Ever since the Wright brothers first took to the air – albeit for a few dozen feet – the dream of conquering the skies has cast a shadow that seldom gets mentioned in all the romanticism that surrounds flying. That shadow is death or at least the imminent danger of it. So what of it? How dangerous has flying been over the past century since the maiden voyage of man-powered aircraft?
What comes up, must come down. Sometimes unintentionally.
Crashing a plane is not a common occurrence – at least not nowadays. A very rough estimate of the odds of an aircraft going down due to an emergency is about 1 in 11 million, meaning it would take the most lucky of us quite a few lifetimes before actually experiencing a crash. And yet this figure is ultimately so small due to the large volume of flights in the air – so make no mistake, planes do come down. Maybe not as often as they used to, but it’s something to consider.
Disclaimer: since the entirety of this analysis is based upon crash data (so not the actual frequency of crashes contextualized with metrics such as distance and flight times) – there isn’t any simple way to say which aircraft, manufacturers or airline companies are in fact most safe, or indeed least safe. But this isn’t the goal of the analysis I’m currently doing. What I’m trying here is to put together a picture of what crashes have looked like ever since our first few seconds of air-time. Having said this, I recommend this source provided by a helpful friend, for those wanting to see the history of aviation safety oversight in the United States.
Checking crash frequency from the past century makes a few things stand out. For starters, wars tend to cause spikes in aircraft downings – World Wars 1 and 2 jump out from what otherwise seem like neutral or downward trends. We also see an ascending trend for the number of plane crashes up until the last quarter of the 20th century. The logic with this is pretty straight forward; the data seems to suggest that as planes and other aircraft gained more traction as a means of transportation (so not just for rich folks and the army) the number of crashes has also gone up – and it isn’t until the late years of the 1900s that we see this trend taking a dive. Conversely, this late downward spike can be attributed not to a decrease in the number of flights – but more to better safety standards.
In recent years the trend has continued on a downward slope, albeit with a few hiccups. In 2010 there were 167 recorded plane crashes worldwide, in 2011 we saw 155, 156 in 2012, 139 in 2013, 122 in 2014, again 122 in 2015 and in 2016 we’ve had a total of just 85 plane crashes worldwide.
Nobody wants planes or other aircraft to go down – much less if people die as a result. Given the severity of even minor slips in safety while flying, it’s become an imperative for most modern airline companies to continue the development of better and safer aircraft – with improved design, electronics and control systems. This causes for fewer planes to go down (even given the growing number of aircraft deployed worldwide) – but increases in survivability tend to be hard to pinpoint.
The available data points suggest that during the past 100 and some years of manmade flight the death toll from aircraft crashes has shifted along with the ups and downs of how many crashes have occurred. Furthermore, when looking at survival rates we see that there aren’t any real improvements or retrogressions in the trend. How can this be? Shouldn’t better safety standards and flight policies also make for better chances of getting out of a crash alive?
I’ll get to this with more detail in a moment, but I believe it’s safe to assume that the reason why we see the survival rate of crashes jumping up and down around the 20% mark is due to the nature of flying itself: hitting the ground at high speed in a machine built for the opposite makes any improvements to safety standards very hard to feel. Crashing is very dangerous, and you can’t really control what happens when aircraft go down.
But safety standards enable you to try and minimize the risk of planes falling in the first place. Learning from mistakes might not be the best way to improve, but it has changed the way planes are designed in such a way that survivability is treated with exceptional priority, all the while reducing the chances for human or mechanical errors that might down the aircraft.
Is there such a thing as a bad time for flying?
A few calendaristic months have stood out during the past century due to their proclivity for hosting crashes. Conversely, it’s those very same months that see the lowest crash survival rates. November is the month which has seen most people die on account of aircraft crashes, followed closely by March, August, and December. With so little difference overall it’s plausible that a single downing of a full 747 might change the ranking, so it’s fair to say that the overall danger is more spread out than definitive for a single month.
What’s more important to see out of this is that the few months that stand out also coincide with the most dangerous times when it comes to weather: November and December are winter months in the northern hemisphere, so it’s presumable that poor meteorological conditions had a say in these months spiking in terms of how many victims they claim. August is also a month predisposed to adverse weather – as it’s the peak of hurricane season in the Atlantic Ocean.
Looking at days we see little overall variance, but there does seem to be an apparent trend by which accidents in the middle third of the month host more fatalities. By extension, the days with the highest survival rates among people that are in aircraft crashes tend to be towards the end of the month, with the lowest survival rates being smack down in the middle.
An hourly look shows something that might be at first counterintuitive: most crashes take place either at dawn or at dusk, with very few crashes (comparatively) during nighttime and early morning hours. While it’s tempting to say that this is because fewer flights take place during those specific time slots, it’s a much more plausible explanation to say that it’s specifically during early morning and evening that people are most tired. Extending this argument would mean that it’s due to tiredness that human error rears its ugly head, and hence more crashes take place in the timeslots where either ground crew, pilots or other aeronautical staff are more prone to committing mistakes.
So why do crashes happen in the first place?
Having reached this point, a few more questions come into play that haven’t yet been tackled. One of them underlies almost every crash investigation since the beginning of human flight: why did the thing crash in the first place?
In order to answer this, I took all the available crash descriptions since 1908 and up to 2009 and looked for repeating patterns. A look at single word frequency shows how most crashes include, either as an active or passive element, things like the pilot, the crew, the cargo, natural elements such as the landscape or the weather conditions, airport elements such as runways, but also active elements that hint at most common situations when crashes occur (i.e. maneuvers such as landing). From what descriptions are available, it seems that landing and taking off are some of the most common times where planes and other aircraft crash.
Still, looking deeper into this I’ve found that pilot error is the most commonly cited reason for planes going down. Since the 1960’s a total of 640 crashes (or 58%) were later attributed to piloting mistakes. And yet it isn’t pilot error per se that causes some crashes that are attributed to something he or she has done wrong: improper communication is often the reason behind the spark that causes fatal crashes to occur. This can be caused by a lot of things – but one of the most commonly cited ones is fatigue caused by the disruption of one’s circadian rhythm, insufficient sleep and long hours at the helm.
Mechanical issues are also a major factor (195 crashes, or 17%). Engines, for all their worth and all the tech that goes into them, tend to fail for the same few reasons over and over again. These include fuel starvation, fuel exhaustion, foreign object damage, engine breakdown due to metal fatigue, improper maintenance – or even mechanical failure as a result of manufacturing defects gone unnoticed.
Other causes include weather conditions (63 instances in the dataset, so 6%), and it’s because of the perils of the weather that all commercial flight plans require ground dispatch to constantly check meteorological conditions as the aircraft are in the air. It’s their job to let pilots and crew know of any impending dangers that might force them to change route.
Snow (and ice) are major contributing causes to a vast part of crashes. That’s because icing or frost built up on a plane can hurt its ability to create lift – and while this isn’t that big of an issue on modern planes (because their fuselages and wings are swept backward), it can become a major problem on older aircraft that tend to have straight and rectangular wings and tail ends.
While I’m on the topic of weather, I’d like to also highlight that for the most part planes aren’t affected by lightning. Big planes do get struck by lightning twice per year (on average) and come away clean almost all the time – but new research would suggest that there’s more to thunderstorms than first meets the eye. While it’s true that most planes are safe and equipped sufficiently to handle negative polarity lightning strikes (which are the overwhelming majority worldwide), there are considerable doubts as to whether aircraft are prepared to handle positive polarity strikes (which are much more powerful, but so very uncommon that it wasn’t until 2006 that safety standards were adapted to fit the threat they pose).
Other reasons planes go down include impact by foreign object debris. These include anything that might interact physically with the aircraft, that isn’t meant as a part of the aircraft itself. While this would normally encompass things like volcanic dust, hail, and even birds (yes, it’s not that uncommon for wild ducks and geese to get sucked into jet engines mid-flight) – it also includes forgotten tools from ground crew or even debris on the landing strip.
The people’s story: what crash data can tell us
It’s redundant to say that over the past century a lot of the people unlucky enough to be in aircraft crashes had also perished, but what the data shows us is that there’s some room to maneuver in this perspective. While it’s true that crashes seem to kill all their passengers (especially when the total load is under 200 people) it’s also worth pointing out that a lot of people survive. The below depicts this, as there are two broader trends that become plausible: either everyone dies in the event of a crash, but also that few to none do so.
The FAA developed new tests in the 1980s to test parts of aircraft that are more prone to collapse (such as cabin ceilings, cabin walls and aircraft partitions). Following this, in 1988 all aircraft manufacturers were required to use safer materials that spread heat out instead of concentrating it, also making for less smoke in the event of a fire. These regulations also caused changes to the interior design of all air-ready aircraft – all of which have had a great influence and have increased survivability.
So while looking at survival rates, what we see further points to this: crashes are either really bad or relatively harmless; and maybe we can attribute the harmless part to tougher flight safety standards in the last quarter of the 20th century. Smaller planes with less than 200 passengers seem to be the biggest gambles when it comes to surviving a crash, but as the size goes up what we see is a discrepancy – aircraft crashes either kill everyone or little to no people at all.
The golden rule issued by the FAA is that all airplanes be set up so that evacuation takes no longer than 90 seconds. This process is guided by lights on the floors and ceilings of aircraft cabins and reflectors or other markings within the cabin itself that show people which way to take in order to reach emergency exits. These relatively simple improvements alone are credited with increasing evacuation speed by as much as 20%. And it all started in 1986, a year when all American issued planes were fitted with emergency lighting.
Cumulatively, all the safety standards existing today bring the odds of being killed on a single airline flight to about 1 in 4.7 million (but on safer airlines, it can be as good as 1 in 19.8 million).
One little-known point is that survival rates are greatly improved by good seat design. What this means is that if the seat can remain in place, even in the event of a violent crash (and also absorb some of the impact energy), then people stand a fighting chance to get out of a crash and live to tell the tale. This is why seat design regulations have come from being able to stand 6G forces in the early days to 9G in the 1950s – to a hefty 16G in the planes of today.
An interesting point is that survival rates also fluctuate depending on seat positioning within the aircraft:
- First / Business class: 49%
- Ahead of the wing: 56%
- Over the wing: 56%
- Rear cabin: 69%
But even so, looking at frequencies of death makes for a somber picture. By far, in most aircraft crashes on record, the fatality rate is 100% – with events that have no fatalities very rare.
The geopolitical perspective on aircraft crashes
The United States has seen the most crashes since man-made flight came about, and it also has the most deaths on record: over 16.000 people have died (until 2009) in American aircraft crashes.
But when it comes to fatality rates, it seems that the US isn’t all that bad. The most dangerous place to crash since 1908 is the now defunct USSR – where there was an average crash fatality rate of 92%. By comparison, the United States ranked up 84% – which is quite the difference when considering how much more volume American flights had, compared to Soviet ones.
The quirky relationship between aircraft popularity and crash fatalities
Passenger airplanes were first introduced in the 1930s – a time when the American middle class started aspiring to leisure travel – and since then aircraft manufacturers have constantly put in effort into making them both profitable and safe.
As Alan Anderson, former Boeing Chief Payloads Engineer says: “Accidents involving the current generation of commercial airplanes are rare but offer important insights into advancements in the safety and crashworthiness of airplane design. These advancements reflect decades of innovation and targeted efforts to improve survivability in an airplane accident, especially during takeoffs and landings.”
But over time, a few famous airline companies due to the large volume of their flights have also managed to stand out when it comes to how many people they’ve lost during aircraft crashes.
By looking at the above, a few explanations become plausible. Outliers from the pack include world famous airlines that have vast amounts of planes in the air. These companies score higher victim counts due to more of their planes having crashed (which, of course, doesn’t mean they are more dangerous – it just means that they have a lot bigger fleets). Just by looking at the data there wouldn’t be an easy way to judge which airline is most safe – but the opposite might be doable.
For one, we can see that the US Air Force has had more than double the fatalities of almost all airline companies taken into account (while not technically an airline company, I decided to include the Air Force for a more comprehensive picture). Moreover, the Air Force has a fatality rate (in the event of a crash) of almost 100% – meaning that it’s subject to some of the most frequent and dangerous aviation accidents worldwide. But it’s not the biggest outlier.
Aeroflot Russian Airlines bests all other airline companies worldwide with 7156 victims and a crash fatality rate of 87%.
Aircraft manufacturers also follow similar patterns. The more aircraft a company produces, the higher the death toll associated with its name. The all-time leader in this respect is Boeing, a company that has seen 18569 people die in crashes which involved their planes (again, until 2009). Yet it must be mentioned that Boeing isn’t the most dangerous producer – Russian aircraft tend to lead in this respect with 84%, as compared to Boeing’s 75%.
But just to put things into perspective, the world’s worst ever plane crash was that of a Boeing. A 747-200 to be precise, which crashed over Japan on the 12th of August 1985 and killed a massive 520 people. Only one crew member and three passengers were reported as having survived.
Lastly, I took a look at aircraft models to see which have killed most people when crashing. While it’s fair to say that most airplanes tend to have the same amount of victims overall, I found the mother of all outliers: the DC-3.
Part of the Douglas Aircraft Company’s lineage, the DC-3 was itself built in the 1930’s and made worldwide passenger flights cheap, comfortable and very profitable. Easy to fly, easy to maintain mechanically and versatile when it came to where it could take off and land (be it grass, dirt, water or tarmac) the DC-3 became one of the symbols of the American might during World War 2. In fact, it gained such a good reputation during the war that it became one of the go-to aircraft for commercial airlines after 1945.
(above: the Douglas DC-3)
Even today some 2000 DC-3’s are still airworthy – and it’s due to the overwhelming popularity of the DC-3 (that spans over 80 years) that it’s also seen a lot of crashes – both in times of war and in times of peace.
With all this talk about crashes and fatality rates, I suppose I should end this on an up: chances are, not only will you never be in an aircraft crash yourself, but neither will you ever know someone that will experience a crash.
In a world where air traffic looks set to keep growing, we’re also thankfully experiencing fewer crashes by the year – with the majority of them in light and ultralight planes (so not the commercial flights we’re more accustomed to). But just in case, keep your seatbelt buckled and know your nearest point of exit in case of an emergency – it might save not just your own life, but the lives of those in your immediate vicinity.
Researching this post wouldn’t have been possible if not for me stumbling upon the work of Sauro Grandi on Kaggle – and his linking to the original data source on OpenData. That being said, I’d like to give a big thank you to Sauro and the other guys and girls at Kaggle and OpenData for making the internet a better place.