2020 was an interesting year for aviation. It was dominated by Covid, which saw traffic numbers fall to the levels of several decades before – which is why a review of the accident statistics is an interesting one to consider.

What sort of accidents are taking place?

The primary accidents seen in 2020 are unsurprisingly similar to those seen over the last decade:

• Runway excursions
• Loss of control in flight
• CFIT
• Abnormal runway contact (hard landings and tail strikes)
• Actually missing the runway (undershoot and overshoots)
• System malfunction or failure
• Fire

We wrote a bit about these in a bit more detail not that long ago. We called it the ‘Seven Deadly Things’ and you can read it here.

What are the 2020 stats?

Well, first up, 2020 was roughly the same in terms of capacity as 1998 – a year known for Bill Clinton, the inception of the Euro and the movie ‘Titanic’. Yep, that long ago. So, same traffic levels, but different accident rates – 1998 saw 10 fatal accidents and 24 hull losses compared to “just” 3 and 6 in 2020.

But if we compare the 2020 numbers to 2019 it paints a different picture. Or rather, it is actually a very similar picture. While there there were only roughly 50% the number of flights in 2020 that took place in 2019, there were still 75% the number of fatal accidents. 

OK, this isn’t a very telling statistic since we’re talking 3 instead of 4 and neither is huge, but it does mean the fatality rate and hull loss rate went up per million flights in 2020. It was not a significant increase, but it is enough to suggest that yes, not flying regularly can lead to more accidents and incidents. 

Not really news there then, but something worth considering.

Point number 1 – Lack of flying leads to mistakes

If we take a leap back to 1958 and look at the accident rates through the decades then there has been a steady overall decline, and now we are sitting “comfortably” at under 5 fatal accidents per year, while flights have increased from about 12.5 million (1989 sort of time) to 35.8 million (the peak in 2019).

So, in thirty years the rate per million flights has dropped significantly to around the 0.17 per million flights point, and hull losses to 0.34 per million.

How did it get so low?

Significant leaps have been made in aircraft design over the years and this has had a huge impact on safety levels. Of course, training, CRM, Human Factors awareness and all of that has played a part too, but the major pat on the back goes to the airplane builders. For every silly mistake a pilot has made, they have generally identified it and then helped prevent it by building us better instruments, more robust systems, or things that catch our mistakes for us.

In fact, if you look at the fatal accident rates per million and then break it down into aircraft generation, it has dropped from 3.0 to 0.1 , and 5.4 to 0.2 for the hull losses. So technology is helping us. A lot.

Those big ones – the CFITs and LOC-I accidents – have reduced by 86% and 89% because of technology upgrades from Generation 1 to Generation 4 aircraft. This is down to the introduction of things like glass cockpits, FMW and TAWS systems.

How low can it go?

Can we reduce the occurrences to zero? If not, even with all this handy automation, then why not?

Well, these statistics offer us an answer there as well.

They are taken from across civil aviation, revenue flights on western built commercial jet aircraft that carry over 40 passengers, and also big cargo ones. It doesn’t include non-western built aircraft (possibly because the safety records on them ain’t great), and it doesn’t include Business Aviation.

Why not? Well, because the operational environment is very different, and very different in challenging ways. 

So, we are looking at the accidents which have involved nice, relatively modern commercial aircraft generally piloted by experienced folk going into places they have gone into many times before. And yet they are still managing to get it wrong.

What’s more, we’ve seen how automation is helping – it has brought us down to a very steady level.  So what is going on? We recently published a piece on the ‘Hidden Risks of Automation’, which we think offers some of the answer.

The ‘Problem of the Person’

Unfortunately, the solution to the Problem of the Person is not a simple one.

‘Human Factors’ might give us some reasons – poor decision making, bad workload management, lack of understanding the systems, but none of these really provide the answer to correcting it. The work now comes down to us.

1. Don’t Become Complacent: We have multiple systems put there to provide another layer of safety but we are seeing pilots rely on them as the only level of safety. These systems are a last line of defence though, not the the only defence.

ROW/ROP should supplement good landing performance assessment and stabilized approach management.

TAWS and GPWS systems give us a hard floor that we must not go below, but our own situational awareness should keep us well away from ever having to hear those calls.

Autopilots, flight protections and warnings should be a final alert, but basic airmanship and handling skills should correct our flightpath long before we reach a level that needs those systems to help.

2. Poor Decision Making and Workload Management: None of our clever automation and systems have the ability to think and question for us. So we need to make sure we are doing this, and we need to make sure we are doing it in the right way. Ask the right questions, gather information and use your resources properly.

Ask “What does this mean?” – Diagnose the problem not based on what has happened, but on what the impact and consequence of that failure is.

Ask “What has changed?” –  Review your decisions. Don’t fit new information into the solution you’ve already picked, rather adapt your solution to consider the new information. 

Ask “What do you think?” – Open-ended questions that gather input from someone else might catch things you have missed, or misinterpreted.

3. Just Do better

When we have seen automation and systems reduce the number of occurrences down to this point where the vast majority of accidents are down to human error, there really is no better solution than us Just Doing Better.

But this ‘better’ falls on the whole industry.

Sharing information, experiences, supporting development in others and improving training and pilot resilience. 

There are multiple projects out there:

• IATA and the Flight Safety Foundation have just released their recommendations for reducing runway excursions (GAPRE).
• ICAO are implementing new Runway Condition Assessment and Reporting standards from the end of this year.
• UPRT training is being developed and improved.
• IATA and ICAO Evidence Based Training development is shifting the training paradigm to train competencies rather than practicing solutions to singular events.

At the end of the day, aviation has grown progressively safer and more efficient over the last few decades, but the trend is flattening out and the same events seem to be occurring, for the same reasons. The ball is now in our court to try and fix the remaining issue – because, as harsh as it sounds, that issue is us.

Fancy reading some more?

• We got a lot of our info from the Airbus Safety Analysis report, and you can check it out here.
• The Global Action Plan for Preventing Runway Excursions is full of recommendations. You can see the report here.
• Here’s one we wrote earlier on Unstabilised Approaches which are one of the most common precursors to runway excursions and abnormal landing events.

Over the past decades our industry has undergone an automation revolution.

Basic autopilots from eras-past were little more than wing levellers. Today they are sophisticated computers capable of awe-inspiring accuracy.

The industry has welcomed automation with open arms. And it’s no surprise. The vast majority of aviation accidents are caused by us, humans. Mechanical failure on the other hand only accounts for less than a quarter of all accidents.

So for operators and manufacturers alike the benefits of automation are clear – safety and efficiency. We are simply not as predictable or consistent as a computer because we are human. And automation has become a major line of defence.

But herein lies the problem…

It’s easy to see that a pilot’s role in the flight deck has changed forever as we interact with higher and higher levels of automation. Some might even argue that we are being progressively designed out of the cockpit completely and to some extent this may be true. Whether we like it or not, full autonomy is coming. Take the Xwing Project for instance – their concept can be retrofitted to conventional aircraft enabling them to fly without a pilot.

But right now the more pressing issue is that our role continues to transition more and more from flying airplanes to managing automation. Put it this way. A recent study found that across a large sample of flights aboard the Airbus A319, pilots were spending on average only 120 seconds manually flying each flight. And that was the middle of the curve.

This creates a unique set of risks that the industry collectively needs to better address.

Good Automation

By no means is this an attempt to detract from the positive impacts that good automation continues to have in our skies. The benefits are no secret. When used as intended it is a huge work-load reducer. It allows us better flight path control and liberates us from repetitive and non-rewarding tasks – something humans are known be no good at. We become less fatigued and have more capacity to deal with other things.

It also works in unison with systems like ECAM and EICAS to better help us manage things when something goes wrong.

Automation has also improved the skies we fly in. Fantastic things like RVSM and PBN have allowed us to fly closer together and make better use of crowded airspace. While around the world minimas grow ever closer to the ground thanks to things like RNP approaches where automation can help us ‘thread the needle’ in some one the world’s most challenging approaches.

Take Queenstown for example. The notorious airport down in New Zealand boasts beautiful scenery but a reputation amongst pilots as being one of the most demanding in the world due to the intimidating terrain that surrounds it. RNP approaches have dropped minimas from over 3000 feet off the deck to less than 300. And now you can land there at night.

Bad Automation. Here is where things start to go wrong.

All positives aside, automation is also having an effect on us pilots. And it is important to remember  just that – we are still pilots. We must never lose the ability to fly without automation. Back in 1997 the late and well-respected Airline Captain Warren Vanderburgh saw it coming and coined the phrase you are no doubt familiar with – Children Of The Magenta Line.

This remains true to this day. If we become too reliant on automation, avoidable accidents happen. Here’s why.

It Erodes Skills.

Slowly but surely automation is chipping away our manual and cognitive flying skills. You know the ones – your stick and rudder. We are being actively encouraged to keep automation on and control our trajectory through it. Do that for long enough and we begin to forget how to do it the other way – with our hands, eyes and feet.

It Distracts.

Because we are so used to flying our airplanes through automation, when something unexpected happens such as short notice changes from ATC our immediate response is to try and figure out how to make the automation accomplish it. We go heads-down precisely when we should be going heads-up – and the clock is ticking.

It Confuses.

Chances are if you have operated anything with high levels of automation, at least once you’ve uttered the infamous phrase “what’s it doing now?”

And yet still we are reluctant to turn it off. As soon you identify that the aircraft is not going where it should, that’s your cue to intervene. The minute you don’t, you are simply along for the ride. Pilots around the world would agree, this is never good enough.

Mode confusion is another. Modern automation features many different ways of achieving the same outcome, but with subtle and sometimes dangerous differences. We need to understand the limitations of each one because if we don’t, we know that tragedies can happen.

A little known incident in Australia serves as a good example. Snowbird, an Airbus A319, was on approach at YMML/Melbourne airport on a clear calm evening. A tired but highly experienced crew were flying an unremarkable STAR and ILS approach at the highest level of automation. All was going well until the pilot flying reached up to arm the approach in a dimly lit cockpit. He pressed the wrong button. Over the next 39 seconds chaos ensued.

What followed was a series of rapid fire mode changes, confusion and attempts to salvage the approach through the automation. Three EGPWS warnings were triggered and an altitude alert issued by the tower as the airplane reached just over 1,000 feet off the deck at 315 kts before they regained their situational awareness and executed a missed approach.

After the incident neither pilot could recall exactly what happened, what modes they had engaged and neither had heard any of the EGPWS warnings. The automation had performed flawlessly throughout by providing the crew exactly what it was told to do. When it all went wrong, it seems the pilots were reluctant to turn it all off.

It Startles.

Automation is designed to give you back control when something goes wrong. For crew our first indication is usually a loud aural alert and a flashing red light. For systems that seem to operate flawlessly flight after flight, day after day, the affect can be startling.

Pilots are suddenly given full control because we are supposed to be the ultimate fail safe.

We are not even supposed to be there unless we can fly our aircraft manually without hesitation. But the problem is we are not used to flying manually anymore. We are used to flying through automation, so when it’s suddenly not there it’s like going back to school.

There have been a number of instances where pilots have been faced with failing automation and have been unable to keep the aircraft flying safely using manual control.

Air Asia Flight 8501 is a good example. To get rid of a nuisance alert the crew pulled a single circuit breaker to one of the aircraft’s flight control computers. As an unintended consequence the autopilot disconnected and the aircraft transitioned into a degraded mode of flight where the automation was no longer available and flight protections were removed. It had done what it was designed to do – hand back control to the pilots.

Tragically the pilot flying, startled by having to fly manually in a degraded mode, stalled the aircraft from straight and level flight. The crew never managed to regain control.

As an industry our approach to how we interact with automation has to change.

Automation dependency is not a new issue. But as automation becomes more sophisticated and complex we have to continue to manage how we interact with it.

It was never intended to replace our core skills and abilities as aviators, only to better support them. Like the image below our core ability to fly manually is supposed to be a constant.

But there are some ways to help.

SOPs. They must be flexible enough to allow pilots to turn the automation off when it is appropriate. You have to give pilots the freedom and confidence to use their hands and feet. Six months between sim sessions is too long.

Training. Evidence based training is revolutionising our sim sessions. There is opportunity there to encourage manual flight. To turn it all off without warning and give us the much needed confidence back.

Monitoring. We need to encourage active monitoring so that we can intervene quickly if we need too. We should always be mentally flying the plane even if an autopilot is flying. One way to do this is by keeping our hands on the controls during dynamic phases of flights. It is a tactile reminder that we are still in control and can take over at any stage.

Practice. It makes perfect. It’s what we got into this game for. When conditions are right and workload low, take the opportunity to turn it all off. It’s right there waiting for you again if things get busy.

Automation is here to stay.

What matters is how we use it. We cannot allow it replace our abilities to fly an airplane without it because for the foreseeable future we will still be the ultimate failsafe.