Turkish airliner crashes at Schiphol

[MU3001A]

bobcat4

If there is a configuration nobody would use, and if that configuration could be dangerous, I would opt for a removal of that configuration. Single channel approach is of course fine. But could anyone please tell me of what use the auto-retard is in that configuration? Being to lazy to pull the levers to idle is not an argument (IMHO).

Very well put.

No additional warning/checking system required. Simply disconnect the unneeded A/T retard function for other than autoland approaches. Problem solved.

[Fred Bound]

finfly1

"confess to being rather bored by all this poring over and implied criticism of the aircraft systems. It seems that one Radio Altimeter failed - so what?"

So it told the a/c it was on the ground, reduced thrust and crashed and killed people.

The point that you seem to be missing is that the failure of a Rad Alt and it's subsequent implications shouldn't have caused this accident.

The RadAlt/AutoThrottle anomaly and subsequent loss of airspeed should not have ended the way it did had someone been paying attention to what was going on.

That's the point.

[Whippersnapper]

Quote BOAC:
Here's a thought for the REAL 737 drivers out there. If, as it appears, we really do have to cater for lack of speed monitoring, why not push for an immediate change to min speed rev to allow it on a coupled GP, down, say to 100' (as I understand AB) or to FLARE active?

Of four airlines I have flown the 73 for (300-800), only one permitted the use of the AT during the approach after disconnecting the AP, and that was with manual inputs by leaving the AT on but deselecting SPEED on the MCP. The idea was that it would continue to give underspeed protection and the automatic GA thrust settings if required, but in the ARM mode would otherwise not interfere with manual thrust setting. The other three airlines forbid that use of the AT, rigidly decreeing that it must be completely off when manually flying. The reason I was given by the first company was an accident along the lines of this one, involving a stall and crash. I don't know any of the details, but it looks like a similar incident (though perhaps nothing to do with the RA) happened before.

As for redesigning the system as per your question, I suppose it's because that altitude would have to be input by the same RA that caused the glitch in the first place, or baro height above the threshold, which is awfully close to Cat III minima if there's any altimeter error (cold corrections too) and is at the Cat II minima. I'd want to see whether the RETARD mode was armed or not before getting so close to the tarmac. Another possible reason - I tried a simulated Cat II autoland at an airport carrying out Cat II trials, and the topography caused a bit of trouble: at 500'RA, we were only 180' above the threshold. Perhaps that could cause the AT to be unable to arm or engage RETARD in your scheme?

[Profit Max]

bobcat4:

Come on... Going from +1950 to -8 feet in say one second is 117480 feet per minute. Well, I would say that is a very impossible sink rate. Isn't that supersonic? Anyway, it not a normal condition. And this sink rate is based on the assumption that the RA outputs its measurements once every second. The real output rate is probably a lot faster, which makes the sink rate even greater.

The simple solution is to have the AT remeber the last reading, and the timestamp this was read. Then it should calculate the sink rate. If it's close to supersonic or some other far out speed it will simply flag the RA as unusable/faulty. What is the aircrafts envelope regarding sink rate? That would be a good limit to set for a sanity check.

The change in radio altitude is not equal to the sink rate! Think of terrain - tall buildings, cliffs, hills, etc. If steep enough, this could lead to very high "sink rates" according to your calculation.

Guys, you essentially want aeroplanes to fly without pilots. I am not saying that it will be impossible forever, but consider this: There are only a handful of underground/tram/light railway systems worldwide that are driverless. These systems are magnitudes less complex (especially in the case of a failure) than an aeroplane, and still, they are not in use everywhere.

[grebllaw123d]

Kpt40,

You have a point - I read the report again and it states that the descent was performed on autopilot - nothing about the approach!

Earlier posts have indicated that the aircraft intercepted the LOC at 4.8 NM at 1900 ft.,i.e. about 300 ft. high.

I remember that there was a discussion many posts ago about intercepting the GP from above on autopilot on the 737.
As far as I remember, it is not possible!??

737 drivers please correct me - if I am wrong!

brgds

[BOAC]

Originally Posted by whipper

the same RA that caused the glitch in the first place,

- yes, I'd seen that as a problem in this accident, but it was a suggestion for all approaches. Would it not improve overall approach safety?

The 500/180 you quote should resolve itself over the threshold.

On another post, I also cannot really see why Boeing designed A/T retard for single channel ops.

[GXER]

Is it reasonably to be expected that professional flight crew will recognise and compensate for autothrust malfunction which (if not corrected) results in airspeed decaying below that needed to sustain flight?

It is not <<reasonably to be expected >>... it is a basic requirement.


Simple question. Simple answer.
True. Hopefully I've given it

NoD

Yes - thanks. I suspect it's also a "basic requirement" that a train driver should not blithely ignore a red light. If he does, a crash is likely to follow.

It is possible (at risk of unintended side effects) to build in safety mechanisms that protect against even the most fundamental operator errors (I think that has been done, at massive cost, in some cases). Whether the incremental reduction in risk justifies the cost is debatable.

[bubbers44]

Hopefully an RA failure won't be an excuse for a professional crew to crash an airplane. At least not at 2000 ft. If it failed and caused the throttles to retard it should have been a nonevent at worst case requiring a manual approach. Hopefully everybody flying can do that.

[africangenesis]

autopilot should warn?

Safta's sim and Colgan 3407 both seem to involve the autopilots having the AC in extreme trim before stick shaker. Presumably the AP "knows" when it is having to going to extreme trim, is an opportunity for a warning being missed? What should the trim threshold be? Should it be selected at a point at which an attempt a full power recovery might increase the stall? Or are extreme trims too normal for warnings to be anything other than distracting?

A narrowly focused human pilot having to fly without throttle woiuld presumably speak up, "Hey, I'm having to go to extremis here, HELP!" The autopilot would seem to have all the info it needs to issue this kind of warning.

[MU3001A]

BOAC

I also cannot really see why Boeing designed A/T retard for single channel ops.

Perhaps they didn't but just looked to the simplest engineering solution without regard to malfunctions like this one that could be covered by MEL's and SOP's. Malfunctions that so many commenting here believe should be the sole preserve of pilots to rectify in the heat of whatever moment the failure/malfunction occurs.

[GXER]

@ant1

Simple? answer:

Crews should'nt perform, among other things:

CFIT, land gear up, collide with other aircraft, you name it

Still regulatory bodies decided that we should carry GPWS, gear horn, ACAS ... and that they have to be conveniently designed and reliable.

Those are reasonable expectations; but, in the hierarchy of "essential conditions for flying", sufficient airspeed is, surely, top of the list.

[Feathers McGraw]

Very interesting reading Safta's comments about the 737-300 SIM.

Now since he mentioned the way that the trim operated, gradually speeding up as the AP tried to follow the GS, the trim wheel would be making a racket wouldn't it?

So, assuming the same thing happens on the -800, and the trim actuation is as noisy, how could this have been missed or ignored by 3 pairs of ears?

[ant1]

I agree, simply add speed to the above list. You can also add the baro alt comparator to the items the 737NG carries, probably for some good reason.

Now I think there seems to be a good reason to consider the need for a RA comparator.

Whatever the crew implication, it bothers me to think that possibly this simple feature would have saved many lives on that particular day. Before all this discussion I would have thought that disagreements between RA were monitored and classified as failures.

WE are humans, do I need to remind you examples that you very well know:

AA failing to retract speed brakes in Colombia during pull up

I could go on forever. I will never say never, will you?

[vino]

If the a/c was on A/P, why did it overshoot the g/s intercept by 2/3 of a mile(200 feet).Based on my ruff numbers it had to hit a 4 degree glide to catch the g/s. Are we sure that the a/p was on?

May well be using V/S as pitch mode, with GS armed (or maybe not). That would allow LLZ tracking but explain vertical offsets.
Typical technique to capture GS (from above) would be V/S, as Lvl Chg won't command sufficient V/S, particularly while decelerating.
Capturing GS from above is not recommended (I think FCTM refers) but not generally forbidden by most carriers as it is too restrictive when less than ideal vectors/descents are provided.
I think we (pilots) are all fairly familiar with this scenario

[RatherBeFlying]

Should A/P Trim authority be limited?

Safta's sim and Colgan 3407 both seem to involve the autopilots having the AC in extreme trim before stick shaker. Presumably the AP "knows" when it is having to going to extreme trim, is an opportunity for a warning being missed? What should the trim threshold be? Should it be selected at a point at which an attempt at full power recovery might increase the stall? Or are extreme trims too normal for warnings to be anything other than distracting?

Perhaps it would be better for the A/P to run out of available trim and allow the a/c to descend below cleared altitude or g/s or selected v/s than allowing trim to be at a point where full power will result in an uncontrollable pitchup to stall.

There are of course problems inherent in allowing an uncommanded descent, CFIT and traffic conflicts come immediately to mind.

I suspect the certification authorities will be revisiting these issues.

[GXER]

I could go on forever. I will never say never, will you?

No. But - possibly (working from the comments of many professional pilots on this thread) - this accident includes features of crew reaction (or inaction) that are so difficult for professionals to explain that it is reasonable to discount the risk/probability of repetition. That is not to say that systems safeguards should not be enhanced, if that can be done efficiently.

Is it not possible to contemplate that a single accident (possibly this one) may include components that really are, in reasonably foreseeable circumstances, massively unlikely to be repeated? Put more succinctly - "hard cases make bad law".

[Flyinheavy]

@Kpt40

Based on what You wrote here, the crew would have had to fly a G/A as the app was not established when passing 1000' AGL in IMC.
You could go on discussing this terribly accident forever, it will always come back to the same: the crew did not do what they were supposed to, namely FLY THE AIRCRAFT.
Thinking of how to discriminate between a 'good' and a 'bad' RA also does not lead to anything, unless You put a 3rd RA (like on the 47) into the system.

To all of You who hold Boeing responsible to create more sophisticated automation: what are You talking about? Are You aware that any of the highly computerized A/C of today present more or less the same number of software errors. Means the more software You put into a system, the more software problems You will create. To my opinion we are far away from beeing able to create a tecnical system that will be 100% failsafe.

Besides, isn't it the reason that we sit in 0L and 0R as pilots to prevent that things end up in an accident when any of the systems fail? To my philosophy we are paid to provide air transportation using our sound judgement exactly for situations like that, the day by day ops is already leading keen young airline executives to call for downgrading our profession to more or less system operator status. Hudson River was a splendid example for what I mean, that crew earned the salary for all US Airline crews for a year. I would also say it is not missing sensitivity to call it a gross error if a crew did not acomplish the task that they were supposed to.

As to reaction from the Turkish side, I would compare it to the sick reaction of Brazilian authority when prompted to improve ATC, this in connection with the findings of the 2006 midair. For those who do not know, Brazil ATC is in the hand of military, looks like this is painting a certain picture.

I think it would be a high time for international pilots association to attack these 'cultural' problems in a more agressive way to the benefit of the aviation industry.

[YRP]

I think software developers reading this thread, and people who have suddenly decided they could make software developers, should understand that far, far better software developers than you thought long and hard, and worked on this system design, and developed it in the light of experience, and have produced a system that has worked fine for all the world for 25 years.........except for 1 flight.

Rainboe, if you look closely, I think it is not the software developers or system designers but those who think they could do it better ("how hard could it be") who are suggesting the system should be changed.

As you say, anyone who has done system or software design for a fault tolerant system knows just how hard it can be to diagnose faults reliably. It's easy enough to say that a couple lines of code would have fixed this specific situation. But it is much harder to spot all the situations that wouldn't be fixed and would be created by the change.

[cargun]

Would TOGA also be fed by the faulty RA 1 in this case? It has been stated in the Boeing manual that the RA data is used for TOGA calculations. Any idea?

[Flyinheavy]

@cagun
You can apply the TOGA anytime and the AFS will react accordingly.Except when you already deployed the reversers.
Means even after touch down on a CAT3 autoland.
Fh