Pilot error blamed for Emirates near disaster at Melbourne Airport

Herald Sun ^ | 19th December 2009 | Ben Packham and Geoff Easdown

[naturalman1975]

A NEAR disaster involving an Emirates jet at Melbourne Airport was the result of human error by two apparently alert pilots, air safety investigators believe.

The March 20 scare, when an Airbus A340 struggled to get airborne, was caused by an "inadvertent" keystroke on a flight computer.

The error meant the Dubai-bound aircraft was flown on the basis that it was carrying 100 tonnes less than it actually was, the Australian Transport Safety Bureau said.

Its tail hit the runway five times during a botched takeoff.

The ATSB's interim report said the first officer "inadvertently inserted a takeoff weight of 262.9 tonnes, instead of 362.9 tonnes", and the pilot failed to pick up the 100-tonne shortfall, leaving the aircraft with insufficient takeoff thrust.

[saganite]

The FMC would filter all pilot inputs based on what it thought the weight of the plane was but I would think they could regain control of thrust by disconnecting the auto throttles.

[PapaBear3625]

I can't believe they didn't do an accelleration check. Ye gods, we were doing those forty years ago with stopwatches. Mechanical stopwatches.

Elementary software design involves the realization that humans WILL feed in bad data from time to time, and trying your best to avoid disaster WHEN that happens.

In this case, adding an accelerometer to the plane's sensors (if they can put them in Wii game wands, they can afford one on a plane) and having the software notice when acceleration doesn't match up with expected runway length and sounding an alarm might have saved an accident or two.

[PapaBear3625]

All they had to do was push up the power when they saw the boundary fence coming up.

A competent pilot should have noticed their airspeed was inadequate WAY before the boundary fence. A competent and alert pilot should have noticed inadequate airspeed before the plane covered a quarter of the runway and either hit max throttle or abort takeoff before reaching a critical point.

[saganite]

The only problem with aborting is that the abort speed is predicated on several factors, one of which is weight. If you are not accelerating as fast as you should be you will reach your computed abort speed further down the runway leaving less room (and possibly not enough room) to accomplish the abort. You’re absolutely right about pushing up the power though. Once the end of the runway was coming up and you’re still not airborne the only real option is to push up the power.

[taillightchaser]

IF IT AIN’T BOEING, I AIN’T GOING!!

[RossA]

This is 100% a design fcuk-up. Ask any engineer: how can you justify a typo causing (potentially) 300 deaths and $100,000,000 in damages.

The computers knew how many people were on board. They knew how much the fuel weighed, given the flight plan. How can they not know the total weight to within 20 tons? Let alone allow a 100 ton error?

[JohnInSoCal]

I fly airbus on jetblue I didn’t know they had to enter proper weight in order to take off that’s crazy. If the pilot gives full throttle it should go full throttle, why put a govenor on a plane ? To save some fuel mileage ? I believe there was an earlier incident where the airbus wouldn’t let the pilot fly the plane properly and crashed. Taking control away from the pilot and giving it to greenpeace carbon counters is a bad idea.

[ArmstedFragg]

I wonder how you say, “first the V’s, then the rotate” in their lingo?

[behzinlea]

I just read Capt. Chesley "Sully" Sullenberger's memoir, "Highest Duty." He recounts lessons his father instilled in him in his youth, including one my own father taught me: measure twice, cut once.

That same attitude would have prevented this mishap.

[saganite]

Nope. The computers only know what they’re told and if the pilots input the wrong data the computers don’t have an independent method of checking. Total number of pax onboard is not a data entry although maybe it should be. That way the bullshit flag would be raised when an absurdly low number was entered for the total weight. The only entry is the payload weight and if the pilots enter that in error there’s nothing (to my knowledge) that will catch it.

[MHalblaub]

"Different airline."

No, that picture shows a failed engine ground test prior to delivery.

What you see is Airbus homework.

[wolf78]

The computers knew how many people were on board. They knew how much the fuel weighed, given the flight plan. How can they not know the total weight to within 20 tons? Let alone allow a 100 ton error?

Nope. Or at least not necessarily. Usually this is done on two different systems: One on the ground for check-in, luggage and connections and one in the cockpit. With older planes the check-in system will generate a load sheet, which is then used to calculate the necessary fuel, thrust etc. (all this is done on the ground) and the pilot is given the final data. With the the newer glass cockpits you can hand a load / weight and balance sheet (short form) directly to the captain, who types in a few numbers into the onboard computer, or you could send the data via a secure connection.

The computers onboard don't care about which seat is taken. It could be done (e.g. seat occupancy sensors etc., still doesn't account for freight), but 1990s generation airplanes (like the A320 or the 737NG) don't have it.

And no, this has nothing to do with Airbus vs. Boeing. If a Boeing pilot uses wrong data (e.g. all you need to do is mix up printer numbers and send the load sheet to the wrong gate - if the pilot doesn't check the actual flight number on top, he won't notice when an airline flies a lot of the same type aircraft like e.g. Southwest), you run into problems. That's why you have to check at least twice - on both Boeings and Airbuses.

[wolf78]

The computers knew how many people were on board. They knew how much the fuel weighed, given the flight plan. How can they not know the total weight to within 20 tons? Let alone allow a 100 ton error?

Nope. Or at least not necessarily. Usually this is done on two different systems: One on the ground for check-in, luggage and connections and one in the cockpit. With older planes the check-in system will generate a load sheet, which is then used to calculate the necessary fuel, thrust etc. (all this is done on the ground) and the pilot is given the final data. With the the newer glass cockpits you can hand a load / weight and balance sheet (short form) directly to the captain, who types in a few numbers into the onboard computer, or you could send the data via a secure connection.

The computers onboard don't care about which seat is taken. It could be done (e.g. seat occupancy sensors etc., still doesn't account for freight), but 1990s generation airplanes (like the A320 or the 737NG) don't have it.

And no, this has nothing to do with Airbus vs. Boeing. If a Boeing pilot uses wrong data (e.g. all you need to do is mix up printer numbers and send the load sheet to the wrong gate - if the pilot doesn't check the actual flight number on top, he won't notice when an airline flies a lot of the same type aircraft like e.g. Southwest), you run into problems. That's why you have to check at least twice - on both Boeings and Airbuses.

[wolf78]

I fly airbus on jetblue I didn’t know they had to enter proper weight in order to take off that’s crazy. If the pilot gives full throttle it should go full throttle, why put a govenor on a plane ?

That's not how it works. You are talking about two different things. One is the onboard (I think it's a laptop for the A320) computer that does the weight and balance calculations. The same kind of calculations that decades ago were done manually on the ground. A pilot of a 1960s Boeing 737 also gets takeoff thrust recommendations. The only difference is that today the pilot can be given the raw data (e.g. just the weight / load for aircraft sections A,B,C and you're almost done) and do the stuff onboard, which saves time.

The other thing is flight envelope protection, which is a safeguard against e.g. intentionally stalling the aircraft in mid-flight. It only prevents actions that would threaten the structural integrity of the aircraft or make the plane fall out of the sky, not going full throttle on takeoff. Programming to slow an acceleration into the autopilot can of course be overridden manually.

To save some fuel mileage ? I believe there was an earlier incident where the airbus wouldn’t let the pilot fly the plane properly and crashed. Taking control away from the pilot and giving it to greenpeace carbon counters is a bad idea.
This could refer to Air France Flight 296 or American Airlines Flight 587