Pilots need an automatic wake-up call
Fatal crashes of passenger aircraft in Libya and the south Atlantic have raised alarming questions about the interface between pilots and the new highly automated flight decks. Alex Fox talks to training organisations and safety experts to find out what trainers can do to help.

On Monday June 1 2009 Air France flight 447, an Airbus A330 en route from Rio de Janeiro to Paris carrying 228 people, crashed into the Atlantic Ocean. Until earlier this year the cause could only be speculated. However, at the beginning of May, during phase 4 of the search and recovery operation, the wreckage and, importantly, aircraft flight data recorders (FDR) and cockpit voice recorders (CVR) were located 4,000 metres below the ocean. What investigators found has brought new resonance to the on-going debate about pilot training and flight automation.
In May 2010 another reported loss of control accident occurred when Afriqiyah Airways flight 771, also an Airbus A330, crashed short of the runway at Tripoli International Airport in Libya. While investigation efforts have been hampered by the on-going civil war in the country, it has been reported that the pilot disengaged the autopilot to bring the aircraft in for a manual landing. When the pilot realised he was in trouble he tried to climb and re-engage the autopilot but it was too late – the aircraft crashed 900 metres short of the runway, killing 103 of the 104 people on board.
While this accident occurred at low-level, unlike Air France 447, it is the apparent manual mishandling of the aircraft that bears similarities.
A US advisory board recently concluded that long hours flying on autopilot might have dulled the skills of airline flight crews. The board recommended that airlines build into existing operating procedures more manual flight time for pilots to counteract the problem.
The concerns of the board are highlighted by a number of accidents that have occurred during the last decade; with perhaps the Air France crash being the most well known. The accident occurred over the Atlantic off the coast of Brazil. When the crew received warnings that the aircraft was stalling at high altitude, after the autopilot disengaged, they failed to recover the aircraft from the stall and rapidly descended from 38,000 feet into the ocean. French investigators found that pilots had failed to acknowledge stall warnings and had also not received high-altitude manual flight training for such scenarios.
The crash of a Colgan Air commuter plane in the US in January 2009 further highlights concerns that manual flying skills are becoming dulled with over-reliance on autopilot. The Dash 8 Q400 crashed in icy conditions after the pilot countermanded aircraft system recommendations as well as airline and FAA standard operating procedures, subsequently leading to it stalling. The pilot then failed to recover from the stall, suggesting he had forgotten, or perhaps not been trained in some key procedures.
The Colgan Air and Air France accidents both involved the aircraft stalling, not in itself an irrecoverable condition of flight. The fact that in both cases pilots failed to recover, even when prompted, and that the Afriqiyah pilot lost control while manually handling the aircraft, could be indicators that manual handling skills are deteriorating. Is this the case? Could automation be the key catalyst in this problem? Or is automation still the safest option?
It is worthwhile noting that in the Air France and Colgan Air scenarios, there is a common misconception that the reactions of the pilots were wrong because they had ‘forgotten’ the correct technique for stall recovery as it is taught during ab-initio flight training – where recovery involves reducing back pressure on the control column (to reduce the angle of attack and unstall the wing) and to increase power to full (to minimise loss of height). However, it must be noted that stall recovery in modern airliners differs depending on the aircraft, configuration and phase of flight.
It is apparent from transcripts of the Air France CVR that the crew appeared unaware as to why there was a loss of control. Flightglobal’s David Learmount suggests that a pilot must do three fundamental things with the information received from the instruments: recognise it, believe it and understand it. It is possible that Air France 447 could have been recovered had the crew recognised and understood early on the situation they were in. However, it is entirely possible that the crew were carrying out recovery actions for another scenario and were so far outside the tested flight envelope that nobody knows if it was even possible to recover. Actions on board suggest the crew recognised most of what was displayed to them; that airspeed indicators were not functioning correctly and that flight controls had changed from ‘Normal Law’ to ‘Alternate Law’ (where flight envelope protection is not given and the aircraft can be stalled) the crew did not acknowledge the stall warning.
Learmount further suggests that the misapprehension that occurred on board after the autopilot disengaged was certainly not inevitable, but it was – and remains – a likely product of the way aeroplanes have changed, while training has not evolved to take account of the differences in the modern flying task.
These sentiments are echoed by Don Bateman, chief engineer at Honeywell and inventor of both the ground proximity warning system (GPWS) and enhanced GPWS – systems that have significantly reduced controlled flight into terrain (CFIT) accidents since being implemented in the 1970s. Speaking at the 2011 Flightglobal Flight Safety Conference in London on the subject of loss of control (LOC) accidents, Bateman suggested that the industry as a whole needs to “revisit pilot training” in a “practical and low-cost manner” and also suggested the “startle effect” is a huge factor in LOC accidents. In almost every instance the crew were not expecting to lose control and the issue also needs addressing through proper training, he concluded.
Bateman suggested a number of engineering solutions could also help, stressing these must also be “simple, practical and low-cost” and should “utilise existing cockpit systems”. Solutions such as visual recovery arrows displayed on the attitude direction indicator (ADI); revisiting frequency-separation attitude displays, which give pilots instant feedback on control inputs; and the use of synthetic vision systems (SVS), which display the outside world in low-visibility conditions on the primary flight display (PFD) are just a few ideas Bateman believes can improve the man-machine interface and reduce LOC accidents.
From a flight training organisation perspective, it is again training that is believed to be unsuited to the modern airliner cockpit. Bill Roe, managing director of the Saudi Aviation Flight Academy (SAFA) and type-rated pilot on the Phenom 300 (a highly automated aircraft), believes the issue is rooted in training. From ab-initio right through to airline recurrent training, there needs to be new focus on the fundamentals of flying, he said. If the autopilot fails a pilot must be capable of hand-flying the aircraft and, in all but the worst case scenarios, power and pitch will still enable an aircraft to fly safely. An aeroplane is still an aeroplane and the principles of flight and aerodynamics still apply.
Roe also acknowledges that analogue instrumentation is a thing of the past in most modern airline fleets today and, as such, “we need to find a way to teach the basics in a glass cockpit”.
That is not to say, though, that pilots should be flying on autopilot immediately. “Pilots must be proficient in manual handling before being taught to use cockpit automation.”
Roe noted how many Gulf carriers have begun adapting SOPs; encouraging pilots to hand-fly sectors to keep manual skills current. In fact, according to Captain Martin Mahoney, SVP flight training at Emirates Training College, Emirates worked with the automation working group in the US to create an automation policy, introduced in 2010, which encourages pilots to consider the threats on the day before deciding what level of automation to apply.
Emirates also conducts a lot of simulator training, over and above the regulatory minimum, of which much focus is placed on manual handling. Emirates believes it is important that sufficient simulator time is given to manual handling skills as well as flight deck automation. “Pilots have to be knowledgeable, proficient and competent in flying today’s advanced aircraft in both modes,” said Mahoney.
Marwan Atalla, CEO of Ayla Aviation Academy in Jordan, said: “Automation can always be turned off. However, it is company regulations that dictate when pilots must use it – with most requiring pilots to use autopilot shortly after take-off until shortly before landing. It might be the right thing to do but then it has to be offset by additional manual training.”
The problem with this is that it will cost airlines more money – even going back to grass roots flying in light single/multi-engine piston aircraft incurs additional costs – and while comparatively cheap compared to an airliner, it is still an extra cost that many airlines would feel aggrieved by, even more so in these days of reduced operating margins.
Some commentators have suggested pilots spend more time hand-flying in everyday operations. Rory Kay, executive air safety chairman of the Air Line Pilots Association (ALPA) has stated that pilots need more time hand-flying aircraft to alleviate the dulling of manual flying skills caused by over-reliance on automation.
United is one airline that actively encourages pilots to operate manually and Kay often hand-flies the aircraft below 18,000 feet – something Roe advocates as he, too, flies manually from take-off to cruise altitude before engaging the autopilot.
Critics of this approach would argue that this is more expensive as hand-flying tends to be less efficient and that it is safer to let the autopilot handle the aircraft, with fewer ‘level busts’ and flight path deviations. Indeed, in some airspace, where reduced vertical separation minima (RVSM) apply, use of autopilot is mandatory.
Captain Lloyd Watson, commercial director of type training at Oxford Aviation Academy, also presented at the Flight Safety Conference and firmly believes that a rethinking of the content of type training and recurrent training is needed. After conducting a gap analysis he found that the cost of training is key. Whether it is the airline or the student that foots the bill it seems that in most cases only the minimum amount of training, as set by regulators, is undertaken to minimise the final cost.
Currently there is no requirement for type ratings to include training in such things as PRNAV, use of head-up guidance or economical flight procedures; all of which are used in modern cockpits. Subsequently, these systems are usually not trained for during type-training – to save money – and this is where the gap emerges. The resultant gap is then bridged during line training and on the line. This poses the question, is the line really the best place to learn how to operate these complex systems? Captain Watson believes the answer is no and that ultimately the problem lies with regulators who set minimum requirements for license and type-rating issue.
The Gulf Aviation Academy (GAA) agrees: “An aircraft with passengers on board is not an appropriate training medium. Skills are developed, enhanced and rectified through practice and training and the best and safest training tool for that is a simulator.”
GAA further argues that pilots are exercising manual flying skills less frequently because modern aircraft are designed for, and pilots are encouraged in, maximum use of automation. If appropriate measures are not taken to ensure fluency this could become an issue. In terms of incorporating more manual flying in everyday operations, GAA doesn’t believe this would be of benefit, stating that pilots require practice in simulators under abnormal, degraded environmental conditions, ensuring that in addition to sharpening piloting skills, pilots will build confidence as well.
In terms of adapting training GAA notes that the industry is already introducing programmes such as the advanced qualification program (AQP) and evidence-based training (EBT) where training and checking is tailored and customised to individual operator’s needs to address unique deficiencies.
Former Royal Jordanian captain and personal pilot to His late Majesty King Hussein of Jordan, Captain Bader Steityeh, also believes a gap exists in airline pilot training, He said: “The problem lies not in automation but in bridging the gap between a basic aircraft pilot and fly-by-wire technology. When manufacturers first introduced fly-by-wire aircraft, the type-rating course was fully comprehensive and dedicated a lot of time to showing pilots the difference between conventional and fly-by-wire aircraft.”
Additionally, said Captain Steityeh, type-rating ground school and refresher courses were conducted by instructors who, through dialogue, could pinpoint weaknesses in pilots’ knowledge. Now computer-based training has replaced classroom instruction and any shortfall is not picked up unless a student fails the computer-based exam. Also, owing to airline financial constraints, the expanded type-rating course was cut by almost half, giving less time for pilots to become familiar with the intricacies of fly-by-wire aircraft.
However, Captain Steityeh goes on to say: “Automation is still the safest option provided pilots are trained not only in how to use it, but also how to correctly identify any problems that may arise. In my opinion, safety must not be jeopardised by cost-cutting.”
Harry Nelson, experimental test pilot and deputy SVP flight and integration tests at Airbus, added: “While there have been great advances in aircraft technology over the last 30 years which have had a very positive effect on the safety of air travel, it would seem from recent events that, with one or two exceptions, in general the training of today’s crews has not kept pace with that rate of change... some of the recent incidents and accidents have revealed significant weaknesses to the general approach being taken in the worldwide training scene.”
He further argued that too much focus is placed on testing where it should be on training.
Hank Krakowski, former COO at FAA Air Traffic Organisation and former VP flight ops at United Airlines, believes automation is not solely to blame but that “certain operational environments also facilitate a ‘dulling’ of basic flying skills – long-range flights have limited opportunity for hand flying with insufficient frequency of take-offs and landings.” He added: “Like any skill, if unpractised, atrophy is inevitable.”
Richard Ellis, flight safety manager at Florida-based Spirit Airlines and former 707, 747 and A320 pilot, who describes himself as “one of those old guys... the kind that grew up on ‘steam gauge airplanes’,” believes there is an over-reliance on automation in the modern cockpit.
“We are now sometimes completely subservient to the whims of the automation on these ‘new airplanes’ and seem to abrogate authority to the aircraft as if it’s a living, breathing, authoritative entity,” he said.
Ellis acknowledges that modern aircraft have far better safety attributes but argues these attributes have led to pilots “losing the professional abilities [they] strive so hard to master in flight school”.
He further maintains that modern fly-by-wire aircraft lack the tactile feel of older aircraft, which may be an obstacle to better manual flying. He accepts that “we live in a different era, with different aircraft and certainly different attitudes”, but argues, “two things will never change – we have to understand that basic airmanship will always save the day and that we, as professional pilots, are still captain of the ship”.
Stay up to date
Subscribe to the free Times Aerospace newsletter and receive the latest content every week. We'll never share your email address.