Asiana flight crash at San Francisco

[dba7]

7 of the 12 cabin crews were KNOCKED out unconscious.

2 were ejected (not sure if they were one of the 7).

And on top of that, 2 cabin crews were pinned by escape chutes (I read TWO) that were inflated inward. Not sure if those 2 were inflated intentionally or not but the fact is 2 were inflated inward and pinned 2 cabin crews. Again not sure if these 2 were knocked out or not.

Despite these, no passenger was left behind inside.

Some passengers stated that no crew appeared and had to take matters into their own hand. Well now you know why.

[roving]

Yet according to cockpit voice recordings, Lee and his co-pilot did not say a word to one another about their increasingly dangerous situation until less than two seconds before the plane’s tail section clipped the seawall at the end of the runway, slamming the fuselage to the ground.



Asiana Airline pilots 'stayed silent until two seconds before crash' at San Francisco airport - Americas - World - The Independent

[suninmyeyes]

A few more comments.

Taking out the autopilot at 1600 feet was entirely appropriate for this long haul flight. Making small corrections in pitch and roll via the automatics has a lag and is not easy on a visual approach. It is easier to disconnect the autopilot and point the nose where you want it.

The 777 is a large aircraft but quite light on the ailerons and it is common for people new on type to over control a little and take a bit longer to nail the centreline.

Although we are allowed to build an approach in the FMC for approximate glide slope guidance we are not allowed to fly the approach in VNAV unless it is in the database.

In thousands of hours on the 777 and hundreds of approaches I have never flown an approach where the autothrottle has failed to maintain the Vref but then I have always been aware of the flight mode I was in. The only time I have ever disconnected the autothrottle in flight was for a TCAS RA. We are not allowed to practice autothrottle disconnected approaches in the 777 as Boeing do not recommend it. We do practice them in the simulator though.

In the scenario described with V/S of 1500 fpm the autothrottle would have maintained the speed at Vref plus 5 knots once the speed had come back to that point. So at some later stage the mode must have been changed to one where the autothrottle stayed in "hold" and maintained idle thrust. FLCH or VNAV SPD would do this. Or as has been mentioned earlier if the handling pilot is resting his hands heavily on the thrust levers this death grip could prevent the thrust levers from opening up.

The low speed warning on the Eicas would still work, even if in FLCH, but probably came at about 150 feet when with the low speed and rate of descent and with the drag of flap 30 and gear down it would have been too late to be of any use.

As in all accidents there are a number of factors. I think the most significant one and the main technical cause of this accident is failure of the autothrottle to maintain the target reference speed for as yet undetermined reasons and the failure of the operating flight crew to realise until it was too late.

[Greenlights]

Guys, I really know that the explanation of a crash is complex.
but come on...

all 3 did not check the profile, all 3 did not check the speed, all 3 did not check vertical speed, all 3 did not check high nose up attitude... what else ?

so, in this case, please tell me : what the hell are they looking at ?

Next time, they should invite a passenger (instead of another pilot) in the cockpit, he may notice something at least.

if just one automation does not work and pilot can not notice and fly the plane...well...let's pray that engineer build perfect automation since pilots are useless in the worst case ?

[Airbubba]

The fact that the approach was not thrown away at 500 feet for not being stable is of concern. 'Hundred Percent Please', #1398, tells us

Several U.S. carriers have gone to requiring stable approach criteria (not 'approaching stabilized') met by 1000 feet agl even in VMC conditions. Anybody know what the actual current policy is at Asiana?

[Mr.Bloggs]

I fly wide bodies to and from the USA. Landing off a visual at the end of a 10-11 hour flight is not easy, period. That's part of why we earn biggish salaries. US ATC do often instruct STUPID 180 to 5 requirements. I simply don't comply and slow down to approach speed by 4 miles at latest, to ensure a stable approach. English is my mother tongue, and I find the US staccato and non-standard communication difficult to understand. And they do not like to take no (negative) for an answer. Which for the Korean guys cannot have been easy. The command gradient at Asiana sounds over-steep as I have already written. And the lack of provision of a fully functioningl localizer sounds like someone at SFO and the authorities have a lot to answer for.

There has been a lot of nonsense written by people who do not fly wide bodies long-haul for a living. It is sometimes excruciatingly tiring and technically difficult. Even for the most accomplished of us. Sadly the public perception of this type of work is based on ignorance and media hype.

I am sure this accident was avoidable, whether or not the A/T was functioning. However the margins are slimmer than many of you would wish to believe.

[Knot Apilot]

Originally Posted by jolihokistix

Meanwhile as the debate rages here, the Chosun Ilbo and the Korea Herald seem to be, perhaps quite naturally, suggesting mechanical failure:
Automatic throttle faulty in crash: pilot-TheKorea Herald

I think this is perfectly appropriate as I think the function of auto-throttle will be a major, if not THE major contributing factor in this accident.

In my opinion, auto-throttles are dangerous. Either one of the two pilots should be manually and personally in charge of the thrust levers during landing, unless doing a fully automated. All on, or all off!

Originally Posted by dba7

7 of the 12 cabin crews were KNOCKED out unconscious.

2 were ejected (not sure if they were one of the 7).

And on top of that, 2 cabin crews were pinned by escape chutes (I read TWO) that were inflated inward. Not sure if those 2 were inflated intentionally or not but the fact is 2 were inflated inward and pinned 2 cabin crews. Again not sure if these 2 were knocked out or not.

Despite these, no passenger was left behind inside.

Some passengers stated that no crew appeared and had to take matters into their own hand. Well now you know why.

Aye. The crew acted extremely professionally and heroically here. Tremendous job. They deserve plaques!

Even the captain had a fire axe in his hand and was chopping down inflatables - he didn't dwell on any possible mistakes in the flight deck and jumped immediately to action.

[Al Symers]

Hypothesise all you want folks, until you have worked in a Korean cockpit you will never understand the weird dynamics. This is not the first accident caused by culture/political correctness, nor will it be the last.

There is no Korean term for common sense.

[TOTitan]

"And about the low-skill pilot with thousands of hours, I'm reminded of the American pilot who was in command of Air France 447 at the time of crash"

It was interesting to read a Korean-Americans outlook on this. Everything you wrote was good except this. AF 447 had an all French crew.

[aston7]

Korean FOMs (similar to KAL and Asiana) state:
Stabilized Approach Criteria


All flights shall maintain the following conditions at 1000feet HAT in IMC,

500feet HAT in VMC.

• Complete Landing Briefing, planned Landing Configuration and 'Before Landing Checklist

• Target airspeed in the tolerance and Power Setting to maintain the airspeed

• In Trim on Correct Lateral/Vertical Flight Path

• The rate of descent, not greater than 1000fpm

All above is a crosscheck procedure. If NOT STABILIZED at 500f AGL - "GO AROUD" callout from PM! If one of the parameters not within the limits - mandatory callout - "SPEED" "SPEED" "SPEED" !!!
Of course we know all this would work in a proper CRM environment.

[MountainBear]

@ aa5bpilot

Assuming that the data behind the charts is good, that is some fascinating work. Since the difference between the July 4th flight and the July 6th flight starts at 4 NM out it seems as if the NTSB would want to talk to the pilots of the July 4th flight to find out what those pilots were doing...thinking... differently than the accident flight because it seems that moment is where the accident crew got behind the airplane.

Taken holistically, what leaps out to me from those charts is that the difference between a hard landing and an accident is a game of inches, a matter of a few brief seconds. The accident crew was off but they were not off by much.

[MPN11]

Another question, from a full-career ATCO.

I understand that the airspace is complex, but who ever devised a visual approach procedure that initially places aircraft above the notional glide path and then requires a 13 (?) mile scrabble to get on glide path whilst trying slow down, configure, stabilise etc.?

[SASless]

There has been a lot of nonsense written by people who do not fly wide bodies long-haul for a living. It is sometimes excruciatingly tiring and technically difficult. Even for the most accomplished of us.

Yes.....it sure seems so! There also seems to a bit of the same quality material written by some who fly wide bodies long haul too.

Please do explain how a functioning Localizer gives one Glide Slope data will you?

[Faire d'income]

the NTSB reported they indeed were asked by SFO ATC to maintain 180 to 5 miles

In a 777, while doing a visual, >300' high, after a long haul flight?

Add in anything else such as a full load or tailwind aloft, or a minor distraction such as a new type, and anyone would be at risk of a messy approach, in any airport.

All those pointing to a go-around as the blatantly obvious answer are of course correct. That can't be argued with in hindsight.

This will be all the brilliant desk pilots answer too.

However, put your mind back to when you are still on the first approach, following such a go-around, which of the following changes/improves?

Still a 777, still a visual, still probably the same daft ATC speed instructions, still after a long haul flight but now + a go-around. If there was a tailwind and/or a full load they won't change either. Your mind may try to persuade you that it will simply all happen again.

Before the hindsight, this weighs heavily on the mind too, especially when you are tired.

[Airbubba]

The accident crew was off but they were not off by much.

Uh, I'd say they were off by quite a lot.

Four red PAPI's and 30 knots slow on short final is not something most of us will ever encounter in our careers of international widebody flying.

[Mr.Bloggs]

A functioning localizer does not give a glide slope but allows an automated approach to be performed, albeit to a higher minimum than a Cat 1 ILS. A visual approach is rarely performed using full automation inside of 5 miles.

[deepknight]

Well below V-ref, go-around thrust asked of two powerful under-wing engines - result: major pitch up that only serves to make things worse?

[WillowRun 6-3]

Robotics, and/or Automation

Chaps (I'm legal counsel, always wanted to see my name in In the Front Office in Av Week, but that's for a different board - I posted a ton on the comments sections on the Wall St J articles on the 787 battery deal and grounding - & point of this is, I make no claim to "getting" the flight dynamics or piloting dynamics. Yes, I googled CRM & FOQA and the tech discourse on flying approaches could give some atty's FLCH Trap for lawyers (which is when a lawyer stops thinking that she or he knows it all). Anyway.)

Two comments. One, relative to dissemination of info, by NTSB chair. Agree that advent of www & social media as primary communication format(s) has greatly, make that very greatly, accelerated matters. Evidence: when NTSB released the prelim factual report on the JAL 787 batt incident at BOS, the report had links to everything- right down to the interviews of the cleaning crew who alarmed the batt burning. Imagine that level of info accessibility to the public, everywhere where there's a www connection, in the equivalent length of time after a Dutch pilot drove a '47 down the runway a bit too soon at Tenerife (yes, I read the entire report, back then, in "sole private sponsorship and independent funding status"). Sometimes a look out the info dissemination window is a good slope check too.

Second, I could not bear to read Holman Jenkins column in the Journal today. He's the cat who wants robots to fly planes, you know? Is there a thread dedicated just to his analysis?

[Etud_lAvia]

For what it's worth, I remember learning about 11 seconds as a "rule of thumb" for a high-bypass to spool up from idle to TOGA. Obviously, this will vary among engine types.

Of course, in a stabilized approach, it would almost never happen that engines reach idle, so throttle response would usually be a lot quicker.

[wiggy]

I think this is perfectly appropriate as I think the function of auto-throttle will be a major, if not THE major contributing factor in this accident.

Maybe..I think on current "evidence" crew performance will be at the top of the list (no surprise there) but.....

Once upon a time when I was converting to the 777 a wise trainer said to me " great aircraft, but because of the full time auto-throttle you really have to work at keeping the speed from your scan and watch out for the FLCH trap"....

Now he was aware of the FLCH trap, many people in this place knew of the FLCH trap (and can quote FCOM 2 chapter and verse) ..it was a "known, known", a bit like the AF pitot probes in icing.... Now that to me begs at least two questions:

1. Is the FLCH "trap" or the autothrottle "going to sleep" somehow an unavoidable characteristic of the autothrottle logic or, if it isn't

2. Shouldn't err, somebody, have had another look at the software and:or hardware?

Just asking?? Opinions, info? Please don't shoot the questioner.