Faulty RadAlt brought down Turkish Airliner

[Chuck Ellsworth]

How many of you have flown in the Arab world?

It still gets me when we hear ""landing in 15 min, Inshalla." on the radio

I have and I firmly believe that my belief in trusting Allah and my never failing to include " Inshallah " in all my transmissions and after doing any check is what has kept me safe and still here.

[Jastapilot]

Both autopilot A and Autopilot B have to be engaged to do an autoland,which will enable after a self test,the flare maneuver.If only one autopilot was engaged.the aircraft would not have pitched up(ie flared)you can not do an autoland
in a NG with only one autopilot engaged.So with one autopilot(B) engaged,the only thing that in probability,should have happened would be the auto throttles would have retarded at 27'RA.as per the malfunction.So the aircraft was probably in a dual channel approach.both autopilots engaged.

actually, I don't think that's entirely true. I've seen a single channel autoland. And, boeing has released the info stating that B system was the only autopilot engaged, and it tried to autoland due to erroneous rad alt inputs.

If certain conditions are met, the plane will autoland on one autopilot, and judging from what these idiots were up to, I think the conditions were met.

[MUSKEG]

So what will happen now. Boeing will have to build in a fail safe into the autopilot. This is so freakin stupid that it is beyond words. This is one accident that hangs directly on the shoulders of the crew, no where else. But just watch and see, engineering will have to make these aircraft even more idiot proof so that some of these people that HR identifies as suitable candidates can actually get the job done. Heaven forbid we ever put the blame where it needs to be. This is now the second accident in 3 weeks where the use of the autopilot has lead directly to the crash. The mentality that the autopilot can do it better than crew is true, but garbage in garbage out. Maybe it's time to ban the use of autopilots below 10000 feet. Some would actually have to learn to fly. End of rant.

[Pratt X 3]

The digital flight data recorder indicated, that the autopilot B and autothrottle was being used for an ILS approach to runway 18R. The right hand radar altimeter was providing proper data, while the left hand radar altimeter provided faulty readings of -7 and -8 feet. The autothrottles, using the left hand radar altimeter, transitioned to the flare mode and retarded the throttles to idle thrust.

My comprehension is that they were not conducting an AUTOLAND; just using AUTOPILOT B to fly the ILS. The AUTOTHROTTLES were engaged to maintain the selected speed, however, due to the faulty RADAR ALTIMETER, they went to idle too early. The plan was probably to use the AUTOPILOT and AUTOTHROTTLES to fly the approach and conduct a manual landing. Is it possible to handfly with the AUTOTHROTTLES still engaged and/or land with the AUTOTHROTTLES engaged?

[rd1331]

Is it possible to handfly with the AUTOTHROTTLES still engaged and/or land with the AUTOTHROTTLES engaged?

Yes it is possible. The FCTM suggests disconnecting the Autothrottles when the autopilot is disconnected, but is only a suggestion. And the reason most airlines do not follow that is that the autothrottle does a much better job holding airspeed than a person. Hence why SOP is to bug a speed of 5kts great than Vref with the autothrottle engaged, whereas without it you must add a minimum of 5 knots plus half the headwind and all the gust to a max of 20kts or 5 kts below flap speed. So in the end you could possibly be approach 15kts faster with the autothrottles disengaged than with them engaged. Another great feature of the autothrottles and why you could keep them engaged with the autopilot off is the G/A feature, if you press the G/A buttons without the autothrottle engaged you have to advance them yourself to G/A thrust. With it engaged, one press advances the throttle to a reduced G/A thrust and a second tap advances them to full G/A thrust, normally you only use the reduced.

The aircraft is not certified to do an Autoland unless both Autopilots are engaged. There are multiple criteria that must be met for an autoland to be accomplished. From what I see is they just had the B autopilot and autothrottle engaged doing a normal ILS approach and not an autoland. Even on a non-autoland the autothrottle retard at 27ft.

Fly the Airplane.

[crazy_aviator]

., im calling you on that last remark, and raising you 50 BUDDHA statues !! Honestly, this flight had nothing to do with Allah or Buddha
or whoever ! it comes down to Pilot ERROR ! The thread heading should (likely)read " Faulty pilots brought down Turkish airliner ASSISTED by a KNOWN faulty RadAlt.!!! "

[wooden spoon]

so why was there a pitch up(ie a flare manouver as per the autoland procedure)As in a dual autopilot approach,autoland procedure.The fma f(light mode annunciator) would have given them indications that they were in dual channel by around 2400 ft and below.as well as both autopilot A and Autopilot B lights on the MCP illuimated. ???

[Pratt X 3]

What pitch up? Can you point out where it is written that this aircraft pitched up? The latest report talks about the AUTOTHROTTLES being in the flare mode, not the AUTOPILOT. By that report, an AUTOLAND was not being conducted so there wouldn't have been an AUTOLAND annunciated.

[dji]

does the 737 call out retard out loud or is it airbus only? would be pretty hard to miss

[MUSKEG]

Retard is an excellent discription.

[swordfish]

They had almost 2 minutes to realize something was amiss, click off, and advance the power levers/thrust levers/throttles (whatever they are called in Boeing-speak) and fly MANUALLY out of this mess.

No matter what config the autoflight systems were in they had plenty of recognition- and reaction-time, and failed to respond. The plane was flying the pilots, not the other way round. Regardless of the rad alt error - and they should never have been using it - pilot error is clearly indicated, not Boeing error.

I hope Boeing doesn't come up with some knee-jerk response to an already well-designed and proven aircraft.

[swordfish]

One more thing...the thrust levers actually MOVE in response to autothrottle commands in a Boeing - unlike a 'Bus. So you can "see" them retarding.

[Chris]

Wouldn't the A/P have to pitch up to maintain the glideslope if the throttles went to idle?

(which would explain the rumours of the aircraft pitching up)


While I agree that this was pilot error, due very likely because of poor training.....it's interesting to note that the airlines training is contracted out to Alteon; which is wholly owned by Boeing.

[MUSKEG]

You can be certain that they just came up with a new emergency item for the simulator. But it's the old complacency bug that bit again. Hey we've never had a problem before and we don't certainly don't anticipate one now. You are grilled with ever emergency imaginable in sim training, and you leave there confident in your abilities but also quite certain you will never ever see most of them for real. Again the faulty equipment was a relatively cheap component. The weak link was the human factor. The lawsuits on this one are justifiable and could very well sink this carrier, given the global climate.

[swordfish]

Not knowing what the "experience level" of the crew was (I believe the copilot was doing [equiv to] line indoc) it begs the question of their confidence in manual flying, rather than the issue of their training standard. After all, they passed a ride on type, so their initial competency was corroborated at that time.

i.e. "we can solve this issue by going to selected mode (I am assuming they were in managed), twiddling some knobs, and the FMC will do the rest". You have to have a certain level of confidence and assertiveness (bred from experience), to click off and fly manually at any point in the approach. This is hard to do when you're new to type, of haven't spent a lot of time hand-flying the aircraft.

[Jastapilot]

Both autopilot A and Autopilot B have to be engaged to do an autoland,which will enable after a self test,the flare maneuver.If only one autopilot was engaged.the aircraft would not have pitched up(ie flared)you can not do an autoland
in a NG with only one autopilot engaged.So with one autopilot(B) engaged,the only thing that in probability,should have happened would be the auto throttles would have retarded at 27'RA.as per the malfunction.So the aircraft was probably in a dual channel approach.both autopilots engaged.

actually, I don't think that's entirely true. I've seen a single channel autoland. And, boeing has released the info stating that B system was the only autopilot engaged, and it tried to autoland due to erroneous rad alt inputs.

If certain conditions are met, the plane will autoland on one autopilot, and judging from what these idiots were up to, I think the conditions were met.

Once again, it is possible the plane went into autoland function, and I'd say likely. I still don't understand WTF those guys were doing up there while the airplane flared and stalled a few hundred feet off the deck.

[joco]

Once again, it is possible the plane went into autoland function, and I'd say likely. I still don't understand WTF those guys were doing up there while the airplane flared and stalled a few hundred feet off the deck.

Remember the Human Factors, Decision Making Process the Invulnerability "It can't happen to me" attitude?
That's all I say.
joco

[2R]

Did they have enough fuel for a go-around ?

[joco]

The reports indicate that the fuel was not an issue.

[YCL Boy]

Excellent Video of the Accident in question:

http://link.brightcove.com/services/pla ... 3635223001

[wallypilot]

great video that sheds some light on the happenings in the flight deck, but still....it's hard to believe they let the airspeed decay unnoticed, and that for such a long time they didn't see the anunciation of the AutoThrottle mode "retard". And with 3 sets of eyes. Brutal.

[xsbank]

Wasn't the radalt, it was the pilots. 3 of them, to boot.

"Check speed." "Correcting."

Or, if you are not ready for the approach, don't accept it. Situational awareness was absent. Minor mechanical failure, 20 dead. Duh.

[wallypilot]

"Check speed." "Correcting."

ya, brutal...total lack of flight deck discipline and situational awareness.

[Flying Nutcracker]

A/T can go to RETARD in any A/P mode if the captains RADALT goes below 27'.

I believe this was a single-channel ILS to a manual landing. They were approaching the glide path from above because of a bad vector, strike one. They couldn't configure until the very last moment because they were high. A 737-800 is hard to slow down on a 3 degree slope, not to mention if you are flying a steeper path to get on the slope, rendering you unable to configure. That is, unless you are smart enough to realize the situation and configure early, making it easier to descend to your profile on speed. Strike two. Because they were configuring late, I bet and assume, they were preoccupied with checklist items at a point were all focus should be on the flying. Nobody caught the FMA, which really is easy to forget if you're busy, no hands on throttles, and no monitoring. Strike 3. This sounds more like a perfectly normal approach until they are caught high on a bad vector and then try to save the approach, letting the airplane get ahead of them. It doesn't work very well in a slick jet aircraft! The A/T was definitely a contributing factor, but this has LASO and poor CRM and use of SOP written all over it!!!

The fact that they were not even configured at 800', still slowing down at idle should have been a go around if the statement about their SOPs are correct!

Furthermore... even with a dual channel approach the integrity of the RADALT is not checked, except by visual comparison. The ILS deviation monitor system is checked at 1500' RA, but ONLY if both LOC and G/S has captured, FLARE ARMED is annunciated. Because they did not capture the G/S until about 1300', this test would most likely not have occurred on this approach. If the FLARE is not armed by 350'RA, the A/P will disconnect. Would not have saved the crew in this particular instance.

One thing to note is that FMA will announce a green RETARD when reducing N1 towards idle, and then go to a white "ARM". In this case, it would have stayed green all the way. If you're busy, green usually means good. A bit of a trap in this case...

I will add one more thing regarding RADALT. The A/P will only disengage 2 seconds after LOC/GS capture if one RADALT is INOPERATIVE. In this case the Captains RADALT was "working" in the sense that it was actually reading.


I read this accident as a result of a crew trying to save an approach from a bad vector, leaving them high and fast on profile, not configured to slow down and ultimately 3 pilots preoccupied with one task... configuring. What do you call that... oh yeah... fixation. And I believe fixation is one contributing factor to Loss of Situational Awareness! And of ALL DAYS to have a RADALT problem leading to a premature RETARD by the A/Ts... Swiss Cheese sometimes lines up if you just stare at the holes long enough!!!

[115B]

There is no excuse for waiting 100 seconds before doing anything. Hands should have pushed the throttles through the instrument panel when they moved back at the wrong time. Of course if you cannot fly the airplane without the autopilot, you are going to die.

Training at Boeing (Alteon), I asked the Instructor if he would put his wife and kids on another airline (not Turkish) for whcih he also was doing training and Line Indoc; "I wouldn't leave the flight deck when I was on that airline" was his answer.

Just cause you can buy or lease the airplane doesn't mean you should be operating it.

36 plus years of major airline flying including Line Indoc. Lots of current airline pilots have never had the autothrottle off; many of them are going to have a bad day in a failure situation. When airplanes didn't have functioning autothrotles, pilots could maintain speed as well as todays autothrottles; a good pilot is better than the autothrottle in some conditions; if current.