Thanks for confirming that my first impression was the correct one!
The trick is that every take off is different and so it is hard to get an intuitive feeling for what is right and what is wrong.
Due to variations in the aircraft weight and environmental factors such as temperature, pressure, and wind component, a take-off from a particular runway may be limited by the length of the runway, the ability of the brakes to absorb energy during an abort, or obstacles many miles away on the take-off flight path. A take-off at max thrust, that gives good acceleration, may be closer to whatever the limit is than a very reduced thrust take-off that has you trundling down the runway like granddad going for a Sunday drive.
Another factor is that, unless you’ve suffered an engine failure, or had to abort the take off at high speed, or made a gross error, you may never know that a take-off didn’t achieve the necessary performance. The take-off performance is designed to account for a significant failure during the take-off. If there is no failure then the extra performance you get will tend to mask all but the grossest of input errors.
An Emirates A340 came very close to crashing during take-off from Melbourne, Australia, because the First Officer inadvertently used 262.9 tonnes for the aircraft weight in the performance calculation instead of 362.9 tonnes. A whole 100 tonne error and yet they still managed to take off without losing anything other than their jobs.
https://www.atsb.gov.au/media/3531728/ao2009012_full%20report.pdf#page23
Realistically you need a computer system that can compare expected performance with actual performance.
I read somewhere (Freakonomics, I think), that closing LGA would actually reduce delays at the other New York area airports. It seems counter intuitive that having fewer runways would actually improve air traffic, but apparently the problem is that the approach to LGA crosses the approach to JFK (I don’t recall if it affects EWR as well). So ATC can’t land as many planes at JFK per hour as they’d otherwise be able to, because they have to allow for separation from planes landing at LGA. If LGA closed, traffic could flow to the other NYC airports much more efficiently.
But you’d never be able to get public support for closing LGA, because it’s the most convenient airport for high value business travelers in Manhattan traveling to the airport by taxi or car service.
My home base has a 6300’ runway. We fly a variety of A320/A321s. On a standard day with no rain or wind and taking off towards the hills, all our variants can take at least a full load of cargo and passengers plus 10 tonne of fuel. We occasionally run into limits on full international flights when we may be wanting to take more fuel. So it’s something that needs consideration but it is rarely limiting. Singapore flies their A350s out of here as well, I’d expect they’d be very limited.
Taking off in the other direction the runway is slightly longer and there are no hills to clear. The domestic A320s have no limitation while the internationals have a very slight limit on take-off weight but not one you’d be likely to run up against.
Edit: forgot to make the point of the post. A 7000’ runway would not normally be limiting for a midsized jet like an A320 or B737.
This seems more the romantic ideal of someone who loves manual flying that a realistic path to greater aviation safety. It seems far more likely to me that the future of commercial aviation is that technology and automation improve until we simply pass the point where the presence of a human pilot with any authority to intervene is a net safety benefit.
For sure what you propose is the likely end state. Just as it is with automobiles, trains, ships. And for sure the aircraft and avionics manufacturers are pushing the state of the art that way as best current IT/AI technology can support.
But I don’t think that’s soon enough to solve the immediate problem.
The current situation in worldwide airline aviation is that pilot skill is plummeting faster than automation ubiquity and reliability is expanding. That’s presages a big dip in safety over the next 10-30 years.
Some combination of COVID & climate change mitigation may put a major crimp or even halt to the breakneck expansion of airlines in 2nd & 3rd tier nations. If so that will buy time for a smaller crop of better pilots to fly until they’re replaced by something akin to an AI.
But even today we need better flying skills from even first tier national pilots. The machinery still breaks and the computers still make dumb decisions and have dumb bugs. Weather still happens. Both ATC and other pilots make mistakes. All of which are correctly handled by the pilots. Except for the occasional brittle failure where they fail badly in their crucial role as the ultimate general-purpose backup for everything.
What's easy to forget in these analyses, and the same applies to self-driving cars, is that every one of the things pilots / drivers do *right* must also be built into the computerized replacement with equivalent reliability. If not, we'll see what we have seen so far with self-driving cars: they make mistakes that (almost) no human would make. Said another way, in those specific situations they've demonstrated less reliability than the known-fallible humans they're meant to replace.
More simply stated, until the AIs are as good as current drivers / pilots they won’t be adding to safety. It’s only when they can do everything we do as well as we do it, and also do more / better where the humans fail, that they’ll be providing a net benefit vs humans with all their well-known shortcomings.
RIght now getting to the as good as level is a very tall order.
Once we do get there, or very near there, beating the rest of the pants off the pilots / drivers will be easy. Which is why so many boosters (including me sometimes about self-driving cars) focus on that last exciting part while assuming away the difficult first 98% of the journey to true AI, even driving / piloting-specific AI.
Meanwhile, we need better pilots qua pilots. Not as rote airplane operators or computer-watchers.
How do you all feel about air shows? I wouldnt go or take my kids for the same reason I wouldnt a circus. Yes, mistakes are very rare…but the tension is very real. For me at least.
As to me … I’ve been to many. I’ve also been to many air races.
I’ve watched 2 men die on different days while they were flying an airshow. And one in an air race. I’ve also watched fatal lightplane crashes at ordinary airports. I’ve seen pedestrians killed by cars and by trains. I’ve been knocked off my motorcycle at speed. I’ve also attended the last days of the elderly and the severely ill. I’m nobody special; if you spend enough years on Earth you’ll see all the same.
Death is one to a customer. Always has been; always will be. Even eating breakfast is risky.
A well-run airshow is meant, like most tension-causing entertainment, to appear far more risky than it actually is. Which is not to say the risk is zero; it clearly isn’t. But if they’re creating that tension in the audience, they’re doing a good job.
But yes, just like standing near a busy street corner, or (pre-COVID) volunteering at a hospital, don’t go to an airshow unless you’re prepared to see whatever happens. Because it just might.
Seems to me one issue with having backward-facing seats (aside from the fact that it would be unpopular with passengers accustomed to looking forwards) is that, during takeoff acceleration, any passengers or things not strapped in, would be prone to parting ways with their seats.
Whereas, in the current setup, even if you’re not strapped in, the acceleration of takeoff still keeps you in your seat anyway since it presses you in.
(Of course, it could work just the other way in the deceleration of landing.)
Thanks for the added insight. I guess that also explains why I have generally enjoyed my few experiences flying in a 747 more than any other flights; it’s not so much that the big jet was capable of greater acceleration on take off (my favourite part of the flight), but rather that the bigger vehicle probably needed to use that power for a safe departure whereas a more modern twin-engine jet, not so much - they can afford to burn less fuel and accelerate more gently to take off speed, using the same length of runway.
Yup, the scandal in Pakistan shows that we have a paradoxically dangerous situation where the current level of technology protects incompetence sufficiently during routine flights that unqualified pilots can fake it for a long time. We’re certainly not anywhere near a point yet where the lack of traditional flying skills isn’t a major safety concern.
That is an excellent summary although I’d say “under-qualified” rather than “unqualified.”
With the non-trivial addition that even originally competent pilots can become waay too rusty in environments where automation use is almost always required and also where complacency can creep in after years of trouble-free ops. This is a legitimately tough problem we’re all wrestling with.
As to the scandal …
As discussed in the dedicated PIA 3803 thread, those Pakistani guys did NOT cover themselves with glory. Rather the opposite. And neither did the guys who crashed two perfectly flyable MAXes in Indonesia and Ethiopia. Nor the Air France A330 nor the Asiana 777 nor …
But at the same time it appears pretty clear that the headline about Pakistani fake pilot licenses you quoted has been pretty thoroughly debunked. Yes, the CAA official really said what the article reported he said.
But, like many a Trumply pronouncement, any connection between what the official said and fact-based truth is pure luck. The sizzle of scandal sells far better than the steak of headline seeking lazy bureaucrats jumping to conclusions.
Could you expand on this? Upon googling, I’m still seeing numerous stories from reputable sources like CNN & BBC about fake licenses and 150 suspended/sacked pilots, with no retractions. What are you saying has been debunked?
I hope no-one minds if I resurrect this thread to continue this hijack - I just read this report today of a similar issue (and which refers to previous occurrences - warning, pdf): https://assets.publishing.service.gov.uk/media/5f2173e08fa8f57ac3af2d31/Airbus_A320-214_G-EZWE_09-20.pdf
Clearly in this case the training emphasis on using TOGA thrust at the first sign of trouble had not taken effect. Possibly relevant was that the commander was a hugely experienced 57 year-old - maybe pilots nearer the start of their careers are more steeped in this. It appears both pilots recognised an issue with 900m of runway to go, which I think translates to about 10 seconds of time at the speed/acceleration involved.
Once again, focus is rightly on improving the pre-flight routines to avoid the problem in the first place, but it will never be possible to completely eliminate such errors - and of course, the fewer such errors occur, the less likely the pilot is to be equipped to recognise and respond to it.
Also referenced in the report is the automatic detection system mentioned earlier in the thread, along with the comment that it was borderline whether such a system (had it been installed) would have kicked in. Which seems about right - what we have here is an interesting converse to the usual situation in an actual crash, when there are a number of “what if” scenarios in which we don’t know whether a different action would have prevented or mitigated the accident. Whereas in this case there was no accident, so we know that such a system was not in fact needed to prevent a crash. And had it been fitted and the situation was slightly worse, it probably would have activated to prevent disaster.
In the long run, again as already stated, more automation is probably the way to go - it doesn’t matter how good a pilot/driver you are, humans are going to be prone to lapses of concentration at some point. Most of the time these are not serious, let alone fatal, occasionally they are, and they cannot be completely trained out of the system. However at the moment we might be in a bit of a dangerous place whereby the computers work brilliantly 99.99% of the time, which leaves pilots potentially in trouble when freak circumstances occur (because they are so used the humdrum normality of commercial flight). See also AF447, the B737 Max cases, and many, many others.
One thing that stands out about the AF447 incident is that, despite the stall warning sounding (“Stall, Stall, Stall”) over 70 times, the first officer did not connect the mental dots that a stall warning calls for pushing the nose down.
So maybe instead of just saying “Stall,” a warning system ought to also include the recommended action "Stall, push nose down!" (like how some terrain warnings already say “Pull up”)
The problem with an aural warning like “STALL STALL”, or anything else you want to make it, is that under high workload your hearing essentially shuts off and the warning is ignored. I don’t know what the answer is.
Also, my understanding of that incident was that due to various circumstances, the pilots felt for the majority of the crash sequence that the stall warning was sounding incorrectly. Or in more general terms, their mental model of the aeroplane’s flight did not match the reality and that caused them to reach incorrect conclusions/make the wrong inputs. This is also a common factor in air disasters that can never be completely eliminated.
"thro
Correct. That was because the stall warning system was malfunctioning.
Or more precisely, because of how two operating-as-designed pieces of software interacted in a way the engineers never considered, but which was completely predictable in 20/20 hindsight, the net effect was the stall warning was backwards from what was really going on.
The pilots correctly deduced that it was meaningless. But they couldn’t solve the rest of the puzzle before the ocean arrived.
Unlike the MAX’s ill-considered decision to use only a single AOA input to MCAS, these latent defects in Airbus systems have never been addressed as needing to be fixed. American almost lost an A300 back around 2000 due to a broadly similar issue.
Summarizing mightily, different subsystems have different tolerances for when they decide a sensor input is so weird that it should be disregarded as defective. Meanwhile other systems are still believing the defective input. Last of all, it’s possible to get the airplane into flight situations where the input is accurate, but the engineers assumed the airplane could never be put in that situation, so the perfectly valid inputs are ignored as invalid. But only by some systems.
Once the higher-level avionics get confused and alarms are going off (or not going off) wrongly and defective information is displayed while valid information is removed, it takes superhuman effort and perhaps magical clairvoyance to determine the actual aircraft state and then maneuver to get it back towards normal.
Trying to put an AI in charge of a scenario like that would certainly not do any better in that situation than a human. And would probably do worse.