[layman] It seems bizarre to me that in a highly automated environment such as the avionics on a modern airliner we are feeding different radalt values into different systems, allowing them the opportunity to make decisions on the same manoeuvre based on different height data.
Can systems be designed so that both radalts feed into the same system and that a single height is derived from that data? After all, the aircraft only has one height AGL. Comparisons can be done based on the individual values within set tolerances and any mismatch flagged to the crew or to systems so that the feed from a faulty source can be eliminated. All aircraft systems are then singing from the same song-sheet, as it were.
[/layman]
In the cockpit 3 people were present, the captain, seated left front. To his right was seated the F/O for whom this was a training flight. (The F/O held all licenses/qualifications) Another extra F/O was seated in the middle of the cockpit.
For your info, The Trainee as everybody calls him is a Fully Licensed and qualified Co-pilot who has passed all of his tests. The only thing he has not is Line flying experience. What he was doing on this flight and what hundreds of other pilots are doing every day is Line Training. Getting training/experience on the line under supervision of a Line Training Captain. The other Co-pilot on the extra seat in the cockpit (Also called jumpseat) is there for safety reasons. He is the so called Safety Pilot. If the Captain would be unable to perform his duty's he should be able to safely land the aircraft.
Why this safety pilot has not intervened I do not know. This might/could be a Cultural problem though.
Most misinterpretation of the last two pages stems from the sentence
"the authrottle went to "retard" mode and the throttles then stayed at idle"
Most of you mean that it went somehow into an autoland mode.
This isn't necessarily so. AT can also go into retard mode if speed demand is lower than actual. This is the case in LVLCH mode or VNAV approach. 737 rated people please confirm. They were on the GS but that doesn't mean that their AP was tracking the GS! Two different things! If you check the ADS-B data you can see that they never accuratly tracked.
The most probable explanation is that they just forgot that they left the engines in idle. They were doing the checklist and maybe discussing other things. Get real, guys. See the obvious!
And again, the only thing they did wrong was not doing a Go-Around at 1000ft.
The press seem to be touting Rad-Alt failure as the primary reason for this crash..... when it was only a contributory factor. As a Training Captain you have to be capable of doing it all yourself under the arduous condition of actually having someone next to you who is working against you. (It would be easier for the RHS to be empty!)
For whatever reason the approach becomes rushed - I can understand that with the enthusiasm of the Schipol controllers together with such low ILS platform altitudes that can very easily happen.... Anyway the approach is rushed, the Training Captain becomes overloaded, maybe he takes the thrust into his scan at some stage but it is at idle- just where he wants it during a rushed approach.
They continue down the slope- maybe he noticed the RA1 was reading rubbish but they're not Autolanding off this so it should be no big deal. At some stage it all comes together and he sits back and breathes a sigh of relief - he has averted yet another GA from a rushed approach. Unfortunately the thrust is at idle - maybe obvious from the FMA, maybe not.
Whichever way it happened the approach never became stable and should have been thrown away long before the stall warning sounded.
Why did the TC not intervene earlier? I have sat and watched my more inexperienced colleagues make their mistakes in the RHS, some of them have been enough to give a lesser mortal an apoplexy. Perhaps the TC was incapacitated ?
Yes they should have picked it up, but this is not a classic 737 it's a new NG....
What sort of modern airliner design permits a single benign failure of an autoflight system to send the aircraft towards a fatal stall if not picked up?
Was said this happened on two previous flights according to the FDR. Guess they will be looking at this to see the previous crews write ups and what actions were taken concerning this R/A. This R/A failure was only a contributing cause to this accident. Better monitoring and situational awareness could have avoided this. Three sets of eyes and no one noticed the airspeed dropping well below VREF.
The Dutch Safety Authority says the plane was landing on automatic pilot and the problem with the altimeter, a device that measures altitude, led to a loss of airspeed before the crash.
The Boeing 737-800 carrying 135 passengers and crew came down in a muddy field less than a mile short of the runway at Amsterdam's Schiphol Airport shortly before it was due to land on February 25.
Chief investigator Pieter van Vollenhoven said the aircraft had twice before experienced problems with its altimeter.
Boeing has been instructed to warn clients, he said.
At 1,950 feet "the airplane's left radio altimeter suddenly registered a change in altitude" of negative eight feet.
"It didn't only register it, but passed it on to the automatic steering system," Mr van Vollenhoven said.
He added that it was not unusual to land a plane on autopilot.
According to conversation recorded between the plane's captain, first officer and an extra first officer on the flight, the pilots noticed the faulty altimeter but did not consider it a problem and did not react, he said.
Gas to the engines was reduced and the plane lost speed, decelerating until, at a height of 450 feet the plane was about to stall, and warning systems alerted the pilots.
"From the black box it appears that then the pilots immediately gave gas, full gas, however it was too late to recover," Mr van Vollenhoven went on.
The pilots had been unable to see the runway at the time the plane began its descent due to weather conditions - cloudy with light rain.
The plane fell into a freshly plowed field, striking the ground tail first and breaking into three pieces.
Those killed in the crash included five Turks and four Americans.
Last edited by brandtzag : 4th March 2009 at 15:20.
The 737 does a single channel autoland as smooth as a dual channel autoland. You just get no mistrim, no FLARE announciation and you don't have the needed redundancy, but it works
System behavior in this failure scenario may be considered a contributing factor, but nothing exempts 2 qualified flight crew and an aviation professional in the middle staling a 737NG on final. This is nothing like the Everglades accident; the P1 inadvertently disconnected the autopilot, the aircraft was under manual (read: no) control. the commonality is lack of situational awareness.
And re redundancy and failure detection: if you can't decide from two, you need three.
Absolutely.
If my understanding is correct, the RA's are not talking to each other. Maybe they will in the near future?
One can hear the cogs whirring in the wider community already....
If only the aircraft had more automation, it might have been able to assert itself when the pilots failed to assert themselves.
But wasn't it a failure of the automation that started the whole process in motion?
The system is only ever as strong as its weakest link, be it man or machine.
The accident is a classic example of why pilots are needed. In fact, maybe a similar situation had occurred and been successfully overcome by the very pilots who placed the defect in the Tech Log previously?
It is just that these successful outcomes are never reported.
Increasingly, in the modern flightdeck whose reliability and capability is staggering, we are the last line of defence. The mandate is simple:
First of all I'm a software engineer from Turkey. I have been reading this forum after plane crash.
With all that sensors and their backups, how an autopilot system can not detect a faulty RA sensor and lead to a crash? ( I'm not denying crew side of the problem ) People do mistakes that's human nature. So we use computers with pre-defined rules to make some decisions to eliminate human factor. I mean come on Boeing, am I asking for too much?
It would be interesting to note the difference in statistics of accidents of aircraft witth the minimum flight crew and those with extra crew members in the fligh deck.
Could it be a possibility that senior training captains spend a lot of time doing office work on in the simulator and may loose touch with the challenges of the basic operation?
"Because of this sudden change, in reality the aircraft has responded as if it were at a height of a few meters above the Polderbaan, by closing the engine power.
It appeared as if the automate – with retarded engines – was in the last stage of flight.
As a result the aircraft lost speed.
In first instance the crew did not respond to this development.
Because of the loss of speed the aircraft arrived at it its minimum flying speed (stall situation) causing warning signals (shaking of the control column at a height of 150 feet).
From the black box it shows that then power was given again immediately. However, for the reinstatement of flight this came too late, the aircraft was too low with the result that the Boeing crash 1 kilometer short of the runway.
From the black box – which can record 25 hours of flight- in this case for this aircraft this consisted of 8 different flights – this particular problem had occurred twice before in a same situation before landing.
The aircraft touched the ground on its tail in first instance and subsequently with its landing gear.
The forward velocity at this impact was around 175 km per hour.
And an airplane at this weight should normally have a landing speed of 260 km per hour.
Because of the agricultural land, which consisted of wet soaked clay, the aircraft was arrested quickly (after about 150 m).
Because of the fall and the braking effect of the ground, the aircraft broke into two parts, the tail section got detached and the aircraft split in the area of the business class section.
The landing gear was sheared off, as is intended by design. This also goes for the two engines.
Because of the full power and the abrupt breaking action both engines were projectiled forward for another 250 meter.
Most of the deadly victims were seated at the split in the business class section and the three crew members in the cockpit deceased as a result of the enormous forces of the breaking action, which also resulted from the stuck nose wheel and the forward pressure of the aircraft.
The parts around the wings have remained in tact mostly.
On board were 127 passengers and 7 crew, of which until now there are still 28 in hospitals.
The investigation of the OVV will from now onwards be fully concentrating on the functioning of the radio altimeters and the coupling towards the automatic throttle system (automatic piloting system)."
You understand Dutch (presumably). You are therefor in the position to read and understand the preliminary report released in that language.
The "No corrective crew action taken" comment relates to the very first indication something was wrong, namely when the aircraft was at 1,950ft when, suddenly, left radar alt changed to -8ft and a Gear Warning was given. The preliminary report refers to this instance when it states that "Initially the crew did not react to the issues at hand".
When they did respond (full throttle) it was already too late.
"As a result of the deceleration, the aircraft's speed was reduced to minimum flying speed (stalling situation) and warning signals (the steering column buzzes at an altitude of 150 metres) were given.
The black box shows that full power was then applied immediately. However, this was too late to recover the flight, the aircraft was too low and, consequently, the Boeing crashed 1 kilometre short of the runway. "
The "No corrective crew action taken" comment refers to a moment in the approach phase when such an action still could have made a difference.
PS English Version of Preliminary Report can now be found here.
I would just like to drop a quick note to remind all of you guys that the BirgenAir 757 accident initiated due to captain-side altimeter failure in 1996 was also a three-pilot flight.
...just what we don`t need....face saving versus the blatantly obvious. How can safety improve if you can never admit that human beings make errors....shame on the Turkish media.
1. Turkish Airlines initially confirmed that 'all passengers have survived'
2. Turkish Airlines confirmed not a few hours after the crash that the airplane's maintenance status was beyond reproach
3. Turkish pride or whatever you might call it leads Turkish Airlines president to compare this with the Hudson river landing and their pilots have "martyred" themselves in the process of saving everyone.
4. This leaves anti-Boeing and anti-American sentiments to continue in a place where martyrdom trumps professionalism in any matter of the day, including a flight crew that stalled and crashed a 737 because of a faulty rad alt...
