Slashdot Mirror
Data Entry Blunders Force Air Asia Pilots To Land in Melbourne Instead of Malaysia (mashable.com)
A flight from Sydney to Malaysia ended up in Melbourne after the captain incorrectly entered the plane's location in its navigation system just before take-off, according to a safety investigation, whose conclusion was published this week. Mashable reports:The Air Asia pilots made several errors in entering data into the aircraft's navigation system, which caused them to follow an incorrect flight path out of Sydney, according to Australian transportation officials. While troubleshooting the incorrect flight path, the pilots were unable to fix the issue, and may have compounded it. The aircraft's systems would not allow the plane to be flown in instrument conditions and the weather also had deteriorated in Sydney by the time the pilots decided to turn back. They were directed via radar to a visual approach in Melbourne where they could land safely. The pilots did not believe the airport was located in Malaysia.
Mashable reports...
That summary is such a hodge-podge of disconnected half-facts. A link to a publication written by trained journalists, or even trained monkeys, might be more coherent.
From the investigation report :
The ATSB also found that the aircraft was not fitted with an upgraded flight management system that would have prevented the data entry error via either automated initialisation or automatic correction of manual errors.
Wow. You found us out. You are so smart. Now post something about Bilderberg and the Fed!!!
The same happened in Brazil almost 30 years ago (https://en.wikipedia.org/wiki/Varig_Flight_254):
Varig Flight 254 was a Boeing 737-241, c/n 21006/398, registration PP-VMK, on a scheduled passenger flight from São Paulo, Brazil, to Belém, Pará, Brazil, with several intermediate stopovers, on 3 September 1989. Prior to takeoff from Marabá, Pará, towards the final destination, the crew entered an incorrect heading into the flight computer. Instead of flying towards its destination, the plane flew due west and after some time was over a remote area of the Amazon jungle. Attempts to reach an alternative airport were unsuccessful, and the plane eventually ran out of fuel. The pilot made a belly landing in the jungle, 1,050 mi (1,690 km) northwest of Rio de Janeiro. There were 54 occupants on board—48 passengers and a crew of six; 13 passengers died, and many more sustained serious injuries. The survivors were rescued two days later.
Have gnu, will travel.
Umm, right. You don't just 'unhook' the autopilot and stuff in a new one. This is not a car stereo.
I suppose you could just sneak a 777 off the tarmac a bit ('hey guy's, it's cool') and rip out the cockpit guts and stuff your new black box in, do extensive system checks, button up everything and sneak it back. Without anybody whatsoever noticing.
But I doubt it.
Faster! Faster! Faster would be better!
And everybody thinks they're crap at driving...
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
This is a much more complicated and interesting story than the headline or first glance would suggest. (because as is common, the headline makes it sound like a bunch of bungling pilots from a LCC airline were flailing about stupidly, which is not the case)
The omission of a trailing zero digit in the manual entry of longitude during system initialization caused serious autopilot/navigation problems that were not resolved by automated cross-checks that should've caught it. (Error #1)
Then, as a result of trying to fix/diagnose the problem on the fly, the flight display/instruments were put into a failure/safe mode where only visual flight conditions could be handled (Error #2).
It turned out ok in this case (just a diversion), but if the weather had been poor or other combinations of conditions existed, it could've easily gone wrong. Very interesting...
So are they former or future Slashdot editors?
This is because the paragraph above was posted as a correction to the article.
Otherwise the article gives the impression that the pilots were so incompetent as to not even realise which country they were in.
systemd is Roko's Basilisk.
e.g. You plug in all the numbers for your flight path. It should then display a world map with your flight path overlaid, so you can easily check that the numbers you entered have you at least landing on the right continent. This sort of sanity-checking is common in other fields, like accountants check to see if a discrepancy is divisible by 9 to quickly identify a transposition error.
This is one of the reasons I still advocate doing navigation in nautical miles instead of km. One nautical mile is defined as 1 arc-minute of longitude at the equator, which is also pretty close to 1 arc-minute of latitude anywhere on Earth. So basically you look on your big navigation chart with latitude lines labeled in degrees and minutes, and you immediately have a sense of scale in terms of nautical miles (each degree is 60 nm). You do a bunch of complex navigation calculations in nautical miles, plot it on the chart, and say, "Hey that doesn't look right. My destination is over 30 arc-minutes away, but my calculations say I'll only be sailing less than 30 nm. I must've made a math error somewhere..." Whereas if you do it in km, the sense of scale is not as intuitive and you may not uncover the error until you're far along the route and wondering why the landmarks you were expecting aren't showing up.
This is kind of a false idea since a Boeing has the exact same error potential. I was on a flight one time where a navigator messed up the whole inertial reference system and we lost all positional awareness (just like in this article). We didn't get "lost" because I navigated by other means. However, it was on a Boeing airplane. Airbus and Boeing, and while we're at it: Embraer, Sukhoi, Canadair, etc. have the same mode of failure in their navigation systems.
The Airbus is highly automated but it is very well thought out. In the worse case, it can be flown like a more traditional airplane (Boeing). In fact, all the systems revert to a "direct law" system where the pilots have full authority over all flight control systems and the pilots can break stuff by exceeding limits. Honestly, more pilot error happens with Boeings than with Airbus. Think of an older Boeing (e.g. 737) as an old circa 1980's chevy with electronic fuel injection but nothing more. Think of an Airbus as that plus systems such as automatic breaking that keep you from rear ending another car. Worse case, that automatic system can be shed so that you can crash into the car in front of you.
These pilots made logical choices which ended in a null condition. Unfortunately, this is what happens when people only know their job and not the theory behind how everything works. I doubt these guys were airplane computer engineers so without the deep knowledge of the systems (e.g. at the code level), anything they would have attempted was probably a guess. The manuals also may not have been specific to this situation and thus might have put them in this negative condition.
Nevertheless, I can't help but to wonder if they tried using traditional terrestrial navaids to determine their position or if they were simply in oceanic airspace and inertial reference was their only means of positional awareness.
checks article Yup, Airbus.
Nice selective quoting.
What it actually said was "The aircraft's systems would not allow the plane to be flown in instrument conditions". This basically says the autopilot refused to kick in due to weather conditions and threw controls back to the pilots. The pilots didn't want to land in Sydney due to the aforementioned weather conditions and diverted to the nearest major airport... which was Melbourne.
So Yep, Airbus... did it right.
Calling someone a "hater" only means you can not rationally rebut their argument.
http://www.gcmap.com/featured/...
It would have been nice to see a map in the news article to give some idea to those reading it unfamiliar with the area just how big of a "blunder" this was. From the Great Circle Mapper website I linked to above we see that KUL is about 4000 miles from SYD, and SYD is less than 500 miles from MEL. Given the typical cruising speed of a jetliner they were in the air for perhaps not much more than an hour on a flight that would have lasted 8 or 9 hours. Since they knew right away something was wrong I doubt they were flying much longer than that, maybe 3 hours. If they were flying much longer than that I suspect they would have landed much further from either SYD or MEL, or we'd be reading about a plane lost at sea.
The article makes a big deal about "landing in the wrong country" which I suppose is a big deal if you take off in the USA, headed for Canada, but end up in India. Much less of a deal if you take off from USA while headed for Canada but a technical problem means you have to land back in USA.
I am armed because I am free. I am free because I am armed.
OK, here's what really happened:
The captain entered the wrong coordinates into the Inertial Reference System, which is the airplane's primary reference for position. Normally, there are several ways the system could have detected this:
- It is normally updated by GPS, but if the difference between the two is too large, the sytem considers the GPS to be faulty. Some warning messages were given, but these warnings often occur during normal operation on the ground and then disappear afterwards, so pilots are sort of "trained" by experience to disregard them.
- When applying take-off thrust, the system normally realigns to the coordinates of the runway. But, thanks to genius Airbus programming as usual, the system did not perform the update because the error was too large.
- If an incorrect latitude is entered, the system can detect the error because it measures the earth's rotation vector and the gravity vector during IRS alignment while the airplane is still parked. In this case the longitude was incorrect, which cannot be detected by the system. The earth rotates the same at any longitude.
After Take-off, two problems arose:
1. The position was way off, the plane actually thought it was somewhere close to South Africa
2. The Enhanced Ground Proximity Warning System activated, yelling "Terrain, terrain!", based on the incorrect position (not a big problem, but confusing as hell and creating extra stress for the pilots).
3. The heading indication was incorrect, because the IRS internally knows only the true heading (relative to true north) while heading indications on aircraft systems are relative to magnetic north for historical reasons. The IRS converts between the two using local magnetic variation (the difference between magnetic north and true north) but, because it thought it was near South Africa, it used the wrong value for the variation and therefore displayed an incorrect magnetic heading. This caused the autopilot to turn the wrong way.
The pilots, fearing that something was wrong with the entire Air Data Reference system, applied a procedure for unreliable airspeed (they did not know which instruments they could trust anymore) and turned off two of the three ADR computers. This resulted in a further loss of information displayed on their screens which was not exactly helpful to their situation. The procedure exists to disable certain safety protections that might accidentally activate based on erroneous data (if two of the three computers say the airplane is stalling, the flight control computers will push the nose down and override the pilots' sidesticks) but in this case this procedure was not necessary and actually made their life harder.
Now, if you've lost half your instruments, you don't know which instruments you can trust and which you can't (with the heading being wrong for certain), it's not a good idea to fly on instruments through clouds at low altitude. Certainly not an approach to a runway without being able to see it. So they needed to go to an airport where the visibility and cloud base allowed a visual landing. The closest suitable airport with good weather was Melbourne, so that's where they went.