Asiana flight crash at San Francisco

[OFSO]

'Nother numpty question, how long do those 777 fans take to spool up to max thrust from idle ? I used to ride in a 737-200 cockpit long ago and if my memory serves it was around five seconds on departure.

[Non Zero]

Just four pilots, none of whom was willing to take command early enough. The Cabin staff probably would have done a better job, or a sixteen year old on a computer.

Your being Australian doesn't excuse you from shooting to the pilot community! Having said that you might be right ... but since you were so prompt to shoot try to be as prompt replying my answer

were all 4 pilots in the cockpit for landing? And I'll give an help 2 were seated in the pilot seat! What about the other 2?

[lowvaeater]

FO on 744, promoted to captain on A320, transitioned to 777.
So he has been to KSFO, and is or was familiar with the procedure.

This accident has more news coverage due to having surviver to interview (or to blame?) and it is probably easier to criticize on their flying, CRM skill for their nationality or ethics for some, hence the relatively larger discussion going on.

The A/T servo on 777 is considerably lighter than that of 747 and 737, if you happen to rest your hand on it during idle operation, the servo can have a hard time erecting the lever back up.

Regarding to skill of hand flying, don't be so quick to judge. Their A320 flies some of the more difficult visual and circling approach in the region.

I hope these explains some of the questions you guys have.

[Mimpe]

The situation looked pretty desperate between 30 and 20 secs from impact looking at the speed data.

[Dave Gittins]

There sound to be some similarities to a KAL crash at Stansted fourteen years ago. The Captains defective AI did not show the rate of roll whilst the first officer watched his funtioning AI roll through 90 up to about 150 degrees but was too afraid to question the captains authority.

After the investigation, Air Accidents Investigation Branch of the United Kingdom issued recommendations to Korean Air to revise its training program and company culture, to promote a more free atmosphere between the captain and the first officer.

Korean Air Cargo Flight 8509 - Wikipedia, the free encyclopedia

[Callsign Kilo]

So, thus far we have...

The NTSB stating that the speed was 'significantly' below the assigned fly speed

The NTSB stating that a go around was only executed 1.5 secs before impact

The Cabin Manager stating there were three pilots in the flightdeck at the time of the accident

NOTAMs from KSFO indicating the ILS was U/S however PAPIs were serviceable for RW28L

The weather was CAVOK and the winds were light. It was daylight.

Asiana stating there was no mechanical issue with the aircraft

Asiana stating that the LTC was conducting his first flight as an LTC, he had 3000hrs on the 777 and had landed in KFSO 33 times in this aircraft. The Captain was under training and had 43 hours on the 777. He had landed in KFSO 29 times in another type. Between then they had in excess of 10000 hours

There has been other suggestions of poor training, over reliance in automation, inexperience on type, mode confusion, tiredness, being kept high on profile by ATC. Those and what I will state below may be contributing factors?

I find it unsurprising that people are speculating that this was a cocked up visual procedure which developed into an accident due to poor airmanship. Those within the industry are well versed in the 'issues' that have emulated from flightdecks with steep authority gradients. South Korean operators have came under specific scrutiny in the past. There are numerous accidents to testify to this. This is factual.

Now there's ways of bringing this to the fore. Other methods will have you labelled as a racist, as it appears. However I for one will be completely unsurprised if the NTSB's findings suggest the same.

[Capn Bloggs]

If the crew couldn't see that their touchdown point was climbing up the sight-picture, I'm not certain what assistance a HUD might offer.

On the contrary, I would have thought watching the HUD flight path marker dropping down into the underrun/water would be far more obvious than "their TD point climbing up the sight-picture (aka windscreen)".

[Lonewolf_50]

tdracer

But this crash, along with the Air France A330, show that we still need to train pilots to fly, not just manage, the airplane.

If by "we" you mean the airline industry and regulation authorities, yes.

Every study done, has shown that a managed evacuation is ALWAYS more effective than a panicked rush to the nearest exit.

Thanks for that.
Franzl

The probable obvious cause of this accident seems to be non existent energy management by the crew. There is no excuse that the speed decay in that published amount and time was not recognized until it was too late. ....says an old fighter pilot

Didn't you read a few pages back that fighter pilots are a horrible fit for airline cockpits and CRM?
FWIW, I am trying to find the right files at the "smartcockpit" site to try and understand the "FLCH trap" previously mentioned. If it's a systems issue, it's a training issue, right?

One of the first things said to me when I entered the aviation industry was "It's better to ask a stupid question than it is to explain a stupid mistake". Of course if you are dead, you wouldn't have any explaining to do.

.
I recall about thirty years ago the semi-sarcastic observation that "it's better to die than to look bad." Our safety officer responded with:
"I've never seen a good looking corpse."

Now there's ways of bringing this to the fore.

A sub optimal way to bring such issues to the fore, in a given organization, is to have a wreck. Sadly, that seems to be what gets some organizations' attention where other input does not.

As we used to say in the Navy, with heavy heart as some of us got older ... the NATOPS is written in blood. (That's the operating manual/bible for a given aircraft ...)

Looks like the sanguine penmanship continues ...

A point on CRM and "who is doing what" from the distant past: a couple of decades ago, a crew in a squadron at our base had an engine act up during a training event. The aborted the training event (which was over the ocean) and headed for the nearest field. They came in to do an approach to a runway, rather than a helicopter pad (the bird was SH-60F) and made the approach. With split torques, you retard the bad engine to let the good engine lead. Well, for whatever reason, the PNF retarded the good engine and the bad engine led. Short final, power loss, inside HV diagram bad zone, and a nasty crash. Aircraft lost. Crew badly injured. (PIlots and aircrew).

Cause? Well, the bottom line ended up that a fairly straightforward thing (which all pilots in that squadron had done in the sim many, many times) "who was doing what" went wrong.

Apply that to this accident: a fairly straighforward event, visual approach, and the correct movement of engine power controls (be it by the automated systems or one of the pilots) went wrong.

If you look at the NTSB's pithy extract of speed versus profile, and the target speed (137 knots), at the very least the correct and timely movement of power levers went wrong. (Why will in time be explained, one hopes, by the NTSB's team).

A point that has been raised before in this thread, time and again, also needs to be answered, and that has to do with an industry-wide standard for those carrying passengers: the issue of waving off (go around) a non-stable approach in a timely fashion.

[Agaricus bisporus]

What this all began from was an unstable approach, pure and simple.

We have seen time and time again that if stable parameters are not met at the gates the operator/manufacturer sets then trouble is all too likely to ensue. And without prejudging the issue this event seems to have all the hallmarks of a mismanaged approach being continued instead of thrown away and getting further and further out of hand.

It won't take the NTSB long to figure what happened from the speed/rod/vertical profile we've all seen, the "why?" is going to be discussed for a long time.

[Lord Spandex Masher]

Originally Posted by Lonewolf_50

FWIW, I am trying to find the right files at the "smartcockpit" site to try and understand the "FLCH trap" previously mentioned. If it's a systems issue, it's a training issue, right?

Essentially FLCH will give you flight idle in the descent until the AP captures the selected altitude and the AT will then revert to speed mode. If, for whatever reason, you turn off the AP and try to level off above the selected altitude the throttles will stay at idle and, therefore, your speed will continue to reduce.

[aerobat77]

Reading this thread reminds me why I don't come on here much anymore. Apart from a large number of armchair experts posting a lot of drivel

well, this is the spirit of a open anonymous aviation forum. we should understand that this here does not reflect official investigations or 777 pilots opinion but only a (partly heated ) discussion of aviation fans without any real experience on this type of AC or this particular airport .

you can call it pprune as usual.

nevertheless , respect to the victims should point anybody to be careful in quick blaming by informations based on news reports and internet knowledge.

best regards

[stuminisprite]

thats what i meant captain. if the crew are flying vfr or in bad weather then surely having critical info next to the line of sight would make life easier. any messages could also be displayed in text to correlate to instruments and audio warnings.

[wasthatit]

An approach that is unstable below 1000 feet above airport elevation in IMC or 500 feet above airport elevation in VMC requires an immediate go-around.

Would it be worth considering a 'Mode 8' for EGPWS? Something along the lines of speed VREF to VREF + n, V/S between 500 fpm 1000 fpm at 500 feet otherwise "UNSTABLE GO-AROUND". Like an RA this would be a mandatory instruction.

I wonder how many unstable approaches actually occur each year that do not result in an accident, but are close.

[GarageYears]

Presumably the flights before this one landed successfully without the navaids?

I've seen post after post raising a stink that the ILS was out... has it escaped everyone's attention that the flights prior to this one made it successfully to the tarmac? Unless someone thinks this was the first flight in with the nav out.

[ross_M]

The one point that's still a bit intriguing to me is the nature of the burn. I'd love to see the report on the origins / spread of that fire.

The internal cabin photos reveal a mechanical wreck but not significantly bad fire damage. The roof OTOH.....

Was it really oxygen bottles? Or is the roof area naturally more combustible?

Assuming a fuel leak fire that wouldn't be the shape of the fire damage I'd expect. Then again, I'm no expert.

[Tee Emm]

Just because these were professional pilots qualified to operate the 777 means nothing in the context of this accident.

It matters not that the aircraft was on an unstable approach and if allowed to continue it would almost certainly lead to landing short of the runway. Which of course it did.

What does matter is the realisation that despite the best technical training available, accidents will occasionally happen where experience has shown it that in some societies, it is well nigh impossible to change years of ingrained hard wired ethnic culture...

[Lonewolf_50]

Lord Spandex:

Essentially FLCH will give you flight idle in the descent until the AP
captures the selected altitude and the AT will then revert to speed mode. If,
for whatever reason, you turn off the AP and try to level off above the selected altitude the throttles will stay at idle and, therefore, your speed will continue to reduce.

If one turns off the AP, would not one then begin to move/adjust the power levers manually as a matter of habit/training?

A couple of the posters who commented on the FLCH trap feature mentioned other modes. For my own understanding, getting a better look at the system may help.

[Lord Spandex Masher]

Lonewolf, that's the trap - if you don't turn the AT off.

[suninmyeyes]

Interestingly Zlin has 8000 hours on the 777 and cannot understand why the thrust levers did not maintain the speed. I am sure he has never been in FLCH down to touchdown, quite rightly as that is not a mode you want to be in, but if you were to do that the autothrottle will not maintain speed in that mode if the pitch is manually increased to a pitch above the flight director command.

Having seen this exact scenario of getting low and slow in the 777 expertly demonstrated in the simulator a few times over the years as a warning I would like to make a few comments and explain a few points to those aviation enthusiasts who are having difficulty understanding how it can happen.

You have to have the exact combination of being too high, too fast, rate of descent reducing, speed reducing, autothrottle in hold mode. You have to simultaneously intercept the appropriate 3 degree profile from above and have the speed at Vref and decreasing at about 500 feet. At that point it looks ok as regards height and speed but it goes wrong terribly quickly. If the Asiana had had the stick shaker 20 seconds earlier they could probably have successfully gone around, 20 seconds later and they would have been on the runway having had a heavy landing.

When manually flying the 777 there are still many automatic functions helping you. For instance if you roll into a 25 degree angle of bank there is no need to increase the back pressure on the elevator. If you increase the power the nose will not pitch up. If you are flying a visual circuit you do not need to advance or retard the thrust levers manually at any stage. This is all designed to make it easier to hand fly but it does erode basic flying skills.

The NTSB comments were very much what I was expecting. It is quite difficult to crash a serviceable 777 in good weather but there is one way for a pilot inexperienced on type to do it. Start the approach high and fast in FLCH with 0000 in the alt sel. Do not conform with stabilised approach criteria at 1000 feet, Regain the visual glideslope and Vref at about 500 feet and do not change the mode. The speed and height will drop off very quickly. I think unfortunately they discovered this method.

[Lonewolf_50]

Thank you, Spandex ... a training / systems issue.

EDIT: also, thank you suninmyeyes.