Asiana 777 crashes in San Francisco

[stekeller]

torrence:

I believe the recommendation to be high on the PAPI was specific to the 747 because of the height of the cockpit. Now the 777-300ER is longer than the 747 so the effect may be similar. Asiana was the shorter 777-200ER, but the pilots would have been trained on the proper aiming point for their aircraft. An interesting note is that the captain in the left seat (being evaluated) was transitioning from the 747-400.

- Stekeller
KORD

[torrence]

Makes sense, thanks.

In any event, from what I've seen in the public reports, it appears that the pilots were not very comfortable with flying a normal visual approach in good conditions.  Along with the Air France crash this may be another example of failures due to not training in basics of manual flight techniques, resulting in over-reliance on and/or misunderstanding of the limits of automatic modes.

Cheers,
Torrence

[IefCooreman]

Sorry for the late reply, but it took me quite some time to write down a "relevant" answer. On the PAPI: 777 is normal as any jet, 2 reds, 2 whites. It's not a double decker.

Do I understand the reaction of the pilot? Yes. Do I understand the reaction of people flying Cessna's? Yes. But it's a completely different world and the cause of the crash is not really clear yet (I think there is still a full report to be published), and the case itself (together with the AF A330, and the recent UPS 767) is still discussed everyday, even on flightdecks. Some thoughts...

1) it was a training flight. Every pilot is nervous for those flights (some exceptions exist). A guy with plenty of experience on 747 transfers to 777. Different for him? Yes. 777 actually has to be flown completely different (pilots in some companies with 744's, who got 748's, complained about the "itchy-ness" of the bird, which is nothing more than fly-by-wire than needs a different approach in stick movements) and the 777 is also much more of a glider than the 747. The hot-and-high situation on the approach is either caused by this, or is caused by ATC and makes the situation only worse. If you have the experience flying a 777, any approach is easier on the 777 than on 747. If you don't have the experience it's the other way around. It's about "getting to know the bird", and even after simulator, it takes a couple of takeoff and landings in extreme situations considering the big difference in weights widebodies are subject to. One can question the training captain yes, but asian culture might have an influence here.

2) autothrottle use: when you transfer from one aircraft type to another, you have to leave some of the stuff from previous types behind because it might be bad on your next. Initially this was the case with the autothrottle: we for instance (Boeing training), never disconnected it during training. The 777 has been flying around for years and it has proven it's reliability. When I asked it to my company, we were allowed to do "autothrottle off" approaches, and have done so in the simulator as well. But it's a matter of training. The captain will probably have received the same simulator training and will have been exposed for a very long time to an environment where you never hardly questioned autothrottle. The fact that there is no autothrottle wake-up below 100ft, is a detail in the book. It never happens because in theory you are supposed to be stable at 1000ft/500ft. And that was not true here (swiss cheese model in aviation safety if it rings a bell...). That same stress might have pushed them to continue anyway (swiss cheese again). This point is now the center of a VERY hot and ongoing discussion in aviation: are we making pilots to depending on automation? I am fortunate enough to fly for a company that allows a lot of manual flying, but not all companies are like that. The more you go east from my home, the more you will be forced into automation. The discussion is not about automation itself though, because automation is a good thing and has improved the safety of aviation.

3) Visual approach. Still acceptable? Well, on a cessna it perfectly is. But long haul is a world where pilots sit in a metal tube for hours and hours and hours and do 2,3 landings a month, flying between busy airports with prescribed procedures that usually end in radar vectors for an ILS, or an RNAV exception. I've done 2 "visuals" in 3 years so far. The first one after 8 hours of flight, where takeoff was at 3 in the morning, the second one after 9 hours of flight with takeoff at 8 in the morning local, but midnight at home (8 hours time difference), at max landing weight. Now the idea of doing a visual is a bit thrilling for me, so adrenaline helps a hand. If on the other hand you are flying for an Asian carrier where automation is the way to go, you will not feel at ease at all. The world you work in is not one that contains visual approaches. On top, the Asiana crew (in the pilots's local time) had done pure nightflight and landed 3 in the morning (if I calculate correctly), which is about as bad as it can get for your body. So a visual approach is an acceptable approach, but it has to be one where the pilot has the choice to do so or not. It's too "of-standard" to bluntly to state that we should always be able to carry one out.

4) ATC using visual approaches standard. Many people who know me will probably know that I consider USA ATC the worst in the world for long haul. The world improves itself every day, USA ATC on the other hand seems to think they are the best of the world and use that to deny any reason to improve their way of working. While their way of working is perfect to allow for a fantastic VFR experience, and their controllers are pretty damn good in "peak times" to get everyone in- and out, their idea that everything in VFR should be applicable to IFR is simply dangerous. I consider New York dangerous for this, I have no experience with Los Angeles airport itself, but some with "Socal" radar. Many collegues - me included - have a standard "protection" that we never tell American ATC we have "traffic in sight", because doing so gives them the freedom to handover separation to us and give us a landing clearance, which is ridiculous. A heavy jet is flown (and monitored by the second pilot) eyes inside up to very low height, except for specific cases like ie TCAS alerts. You have too quickly deviations from altitude that are dangerous. I've had one encounter in JFK airspace with a VFR traffic 500ft above. Nobody seems to realize how quickly a 777 flying at 250kts can climb that altitude if you don't watch INSIDE. You can, with enough experience, start to "fly outside", but again, it requires experience because you are never trained to do so. Only once when doing touch&go's in a simulator, but a simulator is still not very reliable for visual work.

I'm only writing this because I hope it makes people understand how we are trained. I personally love to fly with minimum automation. But the Asiana crash is not really about that only, and I'm actually happy the captain has made that statement about his "stress level", especially because he flies for an Asian carrier. I was actually surprised as well how "easy-going" people were discussing the UPS 767 crash in Birmingham, where (if true), pilots simply busted long going basic rules, where in the case of the Asiana, the trainee (because that's what he was) was tricked by training and a small detail in autothrottle functionality. Sometimes I even think the case would have never happened if he was looking inside to his instruments, instead of outside to a PAPI... (oh yes, on the second visual approach I did, on the ground both the captain and I discussed how poor the visibility of the PAPI was indicating both red and white on the same spot)

There's a lot I could add, but I think this short message should be sufficient... :-)

[IefCooreman]

One note to add, after re-reading my message: when transitioning to 777 from any other Boeing type (except for 787), the autothrottle magic is one of the most "deep" changes you go through in your flying habits, together with the speed trim. Previously you disconnected autopilot and autothrottle simultaneously, this time you are learned to keep the autothrottle all the way and never disconnect. It is exactly this "change" that is one of the main causes it seems. A pilot going through a full training course where autothrottle is never questioned, is here all of a sudden confronted with a small "exception" (that should never have happened if they were stable, of course). But to think that the pilot - who is looking outside - is going to catch that is difficult to tell. I admit as well, that one of the results of the autothrottle system is, you monitor less your speed. This changes when you start to feel comfortable again in your new environment (the 777 flightdeck), but it takes time. You learn step by step.

[Richard McDonald Woods]

Very, very interesting, Ief. Thanks for your thoughts.
I have found myself more and more flying heads down on final approaches in recent months.
Regards, Richard

[Jeroen D]

Thanks Ief for that detailed explanation.

Could you elaborate some more on why US ATC is so bad for long haul? Also, your comments about "traffic in sight". I'm familiar with that having flown in the USA, but does it work differently in say Europe compared to the US?

Thanks

Jeroen

[stekeller]

Hi Ief:

Thanks for all of that valuable input. Nobody wishes to rush to judgment, but this crash is so puzzling because the pilots were experienced, the flight conditions were mild, AND the aircraft appears to have been functioning normally.

My big question is this - I am puzzled by the statements that the pilot flying was unaware or surprised at the autothrottle being disengaged because of 3 things:

1. The mode annunciation for the A/T on the PFD, which he was presumably watching. Or maybe not, because speed was decaying and he did not react in time.
2. The movement (or lack thereof) of the throttles themselves. On the 777 I believe they move (unlike on Airbus FBW aircraft)
3. The sound of the engines - you can hear the power being applied and decreased on approach (at least in the cabin)

In my view, if the pilot flying was looking outside, the sight picture and PAPIs would have helped him. If he was looking inside, the PFD indications, speed tape trend vector, and V/S would have helped him. Back of the clock issues that you mention may explain some mental "paralysis" but it is still quite confounding for me.

- Stekeller
KORD

[Jeroen Hoppenbrouwers]

Don't forget they came in hot and high, engines at flight idle... Nothing will have moved while the aircraft tried to slow down. It went wrong only very late in the descent/approach.

At least part of the surprise was caused by the long flight idle that was expected, in my opinion. Turkish at EHAM suffered from the same issue. Coming in with engines idle, then dropping through the set Vref instead of picking up. You need explicit awareness and almost literally say to yourself "Here comes the A/T... yes... well... hey... *click click* manual" as the speed decreases quite sharply when you go from (presumably) FLCH descent into glide slope capture with engines at idle.


Hoppie

[Will]

Interesting perspective, Ief. I suspect the United States' VFR culture goes back to the overall aviation culture in the country. Almost all civilian airline pilots start out as flight instructors in Pipers or Cessnas for a thousand hours or so, and then build time flying night mail, or hauling skydivers, or doing pipeline patrol, or air ambulance, or some other kind of entry-level work in small planes, followed eventually by a transition to turbine aircraft and then maybe a regional airline before flying big jets. I think something like 5000+ hours prior to airline hire would be fairly typical, much of it in airplanes without much automation. Furthermore, most of those thousands of hours will be into small airports where a visual environment is the rule, not the exception.

I don't know how Asiana hires and trains their pilots. But some countries select their pilots on the basis of pre-employment aptitude tests, and then offer them a job without any previous flight experience at all. They then go from something like 250 piston flight hours, all in a dedicated airline training environment, straight to their airline equipment. That's a very different path from novice to airline cockpit than is what is typical here.

Both training systems are reasonable, and I understand that countries like Korea almost certainly don't have the aviation infrastructure to allow pilots to "work their way up" from smaller jobs to airliners. However, I think the differences in the environments explains the comfort level that the American system has with visual maneuvers.

[Jeroen Hoppenbrouwers]

Ok, this deviates from the thread a bit, but I agree with Will that the US aviation culture is extremely different from at least the European culture.

Case in point: when family or friends from Europe are over, I like to take an unexpected right turn when we pass the airport near our home and drive straight onto the airport. It's KTMB, a simple three-runway 500-craft non-airline training/parking field with tower. Roads, hangars, no food/drink, basically a straightforward industry area. No gate control, always open, just a 30 mph speed limit sign.

As soon as I "crash through the gate", the newbies duck and expect to be shot at. They remain extremely uncomfortable while we tour the airport and repeatedly ask whether this is allowed. I point out that this is just a road, not the flight line, and the flight line is on the other side of that 2 m fence which incidentally has gates that aren't necessarily locked. I have driven my car onto the flight line (i.e. the other side of the hangars) often for work. It's just a car park with wings.

For non-US people, this seems impossible to accept: aircraft are military objects that are defended by people with guns and making photos sends you to jail, or worse.

In such an aviation-averse culture, I can well imagine that pilots don't see any aircraft close up until they are ready to become their F/O, so to say.


Hoppie

[Will]

And now I'll offer some perspective as a pilot. Most of you know my story, so bear with me as I repeat myself. To explain my perspective and the experience from which I come: I flew for a regional airline in the USA and I had 6000 hours or so when I left for a different career. I didn't fly a heavy aircraft, but our speed in the terminal area was the same as everyone else: 250 knots until close to the airport and then the same speed limit as every other aircraft (about 180 knots) until the final landing phase, followed by a slowdown on final to our Vref of about 135 knots. So things like relative speed between aircraft was similar.

My plane was a high-workload aircraft. We had no autothrottle (!) and no autopilot (!!!) so every approach and landing required one pilot to maintain a scan inside the aircraft at all times. For that matter, every moment of every flight required exactly the same thing: someone had to be flying the plane, constantly. Further, we had old "steam" gauges, so it was a complicated scan. A gust could knock the plane into a climb and then within seconds we would be unacceptably high or low, or a slight bank wouldn't correct itself until we were out of bounds. It required constant attention. The non-flying pilot would operate radios (ATC, company, and guard; we had no ACARS), look outside, and generally assist, but they were expected to monitor the flight instruments as well.

So here are my thoughts about visual approaches, workload, technology, and culture.

1. Visual approaches. Highly preferred, since when navigation is up to us in the cockpit, the plane can get on the ground faster. Also, there's nothing more fun about flying than hearing "cleared for the visual" when about 15 miles from the airport on a high extended downwind or base. That was when flying got to feel like flying again; you could strategize and plan and aviate, and actually be in command of the aircraft instead of letting ATC steer the plane for you with radar vectors. I remember using the Golden Dome of Notre Dame university as a marker for the turn to final at South Bend, Indiana, or just using your mind's eye to paint an extended centerline out into the landscape and planning your turns accordingly. This was aviation: managing your airplane yourself, in four dimensions (the three dimensions of space plus airspeed). Sometimes people ask me what I miss about aviation and it was moments like this. Yes, we would back up the approach with an ILS whenever possible. And it was usually possible. But the ILS was reassurance that your aircraft was on the right path in space, as opposed to the visual picture being reassuring that the ILS is correct. If every moment had been that fun, I wouldn't have quit. (But that's another story for another day.)

2. Workload. I appreciate the challenges of a visual approach at the end of a long trip. Believe me, I do. However, to place my opinion in context, envision a day of nine turnarounds (we had no corporate maximum), with maximum allowable flight hours, and maximum allowable duty hours, then repeat times four. So at the end of a four-day trip, I had worked 16 hours duty time each day, with 8 hours flying time each day, with nine legs each day, all flown completely manually (no autopilot, remember). With this type of schedule, you learn to curse the day you decided to be an airline pilot, but you do get a ton of valuable experience: flying, aviating, looking for traffic, getting used to the terminal environment; situational awareness becomes like a second sense. Situational awareness is the oxygen you breathe. To say that one would prefer that ATC relieve us of the responsibility to look outside the windshield is a (literally) foreign concept to me. Looking outside is what you do. It's part of aviating, it's part of situational awareness, it's part of the job. Even when you're fatigued.

3. Technology. Technology adds complexity but reduces workload. Anyone who thinks technology doesn't reduce workload is invited to step into 747-100, or a 707, or for that matter into my shoes as a regional pilot and enjoy some four-day trips, nine legs each day, in bad weather, without an autopilot, without ACARS, without EICAS. Enough said. Still, the NTSB is full of incidents in which human beings hadn't yet worked out the proper relationship between pilots and their technological tools. Sometimes pilots ignored indicators that should have helped them. Other times, they relied on faulty gadgets that led them to difficulty. Managing the technological challenges of modern automation is not simple. Anyone who thinks a 747-400 or a 777 is "easy" because of the automation is flatly wrong. However, the automation at the end of the day is simply that: automation. No pilot should, in my opinion, ever get into the cockpit with an expectation that the automation replaces basic airmanship. The pilots in the United States love the 777. They don't get into the cockpit thinking that the systems are so complex that things like situational awareness take a back seat to flying the plane. They just don't think that way. Airmanship comes first, and it is honed through experience; operating a complicated piece of equipment comes second.

4. Culture: I said it already. Airmanship comes first; operating a complicated piece of equipment comes second. Recall what I said in my previous post in this thread. Your typical American pilot comes to the game with thousands of hours of flying experience, accomplished without electronics to reduce workload, and achieved in complicated airspace in which the pilots had to be as aware of the situation outside their cockpit as the situation within it. I think that pays off in several ways, one of which is being comfortable with the concept of a visual approach. Statistics back me up on the American way of aviating. Remember the insane workload and fatigue of my airline that I described above? We carried 2,000,000 passengers per year, and so far, no accidents, and no fatalities. The safety record of American aviation as a whole speaks for itself.

One final point, off topic, to Jeroen's last post: I flew sightseeing tours at the Grand Canyon back before it had a fence around the runway. The airport at that time was the busiest airport in terms of takeoffs and landings in the whole USA. (Most planes were small so the airport wasn't very large in terms of passengers flown, but it certainly had lots of takeoffs and landings.) Anyway, they didn't build the fence for security reasons, they built it to keep elk and coyotes off the runway. (The coyotes learned to tunnel under the fence, probably because rabbits learned to live inside the fence. The fence did keep the elk out, though.)

[Phil Bunch]

Will,

Thanks for the most interesting (and well-written) reflections on your aviation career and flying.  I look forward to rereading it many times.

Your essay gives me a new perspective on life as a pilot for a regional airline.  I've flown as a regional airline passenger many times in my former work for a large company (I'm now retired), but I never gave too much thought to what the pilot's life was really like.

Also, I believe these types of considerations and experiences apply in a general sense to many careers besides aviation.

[Will]

Phil, thanks so much for your comment.

Upon rereading, I feel I should add just a little clarification for context.

Clarification on my career: I was flying a turboprop, the mighty Jetstream 32. I started as F/O, upgraded to Captain, and then finished my career as an Instructor Pilot in our training department. It wasn't always flying in the gentle airspace over South Bend; every single leg was either into, or out of, the crowded airspace of a major international airline hub.

Clarification on how one can do nine legs in an 8-hour flying day: yes, it's totally possible. Our three closest outstations were each between 30 and 45 minutes flying time away from the hub. These routes were very highly traveled and always oversold, since a 35-minute flight could save a business traveller two or three hours by car. Crew Scheduling loved to bunch all of these turns into one trip since the timing worked out so well. So a nine-leg day made practical sense. We also had our share of "long haul" flights, which for us meant about three hours at FL250 or thereabouts.

Clarification on fatigue: back in my day (the 1990's), we had a 16-hour duty day. Those who are good at math will notice that this leaves 8 hours off, and when you factor in time to get to and from the crew hotel, there was significantly less than 8 hours remaining for actual sleep. Trips were four days long, so there were three overnights per trip, and depending on how long it took to get to and from the hotel, actual sleep time varied between about 6.5 hours to an hour or so less than that. And let's not even talk about the dreaded stand-up overnight.

Clarification on the stand-up overnight: Okay, so we'll talk about stand-ups for a moment. The dreaded "stand-up" was a 16-hour flight day that started at 18:00 and ended at at 10:00 the following morning. It would begin with 5 hours of flying time between 18:00 and about 2:00 am, interrupted by two hours in a hotel at an outstation, and then topped off with another 3 hours of flying from 6:00 until 10:00 the next morning. The term "stand-up" comes from the idea that there really isn't any time to get any meaningful sleep, so you might as well just "stand up" in the hotel. Seriously, hitting the sack for only two hours felt worse than not getting any sleep at all. Crew Scheduling was pretty good about putting a month's worth of stand-ups into one bid so you had time to get your body used to flying on the back side of the clock. But now and then, stand-ups were thrown into a regular line bid. I did literally thousands of hours of this kind of thing, which was excellent training for flying a complicated and fast aircraft, with no autopilot, into and out of some of the most complex airspace in the United States, whatever the weather, and staying up to the task all the time, using both instruments and flight crew in smooth coordination to get the ship slotted and on the ground safely.

Clarification on the similarities between the stand-up overnight and a transcontinental flight: they're pretty similar. Think of a long duty "day", with a few hours of sleep in the middle, with maximum attention needed at the start and at the finish. Just throw in the fact that the stand-up had multiple takeoffs and landings at each end, in the same airspace, and in the same weather, at the same airports, without autopilot, etc., etc., etc.

Clarification on the universality of my experience: My experience was completely typical. I didn't have it any worse than any other civilian airline pilot in America. My kind of flying experience is what typical pilot-applicants bring to United Airlines, or American, or Southwest, or any other American carrier. (The other typical experience is to get to an airline through the military, but I can't speak to that first hand.)

Sorry this has grown so long! Thanks for reading this far.

[SwissCharles]

Back from a rather longish lurking spell  but I am still at it - I plan on being one of the 544 eehhmmm... oddballs, so please count me in Hardy  .  With the exception of the throttle quadrant, I do have most of the parts in stock...

There is really nothing to add to Will's very illustrating and very well written view on the life of the average regional pilot except perhaps trying to illustrate another possible explanation to the various kind of pilots...

Will's and my career are virtually mirroring each others! The things we have in common is that we both did our training in the U.S. (me in California) and flying turboprops (me flying Fairchild Metroliners and Saab SF 340's). And I too gave it up after almost 6'000 hrs to pursuit another career. My active career took place in the late 70ies up to 1985 (shows my age, huh  :shock: ) so a little bit earlier than Will's.

The difference lies in the fact that I did almost all of my productive regional flying in Europe (flew for a well known at the time Swiss regional, who went up in Swiss after the failure of Swissair...). But again there's not that much of a difference.

We didn't have the stand up schedule for one thing and the Swiss FOCA didn't allow for the scariest of rostering wishes, the airline would come up with. But the kind of flying we did, was very similar to what Will so well described. Another exception: European Operators in general did equip their planes a bit more 'posh' than their American counterparts. We had autopilots and in the Saab even autothrust - both having a tremendous effect on lowering the overall workload.

But even then - and that's what I like to point out here since I am not sure if this is just a cultural matter of pilot training (US <-> Non-US), or even more a question of a pilot's personality - you had two clearly discernible and rather different kinds of pilots:
Those leaning rather heavily on automation and those who didn't. When things started to get ... interesting (say due to weather or due to some kind of malfunction) you could differentiate between these two groups very easily: The former (also called the System Managers) kept the automatics on (after having them selected on an eye blink after gear retraction  :roll: ...)  the latter (sometimes also called Aviators) turned them off.

As we now see, this trend somehow grew into the current state of affairs. My take on the reason for this: Airlines are trying to save costs in ever new ways and training system managers is probably (far?) more cheaper than to train aviators. And what you pay is what you get (pay dirt - eat sh!t comes to mind). Nowadays they took this even further, if you look at Pay to Fly schemes and such.

For me it is almost incomprehensible how a crew can fly a perfectly serviceable Triple Seven into the ground in CAVOK at a big airport with a decent runway.
This is not in any way to tell how great a pilot I am or was (had my br#wn trousers moments as well  :shock: ) But being able to judge ones glide angle and approximate hit-point seems soooo very basic that I am appalled at the lack of general airmanship - or the lack of airmanship-instilling training of the airlines - as displayed in this absolutely unnecessary crash. But what  can you expect when a return ticket to even far away destinations can be had for an apple and a song.

Sorry for this long post and for extending the off-topic leg a little further. Turning inbound now  

Charles

[Phil Bunch]

I just came across this article by the NY Times on the Asiana San Francisco crash.

http://www.nytimes.com/2014/04/01/us/asiana-airlines-says-secondary-cause-of-san-francisco-crash-was-bad-software.html

I was surprised to read that the crew was confused/lost and that the airline has filed a complaint with the FAA about how the 777's flight director/autothrottle/whatever operate.  How can a crew not know how to operate the primary controls in their airliner?  They claim they had not been trained on how to properly use the 777's system.  I believe they would have had to demonstrate their ability to handle the 777 under all sorts of stress situations in the simulator, and to know when to abort a landing.

------------------------------------------------
Here's a key excerpt:

In San Francisco, the prime cause was quickly clear; even Asiana faulted its crew for failing to notice that the airplane was flying far too slowly to stay in the air. But it is also blaming "inconsistencies in the aircraft's automation logic."

The carrier said Monday in a filing with the National Transportation Safety Board that bad software design "led to the unexpected disabling of airspeed protection without adequate warning to the flight crew," and that a system to warn the crew of low airspeed did not sound soon enough. The airline also said that the approach ordered by air traffic controllers "led to an excessive pilot workload during the final approach."

Boeing has focused on the crew's failure to maintain proper airspeed, which is expected to be listed by the safety board as the probable cause of the crash. Asiana's filing is an effort by the airline to have the plane's design characteristics listed among the contributing factors. The board's conclusions are not admissible in court, but its ranking of factors often influences how a carrier's insurance company and the plane's builder apportion the damage settlements or court judgments.

In the Asiana crash, the crew believed that an auto-throttle would manipulate the engines to keep the plane's airspeed in the range needed for a safe landing, somewhat like the way the cruise control in a car will adjust the throttle to keep the speed constant. It later became obvious that because of a quirk in two tightly linked systems, the autopilot and the auto-throttle, and because the crew had manually adjusted the throttles at one point, the auto-throttle had gone into sleep mode.

This characteristic is well known, and occurs when the autopilot is in a mode called flight level change, which is abbreviated on the relevant cockpit button as "FLCH." The abbreviation has given rise to a nickname the pilots use for the characteristic, "the flitch trap." After the crash, safety board investigators were told by the specialists who train Asiana pilots that not only were the three men in the cockpit warned about the flitch trap, but that they had also been told that they could expect it while landing in San Francisco because the typical approach at the airport required a fast descent and extensive use of the autopilot in different modes.

But Thomas Haueter, a former director of the office of aviation safety at the board and now a consultant to the airline, said the flitch trap was "a bit of a setup for the unwary."

If the crew had turned off the auto-throttle, he said, a separate system would have kicked in to keep the engines running hard enough to prevent aerodynamic stall. But with the auto-throttle in sleep mode, "there's no protection at all, you've got nothing," Mr. Haueter said.

The Asiana pilots union, in a separate submission to the safety board, said pilots were not trained on this characteristic of the 777.

The airline's submission notes that in 2010, when another Boeing plane with a similar auto-throttle, the 787, was being certified by the Federal Aviation Administration, the F.A.A. raised concerns about the way the throttles went into sleep mode. But Boeing declined to make a change and agreed with the F.A.A. to put a warning into the pilot manuals.

After the crash, when test pilots from the F.A.A. and the airline tried to fly the approach that air traffic controllers had given the Asiana flight, they had severe difficulties doing so while following other rules, according to papers filed with the board.

The airline's submission also acknowledges other errors by the crew. For example, Asiana, like all big carriers, requires that the pilots be on a "stabilized approach," lined up horizontally and vertically, at the proper speed with the flaps extended to the proper degree. If the crew is still making adjustments below a certain altitude, the pilot is supposed to break off the approach and go around for another try. In this case, one of the pilots called out that the plane was too low.

Boeing's submission said that "all airplane systems were functioning as expected prior to impact and did not contribute to the accident." It added that the crew had cues that it should have stopped the approach because of the plane's speed and because the thrust setting was incorrect.

[farrokh747]

http://avherald.com/h?article=464ef64f&opt=0

details here -


quote:

On Mar 31st 2014 Asiana submitted a 46 page document, looking like and easily confused with an official final investigation report, stating that the crew was conducting a high energy approach (from a high and fast posture). At 1600 feet AGL the flight director/autopilot changed to FLCH, however, none of the pilots recalled pressing that or related button, the automation went into go-around mode accelerating the engines for the go around and the autopilot going to acquire the go-around altitude. The pilot flying disconnected the autopilot, called "manual flight". As the aircraft was still high and fast, he pulled the thrust levers to idle, which changed the autothrust function from thrust to hold disabling airspeed protection and automatic wake up function. At 500 feet AGL the aircraft was on glide (PAPI two red, two white) at 135 KIAS 2 knots below VREF still within stabilized approach criteria. During the next 17 seconds the airspeed decreased to 118 KIAS and the aircraft descended to 200 feet AGL, the PAPI now showed 4 reds (below glide). 7 seconds later a quadruple chime sounded, the pilot flying advanced the thrust levers to go around and called "go-around", pitched the aircraft up by about 10 degrees, the engines however were still at idle and the aircraft continued to lose altitude - the stick shaker activated 4 seconds prior to impact - and impacted the sea wall about 11 seconds later.

Asiana complained: "In almost all situations, the B777's autothrottle system supports stall protection and ensures that the aircraft maintains a safe airspeed. However, if a plane is in FLCH mode, the autothrottle will enter HOLD mode if: (1) the throttles are at the aft stop (i.e., engine is in idle); or (2) the throttles are manually overridden. In HOLD mode, the autothrottle is engaged (on), but it does not provide any input to the throttle levers or engines. The servos are disengaged from the throttle levers. Therefore, while the autothrottle is technically on, it is providing no service to the flight crew. ... Boeing makes clear that the autothrottle is an essential tool, stating that "[a]utothrottle use is recommended during all phases of flight," including "[w]hen in manual flight." ... In contrast to the repeated references to the comprehensive airspeed protection provided by the autothrottle system, the Boeing 777 FCOM contains only a single, one-sentence note which can be read to suggest that autothrottle will not support speed protection when in FLCH mode. The note reads: "When the pitch mode is FLCH or TOGA, or the airplane is below 400 feet above the airport on takeoff, or below 100 feet radio altitude on approach, the autothrottle will not automatically activate."

[farrokh747]

actually, this does not make sense to me -

this "filch trap" should only be relevant below 100ft RA on approach - they were already way below  vref at 200 fr RA....    (118kts at 200ft)

What mode selection, or disconnect/disarm combination of the AT system would result in the acft loosing all speed protection?

fc

[Jeroen Hoppenbrouwers]

But Boeing declined to make a change and agreed with the F.A.A. to put a warning into the pilot manuals.

Although this may sound like Boeing simply did not want to put any effort in this "trivial change", mind you that changing the fundamental software in an autothrottle computer means a lot more than changing a line here and there.

This is DO-178 design assurance level A or B software, as far as I know. Touching it means a few man years worth of paperwork to document exactly what you want to change (requirements update and review), an extensive safety assessment, full documentation of expected impact on safety and any other aspect including other documentation, setup of tests to test the effect before and after the change, documentation of required lab and flight tests, a complete planning and then you can submit to the FAA what you intend to do. Note that no software specialist has been involved yet.

Then you make the change, and painstakingly work through all of your A/T computer code to assure that all your requirements, tested by all your tests, do touch all your code statements, conditions, and modified conditions -- basically all words in your code must have been touched at least once during the full test programme. Oh, detail, the A/T is part of the FMC (I am not sure this is also true on the 777 but it is elsewhere). Oh, so we need to just re-test all of the FMC code.

Then we submit it to the DER for review, and file for flight test requests to put the thing through all possible permutations of modes that may get us to the changed line of code. Just a few dozen hours of flying approaches in a 777 with three experienced test pilots at the helm and a few engineers in the back to log every wink of the autoflight system.

The van to take the resulting paperwork to the FAA needs to be a Super Duty variant. Only senior managers having the right stuff will dare to put their signature on it, both on the FMC manufacturer's side, Boeing's side, and the FAA's side. Expect a bit of politics here. And a few careers at stake. Somebody will delay the project until he retires.

Five years later, you are done and you can propose the FAA to roll out a worldwide FMC code update. It will require a worldwide flight manual update, retraining of flight crew and trainers, reprogramming of simulators which do not use the actual FMC for cost reasons, re-certification of said simulators, etc.

Alternatively, you put a "WATCH OUT!" sign in the manual and call it a day.



Hoppie
just a level D guy, and not anxious to get any higher just yet

[CarlBB]

Interesting insight, thanks Hoppie.

[Phil Bunch]

Hoppie,

Your summary was most interesting to read.  Boeing must employ a huge number of people just to manage the review and approval processes for airliners - it's mind-boggling to think about the paperwork required to produce a new airliner!  It must have taken "100,000 person-years" of paper creation to obtain ETOPS approval, for example!

I assume that the intensive approvals for aviation-related hardware and software is a substantial reason why even the simplest airplane part is extremely expensive.  

Would a similar certification process be required for other countries or does every other country just take the FAA's approval as "good enough"?  For example, from distant memory, I *think* regulated medical products can often use an abbreviated or accelerated approval process after they have obtained approval in another major country or region.
-----------------------------------

Some background - I worked in medical imaging systems R&D for a large company for my whole career, and we had to deal with the US Food and Drug Administration (FDA) as well as national and international standards.  However, the products I worked on were not highly regulated, unlike medications or implanted surgical devices.  As long as your products were both low risk and generally similar to long-standing products, there was no need for extensive clinical trials,, etc.  Most of the problems we had to cope with were related to minor but not ignorable problems that slipped through manufacturing QC or that developed under real-world customer use conditions.  Sometimes this caused a very expensive product recall, but this was fortunately a rare event.