Friend and foe
Automation anomalies
Ever-advancing technology has delivered aviation many benefits in the realms of safety and efficiency, perhaps no more so than in the advancement of flight deck automation.
Autopilots have grown to share the workload and allow the crew a greater monitoring role in lieu of manipulation. However, the relationship between man and machine can be a double-edged sword.
This content is available exclusively to Australian Aviation members.
A monthly membership is only $5.99 or save with our annual plans.
- Australian Aviation quarterly print & digital magazines
- Access to In Focus reports every month on our website
- Unlimited access to all Australian Aviation digital content
- Access to the Australian Aviation app
- Australian Aviation quarterly print & digital magazines
- Access to In Focus reports every month on our website
- Access to our Behind the Lens photo galleries and other exclusive content
- Daily news updates via our email bulletin
- Unlimited access to all Australian Aviation digital content
- Access to the Australian Aviation app
- Australian Aviation quarterly print & digital magazines
- Access to In Focus reports every month on our website
- Access to our Behind the Lens photo galleries and other exclusive content
- Daily news updates via our email bulletin
The tragedy
I recently flew the simulated flight profile of the tragic loss of Turkish Airlines Flight 1951 at Schiphol Airport in 2009. The accident saw a Boeing 737-800 crash short of the runway, killing the flight crew, a flight attendant and four passengers and severely injuring scores more.
The aircraft had begun its approach with the autopilot engaged from a ‘high and fast’ situation, where the ‘autothrottle’ was commanding idle thrust as the crew tried to recapture the descent profile.
At about 2,000 feet the captain’s radio altimeter failed, decreasing from a reading of 1,950 feet to -8 feet. One of its functions is to signal the throttles to close at 27 feet above the ground during an automatic landing. But the 737 was still at 2,000 feet.
The crew manually advanced thrust as the speed decayed initially but with the autothrottle engaged and receiving false information, the thrust levers closed again. As the autopilot-guided aeroplane endeavoured to maintain its descent profile by ‘holding’ the glide slope, its speed continued to drop.
Over the next minute the speed bled back to only 83 knots — more than 40 knots below its planned approach speed. At 500 feet the stick-shaker signalled the impending stall but despite the crew’s efforts, the aircraft hit the ground a mile short of the runway.
To sit in a flight simulator and watch the situation develop is chilling. As the speed winds ominously back every fibre wants to push the throttles forward but for training purposes the tragedy is allowed to run its course.
On a subsequent approach, forewarned and forearmed, a recovery is initiated at the point that the Turkish Airlines crew tried to salvage the situation — and knowing what is going on and responding immediately, the recovery is only just successful at such a low altitude.
The overwhelming question that stemmed both from the simulator session and the accident investigation was — how did the crew allow the aircraft to become so slow without corrective action? Inattention has been cited as the obvious factor but even so, how did a modern jet airliner ever end up in such a predicament? Unfortunately, the Turkish crew is not alone.
More evidence
Accident records are littered with instances of automatic systems either failing or operating normally but being mismanaged by the crew. An Egyptian Airlines Boeing 737 was lost within minutes after takeoff in 2004 when it plunged into the sea. The captain had been an experienced fighter pilot but had less than 500 hours on 737s.
Initially, terrorism was suspected but the debris field on the sea floor was compact and showed the aircraft had not broken up in flight. What the cockpit voice recorder would reveal was that the captain experienced spatial disorientation during the initial turn after takeoff in the dark, wee hours of the morning.
A simple enough human error but what was more significant was the numerous calls of “Autopilot!” as the captain virtually ordered the automation to comply. Unfortunately, the autopilot had not fully engaged but neither the captain nor his co-pilot recognised this and neither intervened to ‘fly the aeroplane’ until it was too late.
The aircraft was banked at 111 degrees with 40 degrees nose down before the call of “No Autopilot!” was made. The 737 hit the water at slightly shallower angle but at 416 knots.
The same model 737 was involved in a close call at the UK’s Bournemouth Airport in 2007. Like the Schiphol accident, the aircraft was on approach with ‘autopilot coupled’ to an ILS and the thrust levers at idle. In this case, the autothrottle failed while the levers were retarded and the crew either did not receive or notice the ‘Autothrottle Fail’ light flashing.(faults were found with the system).
The speed subsequently bled back at about one knot a second until it fell well below the nominated approach speed. At 125 knots the captain called “Speed!” and the co-pilot, who had control, advanced the thrust levers. The stick-shaker activated and the captain took over and began a go-around without delay.
Even so, the Boeing’s underslung engines caused a strong ‘pitch up’ to 22 degrees as the power increased and brute force was required to keep the nose down and get the aircraft accelerating (Boeing’s manuals state that forward trim may also be required to get the nose to lower in such extreme cases as full forward stick alone may not be enough).
Even so, with 98 per cent N1 generated by the engines, the nose reared up to 44 degrees at its worst point before pitching down. The aircraft slowed to a only 82 knots but fortunately, the aircraft was recovered and a second approach was flown with a safe outcome.
Friend and foe
These accidents may portray the Boeing 737 in a poor light but this is not the case. I have about 8,000 hours on various models of the 737 and it’s a great aeroplane. However, when the issue of automation is raised, so often the Airbus is automatically held to account for the problems that arise.
showing the tremendous leap in technological advancement brought by modern EFIS flightdecks. New types
such as the 787 and A350 (pictured above) evolve the digital flightdeck far further. (Boeing & Airbus)
The Airbus family certainly has revolutionised flight deck automation and called for a different understanding of flight deck systems — but the potential threats extend to all aircraft types.
Man and machine are fallible. Both are prone to failings and being misunderstood, even when they are doing the right thing. Furthermore, the way in which pilots and automation interact, or ‘interface’, can be as challenging as flying the aeroplane manually. Critically, technology has advanced at such a rate that the human factor has at times struggled to keep pace and harmoniously co-exist within the confines of the flight deck.
What has not changed is the fundamental need for pilots to be ever vigilant. Just as human incapacitation can be subtle so too can failings in aircraft systems. Crews need to be wary of any aspect of the operation diverging from the norm, whether the autopilot isn’t behaving as expected, or the speed is decaying below what is needed.
The reasons can be wide and varied from incorrect pilot input to system failure or lack of systems knowledge. Whatever the reason, the basic adage of ‘fly the aeroplane’ never waivers.
However, herein lays the paradox. While automation has enhanced safety by freeing the pilot’s brain space from manipulation to facilitate management, many of basic skills surreptitiously have been eroded.
The ‘hands on’ aspect of piloting has become a smaller portion of the operation and yet when automation fails the crew, these are the very skills that are needed. Additionally, the incredible reliability of automatic systems has allowed a degree of complacency to encroach on the pilot’s older personality trait of healthy suspicion.
As pilots we need to find the balance. Automation needs to be used to facilitate safe flight management but can never replace vigilance or the pilot’s prerogative to intervene. If the aeroplane is slow, give it more power and if it’s high or low on approach and the automatics are struggling — fix it by flying.
Like any relationship, it is about how we interact and where we set our boundaries. We cannot compromise the relationship by being distracted nor can we expect too much of the other party.
Regardless of whether the automation or the human is at fault, ultimately the buck stops with the pilot, the final filter with the ability to interpret and respond in a manner that may not be able to be programmed into a computer.
It is impossible to foresee every scenario and eventuality when designing a system. The good news is that this shortcoming in automation should keep pilots employed for the foreseeable future. The challenge is that the role of the pilot in an age of ever-increasing technology calls for the constant revision of skills and processes.
Regardless of aircraft type, pilot experience, environmental conditions or subtle systems failures, distraction or pilot mismanagement can cause automation to turn and bite the pilot.
The very systems designed to assist the crew and enhance safety ultimately can have the opposite effect. Automatic systems have so much to offer but the relationship must always be treated with healthy respect. Failing to do so can leave the aeroplane and crew trapped in a corner from which they might not be able to escape.
A terrible situation that can arise when a friend becomes a foe. There but for the grace of God . . .
Want to see more stories from trusted news sources?
Make Australian Aviation a preferred news source on Google.
Click here to add Australian Aviation as a preferred news source.