Airbus crash/training flight

[bearfoil]

I think jcjeant has a strong point here. With FBW, training "can be" shortened, it inspires confidence, sometimes without the mode to back it up, (BOAC), It is seductive in that "stable flight" is actually cheaper than manual (in the long run), etc.

For me coffin corner is all vectors of flight are maxed, and cannot be stretched, or modified, without upset. Too fast? Pull up. Can't, Stall. Too slow? push yoke. Can't, overspeed. Want to turn? Sorry, drop a wing and roll into a spiral. Etc.

The very best format for avoiding problems with CC is either one, FBW, or Conventional. The edge? Neither one. Where a pilot is not as precise as Otto, neither is Otto aware of the potential disaster, just as A Prot can be compromised by another "protection". "Don't touch" is a human skill, evidently not so much with Otto, except when it gives up on the a/p, and runs for cover. (447)

The word "Protection" I find offensive. Can we lobby for a refit of nomenclature?

[infrequentflyer789]

Originally Posted by jcjeant

Hi,

Why they will learn now .. when they don't learn before ??

What was there to learn before that we/they didn't already know (then)?

• Sensors can fail (especially if damaged in maintenance) - already known.

• Water in AOA sensors can damage them - known, but maybe to be sure, we should ensure that they are to be covered in washing procedures (oh, already done). Note that we also already knew that remembering to tape sensors up, but forgetting to untape them is also fatal, and that these sort of maintenance failures affect both FBW and non-FBW aircraft. Fatally.

• Triple-redundant system cannot cope with multiple sensor failure - known mathematical fact (see previous discussion on thread).

• Alpha prot will not activate correctly in the event of multiple AOA sensor failure - known consequence of above. By design (and cannot be designed any other way).

• Sensor failure can be identified by both a check flight and "maintenance records" - believe this is exactly as designed.

• Aircraft will fail the alpha prot test on check flight in the event of multiple AOA sensor failure - as designed.

• The failed test will not have any other consequence for test flight (not even a stall warning) - which is as per the design of the test procedure.

All these are still known, still valid - even the last one. In this case the crew conducted a different and far more dangerous test in which the inevitable consequence of a failed test was a stall (or approach to) with trim up, which it appears they did not know how to recover. In the light of the latter point, the inevitable consequence of a failed test was a crash.

Why test a protection system whilst relying on it to save you ? - better to not test it at all and hope it works, you'll live longer (on average, whether it works or not).

Maybe we should issue a new procedure that says "for test flights, follow the test procedure, it's really important" (and same for maintenance). Will that fix the issue of people not following the procedure ? Will they learn when they didn't before ?


Finally, on learning, the fatal LOC, the final hole in the cheese, was failure to control pitch / trim, in (approach-to) stall. When will we learn that this issue is not a FBW issue ?
Quoting the AAIB (regarding the 737):

Safety Recommendation 2009-045: It is recommended that Boeing clarify the wording of the approach to stall recovery Quick Reference Handbook Non normal Manoeuvres to ensure that pilots are aware that trimming forward may be required to enhance pitch control authority.

This isn't the only other incident either. There will be more if people don't learn, and mis-characterising it as a FBW issue may stop people learning.

[infrequentflyer789]

Originally Posted by jcjeant

Hi,

Seem's the "know datas" about AF447 put the flight very near "coffin corner" and the margins for speed or altitude were very thin .. at least.

Coffin corner theory was de-bunked extensively on the 447 thread.

Margins may have been insufficient when you add in flying through a large CB... but that isn't "coffin corner".

Wy emphasize on FBW ?
Cause FBW is know for keep the plane flying (more easy than a full manual flying plane) on the limits or to not go of the limits

In transport jets it will be autopilot that flys it in cruise - whether the jet is FBW or manual. Current autopilots are not sentient and do not have the concept of the plane being "easier" to fly, so FBW or manual makes no difference.

U2 flies far far closer to coffin corner than commercial jets, and it is not FBW (I think - obviously I do not have U2 design documents...).

[bearfoil]

The U-2

Is an odd duck, light, very high aspect ratio, and plenty of Power/weight, it routinely operates at very high altitude, frequently nibbling for hours at Stall, overspeed, upset, etc. Coffin Corner is quite familiar to its pilot. It is common for overspeed and Stall to be separated by less than two knots. Not really an accurate comparison. I don't recall in the 447 thread, (either one) where CC issues were completely rejected?

rgds
bear

[infrequentflyer789]

Originally Posted by bearfoil

These aircraft are not sentient beings, they are complex machines, machines that deal out dreadful consequences when operated in less than correct ways.

Such has always been the case. One additional thought: when you design out (or "protect" from) the simple failures, you are inevitably left with the complex and "interesting" failures (and typically with a more complex system).

Not to blame the mechanical, but understand the consequence of progress and complexity.

[infrequentflyer789]

Originally Posted by bearfoil

The U-2
[...] Not really an accurate comparison.

No, but intended to disprove theory that it is FBW that allows you to fly closer to the line - since it doesn't have it.

I don't recall in the 447 thread, (either one) where CC issues were completely rejected?

I do recall posts that actually calculated the margins based on the reported weight, fuel, height, flight-time etc., and came up with reasonable numbers.

That does depend on what you define as a reasonable margin and what is CC, and what is reasonable in a normal cruise may not be good in a CB with unreliable IAS. On the other hand, both CBs and AS problems can and have killed, without needing help from CC.

Only the recorders will tell us (and sadly I don't think they'll be found anytime soon).

[bearfoil]

Well, CC is a stretch as re 447. Stall and upset are not, and the U-2 can be flown (is) manually on the edge. It takes a lot of training. Not long ago, a U-2 pilot did something most pilots would have said was nearly impossible, he Stalled, and Spun, while turning Final at Beale. He crashed in town, I think, but with the lift the a/c has, and the glide, there is no excuse for stalling a wing. Especially when one can control the a/c at the razor edge in thin air. Beale is virtually sea level, and the weather was nice. This aircraft gives up and drops a wing on the Runway slow enough for E-1's to run along it and grab the tip, to keep it from scraping concrete. Landing is actually the most difficult manouver in the a/c's quiver. It was designed w/o focused attention to landing, other performances were critical. It simply does not want to stop flying.

FBW actually does make possible flight beyond the safety limit for manual control. It isn't advisable, but as BOAC has said, it promises alot. At the edge of a Stall, the FBW machine won't burp, or sneeze or suffer nervous muscle contraction. Manual limits are grosser, this is why I think numbers can be calculated for this test, and flown down to prot (calculated prot). The computed number is accurate, and should be relied on, but the crew here did not take the time to calc this low speed number.

It really is as BOAC says, and I don't think he would disagree that FBW can glitch, meaning the pilot must be schooled in Glitch response. I disagree that the glitch is in any way different from any other machine. The fault lies in unjustified reliance on the machine, based on its representations, instead of scepticism, and proper training/response. IMO.

bear

[HazelNuts39]

I don't recall in the 447 thread, (either one) where CC issues were completely rejected?

AF447's CC was at FL450. It was cruising at FL350.

regards,
HN39

[BOAC]

I think bear and I are pretty much on the same sheet - my view is that:-

1) There are software deficiencies which need to be addressed
2) There are training deficiencies - ditto (see my thread about computers on Safety forum)
3) There are deep-rooted misleading 'salesman's falsehoods' which need to be eradicated from human memory in conjunction with 1 and 2

I trust I am not mis-understood here - I definitely see FBW and software assistance as the logical and safe way forward. I loved the 'automation' that gradually eased my work-load (but sometimes increased it!). It is the road that we have been seduced into following that I take issue with.

..and can we PLEASE stop talking about 'coffin corner'?

[olster]

I remember from my tenure on a FBW Airbus type (fairly early days) some startling isues which would corroborate with previous posts.Firstly ,very poor training from instructors with limited or no time on type.Secondly,far-fetched propaganda from Airbus hq which did not fit the facts.Not disputing that the Airbus FBW types are not fantastic feats of engineering but there are and apparently remain , traps for the unwary.My view became that any aircraft that can cause the demise of the airbus chief test pilot (RIP) can do the same to me with ease.

Cheers

[bearfoil]

olster

To be fair to AB, the demise of its chief of test is most likely statistically more probable than a low time line pilot. The line pilot should be very familiar with the fat part of the Bell curve re: handling. The Test Pilot flies in the corners, challenging the a/c's limits, virtually daily. He has a different mission, and it is not without enhanced risk. Insurance is more expensive for him for a reason. This is not to say his job is dangerous, only that he encounters more risk. His skillset is different, not necessarily "better". IMO

In this accident, a "test" was undertaken without the numbing prep work, and executed by a pilot who was out of his domain. The instructive part of the accident has to do not with AB "problems", nor is it to do with poor pilotage. It shows the results when preparedness is sacrificed and substituted with overconfidence that had no foundation in fact. imo

bear

[bearfoil]

I think in and of itself, flight is quite simple. I have the persistent feeling that Captain Hank of old, the guy who flew water a/c through 747s for PanAm could fly the AB or the Boeing with equal confidence.

Let me be honest here. I think in some cases, perhaps in many cases, complexity is devoid of return, and sold for its profit or perceived benefit. I try to put Hank in each of these accident circumstances. 447? Switch off a/p and hand fly. Unreliable a/s. Pitch and Power, and by the way, where is the A/H? Alpha Prot? I need a box to "save" MY a/c from upset? Etc.

If an a/c is perceived to be superior, Fine. Don't forget to show its pilots how to manage it in difficulty. Don't make the goal transcendence above the pilots. No machine is perfect, if it can fail, it will. Don't potentiate a surprise when the a/c does something "impossible". Don't let engineering ego kill people. False confidence is a poor substitute for preparation?

bear

[Mr Optimistic]

Fair enough

...have taken the hint

[The Ancient Geek]

Bearfoil:

This thread is now over 1500 posts long, mainly with people pontificating about automation. Lets get back to the facts as revealed in the report.

The crew attempted to conduct a test at low level despite the test specification warning that the test MUST be carried out above 10000 ft AGL because failure could result in a stall which could take a lot of altitude to recover. The test failed. The aircraft stalled. They had insufficient height to recover.

FBW is irrelevant. Automation is irrelevant. They screwed up by also having maximum nose up trim and TOGA power applied. They put themselves in a situation where ANY aircraft (except possibly an AN-2) would have killed them. Stall recovery is a fundamental piloting skill, prompt and correct recovery action might have saved them but the ground was coming up fast.

Time for stall recovery to be a mandatory currency item at every simulator visit ?

[Bis47]

Stall recovery

Originally Posted by Ancient Greek

The crew attempted to conduct a test at low level despite the test specification warning that the test MUST be carried out above 10000 ft AGL because failure could result in a stall which could take a lot of altitude to recover. The test failed. The aircraft stalled. They had insufficient height to recover.

False.

A. They flew into a full stall, with quite a wing drop and recovered
They recovered from that stall, loosing 1.000 ft. or so (one thousand, not ten). Speed was up, wings were almost level, they were climbing back to initial altitude. For a few seconds, everything was looking as being under control. They used quite normal recovery actions : stick forward, "throttle" forward. Student pilot skill.

Next immediate required student pilot skill was to trim forward.
This is were the crew first failed.
Why?
Because :
- They were not used to operate manual trim, since AB is on auto-trim in normal operation. Thousands of flying hours without ever touching a trim switch or a trim wheel ... So the habit (and the skill) of using the trim "manualy" was lost.
- They were not aware (obviously) that the auto-trim feature was inop.
I see here a systemic failure :
- automation set one step too far, reducing piloting skill below that of a 5 hours student pilot.
- lack of pertinent training

When the trim upset put the aircraft in an unusual "nose high" attitude, the skill required to recover was at the commercial pilot level : "recovery from unusual attitude" (visualy or by instruments) :
- roll to high bank,
- reduce power/thrust
- and level off carefully when the nose start dropping

This requires training, and recurrent training, and specific training on each aircraft type. This requires good pilots and good instructors and good training programs.

It seems to me that the lack of flying skill was one big factor in the fatal issue ...

Strange, when one consider the captain was qualified as TRI/TRE ...

[safetypee]

Bis 47, a good analysis, but not ‘crew failure’ … … a deviation from expected behavior, the circumstances of which provide many safety lessons.
Expanding a few of your items:
they were not aware … auto trim. Apparently so, but the crew may not have considered that auto trim would/could be inop – poor knowledge, failure to consider alternatives, or failure to recall valid knowledge in the situation.

The systemic failure perhaps includes the lack of guidance on flight testing / system checking. Although AB had guidance would an operator have a copy – it would be prudent to request guidance.
Did the national authority have guidelines or even hard boundaries – the UK CAA did, perhaps these should be used by EASA.

The missing ‘flying’ skill was the ability to think about the test and issues which were to be checked; a TRI/TRE qualification or thousands of hours does not automatically infer an appropriate mode of thought – to see the bigger picture.
It’s not the skill in recovering from a post stall upset, it’s the skill of avoiding a stall or the post stall consequences which is important.

bearfoil provides a lustrous view of test pilots (#1503), but it is the provision and use of the skill set which aids safe operation. A tp’s ‘flying’ skills involve the same human thought processes as available to all pilots. These skills might be used more often and also in non-normal scenarios; tp’s are taught to think that way.
When and how are line pilots taught to think?

[p51guy]

The lack of pilot intervention and using basic stall recovery procedures has been repeated over and over on this thread. It is always the pilots responsibility to make sure the automation is doing what he wants, if it isn't just take over manually. These guys didn't.

[Bis47]

Prevention, cure and judgement

It’s not the skill in recovering from a post stall upset, it’s the skill of avoiding a stall or the post stall consequences which is important

Prevention is obviously of prime importance.
However, by disregarding the "cure" side, one let the last hole open (in the chain of failures).

When and how are line pilots taught to think?

That is a good question. Some answers :

- Pilot selection : select people with good education, check theyr "thinking" ability. This is not in phase with the current philosophy of selecting "standard" (average) personnalities. One of my best student was rejected for being "too good" at thinking. He was also very, very good at "handling" the controls ... (Got a job in a less pretentious company without problem.)

- Continuous education : lectures, reading accident reports and analysing failures, recurrent training with the appropriate briefings and debriefings ... From first copilot job to captaincy, many many opportunities arise to learn facts and to learn thinking, decision making etc; still a few years and many more opportunities before becoming an TRI. At the TRI level, scenario based training make it mandatory to think and to teach thinking ...

That is theory ...

I'am a strong supporter of the thinking pilots, and of all the means to improve thinking ability. It is a long quest, and if you care about it for your children or grand-children it is a long term investment ...

In the real world of the airline Industry?

I'm not sure that the management does care. Including at the flying department managment level.

Finally :

Maybe the captain was a good TRI/TRE with great thinking ability. And he was just "not in the mood" that day. Brain performance reduced for any reason. Human "thinking" failure ... That is. Brain power more often than not is a weak link in the chain ...

I keep thinking that in a stressfull context, basic piloting skills - skills that do not require thinking - can save the day.

[PBL]

Originally Posted by BOAC

There are software deficiencies which need to be addressed

What? This is a new one, coming out of nowhere. What sort of software deficiency is proposed? Where did any software fail to fulfil its requirements?

PBL

[olster]

Bear -good points and very well made.I certainly understand that a test regime has different risk to line operation.Perhaps,I was guilty of a glib remark ref the tragic accident of the A330 at Toulouse.However,I still would say that there are(or certainly were) traps for the unwary in the Airbus fbw series which were not instructed well,if at all.I appreciate that this may or may not be relevant to Airbus test flying.

Cheers