abdicate too much responsibility to on-board computers, partly because most current training programs emphasize that automated systems generally are more adept at handling in-flight emergencies.


Crews also must perform various other checks to ensure that the automated systems will issue appropriate orders once they are switched back on.


When speed sensors feed unreliable data to flight-management computers, EASA said the result can be autopilots issuing abrupt and "inappropriate" climb or descend commands, which may "constitute an unsafe conditions." Regulators patterned the directive after a separate safety warning issued by manufacturer Airbus, a unit of European Aeronautic Defence & Space Co.


The move comes after a series of dangerous airspeed-measurement failures and unpredictable or dangerous autopilot commands on Airbus A330s in recent years. The sequence of events that led to the June 2009 crash of an Air France Airbus A330 flying to Paris from Rio de Janeiro apparently started with unreliable airspeed indications and autopilot difficulties. No formal cause has been determined for the crash, which killed 228 people.

............................................................ ......................

Fine words telling us what we already know. Pity the regulators don't insist that operators under their control come up with a viable solution to automation dependancy

My understanding of what should then happen in this eventuality is that the pilots should manually maintain pitch and roll attitude and the normal power setting for the stage of flight while considering their options for continuing the flight with degraded airspeed information.

This is easier said than done for pilots taken by surprise - especially at night, turbulence or cloud, especially when they are at high altitude and heavy. In these conditions a limited safe speed range is available requiring pretty careful and accurate handling in circumstances where the pilots (necessarily) will have little recent currency and practice.

In my view following such airspeed anomalies it would be better for the 'system' not to disengage but to degrade to a basic attitude and thrust hold mode accompanied by a pretty serious warning about what was going on.

..........Automation ........ the word fires up a battle among some others. but... again it should be only an assistance and aid measure to cope up with the whole scenario and not just the whole means of the unnecessary reliance on the circuit boards ........

Agreed .. skills dont equal a programme :}

John Farley:

Sometimes, hand-flying using basic attitude instrument flying skills and a global view of all available information will usually best sort out which is invalid information. (well, perhaps not with the lack of basic skills that seem more prevalent these days.):eek:

The accident report of the December 1, 1974, crash of a NWA 727 in New York should be required reading for all of today's active crewmembers. They forgot to turn on the pitot heaters and when "the stuff hit the fan" they got fixated on the bad information to the exclusion of attitude instrument flying.

The two 757 accidents in South America ( Aeroperu 603 and Birgenair 301) are closer to the modern magenta-line era and are sickening.

The EASA AD discussed in the newspaper article is available here

"In my view following such airspeed anomalies it would be better for the 'system' not to disengage but to degrade to a basic attitude and thrust hold mode accompanied by a pretty serious warning about what was going on."

Whether is is medical or aeronautical, automated systems (patient ventilators for example) should degrade gracefully, sustaining basic functions rather than just cutting out. The greater reliance on automation the more graceful the degradation needs to be rather than dumping the surprised operator directly into a manual/raw data situation.

That's all good and well, if the system is able to accurately define what is causing the problem. If it's not an input error, not a malfunction of the autopilot computer itself, then one is left with other problems that can produce similar results, but which will continue to be a problem should the autopilot degrade to a lesser state of control.

Eliminating the autopilot completely takes all the problems out of the look, returns control to the pilot, and allows the matter to be resolved using a procedure.

Attempting to simply fall back to basic autopilot functions when the issue is a runaway stabilizer trim, for example, may only escalate the problem. Taking autopilot away and allowing the crew to manually fly eliminates the trend, allows the crew to restore equilibrium, and then address the problem.

I agree wholeheartedly that stepping "back into the loop" in turbulent night conditions in the top of a cloud layer or a mountain wave can be a challenge, but then isn't this the reason for having a crew on board in the first place? Fly and control the airplane. Make decisions that the automation cannot. Evaluate. Troubleshoot.

Raw data training these days is often seen as redundant, and in many cases, "raw data" still includes use of the flight director. Crews should be comfortable transitioning between manual flying, and automation, equally proficient in both, and this practice should be approved and encouraged by operators.

Chaps

I well appreciate that by manually flying the aircraft you may be able to diagnose what info is right or wrong but I would prefer the crew to be able to gather their wits about them first, check what all the attitude indicators are saying and then take over and start investigating.

Is this not better than having the aircraft unexpectedly dumped on you with some turbulence induced roll and pitch away from cruise already happening?

OK some oldies were necessarily trained to cope with this but a lot of people today may have many other useful skills but they don't include tricky handling and before you know it a UA is off and running.

Hi,

Source:
L?hyper complexité des systèmes a des limites qu?Airbus ignore. : Les dossiers noirs du transport aérien

I've never seen a training program that didn't include unusual attitude recoveries and or jet upsets. I finished recurrent training at the end of last month, and it certainly included upset recoveries, and automated, as well as manually flown procedures.

During an approach into Kabul this month, a generator failed, and tripped off, taking that side of the split bus with it, as well as the essential bus. As it was going, the autopilot began react erratically, the control column began to vibrate and shake, and then the yoke began to rock back and forth with a high frequency vibration somewhat like an aileron buzz.

The quick solution was to disconnect the autopilot and handfly, while troubleshooting the problem using checklists and procedures.

The autopilot in the airplane I fly will drop from "command" to "manual," providing basic functions, during certain types of malfunctions. However, when in doubt, get rid of it and fall back to basic flying. It is, after all, something one should have mastered at the student pilot level.

Agreed. Also pray that your FO is more than a 250 hour lad/lass who is straight out of flying training.

The message that comes out loud and clear from the original post is that some of todays pilots, whilst relying heavily on automation, are not doing enough to fully learn and understand how it does, or doesn't work.

How many times have you been in the SIM and seen someone, when given a simple problem, get themselves into an awful mess through a lack of system knowledge?

Maybe it is not so much an individuals fault but more symptomatic of todays flying training?

As an ex flight engineer (now), I've taught glass cockpit for quite a few years.

Observations like 'you are now inundated with information' have now run full circle. For many years it was regarded as a good thing.

I suspect now it's not viewed as being the bonus it was designed to be. Even in my last years of flying, I was conscious that the levels of spacial awareness seemed to be reducing as crews 'followed the coloured lines'.

I think glass cockpits are brilliant, but I hope that we still will keep hold of those fundamental skills, the ability to use mental arithmetic and the motivation to keep those skills honed.

Sadly, I don't think that's happening. I think we're allowing a culture of complacency to develop despite all the courses preaching the contrary.

We'll see. I suspect I'll be the subject of a wave of rhetoric by outraged crew, but I flew for over 35 years and that's what I saw.

Is it time to wake it up again?

http://www.pprune.org/safety-crm-qa-...-aviation.html

(Best ignore all the Islamabad stuff at the end.)

Sir, my need is sore.
Spirits that I've cited
My commands ignore.

(J.W.v. Goethe's Sorcerer's Apprentice)

all that STUFF is built by engineers for engineers

BUT what a pilot NEEDS is:

an easy way to make sure the plane is flying properly

that it is on the propert course

that it is performing properly in terms of mechanical health

that threats to the plane can be dealt with (IE weather radar)

I imagine that much of today's cockpit information could be dealt with in other ways, for example, on takeoff...a computer voice says POWER set, engines nominal

freeing the pilots to look out the window

but, whatever method, pilots, airlines, governmental agencies, must allocate resources to make sure pilots can handle problems, including problems that are handled in an impromptu fashion...making things up as you go if there is no time to do a checklist

Generally if people get into trouble for systems knowledge or lack thereof, it's from failure to follow procedure. When people begin trying to out-think the airplane rather than pick up a checklist and work through a problem, they get in trouble.

I do see people get into trouble by trying to follow through with the automation when they really ought to get rid of it. I had a sim partner once that could fly the airplane very well, but who insisted on doing everything on the autopilot. I understood that he wanted to learn that panel and work it well, and I understood that he'd been counseled to let the autopilot work for him as much as possible in the upcoming checkride. The problem was that whenever he started to get behind things, trying to fix it through the autopilot only made it worse. It might be that the descent wheel didn't start him down as fast as he'd have liked on the approach, and ended up high. Trying to catch up by increasing the rate would see him missing his MDA. Perhaps it was being a little fast or a shirt turn onto the localizer, and then blowing through the approach. In each case, he saw the problem coming, and disconnecting the autopilot would have fixed the problem for him by allowing him to handfly and correct the problem. Sometimes trying to work through something with the autopilot is a little like trying to run a race on stilts. It's easier sometimes, and sometimes more effective, to remove the stilts, shut off that autopilot, and fly.

While I fully appreciate the discussion whether automatics are good or bad or both, and I can understand some people's concern about pilots loosing pilot's skills, I must say that this AD has nothing to do with it!

If you read the AD carefully (and if we are talking about the same one, namely the one HazelNuts39 presented us), it's not about automation but about correct failure finding in case of one or several Air Data failures. This can happen in any aircraft, even in steam-driven cockpits. Needless to say, that in most multiple AD failures, the autopilot disconnects by itself. The problem pilot's routinely have with such type of failures in the cockpit is not the lack of manual skills, but the wrong way of identifying the correct AD source or the insufficient use of pitch and power.

Modern cockpit have even more advantages than older ones, i.e. they offer more independent sources to troubleshoot (e.g. GS indication from FMS).

Absolutely agree. It is the fault of the regulators and company Check and Training that is the root cause. A pilot cannot and should not be expected to work out what he needs to know/what techniques/skills he should have. He didn't design the aeroplane and he didn't dream up the concept of flight. Our industry and society in general gets better because the lessons learned by earlier generations are used by current day society, not by pilots having to individually "gain" all that experience, first-hand, over and over again.

From where I sit at least, there has been a fundamental failure of the hierarchy to ensure that pilots are equipped with the skills and knowledge they need to operate in the magenta-line world. It's not as though this problem has just popped up and people have been caught out. The writing has been on the wall for years but has been ignored.

I disagree. This is all about crews being unfamiliar enough with handflying that they get themselves into bother when they have to so they leave hte AP in for too long. A crew who routinely hand flies in a less-automated aircraft will cope much better with a loss of airspeed because they are used to it (and probably know the powers and attitudes like the back of their hand); they are much more likely to get rid of the AP earlier.

Pull out the QRH on the A320 and you will find a table of power settings and attitudes for unreliable airspeed.

On my previous aircraft I had a good idea of these as I had to apply them myself every time I flew, now I need to refresh my memory as I just don't use them so often.

An advanced autopilot is a major bonus during an emergency as it reduces the workload of simply keeping the aircraft flying safely, enabling more attention to be devoted to dealing with the problem. These days airspace is more complicated and congested. Tolerences are more critical and setting up an approach much more involved.

In the 1960s the controller could probably find you a vacant block of airspace to play around in while you sorted out what was going wrong with your B707, followed by a radar vector straight on to the ILS.

These days you will need to maintain tolerances in a hold and sequence into a STAR, following speed and altitude constraints. Don't forget you still need to run the emergency checklists and procrdures as well.

I certainly wouldn't want to be operating early generation jets in modern airspace.

Depends entirely what sort of emergency or just merely the garden variety "Non-Normal" like a generator, pack, uncommanded pressurisation and so on.

The discussion is more about an undue attraction to automatics like the example I observed when a crew took up an instrument approach holding pattern using the magenta line and heading bug to trace their flight path over the magenta line holding pattern.

Pity that a minor programming error placed the holding fix which was a VOR, some 15 miles from the actual VOR and that over the fix the VOR/DME showed 15 DME which you would have thought would have picked up by the crew as somewhat unusual. At the same time the VOR/RMI needle was pointing to where the actual VOR was 15 miles away.

Fortunately it was in the simulator since the real aircraft would have eventually flown into a nearby mountain caused by the crew blindly following a false position magenta line. The reluctance of some crews to cross check frequently with basic navaids during even a normal flight is symptomatic of the inherent dangers of automation complacency. It is plain laziness.

When in the simulator, you see an experienced pilot furiously programming a visual circuit into the FMC, when in the first place all he was asked for, was a take off climb to 1500 ft, turn downwind at 30 degrees angle of bank, maintaining 2 mile spacing from the runway clearly visible from his seat using raw data and manual flying with no autothrottle, turn base and land - all in CAVOK - then you have now seen the archetypal automatics dependant pilot.

Just out of interest...

Is the birdy (FPV we call it on the 777) purely derived through the IRS's? or is there some ADC input?
My company has done the delightful thing of removing the performance inflight section of the ORH on the 777.

Air Crash Observer: Pilots out of Control on Autopilot
I wonder how many other nations this applies to also and if the bean counters have their way then it could happen elsewhere.

All pilots should have the ability to take over from an autopilot and be able to control the aircraft AND THINK at the same time.

Manual flying skills can only be acquired and kept by practice. I first discovered this as we transitioned to glass cockpits. After a few months of mostly autopilot use I found my manual flying skills were not up to my standards. I then started doing more manual flying and less autopilot use which corrected the problem. Since manual skills may be needed at altitude as well as other times your practice should include cruise too. By manual flying I mean no autopilot or autothrottles.

It must be realized that if your manual skills are rusty it will take more mental capability to fly the aircraft and leave less mental prowess to think about your next course of action or how to correct whatever problem you are dealing with. Your manual skills should be almost second nature. Do what ever it takes to accomplish that and when the day comes and its your turn to do your best to save a ship you will be ready.

Indian aviation is a constant source of entertaining issues, such as widespread logbook fraud.

It's not a "beancounter" issue. No airline attempts to save money by failing to train it's crews on aircraft systems (including automation). It doesn't cost any more; there's nothing to be saved by not emphasizing full useage. Indian aviation has never been the gold standard for training, competence, veracity, honesty, or transparency. In a country where everybody in the business likes to call themselves "commander this," or "colonel that," "wing commander," "air commander, or some other ridiculous and pretentious title, you have a cultural issue, not an accounting one.

The thread is about over-reliance on automation, not about lack of training in it. The problem, if indeed such is the case, isn't lack of training in automation, but if anything not enough emphasis on basic raw flying skills. This may be owing in some cases to operators that insist on the maximum usage of automation to provide the smoothest flight for passengers, to some pilots who simply use it more out of preference or habit.

A change that I noted during my last recurrent session was that we've eliminated the requirement to conduct a raw data approach during the checkride. This wasn't an economic issue, but a change incorporated by the FAA to reflect real-world conditions; new procedures were added with changes in additional authorizations. No training costs were saved, but changes were incorporated. Personally, I'd prefer to see us retain the raw data procedures, but it's not my call. It's not the "beancounter's" call, either.

"Beancounters" don't run training departments.

At the end of the day, those pilots that have spend years hand flying, will not have a mental demand-overload if they suddenly have to grab the controls.

The thousands of hours I spent on Viscounts, Heralds and F27s, left me with an autopilot somewhere in my brain. After 5 years of not flying, I suddenly found myself height-holding to the thickness of the needle, without really thinking about it. I was typical of every single pilot I flew with in those days.

Summer nights on Viscounts without radar were something of a baptism of fire for young pilots. The guy in the left was just glad he didn't have to avoid flack, so when in the middle of a CB, the horizon bar became lost behind the edges, he would think little of it. It was also a good education in containing one's fear.

I would hate those days to return, but what has taken its place is some kind of hybrid operation - neither skilled master of the machine, or possessor of a computer that is capable of calling on a wide enough spectrum of learned data. It is not a desirable compromise.

Then I'm afraid you haven't read the concerned AD. It says:

It is - in this case - not about leaving the AP in for too long, but for re-engaging it again before proper trouble shooting has been done. This is a clear issue of proper failure handling and good CRM and not about over-reliance on automatics.

Dani

Thank you Dani, voice of reason at last. Loose Rivets, you have one too sir.

Loss or lack of manual skills is an issue, but Colgan will not be prevented by mandating 1500 hrs. Why not 1550? Recognizing approach to stall and proper stall recovery is mandatory task pre-solo, at 10 hrs total.

FTOs are historically called schools but they are quite far from it, normal businesses run for profit. Why fight the syndromes and not cure the diesease? Do teach pilots in traning and screen them properly? LH does it right, tight selection, stringent pass rules and 250 FOs are ready for the job in a novice position. Any airline could do the same, should they decide to do so.

Low hour RYR FO in Ciampino landed 737 uneventfully single-pilot, I applaud him/her. The pilot was trained for such task, and performed well. There's nothing heroic about it yet it speaks clearly about his training.

LH again, cross-wired sidestick, FO saves the day by using a learned technique.

Kos hard landing clearly uncovered discontinuity in follow-up on training progress. No amount hand flown of hours will substitute that. Yes, the trainee may be slow to acquire skill and need more hours to reach the target, but ultimately it is the pass-grade process which failed.

Prescribing 500 simulator landings will not cure that, faster learners will be unnecessarily tied to the box at immense cost and pilot who needs 600 of them will still be held short.

My 2c
FD (the un-real)

PS: As noted in the beginning loss of manual flying is an issue. I believe it is not the issue which bends metal. Let's read again on A340 in Toronto, AA in Jamaica, Garuda in Jogjakarta and Southwest in Midway. Or Little Rock, mother of all that (still) goes wrong. Drilling holes in the sky, even manual raw data NPAs in limiting conditions will not improve that. Sure it builds confidence but overconfidence is what kills people. Pilot that can professionally perform V1 cut and OEI landing as well as handle manual reversion is ready to fly the line, ready to learn the line. Provided that his previous training did not leave terra incognita elsewhere in the required knowledge.

- part I I agree with, but the last phrase (my bold) not. We need to identify (and gently coax away) the source of the mindset that needs to re-engage the A/P with 'unseemly haste'.

Exactly as you speak, BOAC, all agreed. The automation is not the problem, it is a just a simple tool. When NDBs came, all learned that the tool has limitations, i.e. the night shift (my english vocabulary reached limits here). Same for ILS false lobes. Training against autopilot misbehaviour is only half the story. Back to basics, it is a triangle: man-task-machine. Once this is explained, cockpit automation becomes just another tool just like a power steering on a car. Those not trained for AP use may face dire consequences one day. Not the consequences of AP, but consequences of their inproper understanding.

FD (the un-real)

Now that will be interesting, manually flying new generation aircraft, designed to be operated with the automatics.

Congested airspace, overworked controllers, poor infrastructure in most of the country and the regulator wants to remove the automation and have everyone flying by hand.

Get rid of all the expat pilots and rapidly promote the youngsters with little experience in either manual or automated environments.

Now where did I put my wizz wheel ?;)

John Farley
I fully agree !!!
I'm glad to see this proposed by someone who knows what he is talking about and who himself surely wouldn't need this help but appreciates that it would improve overall safety.
It would buy the surprised crew precious time to sort things out.

IMHO it is not really helpful in today's airliners that once the regular automation gives up, a possibly mis- trimmed aircraft in dangerous (and potentially unclear) attitude and maybe config is handed back to completely puzzled pilots.

Admittedly it requires reliable attitude information but if that is lost also manual flying won't save the day....

Lack of manual skills
 

This trend has been a long time coming.
I was admonished in BA by the Route Check Capt 10 years ago for hand flying to 10000ft in severe CAVOK departing from a very sleepy Carib airport.
He said on the B744, we should always use maximum automatics to avoid overloading the NHP!.
I believe that on the BA B777 even when manual flying, use of the A/T is mandatory, so you can't win.

Friend of mine undertook his first simulator assessment session (A320)after arrival with an Indonesian low cost carrier. He had 16,000 hours mainly 757/767 and a type rating A320 simulator only. Cleared for take off with left turn (simulator) under visual day conditions. Just before bending the aircraft around to the designated heading at 2000 ft, he glanced over his shoulder in a reflex action to check all clear for traffic.

Immediately the Indonesian check captain stopped the simulator and demanded an explanation of why the pilot checked all clear before commencing the turn. The astonished pilot said it was good airmanship to have a look if visual conditions and that he had done this for years.
The check captain said "You should never look outside for traffic- you should always rely on TCAS to warn you of traffic". Another case of automation gone mad. And they walk among us in Indonesia....

" he glanced over his shoulder in a reflex action to check all clear for traffic. "

Don't know about the rest of this crazy world, but this is EXACTLY how I was trained during my initial PPL flight training!
ALWAYS look BEFORE turning/climbing/descending (as much as possible) - radar/TCAS is great....but NOT fool proof! :ugh:

How the heck does that work in the event of unreliable airspeed ? A/T carries on setting thrust based on known bad air data ?!

Or is it that sometimes A/T will drop out, but you can't force it to drop out ? I thought one of the key issues of the great A v B FBW debate was that B allowed you to override it ?

I guess the buzzards and other large birds of prey have transponders strapped to them !

If that's the attitude, then it's just a matter of time before there's a mid air collision.

just how quickly have we forgotten the Brazil tragedy, in which TCAS didn't work because of lack of transponder signal from one plane.

just be a good pilot...look around...oh and for birds...if you bank sharply, you may take the engines out of the plane of the bird formation.

think about it.

infrequentflyer
 

No need to get worked up about A vs. B or not being able to disconnect the Autothrottle. For normal ops the BA 777 SOP is indeed to use the Autothrottle at all times when it's available and functioning correctly, however in the event of unreliable airspeed indications one of the several initial actions is to disengage the Autothrottle and use "manual" power.

Very true, such basic skills are and need to be "fully automated" tasks. No pilot (or whatever operator of almost any complex set of tasks) would be able to fly efficiently without having these skills as totally action models. The psychological load would simply be to great if every single action would be a considered response to a megaload of sensory inputs.

Having lost your level of "human automation" will not only make you slow to react, it also will make you bad to handle any additional stress factors.