Legends of the stall

image: murmakova | Adobe Stock

Simulators are changing to test pilots in more extreme situations

By Ross Peake

The flight crew went to their deaths not quite knowing what was happening, or why. And they took 225 passengers and cabin crew with them in the early hours of May 31, 2009. The disaster of Air France flight 447 also killed off a way of thinking about flight training, the notion that ’if an aircraft’s control systems won’t let it stall, why bother training pilots to avoid stalls’?

The crew of the Airbus A330 flying from Rio de Janeiro to Paris had a cumulative experience of 20,000 hours but failed to recover from a stall in a sophisticated aircraft. All onboard were killed.

The report into the accident notes the ‘startle effect’ played a major role in both the destabilisation of the flight path and the failure of the crew to adequately comprehend and respond to the situation. During the crisis, the stall warning sounded continuously for 54 seconds. ‘The altitude was then about 35,000 ft, the angle of attack exceeded 40 degrees and the vertical speed was about -10,000 ft/min,’ the report says.

These loss of control accidents have been happening since the dawn of time in aviation.

A few months earlier, Colgan Air flight 3407 also entered aviation history for all the wrong reasons. The Dash-8 carrying 49 people entered a stall on approach to Buffalo, New York state, and crashed, killing everyone onboard. The crash was attributed to improper stall recovery technique and pilot fatigue.

Even before these crashes, a few aviation training experts were worried about trends they were seeing in the quality of training for stall recovery and the integrity of simulators being used for the teaching. Their concern went more broadly to loss of control in-flight (LOC-I) accidents.

Between 2001 and 2011, this type of aeroplane accident was the leading cause of fatalities in commercial aviation. LOC-I accidents often have catastrophic results with very few, if any, survivors.

Mild to wild, with a change of emphasis

A key question raised about the prevalence of these accidents was – were the simulators being used for airline training up to the task? Another – was the training too ‘mild’?

Although stalls and other ‘upsets’ are part of the initial flight training syllabus, they have tended to be approached more in theory and in a limited way in simulators, rather than in in-flight practice in training/conversion on large complex passenger aircraft. This is because of 2 factors: stalling a large swept-wing aircraft is a serious undertaking, with several thousand feet required for safe recovery. It is always a dangerous manoeuvre. And because real-life stall flight data in large aircraft is rare, the simulators used for most large aircraft training have until recently not been programmed to accurately reproduce aerodynamic stall and beyond-stall aerodynamic characteristics. These led to training regimes in which the stall became seen as primarily a condition to be avoided rather than reacted to and managed.

This utopian view of flight safety unsettled some experts, even before the loss of AF 447. A conference in June 2009 on aeroplane upsets and LOC-I just days after the disaster led the Royal Aeronautical Society to initiate a study to investigate the problem and recommend mitigating strategies.

The problem had been formally recognised as far back as 1996, when the US National Transportation Safety Board had made recommendations relating to LOC-I accidents. In 2004 a Federal Aviation Administration (FAA) sponsored working group developed the airplane upset recovery training aid (AURTA-Revision 1).

In the past, well-meaning people have had training programs that didn’t quite do the job.

With reduction in LOC-I accidents a high priority, ICAO developed upset prevention and recovery training (UPRT) Standards and Recommended Practices which prescribe, in part, that operators shall establish and maintain UPRT ground and flight training programs.

As part of these ongoing reforms, CASA Flight Simulation Team Leader John Frearson is part of a team overseeing a major upgrade of simulators and simulator-based training in Australia.

‘These loss of control accidents have been happening since the dawn of time in aviation,’ he says.

‘In the past, well-meaning instructors trained students in simulators to minimise height loss during a stall recovery. In fact, I was shown for the [Boeing] 777 how, at 15,000 feet and reasonable weights, if you stalled the aeroplane, all you had to do was to open the throttles and it was so overpowered it would fly its way out.

‘The Air France A330 crash over the south Atlantic showed that that sort of training was really ineffectual because it didn’t take account of what really happens. There was a lot of thinking, from Airbus especially, such as, “pilots shouldn’t stall the aeroplane and in fact we can make damn sure they don’t, therefore you don’t need that training”.

‘But the answer to this is, in fact they do stall the aeroplane, so you better make sure they get the training. This shows two schools of thought – those who say we don’t stall so we don’t need to do the training, while others say we do stall so we better do the training.’

The existence of these two schools was identified as far back as the late 1960s in D P Davies’ seminal reference book, Handling the Big Jets. Davies argued that in both aircraft certification and pilot training, safety would be best honoured if pilots were given thorough theory and practical training in flight beyond the initial indications of the stall.

Analysis of LOC-I accident data indicated contributory factors can be categorised as being induced by one of the following factors, or any combination of them:

  • aeroplane systems
  • environmental
  • pilot/human.

Of the three factors, pilot-induced accidents represented the most frequently identified cause of the event. Recognising the need to address this safety matter in late 2014, ICAO published amendments to Annexes 1 and 6 to mandate training in upset prevention and recovery. At the same time ICAO set the standards required for UPRT programs and upgraded the capability standards for the simulators used in such programs.

As a response, CASA’s Flight Standards Branch tasked Frearson’s team with preparing a new Advisory Circular – AC 121-03v1.0 – on upset prevention and recovery training.

This training required by CASA follows ICAO requirements and FAA guidelines – that an airline’s flight crew members must be provided with ground training and flight training or flight simulator training to recognise and avoid:

  • a stall of an aircraft or, if not avoided, to recover from the stall (emphasis added)
  • an upset of an aircraft or, if not avoided, to execute such techniques as available data indicate are appropriate to recover from the upset in a given make, model, and series of aircraft.

In the circular, Frearson’s team says that previously, a significant proportion of upset events involved the traditional understanding relating to physical conditions – unintentional pitch beyond +25 or -10 degrees, bank angles greater than 45 degrees or speed inappropriate for the conditions. Recovery training was initiated only after exceeding these parameters, without paying attention to the reasons of these diversions.

‘Current thinking … now includes a wider definition of upsets and uses the established concept of undesired state and the pilot’s awareness of this, regardless of airspeed or specific pitch and/or bank angle parameters,’ it says.

Stall and recovery training was needed to ‘undo years of applying incorrect stall or upset recovery procedures’. It also identified the need to address the use of simulators that could not adequately represent flight characteristics after the initial stall warning indications.

‘Many LOC-I accident investigations revealed the affected flight crew had received misleading information from well-meaning training staff or their organisations,’ it says.

One concern was training to recover from a stall with minimal loss of altitude. This resulted in training practices which emphasised a rapid application of power with the least amount of reduction in angle of attack (AOA) to minimise the loss of altitude, rather than appreciating the importance of reducing the AOA to restore the wing’s capability to generate lift.

Startling and upsetting

Human factors are an overarching and integral part of UPRT, according to the CASA team’s advisory circular.

‘Until recently, initial and recurrent training did not promote and test the capacity to react to the unexpected,’ it says. ‘The vast proportion of training has involved standardised and predictable responses to non-normal events whether they involve weather, systems or human factors issues such as incapacitation.

‘This training, though worthwhile, has implicitly excluded “surprise” and “startle” events and hence has not provided crews with the opportunity to experience events with a significant “surprise and startle” factor. In particular the rapid increase in crew workload and degradation of communications and coordination in sudden events is something that traditionally trained crews have rarely been exposed to. Analysis shows that in response to “startle” events, both pilots can attempt to take control and act with little co-ordination and lose their teamwork focus.’

image: © Civil Aviation Safety Authority

Upgraded back to basics

The circular says most human factors issues required to be included in an UPRT program were encapsulated in the report on the Air France A330 accident.

‘The report noted that the “startle effect” has typically played a major role both in the de-stabilisation of the flight path and in the failures of crews to adequately comprehend and respond to the situation,’ it says.

About 600 to 700 simulators are operating around the world but many were built when stall testing stopped at the first sign of the stick shaker indicating a problem. That meant pilots were not being tested to cope with more extreme upset conditions, such as a bank greater than 45 degrees.

‘We need pilots not to think: “the stick pusher keeps the angel of death far away, I’m fine”,’ Frearson says.

‘We need an instructor to take them into those situations and show them the reality, with surprise and startle.’

About 60 simulators are operating in Australia, with a handful being too old to upgrade.

‘CASA took care to ensure a significant level of engagement with industry in planning actions on simulator standards and training in the belief any action should be done by engagement and education, not by a top-down approach,’ Frearson says.

‘The upgrade of a near-new simulator could be $1 million and the installation of an instructor operating station is not cheap. The result of this engagement strategy was a three-year program of consultation invoving research, workshops, feedback and circulation of drafts of the guidance material.

‘This is not a small exercise for an airline like Qantas – taking pilots offline, giving them a day or 2 at ground school, them some extra sim sessions, as well as training several hundred instructors, half a dozen senior instructors and 2 project leaders,’ he says.

The program to upgrade simulator training is designed to bring pilots ‘back to basics’, Frearson says.

‘If you’d taken my father as a RAAF pilot in 1942 training on a Tiger Moth and sat him in the cockpit [during a practice stall], he would have seen the nose high and felt buffeting and said, with his little bit of training, “push the nose down”,’ he says.

‘That’s all we’re doing – we’re essentially going back all those years, back to basics. We’re not condemning anybody. This is about learning and engagement. However, as a regulator, like other regulators, we have to accept that in the past well-meaning people have had training programs that didn’t quite do the job and simulators that did not provide pilots with the proper experiences.’


*CASA Flight Training Examiner (fixed wing) Janine McMullan and Senior Standards Officer (flight simulators) Quentin Fuller were heavily involved in this project. They conducted the initial research on several visits to the USA where they studied FAA and industry UPRT developments. In more recent times they have travelled Australia conducting evaluations of upgraded simulators and ensuring CASA’s strict standards are met. Due to the travel and financial constraints imposed by Australia’s COVID-19 responses,CASA has added significant levels of flexibility to its implementation timelines.