Jul 4, 2012

AF447 report: Will it cause an Airbus v Boeing fan boy brawl, or something sensible?

The final report by the French air safety investigator, BEA,  into the crash of Air France flight AF447 on 1 June, 2009, into the mid Atlantic on a flight between Rio de Janeiro and Pa

Ben Sandilands — Editor of Plane Talking

Ben Sandilands

Editor of Plane Talking

The final report by the French air safety investigator, BEA,  into the crash of Air France flight AF447 on 1 June, 2009, into the mid Atlantic on a flight between Rio de Janeiro and Paris, with the loss of all 228 people on board, will be released tomorrow night eastern Australian time.

Although the flight data and cockpit voice recorders were eventually retrieved and read from the wreckage field on an ‘abyssal plain’ on the ocean floor, together with more wreckage and many bodies, and much has been revealed in the course of two earlier interim reports by the Bureau d’Enquêtes et d’Analyses the final report is expected to answer some controversial questions, yet provoke more arguments.

Why did two relatively junior Air France pilots fly what at impact was a mechanically fully functional jet into the ocean?

How and why did the pilot at the controls (the pilot flying) put the A330-200 into a steep climb followed by a high altitude stall, which then persisted until it slammed belly first and nose elevated into the ocean with all its control surfaces and engines operational?

Why didn’t the other pilot identify that for all but a short period his colleague had gripped his side stick controller in the fully back position holding the jet in a nose high attitude, other than the obvious yet unsatisfactory answer that the other side stick was not fully visible to him from where he sat?

What caused the pilot flying to reportedly briefly release his grip on the side stick controller and tentatively begin to lower the nose of the jet, which is the start of the process for recovering from a stall, but then resume his apparent fixation with keeping the jet nose high and the engines at full power in a configuration in which the wing was no longer generating aerodynamic lift?

Was there anything the captain, who had been summonsed to the cockpit from his rest break, could have done to prevent the disaster in its final stages, and is it made clear from the recorded conversations that he became aware that the jet was fully stalled?

Why didn’t the pilot flying, and his colleague, recognise that that the jet was stalled, even though audible alarms warned them of the situation for much but not all of the final sequence of events?

Why didn’t either pilot recognise the significance of a cockpit display indicating a steep angle of attack in which the jet was held for almost the entire sequence of events from the disconnection of the auto-pilot following an unreliable speed warning caused by ice-up external speed measuring devices called pitots, and through the restoration of speed data, until it was too late to recover control?

Another way of asking those questions is to ask what circumstances made AF447 different to numerous other occasions on which iced up pitots have caused temporary unreliable air speed warnings which pilots flew through by maintaining the correct angle of attack and throttle settings until the situation passed?

At the combative level, which is a shouting match between Boeing and Airbus supporters, the issue is the level of automation across the entire fly-by-wire Airbus range of airliners from A318s up to A380s, and the claim that Airbuses inhibit pilot intervention in abnormal situations.

However at the more thoughtful level, including industry seminars on automation and pilot training, both Airbus and Boeing have expressed concerns at undue or unintended reliance on automation or computer systems supported flight management procedures in current airline practice.

Managements which no longer have directors or senior executives with an intimate knowledge of piloting and engineering like automation. The notion of instructing pilots to rely on auto pilots as much as possible is openly embraced by managements which come from business schools, not flying schools. And that applies to all Boeing fleet airlines, as well as all Airbus fleet airlines, and those who use both.

Automation in flight, and also in engineering and maintenance support, is seen as cost reducing, and in many respects, this is true. But as many pilots and safety authorities, and cautionary voices in the aircraft manufacturers themselves, have pointed out, the fundamental purposes of recurrent training and flying standards is not to ensure that pilots can fly airliners, but save them when something unexpected occurs, and there is an inexplicable and life threatening upset, or surprise, as occurred on board the Qantas A380 that had an engine disintegrate and blast holes through its wing, knocking out half its hydraulics, and causing structural damage that would have quickly destroyed the wing of earlier generation airliners. Or as occurred in the abrupt loss of control that afflicted a Qantas A330 near Learmonth in Western Australia, or the loss of both engines from bird strike that brought down a US Airways A320 on the Hudson River, each incidents in which skilled and experienced pilots regained control without loss of life, as they did in the severely damaged Qantas 747-400s forced to land with almost no electrical systems at Bangkok, or land at Manila after an oxygen cylinder ruptured and tore open sections of the fuselage and again compromised control surfaces.

In what may be in some ways a template for tomorrows AF447 report, the BEA recently dealt with the ‘surprise’ factors in pilot performance under sudden pressure in relation to an Air France A340 incident over the north Atlantic in July last year.

That report is significant and is available in English here, as that incident saw the pilot flying do something similar to what the pilot flying on AF 447 did, stress similar …. there are differences.

The A340 ran into severe turbulence from a thunderstorm the pilots were not correctly monitoring (according to the BEA) which generated an overspeed warning. On this occasion the pilot at the controls disconnected the auto pilot, contrary to the approved procedure, and flew the jet from 35,000 feet to more than 38,000 feet in a reflex action which flirted with the same high speed stall situation that occurred in AF 447 and destroyed it.

He and his colleague in the cockpit and everyone else on board survived this incident, and told the investigators that neither had any recollection of what they had done in their immediate and instinctive response to the temporary overspeed situation they encountered.

They told the BEA that the surprise of the situation was such that they did not follow the approved Airbus procedure for dealing with that situation, which included leaving the auto pilot engaged, and the BEA established that had they followed that procedure, the jet would have briefly departed from its intended altitude by 200 feet, not more than 3100 feet, and it would never had run the risk of a high altitude stall like the one that the crew of AF447 never identified, and allowed to destroy them and everyone else aboard that flight.

However in its recommendations the report places more emphasis on the ‘startle’ factor in the incident than a reflexive disregard by the pilots for the correct remedial procedure, and points out that the Airbus operating system would have suppressed certain alarms which would have prompted the pilots to reconnect with the reality of their making the wrong responses to the temporary crisis, and thus turning it into a much more serious incident.

It is the first time according to some pilots that the BEA has engaged in findings that discuss the failure of pilots to respond to sudden surprises in the cockpit, and is thus considered an indication that it may be about to shift the blame from pilot performance to a lack of preparation for such surprises in the case of AF447, or to possible flaws in Airbus flight procedures.

We will learn more tomorrow night.

(Visited 1 times, 1 visits today)


Leave a comment

3 thoughts on “AF447 report: Will it cause an Airbus v Boeing fan boy brawl, or something sensible?

  1. LongTimeObserver

    I’ll take Door #2 and Sensible, but be prepared for the fanboi brawl…

  2. John Thomas

    One of the facts that airline management fail to realise or acknowledge is that flying is a skill learnt through experience. This does not fit with most airlines operating models where in response to the looming pilot shortage the high experience employment requirements were dropped in favour of other avenues to crew aircraft.

  3. Worrierqueen

    It wil be interesting to see the report and who they blame. No doubt it will be the pilots and the pitots but it would do the industry a disservice to stop there. There are problems with training, computerisation and the whole design of the airbus itself.

    The pilot left to have a nap when he knew they were entering very turbulent weather. He appointed the junior of the two first officers to take over the plane while he was away but allowed the senior co-pilot to occupy his (the pilot’s seat) which is the seat of the PF (pilot flying). Thus the lines of control were tangled from the start.

    Then there were the problems with the computer when the pitot tubes blocked up. There was no direct angle of attack (the angle between the wings and the airflow) indicator in the cockpit, although it could be derived by the primary flight display. However in the case of AF447, the crew were getting lots of bad data and incorrect warnings and didn’t know which instruments they could trust and which they couldn’t.

    They were told an A330 couldn’t stall but this was only true when the computers were in primary law mode. When the pilots pitots bunged up and the data the computers were getting conflicted, they immediately switched to alternate law that allowed the pilots complete control to stall the plane. Furthermore the computers only provided the stall warning until the angle of attack reached 40 degrees (it should be 13 to 18 degrees) at which point the computers then decided the data must be wrong and switched the stall warning off. Thus when the pilots did attempt to lower the nose of the plane to pick up speed and the angle of attack went to less than 40 degrees, the stall warning started up again, leading the pilots to conclude they were in an overspeed stall ruather than an underspeed stall (they were so close to coffin corner it was difficult to choose between them), so would pull back the stick again and the stall warning would stop.

    Next the A300 is designed to operate to minimise G forces in the plane rather than minimise stall chance (since this is theoretically impossible anyway at least in normal law) and the best way for this was to maintain the plane nose up which the horizontal stabiliser continued to do short of active pilot interference (although there is no evidence AF447 was in a deep stall that could not have been corrected with correct pilot input).

    And this leads to the final design flaw of the A330. instead of the old fashioned stick or steering wheel used in older planes, the A330 and later designs uses a joystick arrangement like in a home computer console. These are to each side of the pilot and co-pilot and thus one can’t be seen from the other’s position. So the junior pilot who appears to have been piloting the plane from the co-pilot seat was constantly pulling back the joystick keeping the nose up and this could be seen by neither the senior co-pilot or the captain when he returned.

    And finally finally, in the older planes, when the pilot pulled one way on the control column, and his co-pilot pulled the other, it tried to do both (strongest winning), giving instant feedback that the pilot and co-pilot were in conflict over the controls. In the A330, the computer simply splits the diferrence. Thus when the senior co-pilot said he was taking over control and pushed his stick forward to push the nose down, the junior co-pilot was still pushing his back and the computer just split the difference doing nothing at all, and worst of all not telling anyone in the cockpit what it had done (or not done).

    There’s a lot more to go in this debate yet and it looks as if our computers are becoming so complex, pilots simply are not capable of working successfully with them in all situations.

Share this article with a friend

Just fill out the fields below and we'll send your friend a link to this article along with a message from you.

Your details

Your friend's details