Investigation Into Schiphol Crash Focuses on Radio Altimeter

Wed, Mar 25, 2009 — David Evans

Articles, Featured

Three pilots in the cockpit failed to notice that the thrust did not increase when the commanded speed was reached, so when the alarm went off indicating the plane’s speed would drop below the minimum, it was too late.

But the issue isn’t solely pilot inattention or distraction; how could the avionics design be certified when it’s possible for the system to cut power in the moments before landing?

“The plane was too low at 150 meters [500 ft]. As a consequence, the plane crashed 1.8 kilometers [about 1.1 mile] before the runway” at Amsterdam’s Schiphol Airport, said Dutch Safety Board Chairman Pieter van Vollenhoven.

Wreckage about a mile short of Schiphol's runway 18R.

Wreckage about a mile short of Schiphol's runway 18R.

The 25 February 2009 crash of the Turkish Airlines B737-800, on a flight from Istanbul, killed nine and injured 86 (six critically) of the 134 passengers and crew on board. Thirty-nine people walked away from the crash.

Not among the walking were the three pilots in the cockpit, among them a trainee pilot in the right hand seat who was flying the plane before the crash took place. All three were killed, forcing Dutch investigators to rely on the cockpit voice and flight data recorders (CVR/FDR) to piece together what happened.

According to a preliminary report by the Dutch Safety Board:

“The [Schiphol] tower gave the Boeing permission to land on the Polderbaan 18R …

“The descent takes place with the help of the automatic pilot, as is normal with Turkish Airlines …

“The voice recorder and the black box, both of which are in the hands of the Safety Board, show that an irregularity occurred during the descent …

“At 1,950 feet the left radio altimeter suddenly indicated a change in altitude – from 1,950 feet to -8 feet – and passed this [erroneous datum] onto the automatic pilot. This change had a particular impact upon the automatic throttle system which provides more or less engine power.

“The radio altimeter normally measures the altitude of the plane above the ground very accurately and can start registering this from 2,500 feet. As already mentioned, this radio altimeter is very significant for providing the appropriate power for an automatic landing.

“A Boeing is fitted with two radio altimeters, a left one and a right one. The black box has shown that this deviation only occurred in the left altimeter.

“The voice recorder has shown the crew were notified that the left radio altimeter was not working correctly (via the warning signal ‘landing gear must go down’). Provisional data indicates that this signal was not regarded to be a problem.

“In practice, the plane responded to this sudden change as though it was at an altitude of just a few meters above the Polderbaan and engine power was reduced. It seems that the automatic system – with its engines at reduced power – assumed it was in the final stages of flight. As a result, the aircraft lost speed.

“Initially, the crew did not react to the issues at hand.

“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 meters) 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 kilometer short of the runway …

“The aircraft initially hit the ground with its tail and then the undercarriage followed. The forward speed was about 175 km per hour [108 mph] upon impact. An aircraft of this weight should normally have a speed of 260 km per hour [161 mph] for landing.

“The aircraft came to a rapid halt (after about 150 meters) [500 feet] as a result of the arable land being made up of boggy clay. The braking caused by the ground meant that the aircraft broke into two pieces; the tail broke off and the aircraft’s hull ruptured at business class …

“The full power and the sudden braking resulted in both engines continuing forwards for a further 250 meters [820 feet].

Engines, under full power, broke free as the airplane hit the boggy ground.

Engines, under full power, broke free as the airplane hit the boggy ground.

“Most of the fatally wounded victims were located near the rupture, in business class, and the three crew members in the cockpit died as a result of the enormous braking forces, partially caused by the embedded nose wheel and the forward movement of the aircraft.”

Misty weather and low clouds meant the runway was not yet visible at the height where descent started. “The plane was in heavy fog. I think the pilots did not see that a problem was occurring,” van Vollenhoven said.

The major problem is that three pilots failed to notice the thrust not being increased when the commanded speed was attained. There must have been a very big green speed trend arrow pointing down the speedtape on the primary flight display (PFD). Subsequently, as airspeed decreased below the commanded speed, one would think the crew would have noticed the following:

• Increasingly higher pitch attitude.

• Less noise (wind and thrust).

• Speed entering the yellow band on the speed tape.

• Speed entering the red band.

• Stick shaker, indicating imminent stall!

The initial indications should have led to disconnecting the autothrottle. Later indications should have led to a go-around decision.

Although a system fault caused the loss of thrust, simply standing up the thrust levers at Vref [reference velocity], without disconnecting the autothrust, and holding the thrust levers with muscle power would have prevented the loss of thrust. The throttle lever position commands the thrust, so overriding the autothrottle is possible. Although the pilot(s) will feel pressure from the system trying to close the levers in RETARD mode, in a later stage go-around (TOGA press), this would have solved it.

Three pilots missed everything until they were at 400 ft. with a stick shaker. The question is why?

It should not be surprising if the crew thought that since they were using autopilot B (the “pilot flying” first officer’s autopilot) then the autothrust system would also logically be reading from Radar Altimeter 2 on the right side, and they had no expected issues due to the radar altimeter malfunction on the captain’s (left) side. It’s clear that they were very surprised at the low energy state of the aircraft once they found themselves in that situation. A certain amount of complacency may also have been injected by the long period during descent the engines were at idle. Of course, this does not excuse the crew’s lack of airspeed awareness or failure to have hands on the throttles for some tactile feedback (or an alerting lack thereof).

There may be good reasons why the two radar altimeters are separate, and why the No. 1 radar altimeter is more important than the No. 2. The No. 1 is probably powered by a transfer bus, the No. 2 by a generator bus, therefore if half the electrics are lost the aircraft can shed the No. 2 while keeping the more important No. 1 working.

This arrangement allows the aircraft to be dispatched under MEL (minimum equipment list) with the No. 2 radio altimeter inoperative, but not the No. 1. All very sensible.

Some pilots may know that No. 1 radar altimeter controls the autothrottle but maybe have forgotten the implications of that association. Indeed, most crews may not be aware of this.

However, most crews would probably notice “RETARD” on the FMA (flight mode annunciator), thrust levers at idle and speed decaying by 40 knots. Also, the pitch attitude must have been unusually high so close to the stall speed. The crew should have noticed this gradual pitch up.

The difference between this accident and the 17 January 2008 crash on landing of the British Airways B777 at Heathrow is that the B777 crew noticed the speed decay from ice-induced loss of thrust and intervened, critically, BEFORE the aircraft actually stalled, thus saving the lives of everyone on board.

From what is understood, the Turkish Airlines B737 would not have flared because autopilot B was engaged and taking its height information from radar altitude No. 2. So the aircraft would have maintained the correct glide slope while the speed washed off.

Firewalling the thrust levers was potentially lethal, given that the autopilot had trimmed the aircraft right into the stall. Did the aircraft pitch up further into a nose-high power-on stall, surprising the pilot(s)? That is the big question. The answer may ultimately increase the damage to Boeing’s credibility (and liability). That is, did the manufactuer “design in” a deadly auto-trim trap? (And was the Continental Express Dash 8-Q400 crash 12 February 2009 the result of the pilot’s failure to manually increase thrust, then adding maximum power; that accident may not be due to icing but just a replicated low-level stall scenario.)

The clues to low level stall in the Turkish Airlines crash are scattered through the initial reporting:

• Fully configured at 3 NM and approximately 1,000 ft and crew distracted by heavy training on the flight deck on final approach – when speed starts dropping and nobody notices.

• The approach looks fairly normal in regards to glide path and speed until 600 feet.

• Around 500 ft the automatics (autopilot and autothrust) are disengaged and proper speed management was not maintained, or the pilot(s) assumed the autothrust was still engaged when it was not. The mistake was only noticed very late down the approach and the flight crew was caught without options in a low energy/low altitude scenario.

• Witnesses on the ground described a nose high attitude, followed by a dive to the ground.

• The aircraft hit the ground tail first in a high rate of descent with low forward speed.

Note the fracture in the fuselage, denoting the area where most passengers were killed.

Note the fracture in the fuselage, denoting the area where most passengers were killed.

When the FDR and CVR information is released, it is suspected that a low altitude stall, or approach to stall, will be confirmed, with an incorrect or incomplete stall recovery. Soon after breaking out of the clouds, with autothrottle not obliging with any drag-opposing thrust, the aircraft stick-shaker would have cooked off below Vref (140 knots/flaps 30º). The surprised pilot at the controls would have disconnected the autopilot and selected maximum power – yet immediately encountered the fierce zooming effect of a low-speed max power pitch-up.

Additionally and fatally, courtesy of the insidious effect of auto-trim, the handling pilot also unexpectedly liberated a yoke-full of maximum elevator backtrim and nose-up stabilizer. That he would have been pushing and fighting that powerful nose-up pitch would be without question. In the very short time available, would the Turkish captain have been able to communicate the nature of the problem?

Few pilots have tried a very nose-high power-on stall. If practiced as an exercise, it had best be done at height. Why? Because at the point of stall the nose pitches down to an alarming “face full of dirt” attitude (particularly unnerving breaking out of clouds, as occurred here).

What happens next? The pilot has a very limited height in which to recover from the ensuing dive. Because of the ground-rush phenomenon, he will be trying very hard. Low level aerobatics pilots know the self-discipline necessary to recover from a misjudged low-level maneuver, in order to avoid going beyond buzz, buffet, judder into the stall and, with performance destroyed, into the dirt. Performance with gear and flaps down, as in the case of the Turkish Airlines plane, will be marginal, even with maximum power.

When stall is due to a panic-stricken pitch-rate, the lift vector is killed well in excess of the 1G stall speed, no matter what the deployment of leading or trailing edge flaps. No further convincing is needed that the ground intervened in the flight of the Turkish Airlines plane, and the tail departed and the cockpit was then counter-pitched down, very hard, into terra firma.

What needs to be determined is when and why the auto-throttle became disconnected or disabled. Radar altimeter spikes? Or was it simply a decision to demonstrate to the trainee a manually throttled approach, which was then overlooked by ingrained habit patterns.

One wonders if this nasty scenario is ever practiced in a simulator as part of a recognized training syllabus. One doubts it. Yet something very much like the circumstances leading up to this crash occurred in 1997, when an American Airlines A300B4 on descent into Miami, with the autothrottle engaged, lost speed and entered a stall. As the National Transportation Safety Board said:

“The aircraft then pitched down, and entered a series of pitch, yaw, and roll maneuvers as the flight controls went through a period of oscillations for about 34 seconds. The maneuvers finally dampened and the crew recovered at approximately 13,000 feet.”

The aircraft had lost 3,000 feet in altitude before the crew was able to recover. That’s about 2,500 feet more than the Turkish pilots had.

Boeing has responded to a recommendation by the Dutch Safety Board, warning B737 operators of the risk associated with an erroneous radar altimeter reading and the potential for the autothrottle to reduce power.

On the accident airplane, it is understood that the left hand radar altimeter was transmitting an incorrect value but was not giving a failure sign, which would have caused various INOP (inoperative) warning lights and flags to pop up. “Landing gear must go down” is hardly an advisory for this deadly lack of airspeed. Boeing will have to evolve a failure mode to cover this lethal falsetto scenario. A pertinent question is why this situation wasn’t uncovered and corrected during certification testing?

“The board’s investigation will now focus fully on the workings of the radio altimeters and the connection to the automatic throttle,” van Vollenhoven vowed. If there is a disagreement of more than, say, 50 ft, between the left and right radio altimeters, it seems that the flight crew should receive an unambiguous message to disconnect the autothrottle.

Comments are closed.

Nolan Law Group