23

It's often mentioned that flying is the safest form of transportation, and no doubt this is true when considering only a long-haul trip. There's no doubt, for example, that driving (or taking a train or bus) 3,000km would be more dangerous than flying the same distance.

However, most flight crashes occur at take-off or landing and so I suspect there must be a break-even point after which flying becomes more dangerous than alternatives. For instance, if a flight is only 300km, would it be safer to drive?

For the purposes of this question, it can be assumed that I'm talking about OECD countries, which by and large have very high standards for both flight safety and the safety of other forms of transportation.

user1205901 - Слава Україні
  • 375
  • 1
  • 3
  • 8
  • 4
    I suspect you'd have to hypothesize an airline flight on the order of a few miles before you could get the numbers to work out the way you're suggesting, and at that point economic reality would intrude, because such fights would be expensive beyond all reason. And no quicker than driving, given time taken to board & deplane & etc. Commercial aviation (*in the developed world) is literally safer than living (in deaths per million man-hours), so even with the higher risks at takeoff & landing, it'll be hard* to get the lines to cross. – Ralph J Oct 18 '19 at 01:11
  • 1
    Would regularly scheduled helicopter routes count? There are and historically have been airlines operating dedicated short (10-20min), regularly scheduled, and standard fare helicopter routes. Example JFK->NYC and SFO->OAK (Historical) – crasic Oct 18 '19 at 16:28
  • 2
    The 737 MAX has some pretty short flights. – Ed Plunkett Oct 18 '19 at 16:33
  • I believe airlines (how many?) do a lot of short flights in Hawaii for the Island hopping. Lots of quick pressurizations and depressurizations cycles. The only incident that comes to mind is Aloha Airlines Flight 243. – jww Oct 18 '19 at 17:32
  • 5
    I'm assuming if your flight lasts less than 100 metres in a jetliner, it would be very unsafe indeed. – Richard Oct 18 '19 at 20:07
  • 1
    This question seems to imply that touch and go's add danger to flying... – Michael Oct 18 '19 at 21:55
  • Safer than what? Buses and cars, no chance. But since Amtrak's inception, it's been a footrace, with the stats rocking back and forth depending on the last big accident... The DC-10 troubles, Gunpow, 9/11+Queens, Big Bayou Canot. Between that last and a phenomenal run of luck for the airlines, they hold the trophy. But Amtrak finally has PTC which will prevent a range of accidents, and air is unable to achieve that innovation. – Harper - Reinstate Monica Oct 19 '19 at 18:06

1 Answers1

38

Note: This does not take travel to/from the airport into account.

The Wikipedia page on aviation safety has a nice table with deaths per journey, time and distance (based on data from the UK between 1990 and 2000):

  • Car: 40 deaths per billion journeys, 3.1 deaths per billion km
  • Aircraft: 117 deaths per billion journeys, 0.05 deaths per billion km

Assuming that car accidents are equally distributed along a journey, we only need to consider the 3.1 deaths per billion km. Further assuming that all aircraft accidents are related to takeoff and landing (i.e. once per journey), we only consider the 117 deaths per billion journeys. While the first assumption is probably reasonable (there are differences between types of road though), the second one is not quite true since aircraft can also develop problems during cruise.

With these basic assumptions, we can calculate when the expected number of deaths are equal:

$$ N_\text{death} = 117 \cdot 10^{-9} = 3.1 \times \frac{d}{10^9 \, \text{km}} \; \Leftrightarrow \; d = \frac{117}{3.1} \, \text{km} \approx 38 \, \text{km} $$

So any flight longer than 38km is safer than driving the same distance.

Let us now try to take the cruise part for aircraft into account. According to this page 20% of all fatalities are due to accidents during cruise. We can subtract 20% from the 117 deaths per billion journeys (so this only takes the parts of the flight into account that happen once per journey). Modifying the 0.05 deaths per billion km is not so straight forward. Since most of the distance is traveled in cruise, I am going to be conservative and just leave this number unchanged, therefore providing an upper limit on the deaths per km.

Now, we can make a more accurate comparison:

$$ N_\text{death} = 0.8 \times 117 \cdot 10^{-9} + 0.05 \times \frac{d}{10^9 \, \text{km}} = 3.1 \times \frac{d}{10^9 \, \text{km}} $$ $$ \Leftrightarrow \; d = \frac{0.8 \times 117}{3.1 - 0.05} \, \text{km} \approx 31 \, \text{km} $$

This is actually lower now, because the deaths during cruise are irrelevant for such a short flight. And this gives the final answer: Flights longer than 31km are safer by airplane than driving the same distance by car.

Bianfable
  • 55,697
  • 8
  • 195
  • 257
  • 8
    Nice answer based on statistic extrapolation. Of course, there is no evidence that the extrapolation is valid for such short flights. I would expect that landing 4 minutes after start is stressful and results in a higher accident rate. – bogl Oct 18 '19 at 07:56
  • 2
    @bogl Good point. While there are shorter flights, I could not find a statistics for how safe these are. But I agree with you, the accident rate is probably higher, although not by orders of magnitude, so the answer should not change too much. – Bianfable Oct 18 '19 at 08:00
  • 8
    The fault in analises like this is always the same: it assumes an "average" driver. But the fact is that most car accidents, especially serious ones, are caused by gross negligence. Of course, this isn't a problem that this answer can solve but just once, I'd like to see driving statistics that exclude drunk drivers, morons who text and drive etc. and see what are the odds of an accident if I'm a person who keeps proper focus on driving. – Davor Oct 18 '19 at 09:42
  • 7
    @Davor yes. Then again, even though most accidents are caused by negligence, even a careful driver still has a considerable risk of getting involved in such an accident. — Anyways, a more interesting comparison would be plane vs train, not car. – leftaroundabout Oct 18 '19 at 09:47
  • 2
    I would guess the danger of driving per km actually increases with the length of the trip, because tired drivers make more mistakes. However, I have no idea how big this effect could be. Maybe someone collects on how long the drivers involved in an accident had been driving. – JiK Oct 18 '19 at 10:07
  • 2
    @JiK I don't have the details handy, but have seen statistics more than once that most car accidents occur near the driver's home. Of course, this could be the end of the journey and not the start. – Aleks G Oct 18 '19 at 10:13
  • @AleksG there is also a difference between accidents and fatal accidents. I was involved in a few accidents, almost all near my home. But I am still alive and so was everyone else involved after the accidents. The most serious accident cost a few thousand euros to fix the cars. I am not so sure that most fatal car accidents occur near the drivers homes. – Josef Oct 18 '19 at 12:30
  • 11
    @AleksG that's because most car journeys occur near (definition?) the driver's home – Caius Jard Oct 18 '19 at 12:44
  • 1
    Short flights may well be more dangerous than long flights per km, especially if the aircraft you are on does frequent short flights. Equipment failures due to low-cycle fatigue depend on the number of takeoffs/landings, e.g. Aloha Airlines Flight 243, on an aircraft that did frequent short hops in a corrosive environment. – moink Oct 18 '19 at 14:53
  • 1
    "Note: This does not take travel to/from the airport into account." which is kinda the problem with this type of analysis. The most dangerous rounds are country roads and if you have to drive down those then you will probablly have to drive down them whether the bulk of your journey is going to be made by plane, train or motorway. – Peter Green Oct 18 '19 at 16:51
  • 1
    @JiK the type of carriage way makes more difference than just tired due to long haul , people are tired (e.g. hungover or sleepy) during normal driving e.g morning commutes, but much of the same features of airline traffic, seperation (median) , spacing (lanes), and controlled entry/exit (onramps/offramps) make highway travel safer despite the faster speeds and longer distances. Many fatal accidents occur due to head on collisions on undivided highways world wide . – crasic Oct 18 '19 at 17:17
  • This also does not take into account the type of aircraft. Large commercial airliners are operated with much higher standards of safety than small GA aircraft. A short flight is likely to be aboard a small aircraft, and therefore much more dangerous than a long flight. – StackOverthrow Oct 18 '19 at 17:47
  • 1
    Doing a measurement based on just "deaths per km" is also kind of iffy, since you can say that the crew of the ISS has them all beat, circling the globe every 90 minutes, thus travels ~682,400 km per day, and so far zero deaths on board. (There's been a few going to or from the station, but if we don't count driving to the airport, then we can't really count those either.) – Darrel Hoffman Oct 18 '19 at 18:06
  • @DarrelHoffman Whereas "Earth" is very UNsafe--4.5 deaths per 1000km (traveled with respect to the CBR). – user3067860 Oct 18 '19 at 18:26
  • 3
    @user3067860 Indeed, the Earth has a nearly 100% fatality rate if you consider how long it's been travelling... – Darrel Hoffman Oct 18 '19 at 18:37
  • 2
    @bogl On the other hand, accidents on landing are often caused by unexpected conditions at the destination airport. For such a short flight, conditions are unlikely to have changed much since takeoff. – Skyler Oct 18 '19 at 18:40
  • It's a theoretical answer. No more, and no less. – bogl Oct 18 '19 at 18:44
  • @DarrelHoffman Less so than you might thing... for example, if we froze everything at this point in time there would be a ~7% survival rate of humans. https://www.prb.org/howmanypeoplehaveeverlivedonearth/ – user3067860 Oct 18 '19 at 18:44
  • Note that the UK has a relatively low number of driving fatalities (even for OECD countries). The US has at least twice as many, 7.3 per billion km, adjusting for that we get around 80km and we are close to driving actually being saver for some short but realistic flights. It is definitely within the margin of error for some flights to be more dangerous than driving and so the "fact" that flying is much safer than other forms of transport is probably quite an overstatement. And that is already when we are comparing to the most dangerous form of transport, i.e. driving,... – Kvothe Jul 22 '20 at 16:02
  • ... when comparing against travelling by train I think it will be more dangerous at much larger (if not all) distances. – Kvothe Jul 22 '20 at 16:03