Plane crash rates by model | Hacker News Reader

Plane crash rates by model

airsafe.com

116 pointshosteur146 comments

That second 737MAX crash was full criminal negligence on the part of Boeing. The first crash was caused by bad design, the second crash was caused by Boeing’s choice to do a big PR push to prevent regulatory action and keep the planes in the air. The families should own that company by now.

> The first crash was caused by bad design

That's underplaying it. The design was negligent, and the whole "secret" MCAS is insane.

I wouldn't want to own a company that did that. Or be forced into a relationship or moved out.

I'm not saying Boeing is a bad company. Lots of workers have put their lives into that company and believe in what they build.

I'm just saying putting ownership into the hands of possible victims seems to put more work and responsibility into their laps, not less.

I've been in abusive situations, and hearing philosophies like that worries me.

I understand it's one small part of what you said and the rest of your comment was helpful.

I think, "The families should own that company by now" is a figure of speech that means Boeing should have paid the families so much money that they no longer exist. I could be completely wrong but that's how I interpreted it.

It’s a figure of speech.

> the second crash was caused by Boeing’s choice to do a big PR push to prevent regulatory action and keep the planes in the air

You mean the intervening time in which every single operator of the 737MAX was told to review emergency procedures? In which every platform for aviation news was focused on the behavior of MCAS and how to disable it? And in that time Ethiopian Airlines grossly failed in its duty to train its crews on MCAS specifically and upset/emergency recovery more generally, as evidenced by the accident flight. Read the NTSB and BEA add-ons to the report; I think it's clear that the crew would have been hard-pressed to recover safely from an serious event.

I really don't want to defend Boeing here, the design of MCAS was obviously grossly flawed, made significantly worse because other Boeing airframes with a similar system used multiple AoA sensors. But given that the effects of MCAS misactivation are similar to a trim runaway, which is supposed to be a memory item for 737 pilots, the inescapable takeaway is that a lot of pilots simply weren't/aren't prepared to handle rare but critical in-flight emergencies that builders and certification authorities take for granted.

I’m happy to see that someone else is aware of the 737’s runaway trim condition. I knew several 737 pilots who used to joke about the “wagon wheels” and considered their noisy operation an unofficial concession to monitoring.

> But given that the effects of MCAS misactivation are similar to a trim runaway, which is supposed to be a memory item for 737 pilots, the inescapable takeaway is that a lot of pilots simply weren't/aren't prepared to handle rare but critical in-flight emergencies that builders and certification authorities take for granted

Similar to runaway trim that pilots were familiar with, but from what I recall from one of the videos where a 737 pilot showed in a simulator what handling it was like and other discussion there was one crucial difference. (I may be misremembering all this though).

After the pilots handle runaway trim and have the automatic trim disabled and have manually gotten it back to what it should be, most would try re-enabling it. Most of the time when they would get runaway trim turning the system off and on would fix it.

And if that didn't fix it and you got another runaway trim, it was no worse than the previous one. You'd fix it. And if you wanted you could try turning it on again.

(From what I've seen on documentaries on other crashes, there are actually a surprising amount of aircraft systems where "reboot it and see if that works" is standard practice for in flight failures).

With a malfunctioning MCAS though there was in effect a ratchet. The down trim added by MCAS did not go away when you stopped it and manually restored normal trim. Each time you decided to try it again to see if the prior reboot had fixed it and it hadn't you got a little more down trim bias. Do that a few times and you end up with so much MCAS down trim bias that the maximum up trim you could do was not enough and you are then screwed.

The MAX needed a bold addition to the runaway trim emergency procedure along the lines of "After recovering from runaway trim, you MUST leave this system OFF".

> You mean the intervening time in which every single operator of the 737MAX was told to review emergency procedures? In which every platform for aviation news was focused on the behavior of MCAS and how to disable it?

Unfortunately the information wasn't enough. IIRC, Boeing didn't provide instruction for what to do if the trim wheels aren't able to be moved manually, because the aerodynamic forces are too high, and the pilots ended up turning electric trim back on, adjusted the trim, and then MCAS triggered again. Yes, there were likely steps they could have taken, such as turn electric trim back off, once the trim was adjusted, and use the trim wheels after that, maybe, but it would have been better if there was a way to disable MCAS and keep electric trim, or if MCAS had been designed and implemented with contemplation of failure in mind.

Boeing issued an unactionable memo to shift responsibility onto the operators.

73n vertical trim design and procedures are suboptimal. There should be an automation trim cutout switch to maintain powered trim control except under the most unusual circumstances/malfunctions. It's like turning off a car's brake boosters when ABS is at fault. They can't be retrimmed in full deflection nose up or down by 2 mere Schwarzenegger mortals in time to recover at either low altitude or airspeed because of surface loading.

The 737 MAX MCAS debacle is overhyped compared to the critical structural parts for the 737 NG (-600+) Ducommun fabricated by bashing and hammering fragile aircraft alloys that should've been CNC'ed. These parts contributed to numerous fuselage separations on hard landings and runway overruns, whereas in the past, aircraft bodies survived largely intact and complete. There was an long-form expose on Al Jazeera that almost no one paid any attention to. The 737 NG is a fragile POS more likely to kill you than an Airbus 32x.

> The 737 NG is a fragile POS more likely to kill you than an Airbus 32x.

TFA indicates the exact opposite.

The 737MAX definitely has a bad record, but this data is somewhat out of date. Very roughly speaking, The MAX has been back in service for 1.5-2.5 years depending on jurisdiction. Boeing has produced 1196 of them. A typical narrowbody flies about 4.5 or so times per day. If you figure two years of service, that gives about 2 million flights a year or about 4 million flights since the ungrounding. Add that to the 0.65 million on the linked page and you get 4.65 million flights.

That gives a fatal rate of 0.43/million flights which is pretty bad, the same as the somewhat notorious ATR42/72, but nothing like the number on the page.

And the ATRs tend to operate from small airports with short runways and low ILS categories (if at all!) I guess that's a bit more dangerous.

I flew at an airport myself where ATRs operated as well, nothing bigger. ILS was only a low category and only available in one direction. It seems a lot tougher than landing a 737 at a high-facility big airport with runways that can accommodate even widebodies.

The simple point for most people is that new design planes tend to have higher numbers and older design planes (where all the issues and quirks have been seen) tend to be very safe.

Removing the "no longer in production" planes and sorting by fatal crash rate per million flights:

  Embraer E170/E190             0.03
  Boeing 737-600/700/800/900    0.07
  Airbus A318/A319/A320/A321    0.09
  Boeing 777                    0.18
  Airbus A330                   0.19
  Boeing 767                    0.28
  ATR 42 and ATR 72             0.44
  Boeing 737 MAX 7/8/9/10       3.08
  Bombardier Dash 8             UNK
  Canadair CRJ series           UNK

Looking at this list I'm struck by the separation between narrowbody jets which generally make shorter flights (Embraer, 737, A318-A321) and widebody jets which generally make longer flights (A330, 767, 777).

The exceptions of course are 737 MAX (we all know why) and ATRs (which I'm not familiar with -- anyone know why they have a high fatal crash rate?)

> ATRs (which I'm not familiar with -- anyone know why they have a high fatal crash rate?)

French-Italian collaboration, what could possibly go wrong?

There's a case for adding some weighting by flight hours/km as the widebody flights are on average longer, but it's less clear what that should be, especially with most of the accident risk being takeoff or landing. On the other hand, the displayed safety record of the 747-400 is exceptionally good. Much of the widebody/narrowbody distinction is just statistical noise (as others have pointed out, a third of the 777 accidents are a single flight being hit by a missile, and another third a suspected suicide) and widebodies are less common aircraft so the accident rates are noisier.

ATRs are turboprops that fly a lot into relatively difficult-to-land small airports. Also, the accident rate under-counts the amount of flights they've actually made by more than some of the other aircraft.

I agree that there should be some mileage weighting; but on the other hand, long distance widebody flights are usually piloted by more experienced crew (they're preferable and flights are typically assigned by seniority).

It would be interesting to see statistics for maydays as well; most shorthaul flights have the advantage of having nearby airports for their entire route.

Along the same lines, I'd like to see the number of planes per model. Are more planes flying less vs less planes flying more. Also, is there a sweet spot in manufacturing? Too few or too many?

ATRs had some design problems with ice accumulation[1] and are in certain ways quite unforgiving, but I suspect that the "problem" is their popularity with all kinds of smallish outfits whose training and maintenance are sometimes lacking. Case in point, this year's Nepalese crash[2], where the airline is prohibited from operating in the EU, and the preliminary report indicates that the plane crashed because one of the pilots effectively deactivated the propellers instead of adjusting the flaps while descending, causing a stall.

[1] https://en.wikipedia.org/wiki/American_Eagle_Flight_4184

[2] https://en.wikipedia.org/wiki/Yeti_Airlines_Flight_691

> The exceptions of course are 737 MAX (we all know why) and ATRs (which I'm not familiar with -- anyone know why they have a high fatal crash rate?)

Turboprop aircraft like the ATRs are generally considered to be more dangerous than jets. I couldn't find a MTBF (Mean Time Between Failure) specific to the Pratt and Whitney engines that the ATRs use, but an AOPA article from 2017 suggests that turboprops in general are over three times as dangerous as jets[1].

Here are some potential reasons that come to mind:

- Turboprop aircraft typically fly at lower altitudes than jets do, where there is more inclement weather and less time to navigate to an airport in case of an engine failure.

- Turboprops are more complex as a whole than jet engines; there is just more to go wrong (variable propellor pitch, for instance).

- Smaller airlines are the main operators of turboprop aircraft. I don't like to suggest that they are less competent on the maintenance front, and certainly don't have any evidence for this, but maybe this comes into it.

- Most fatal aviation accidents occur on landing, and since turboprop aircraft are typically used for shorter flights, they will make more landings per mile flown (nautical miles, of course, in ICAO world).

[1]: https://www.aopa.org/news-and-media/all-news/2017/august/pil...

Another thing about the ATRs, being the only turboprops on the list, is that they are more suitable for small airports, and most of the airports in difficult terrain are small airports. Therefore, you are more likely to see ATRs flying tricky routes than the others on the list.

At least one crash, however (American Eagle 4184) was due to a flaw in the design: ice rapidly accumulated where it could not be dislodged by the de-icing system.

Here in Norway, there's been several fatal flight accidents featuring turboprops in my lifetime that I recall. Two of them were both controlled flights into terrain[1][2], both due to pilot error (leading to a large restructuring of the operations and procedure at the airline after the latter).

A third one was due to wind exceeding the airplanes design limits[3], the airport was known for strong turbulence and wind shear due to location and mountainous terrain, and the airport was permanently closed after.

The fourth one seems to be due to improper maintenance and counterfeit parts[4], causing the tail to fail.

There's also been a few fatal accidents involving the Hercules like the Svanfjell-accident[5] due to a misunderstanding of the height clearance and the Kebnekaise-accident[6], again due to human error as well as maps lacking correct height data for the area.

From what I can see[7], there's few if any fatal turboprop accidents involving the engines. So seems that at least here in Norway, the main reason is they fly a lot of stops on small airstrips located in difficult terrain, as well as being operated by smaller airlines with less organizational oversight.

Though I'm not expert by any means, and things might be different in other places.

[1]: https://en.wikipedia.org/wiki/Wider%C3%B8e_Flight_710

[2]: https://en.wikipedia.org/wiki/Wider%C3%B8e_Flight_744

[3]: https://en.wikipedia.org/wiki/Wider%C3%B8e_Flight_839

[4]: https://en.wikipedia.org/wiki/Partnair_Flight_394

[5]: https://aviation-safety.net/database/record.php?id=19660320-...

[6]: https://en.wikipedia.org/wiki/2012_Norwegian_Air_Force_C-130...

[7]: https://no.wikipedia.org/wiki/Liste_over_flyulykker_i_Norge

Turboprops are fucking deathtraps (3x failure and fatality rate) because of unresolved engineering challenges, i.e., spline drives. Props just falling off midflight.

As a Brazilian, I´m happy to see the Embraer as the first. I hear from pilots that Embraer planes have excellent survivability and glide times.

They're also a great ride! I've been on a few Embraers, and been super impressed with their "feel" as a passenger.

I've taken quite a few flights in E90s and was always impressed by them. I was sad when Air Canada replaced them with A220s -- of course I understood the politics of wanting to use a Bombardier plane.

It's impressive, indeed. But I'd curious to know how it compares to other planes of the same size.

I suspect it's very well. For a notably long time, that chart would have read 0!

It would be worth adding the models with a rate of 0 at the top.

Phenom 300's can't fly without GPS. I'm curious what their rates are.

I think there needs to be a more selective filter if this data is supposed to mean anything. The evaluated incidents include terrorist bombings, the deliberate Germanwings crash, etc.

Would those incidents be frequent enough to make anything but a de minimus impact?

Probably, given how rare any fatality in commercial aviation is.

One of the three 777 accidents is being shot down by a Russian SAM.

Another is the infamous MH370.

Another is the pilot error on Asiana flight in San Francisco.

Another one is the crash at Heathrow, but that was Rolls-Royce’s fault

Off-by-one error detected

The entire dataset is microscopic. I'd say it could make a substantial impact.

A really small datasets shouldn’t be used for statistics. But is it small, considering there are millions of flights registered?

Millions of flights, but less than ten crashes for most, and it was pointing out that even some of those are incorrectly categorized. When you're dividing A by B to get ratio C, you cannot ignore that A is bad data because B is good, it still makes for a C that is useless.

All this dataset provides is that with small enough samples it is too noisy to draw reliable conclusions.

You need a larger dataset if you're dealing with very small probabilities. If you measured millions of events and only saw 1 anomalous event, you have enough data to say that the probability of such an anomaly is extremely rare, but you don't have enough data to compare it to other very rare events.

More concretely: if you flip an unfair coin 1 million times and get a single heads, you know that the odds of getting that heads are extremely low, but you can't yet say what the odds are. It's possible that the odds are one in a million, but it's also possible that you got very lucky or very unlucky relative to the actual odds. You have to have a lot of data in both buckets before you can distinguish.

If I'm doing the math right, there were only 3 fatalities on a 777 if you don't count MH370 or the Ukraine/Russian shoot-down. That would change the FLE from 2.01 to 0.01.

I understand not wanting to count a shoot-down but why not count MH370? For all we know there was a safety issue with that aircraft that caused the crash.

We basically know that it was suicide at this point.

As I recall, there's a fair amount of circumstantial evidence at this point to suggest pilot suicide as the most likely cause.

After almost a decade of investigation the preponderance of evidence points to murder/suicide by the pilot. A safety issue is highly unlikely.

This is a very USA/western-centric perspective. No SE Asia reporting is stating this.

What do they say?

Isn't that because suicide is a taboo subject in many countries? This was at least the case for EgyptAir Flight 990 and SilkAir Flight 185, and IIRC people expressed the same thing about Malaysian culture.

Why? What are the alternatives?

Depending on plane, it could.

For example, 777 has had very few incidents, but its figure is impacted by MH370 and MH17.

747's number is also skewed (by about 10%) due to 3 incidents involving bombing and missile shootdown.

By and large though, most plane crashes are due to human error, mostly by pilot. Terrorism, as well as maintenance or engineering defects are pretty rare.

You said:

> The evaluated incidents include terrorist bombings

From the site:

> Excluded would be events where the only fatalities were to crew members, hijackers, saboteurs

The numbers in the "Events" column are not the same numbers as are on the linked Events page, if that makes sense.

Those statements aren't at odds are they? Terrorist events that results in death of at least one passenger are included, despite terrorist activity (presumably) having nothing to do with specific model of airplane.

It includes fatal accidents caused by hijackers or saboteurs, provided they manage to kill passengers and not just themselves

I wondered about that, because isn't the hijacker technically a passenger too? I wonder if their deaths are excluded from the data.

Think it's difficult to make much of the data even with a filter. A lot of other incidents are due mainly to pilot error, bad weather conditions or inadequate maintenance, but the aircraft don't altogether escape blame for those types of incident. Actual incidents known to be down entirely to bad aircraft design are exceptionally rare amongst late generation aircraft (but some of the more mysterious crashes might be).

The ATR figures are inflated by their flight cycles info being years out of date (and based on knowledge of what ATR supplied to commercial databases, likely incomplete), but you'd also expect them to be higher than some of the other aircraft types purely because of how they're used (flown into smaller airstrips in hilly regions). At the other end of the scale, the 747-400 performs excellently with only two passenger crashes, but it's also had four fatal freighter crashes not counted. I bet the calculated "rate" includes freighter flight cycles though

But above all, airliner fatal crashes are exceptionally rare, and some of the near misses and non-fatal incidents are more indicative of design flaws

My first reaction on seeing the fokker stats was 'well, yeah, it's heavily used in the Alps and Himalayas, of course it's had a lot of incidents'

Ironically, the 727s numbers are probably skewed in the opposite way to the 747 - there were loads of passenger incidents with the 727, but it's been somewhat of a workhorse for freight in the developing world over the last 40 years.

An interesting perspective on these numbers is to consider Concorde. Though it flew several times per day on average, for a couple of decades, Concorde went from 0.0 (perfect) to 11.36 in a single event. By far the worst on the list.

I glean from this data point just how safe aircraft are in general.

This is interesting info, especially with the MAX outlier. But not enough to go on if you want to do "rainman" calculations. You would need to know the operator, their country, their airline safety culture etc.

If you can classify each crash into a % score "due to pilot training issues", "due to design flaw", "due to maintenance skimping". Also need to take into account time. I think if a plane crashed today because of wind changing direction rapidly due to a microburst under a thunderstorm, it is more likely to be called a pilot training issue, but go back far enough and it is a weather issue instead.

Just based on stuff I read. I am not a pilot or expert.

Edit: more clear what wind changing direction is. Not just a gentle breeze from the South moving East!

The MAX is the one plane I will absolutely not fly on. It's nice to see the numbers, but as you say, fairly unsurprising. It's also nice to see that the 787 and A380 have never crashed, since those are my favorite planes.

It's not a bad plane. Every pilot now knows its flaws. Don't forget one flip of a switch would have saved all those lives. The problem was the pilots never knew that switch was even there.

I'd fly on it just fine, I wouldn't feel unsafe. Though I'd regret providing profits for the company that deliberately let those people die.

Isn't it more than the MCAS itself but also the reliance on one sensor (redundancy optional expense) instead of standard of three?!

Well yes both. But either way, if the pilots had known MCAS existed, all they would have had to do was switch it off.

In both cases they had plenty of time to do it.

Knowledge is key, even with 3 sensors it's possible 2 of them fail, though much less likely.

The scandal with the MCAS though was in large part because Boeing went out of their way to hide the existence of this system to save their customers training costs, and therefore win more sales from existing 737 operators. On the contrary, Airbus, for instance, provide detailed guidance for what to do if either one, two or all three of a system like the ADIRS (Air Data Inertial Reference System) were to fail during a flight, even though that is extremely unlikely.

ET302’s crew did switch it off [correctly]. They then switched it back on in an attempt to fly the airplane out of the emergency.

ET302’s crew (the second Max crash) also knew about the plane’s fault mode and almost saved the craft. If they’d have continued to command nose-down manually [after correctly executing the runaway trim disable checklist memory items and then re-enabling the power to the offending system in order to command aircraft-nose-down trim for a short time], they’d have almost surely flown the airplane back to base.

It was Boeing’s fault, but the crew almost saved it and inexplicably re-enabled the fault after correcting it. They were so close…

How safe is it? You need the MCAS to compensate for the change in aerodynamics. So if the pilots flick the switch off, then the pilots would need to retrain to fly this now non-737 series plane? Have they?

"I think if a plane crashed today because of wind changing direction under a thunderstorm, it is more likely to be called a pilot training issue, but go back far enough and it is a weather issue instead."

I certainly hope that a plane would not crash due to wind changing direction...

https://en.wikipedia.org/wiki/Delta_Air_Lines_Flight_191

Fair enough. A microburst on approach would technically be classified as a change in wind direction.

I did say under a thunderstorm! I understand that landing under a thunder cloud, or even cruising under one, is a no no.

"Low Level Wind Shear" is the technical name.

It is the ninth leading cause of landing and takeoff fatalities.

It makes sense when you realize that airplanes can only fly when the apparent wind (their speed through the air) is high enough, and landing and takeoff is when you are crossing that transition zone between high enough to fly, and not flying. If you go from a 10 kt. headwind to 10 kt. tailwind fast enough, it is exactly like losing 20 kts. of speed. Especially on small prop planes, that is more than enough speed difference to go from comfortable approach speed to dangerously close to stall speed.

Commercial airliners started to get onboard wind shear detection ~25 years ago. The idea is that if you get enough advanced notice, you can execute a go-around while you’ve still got some altitude.

When I found out about MAX and what happened with the design, after leaving that industry, I was amazed it had gotten approved.

That number there indicates that.

Wasn't the cause a software bug that was fixed and not a design issue?

It's not a bug. Boeing deliberately chose to mislead the airlines and pilots on what they were doing and used software to hide the fact. On top of that there was no redundancy for the single indicator which triggers the nose down/pitch down. The software issues is downstream of the managerial malfeasance.

It was a regulatory bottleneck prompting an inefficient design that required a software fix to fly, which relied mortally on the inputs from a sensor that was optionally non-redundant outside the United States.

Yep. They were trying to compensate for different feel/behavior to the pilots without additional training. I really don't see how that was approved by a competent agency...

Ended up in those two horrible crashes in perfectly good airplanes.

> don't see how that was approved by a competent agency

The version that flies in America has two AoA sensors. To my knowledge, those planes didn't have a problem.

Don't forget a lot of countries rely on the FAA for guidance because they don't all have the resources or manufacturer access to thoroughly evaluate an aircraft themselves.

Also, even with two AoA sensors the pilot had to switch to the alternate because the MCAS wouldn't know which one was right. This is why usually 3 sensors are used in redundant systems.

Because when you're a DER (Designated Engineering Representative), you're still getting your paycheck from Boeing, and being a fussy DER is not a fast track to success. I can't lay down too much more specifics, but anecdotally, that's how things go, and not just at Boeing either. Seen it at a lot of suppliers too. You're a "good" DER, or you're on the short list next round of layoffs[1].

DERs have been getting more and more sign off authority on increasingly critical core airworthiness systems, stuff no one would have dreamed of delegating to a DER twenty years ago.

The actual full time agency people, well, depends on how much they're fishing for a corner office. But by and large, they were hardasses, thank God. All three of them. Particularly the EUROCAE Dutch guys. Whew!

But none of this is important as the Boeing culture, which . . ehhhhhhyyyyuuuuuchhh . . could use some improvements. To phrase it one way.

[1] There's always layoffs, good times and bad, because everything is funded contract to contract, week to week. I had a goddamn amazing analog engineer get pink slipped, packed his stuff, and then - after all that - he got called back before he got his boxes in the car. "Whoops! We didn't mean it! Turns out no one else in the universe knows how to do this!". Luckily, he was a canny old gent, and responded with a well-deserved, "Sure. I'll come back in. But it's gonna cost ya".

I saw bits and pieces of this interacting with DERs when I was working in the field, and you're touching on what I saw first hand. Thankfully I wasn't working on anything related to the MAX.

I think regulations refering to a competent agency doesn't require any competency in the normal sense of the word.

And perfectly good those planes were not, they had broken AoA sensors.

My understanding was that there was a design issue that was supposed to be fixed by software, and that software in turn had a bug.

The design was not "backwards-compatible" with existing pilot training, but the plan was to use software as a cheap fix. They said pilots would not need new training, and were betting lives on the reliability of software and humans.

I don't see anything wrong with that aside from the buggy software.

Would it be incorrect to interpret this data that Airbus planes have a safer track record? Obviously the 737 has almost twice as many flights, but the crash rate is more than double.

The 737 NG has a lower crash rate than any Airbus on the list at 0.07. Hop on a 737 in the US and this is probably what you'll be on.

Are you trolling? Using 737 NG would be like using Airbus NEO variants, which ALL were excluded from the list in the single aisle category for having a 0.00 crash rate…

The NG has been produced since 1997 and there are essentially no pre-NG aircraft in Western service. The NEO is equivalent to the MAX.

Equivalent to the MAX perhaps except no hidden suicide system on board..

> The NEO is equivalent to the MAX

In what way? That they both have new engines?

I'm pretty sure the similarities stop there. The A320, due to higher wing position, did not require massive overhauls to get the NEO functioning - so the NEO offers overwhelming airframe similarity and flight characteristics.

The MAX, on the other hand, put the engines in a new position, introduced an all new fly-by-wire system to prevent the obviously necessary reclassification of the aircraft, and then cut corners on safety despite the huge amount of re-engineering they had to do.

The NEO has also been out longer, and has never had a fatal crash event...

Airbus makes safer planes, I can't believe people are misinterpreting statistics this hard. "If you simply include Airbus stats from the decade before 737 NG, ignore their NEO and decide it's equivalent to the MAX despite being on the absolute opposite ends of the safety scales, and then argue it's reasonable to compare the two drastically different data sets you've constructed with unjustifiable exclusions, Boeing has better fatal crash statistics!"

Pathetic.

The NEO is a much newer plane with much less history than the NG which has been around since the mid 90s. I'm sure it's safe too.

It is fair to compare the A320 family with the NG. They are comparable planes and have comparable crash stats too. (0.09 vs 0.07)

That’s such a misleading number. The numbers you post EXCLUDE both NEO and MAX crash rates, but meanwhile the MAX would grossly inflate 737-NG rates and NEO flights would slightly depress A320 rates

I highly suggest not citing those numbers

There are 6 times more "events" listed for the 737-NG variants as there are for the MAX, so there wouldn't be much inflation.

(There's also more than 100x more flights with the NG than the MAX)

1/6 the events in a vanishing amount of samples would probably imply the number to be 7/6, and given the passenger counts I imagine that’s an undercount. Therefore the crashes would look more like 0.085 for both Single Aisles in aggregate…

The planes are grouped the way they are because they are mechanically of the same generation, not because they have similar names. The NEO and MAX are newer generation planes than their predecessors.

NEO is not a new generation plane, it’s a new engine option with some engineering work to support the bigger engines but overwhelmingly similar airframe and flight characteristics. They also put sharklets on the wings by default

MAX is not just a new engine option, it’s an entirely overhauled configuration and position which accordingly drastically altered flight characteristics masked by all new fly-by-wire systems and previously half-baked safety monitoring in order to prevent Boeing from needing recertification.

Which speaks perfectly to the culture of safety at Airbus vs the new Boeing management approach that is the hallmark of American late stage capitalism

What matters is Boeing MAX vs everything else. It needs a few years more of testing at least until we can trust that it's safe.

Who is we? Basically every aviation safety regulator has tested and verified that the MAX fixes are sufficient and have returned them to service. Even China.

They also certified them prior to those fatal incidents.

As it is with every commercial jet.

The 737 has also been around a lot longer. The 737NG has a rate of 0.07 compared to the A32x's 0.09. The A330 is 0.19 while the 777 is 0.18. But even this is kind of misleading. All but three of the fatalities (total fatalities, not accidents) due to the 777 come down to intentional acts that would have destroyed any aircraft.

It depends which you are comparing. Rates are usually expressed in average per 100k hours of block time (slightly rosier than wheels up time). Classic 737's serviced and flown by major American carriers and 739's on small foreign carriers today are entirely different beasts.

The 737 spans almost 60 years. The earlier planes had a much higher rate of accidents, which skews the overall numbers given that today you'd have a hard time finding one of those in the sky, especially on a passenger route. The 737NG has one of the best safety records of any class of plane flying today. Aside from a 737 MAX, right now if you find yourself on a 737 it's nearly guaranteed to be an NG, which is just about the safest airplane currently in use.

Early 737s had a rare, transient rudder hard-over failure mode that defied analysis for a long time. When it triggered on final result was a quick spiral-dive into the ground.

Another possible confounding factor is that these aircraft have different operational profiles. The 747 variants were likely be to employed on longer haul routes whereas a 737 will be doing multiple individual flights per day in more crowded airspace.

It would be interesting to see this normalised by landing cycles given that the takeoff & landing are perceived at least within light aviation to be highest risk portions of a flight.

But if you have a specific route in mind then it could be useful. I’m a bit sketched out when flying on a turboprop plane like an ATR 72 but feel more at ease on an A320 family plane, because they will most likely be landing at a bigger airport with better emergency facilities and to a less remote area.

The Embraer 190 shows 1 incident, but when you click to read the details, there were two fatal accidents. Why the discrepancy?

From the details page:

> This is not a numbered fatal event because the Legacy aircraft, although based on the design of the ERJ135 regional airliner, was not airline passenger flight.

This particular crash is widely known here in Brazil. The two pilots from the small aircraft had turned off the plane transponder. They had a mid-air collision with a 737. Everyone at the 737 died.

The two pilots of the Embraer 190 are Americans. They were sentenced to jail in Brazilian court but were never arrested to this day (not sure why).

> The two pilots from the small aircraft had turned off the plane transponder. They had a mid-air collision with a 737. Everyone at the 737 died.

> The two pilots of the Embraer 190 are Americans. They were sentenced to jail in Brazilian court but were never arrested to this day (not sure why).

America believes that the collision was caused by a mistake by Brazilian ATC (instructing them to fly at the wrong altitude), not the pilots turning off the transponder, and that it was an accident not a crime.

If I wanted to know “what planes are safest?” But I also want to correct for a bias like “Crash Airlines fields a lot of TypeX so TypeX seems crashier than it probably is” how would I approach normalizing that data?

You could start by normalising by plane age.

Those are rookie numbers compared to stuff like MA60. First flight in 2000, 110 built, 67 still active, 57 exported, 26 in storage due to maintenance problems, 14 incidents (luckily only 1 fatal)

Wikipedia:

  Of the 57 MA60s exported by January 2016, at least 26 were in storage after safety concerns, maintenance problems or performance issues; six others were damaged beyond repair.

Not quite a glowing track record…

The list is completely pointless. It assumes the issue is the equipment. Maybe list the operator, country, and cause of incident.

What I'm seeing here at a glance is that newer aircraft other than the 737 MAX are significantly safer than older aircraft.

Or easier to fly / pilots better trained / airports and atc having improved.

The Boeing 777 should be the safest as none of the accidents were the fault of the aircraft itself

Cool dataset! Though I would hesitate to draw conclusions from it as a passenger, since the # of events is so low. Someone would need to do the statistics to see if the B737 and A320 series have enough flights to be significant.

? B737 has been flying for nearly sixty years at this point.

“In 2013, the global 737 fleet had completed more than 184 million flights over 264 million block hours since its entry into service.”

https://en.m.wikipedia.org/wiki/Boeing_737

“The global A320 fleet had completed more than 164 million flights over 303 million block hours since its entry into service.”

https://en.m.wikipedia.org/wiki/Airbus_A320_family

Not many lay people are aware of the phenomenal record of the Embraer aircraft. Credit where credit is due! For a notably long time, that number would have read zero!

Why make such a useful table, but don't allow sorting by the columns?

HTMl really needs a grid element that natively supports sorting and filtering.

Surprised the 747 has such a high rate. Subjectively I’d have guessed it would be one of the best cause it’s like a workhorse

I'm curious what that the rates are for Aeroflot's Ilyushin Il-96.

Concorde made 90,000 flights?

Why is that surprising? There were 20 built, and were flown for more than 27 years.

I didn't realize there were so many, and that they were active for so many years. Incredible.

I wonder how would it change by excluding chartered flights and crashes

sometimes ignorance is bliss - I'd prefer not to know the crashes by model and fly without thinking about the statistics

They’re all dramatically safer than driving to the grocery store, fwiw

Are they? Just at a glance, there is about 1.35 deaths per 100 million vehicle miles traveled (VMT). Assuming I live a mile away from the grocery store, I will need to drive 2 miles round trip. That means I can be expected to be in a fatal accident after around 37 million round trips.

The safest plane on the list is the list is the Embraer E170. It has 0.03 deaths per million flights or one death every 33 million flights. So it’s actually more dangerous. And most people will be flying on Boeing 737s or Airbus A320s which are about 3-20x more dangerous per flight.

In any case, my takeaway is that people drive a lot so the denominator is pretty big. And if we remove driving drunk to the grocery store/driving there on NYE then driving is even safer.

Indeed, airplanes are (as often cited) much safer if one counts per miles travelled, but they are certainly not if counting per trip.

Taking a US-centric look: in 2021 there were 42,915 automotive fatalities in the US [0], and the average number of vehicle trips per household is around 1900 [1], and about 124M households in the US. [2] That works out to ~235B trips, putting the death rate at 0.182 per million trips, or roughly equivalent to the per-trip fatal crash rate of the A330.

So you're right, an individual plane ride is not "dramatically safer" than the average car trip (if anything it's slightly riskier), but people take far more car trips, so the risk adds up. Many years, NTSB reports ZERO deaths from commercial air travel [3], while vehicle crashes are a leading cause of death—not quite top 10, but close.

[0]: https://www.nhtsa.gov/press-releases/early-estimate-2021-tra...

[1]: https://www.energy.gov/eere/vehicles/articles/fotw-1041-augu...

[2]: https://www.census.gov/quickfacts/fact/table/US/HSD410221

[3]: https://www.ntsb.gov/news/press-releases/Pages/NR20211117.as...

It should also be noted that you can reduce your individual risk even further than the mean by avoiding the most dangerous behaviors. Don’t be intoxicated, tired, or distracted. And wear a seatbelt. People who drive dangerously are pushing up the average.

And not to make this about the argument in the thread, but there’s not a lot you can do on an individual level to prevent airplane fatalities. (The flip side of course is you can be blackout drunk on a flight and still be perfectly safe.)

So the fatality rate across the B737 family is 0.24 per million flights, the rate for the A320 family is 0.09 per million flights. That's 3 times the risk of death boarding a B737 vs boarding an A320.

If we apply the same logic than people who wanted to mandate covid vaccines for the population not at risk, reducing a minuscule risk by a factor of 3 to another minuscule risk is hugely important and therefore people shouldn't be allowed to fly on those dangerous B737!

If covid deaths were as rare and as non-communicable as airplane fatalities, you would have a good point.

> reducing a minuscule risk

Official stats say COVID killed a million people in US alone and 6 million globally. Thats more than Ebola, more than tuberculosis.

War in Ukraine killed far less, is it miniscule?

Furthermore, we still have people getting repeat infections - and full extent of damage is unknown, but its not pretty. Its quite clear that covid is close to tuberculosis than it is to ordinary flu

> Official stats say COVID killed a million people in US alone and 6 million globally

with a median age of 80. I am talking about the population not at risk (i.e. young, no serious medical condition).

And still the risk of death was much higher than the risk you're comparing it to. And that's ignoring the cumulative effect of infection not to mention secondary effects on resourcing.

Are you serious about this point?

The 737 NEO is 0.07. Earlier 737 models predate the Airbus 320 by nearly 20 years. The equivalent Boeing planes are marginally safer in context.

The 737 MAX has a horrible ratio because it’s very new and had a very serious flaw on launch which caused multiple crashes. Whether or not it’s equivalently safe now is yet to be proven.

The point is that even the 737 max number is minuscule and not worth worrying if you are only flying a few times a year.

People are bad at understanding tiny numbers. The probability of not crashing on the Max being 999,997 out of 1,000,000 I think is a bit easier to process.

Reply on news.ycombinator.com