The airline industry appears divided over the usefulness of training pilots to recover from extreme attitudes. Is it essential or does it instill bad habits?
There is a form of advanced pilot training that some airlines think is vital and others think is unnecessary, while certain sections of the industry think it is potentially dangerous. No-one, however, is indifferent about training pilots to recover from flight upsets - especially since the ignition of an industry-wide debate about whether or not such training encouraged the vigorous rudder deflections that may have led to the November 2001 American Airlines Flight AA587 fatal accident at Belle Harbor, New York. Open debate about AA587 - an Airbus A300-600 in which all 260 people on board died - has been suspended awaiting the final US National Transportation Safety Board (NTSB) report - expected by the end of this year.
No national aviation authorities require airline pilots to undergo aircraft upset recovery training (URT), nor do they require airlines to provide it. But the US Federal Aviation Administration, working with the Flight Safety Foundation and many other agencies and airlines, has prepared an "aircraft upset recovery training aid" - a training programme template which airlines can customise and integrate into their own type conversion or recurrent training systems.
The NTSB recommends the use of URT to reduce the risk of loss of control (LOC) accidents, but, only a few weeks after AA587; it also issued an interim warning to flightcrew discouraging the heavy use of rudder during recovery from flight upsets.
More recently, an NTSB-commissioned University of California study - spawned by concerns raised during the AA587 investigation - caused more ripples in the pond by suggesting that the more sensitive rudder control unique to the A300-600 may hold part of the answer to what occurred (Flight International, 9-15 March).
In June the NTSB recommended that the FAA order a redesign of the A300-600's rudder travel limiter system after its re-examination of a 1997 incident in which an American Airlines A300-600 had stalled and adopted extreme roll angles, although the crew eventually recovered control (Flight International, 8-14 June).
Limiter failure
The re-examination revealed that, during the rapid cyclical climb and descent involved in the incident, the speed changed so rapidly that the rudder travel limiter - which adjusts the amount of travel the rudder can make according to the aircraft's airspeed - could not adjust fast enough. Thus it could, under similar circumstances, briefly allow a wider angle of travel than it should at high speed, potentially overstressing the fin and rudder. The NTSB says this recommendation is not directly related to the investigation into the AA587 incident.
Airbus, which says it awaits an FAA verdict on the NTSB's rudder travel limiter recommendations, has objected to what it sees as the implication in the University of California study - that the A300-600's rudder system was a significant causal factor in the Belle Harbor accident. The manufacturer maintains that these things have only ever happened to American Airlines, and has linked the occurrence with the airline's "advanced aircraft manoeuvring programme" (AAMP), the name American gave to its original upset recovery training programme. Airbus alleges the AAMP encouraged heavy rudder use during upset recovery and, in the case of AA587, full rudder deflections in each direction coincided with the passage of the A300 through the two wingtip vortices of a preceding Boeing 747. It was after those rudder movements that the fin failed and the aircraft went out of control.
Before AA587, Airbus and Boeing had jointly published advice about handling large jets that get into upset situations, counselling against the heavy use of rudder in recovery when aileron and elevator alone - with the help of the yaw damper system - will cope with most recovery manoeuvres. The exception occurs during engine failure at take-off when airspeed is low and the asymmetric power has to be countered quickly.
Loss of control
All these arguments, claims and counter-claims have muddied the discussion about whether, for an airline pilot, training for recovery from unusual attitudes is necessary or desirable. But, between 1993 and 2002, loss of control (LOC) caused more commercial jet fatal crashes than any other accident category, according to a NASA study. Meanwhile, an NTSB study of 20 LOC accidents between 1986 and 1996 revealed the most common cause of loss of control is stalling - a situation not always caused by flight upset.
The NTSB recommends that judicious flight upset training is good for safety but in the UK, however, the Civil Aviation Authority Safety Regulation Group has not advised airlines to run upset recovery training programmes and does not see the need to go beyond existing approved training.
It could be argued that flight envelope protection like that in the Airbus fly-by-wire types would prevent upsets. In fact the system protects against stalling, overspeed or an angle of bank greater than 60° even if the pilot selects it. But fly-by-wire will not prevent the aircraft hitting the surface if the pilot does not recognise he is getting close to terrain or if a flight upset has caused pilot confusion.
In the case of pilot disorientation, although the controls still work correctly, control of the aircraft's trajectory can at least temporarily be lost. An example is the Gulf Air A320 pilot who became disoriented on a dark night with no external visual cues at Bahrain in August 2000. The crew had carried out a go-around from runway 12 and was supposed to be climbing left on to a heading of 300°, expecting radar vectors for another approach, but the climbing turn disoriented the captain and he ended up in a 15° nose-down attitude he had selected manually. Despite a ground proximity warning system alert that continued until the crash, the captain had left the power high and had only reduced the nose-down attitude to 6û just before impact, which took place at about 270kt (500km/h).
A recent NASA study used a Learjet 25B simulator, programmed to perform as a "generic swept-wing large twin-engine transport aircraft", and tested 40 pilots from backgrounds ranging from no upset training or aerobatics to all forms of training. The agency exposed the pilots to simulated "upsets" that had preceded real accidents and found that most pilots could not recover from most upsets.
The conclusion was that, while it is not practical to train for "all imaginable scenarios", identifying and training for a relatively small number of the more likely upsets would be desirable.
The final recommendation is that, in simulators, crews should be exposed "to the conditions that precede upsets, and to the onset of upsets, so they can practise recognising cues that distinguish different classes of upset." Prevention, it seems, is still better than cure.
DAVID LEARMOUNT / LONDON
Source: Flight International
