Plane crash at SFO

[mulletman]

well written and well said. there is perhaps a lot of jargon here that most non aviation people won't understand.

as far as the autothrottle issue is concerned. i can't speak directly to your theory because i have never flown a tripple 7. however, i doubt highly that is the problem. it is true that a 777 is a capable autoland airplane, but there are two issues here that make this theory very unlikely. first, pilots generally don't like to autoland airplanes and will only use it in good weather conditions when either the airplane or one/both pilots needs to renew currency. the second factor is that since the glide slope was inoperative, there is probably no way that the auto throttles were engaged at that point in the approach/landing. generally in this situation, the auto pilot needs to be disconnected by the time ILS minimums are reached and then the aircraft is flown visually/manually from that point. and a third point: when the auto throttle/autopilot is disconnected, there is really loud attention getting aural warning that lets the pilots know that it has been disengaged.

i didn't pull out my san francisco plates to look, but i am pretty sure the ils minimum for that runway is 200 feet. most likely the auto pilot/auto throttles were disconnected at this point or sooner in the approach and it was manually flown into the sea wall by the pilot flying the airplane.


there was something mentioned above about engine problems. the 777 has a ridiculous excess of power. each engine produces about 100,000 lbs of thrust. perhaps even up to 120,000 lbs under emergency power situations. what i am saying is that airplane is capable of flying from new york to london on one engine, executing a missed approach at say hethrow airport and flying all the way back to new york for a landing. all on one engine except for the initial take off. what i am saying is that as long as at least one engine is producing full power, an engine issue isn't likely the cause of the accident. i guess this was dispelled anyway by the initial ntsb report.


the training in foreign countries is far and away different from the training we as airline pilots receive here in the usa. lots of foreign countries have the culture where the captain is "god" of the ship. often they never let the first officer fly the airplane and first officers often are apprehensive to speak up when something goes wrong because of a respect issue and also because the captain if often an asshole that will rip the poor guys head off if he says anything that can be interpreted as undermining the captain's autocratic authority. this is why korean air had so many crashed in the 80's and 90's

assuming pilot error, what really needs to be looked at is how well the pilots were getting along in the cockpit. this can be found out by listing to the cvr of the entire flight. it is possible(actually likely) that the pilot not flying saw the impending issue and didn't speak up in a timely manner.

Good points all around. A couple of thoughts: I'm too lazy to break out my Jepps on my day off, but I'd also assume that there is likely an RNAV solution for that runway, and I wouldn't doubt if it was to LPV mins. As to if you can have the autothrottles engaged for that purpose in a 777, I don't know.

Asian carriers in particular seem to be extremely automation-heavy. Like heavy to the point of using autoland regularly. I wouldn't be surprised at all if this was the first visual approach they'd done in a very long time. There have been multiple reports of Asian carriers diverting to other airports when the ILS was out of service, rather than shooting a visual.

CRM in Asian culture is a whole different ball of wax as well, like you mentioned.

Two things that the NTSB have reported have stuck out to me. First, the flight-idle throttle position. Secondly, apparently the briefed approach speed was to be 137 kts... But it slowed to 102 kts.

[flyboyx]

i never turn on the tv so i have not seen this on the news except for when i was waiting around at LAX the morning after the crash. the automation mentality seems to be a good point. if you let the computer auto land an aircraft on every landing, you really aren't a pilot. you just manually take off and then baby sit the autopilot for 17 hours or what ever the length of the flight. 10,000 hours of fight time doesn't mean shit if you never land the airplane. that actually takes some skill.

every time there is an accident, the events will be picked apart and rehashed to the n'th degree. this is how the industry leans from its mistakes. usually the corrections that are made benefit the entire industry and make it just a little bit safer. as you mention getting the airplane as slow as 102 kts when you should be flying at 137 kts is just an unbelievable error. either the pilot's attention was diverted to something other than the business of flying the airplane or he simply didn't have the skills to scan the instruments properly and make the adjustments in pitch and power to stay out of such an unbelievably undesired aircraft state.

in thinking about this, my theory is leaning toward the 10,000 hour captain not having the skills to fly the airplane with the auto pilot off or even have the skills to land it on a runway using raw/visual data. i would venture to guess that as soon as he turned off the autopilot/auto throttles, everything went to shit. i believe the problem was perhaps compounded by the first officer being too meek to speak up of the gravity of the situation they were in or have the assertiveness to take command of the airplane when he saw the captain fucking it up. actually, even worse, captains probably never let the poor bastard fly the the airplane and he probably never had the experience or skills to take control of the situation in his 2400 hours of flight time. seriously.

here in the usa, it is unlikely that with the training and crew resource management classes that this would happen today. we still have an asshole or two but for the most part, this has been repaired by the training we receive. there are a lot of instances from the 70's and earlier where situations like this did cause accidents and loss of life, however. i personally think the industry here in the usa has done a nice job of learning from and fixing its mistakes. at my company(and all of them i am aware of here in the usa) we actually take turns every two flight segments as to who is flying pilot and non flying pilot. this gives the first officer valuable experience in flying the airplane as well as the captain staying used to running checklists, talking on the radio, and performing other not flying duties in order to complete the mission.

the only way we will know for sure is to wait for the ntsb to publish the findings about this incident.

[thearkitekt]

but the PAPI was working fine.

this is what i had been wondering since i saw the report. First thing I did was check to see if the GS was working, and it wasnt, didnt see any notam about the papi though.

I also question the regularity of the asian crews shooting visual approaches. I'm not a 777 pilot, but I do know of a few, and I know landing currency can be a problem. They fly such long segments that they only have a few trips a month, and there arent many landings between them. Those pilots i've talked to usually keep their currency in the sim, which I would probably bet at least one of those is a GS out visual landing (pretty common in all sim training ive done). These are american standards though, not sure if other countries follow our lead or not.

Mulletman - I just checked and there is an RNAV approach to the runway with LPV mins. my only thought here is that these are most likely very rarely used and could have had errors in the db no one caught? I very much doubt this as they were well below ref, but it could be a factor i suppose if it was coupled with auto throttles and it thought they were high. I still dont understand how you could ignore the visual side of things, but I guess thats why there was an accident.

[mulletman]

Mulletman - I just checked and there is an RNAV approach to the runway with LPV mins. my only thought here is that these are most likely very rarely used and could have had errors in the db no one caught? I very much doubt this as they were well below ref, but it could be a factor i suppose if it was coupled with auto throttles and it thought they were high. I still dont understand how you could ignore the visual side of things, but I guess thats why there was an accident.

One interesting thing I just thought of:

When I was landing on 28R at SFO a few days before the accident, the ILS was out of service then as well. We loaded up the appropriate RNAV as a backup to our visual, and went through the normal 'activate approach, cancel heading' BS that you have to do with Universal-made FMS's.... and the approach was way off. Like full-scale deflection off. We just ignored it and landed, but I sure would like a second shot at that RNAV, just to see if I borked up the programming, or if we were experiencing some sort of predictable error with the approach.

[hoveringuy]

I used to build my own approach in the FMS by linking a 3 degree glideslope with the actual mag heading of the runway to a GPS fix at the touchdown zone.

I went into fields where NOTHING was working (think Africa...) and even though it was only a DC-9 I had crosshairs to backup my visual.

I can't get the stereotype of a 4.0 student that's booksmart but can't tie their own shoes out of my mind.

[mulletman]

I used to build my own approach in the FMS by linking a 3 degree glideslope with the actual mag heading of the runway to a GPS fix at the touchdown zone.

I went into fields where NOTHING was working (think Africa...) and even though it was only a DC-9 I had crosshairs to backup my visual.

I can't get the stereotype of a 4.0 student that's booksmart but can't tie their own shoes out of my mind.

Right. We're allowed to build our own FMS approaches to back up a visual, but it's been since training that I did that. It's so much easier to just look outside and fly the airplane, haha. But I can definitely understand how useful that would be, especially with the type of airports you're talking about.