Email
Password
Remember meForgot password?
    Log in with Twitter
Blog Posted in avatar   Michael Cosgrove's Blog

Air Accident investigations

blog:9963:6::0
By Michael Cosgrove
Posted Dec 18, 2010 in Technology
Planes crash sometimes, that’s just how it is. But although air accidents are terrible events I am fascinated by the way they are investigated. It’s a painstaking and long process and you sometimes wonder how the hell the investigators ever find the reasons for accidents.
And there’s hardly ever just one cause. Whatever the circumstances of a crash, if something else had been different about the plane’s construction, or the pilot, the training, the maintenance procedures, or the rules and regulations, the airport, the air traffic control systems – and it’s often a mix of several - the accident may not have happened. So that’s what the investigators do, they look for all the causes.
Here’s an example.
On August 16 1987, Northwest Airlines flight 255 was preparing to leave Detroit Metropolitan for Phoenix. The plane was one of the McDonnell Douglas MD-80 series, which is a bigger version of the DC-9. The plane was captained by a veteran pilot and he had a very experienced co-pilot. There were 155 passengers and crew aboard.
As the plane got ready to taxi out the weather got worse and violent storms were predicted. The flight was late and as it taxied out towards its designated runway the increasingly high winds obliged the tower to instruct the pilot to take off from another runway instead, that which put the flight even further behind schedule.
He finally began his take off roll and found that the auto-throttle wasn’t engaged – and it needs to be engaged for takeoff – so he engaged it and continued his roll and took the plane off the ground. Once in the air though it couldn’t climb higher than 50 feet and it rolled from side to side. Then part of the left wing was sheared off by a lamp post at the side of a highway which was just a small distance from the end of the runway. The plane struggled on a couple of hundred yards but was unflyable so it crashed onto the highway and skidded along it before smashing into an overpass and exploding. Only one person – a four-year-old girl, survived. The flight had lasted just a few seconds.
Investigators first suspected the weather because of warnings about extremely violent downdrafts of wind called microbursts, which are capable of bringing even the biggest plane down. But after checking the weather data they found that that couldn’t be the cause.
The next possibility they considered was that the plane was overweight. Planes calculate their permitted takeoff weight limit with respect to a lot of parameters including weather and runway length. The runway they took off from was shorter than the one they had originally calculated for so maybe the plane just didn't have enough speed before it was obliged to rotate - leave the ground - to avoid running off the end of the runway? But the weight was found to be within the limit for the runway so that lead was also a dead end.
Then they found that the flaps handle on the central pilots’ console they found in the wreckage was in the retracted position. This was crucial. Flaps are the bits that run along the rear end of wings which extend during takeoff and can be moved up and down. There are similar elements along the front or "leading" - edges of wings, called slats.
Extending them during takeoff increases the wing’s surface area which means that more air pressure is applied to it as the plane accelerates and that helps it take off. Almost no modern plane can take off without flaps and slats extended and the fact that the handle was at retracted suggested that neither flaps nor slats were extended, thus making the plane impossible to fly. But other pilots waiting for takeoff said they saw them extended so the investigators needed more evidence.
They got that information when the Flight Data Recorder analysis got back to them. The data showed conclusively that the handle had not been moved to the extended position and that the other pilots’ impressions had been wrong. They also found corroborating evidence in the fact that the cables that make the flaps move had been severed by the lamp post and that they had been severed whilst in the retracted position. So the pilot had made an almost unbelievable blunder. Or at least that is how it appeared.
They simply could not believe that such an experienced pilot would have made such an enormous mistake if everything else had been OK so they looked further and discovered by listening to the pilots’ conversation on the Cockpit Voice Recorder that they had forgotten to do an essential part of the checklist routine. Many checklists involving hundreds of checks have to be done before a plane takes off. But the checklist they had inexplicably forgotten was absolutely crucial because that checklist has to be done to configure the plane for takeoff.
Why did they forget? Further analysis of the pilots’ conversation showed that what with the bad weather and the increased workload involved in changing runways unexpectedly they just simply forgot it as they discussed how to get to the other runway.
But there was a problem with that theory too. Why didn’t the pilots realize that the fact that the auto-throttle was off during the takeoff roll indicated that they may well have missed something? But they continued once they had switched it on, probably because they were late and everything else seemed OK.
But even then there was a problem. Even if the slats were retracted and if the auto-throttle wasn’t working, there should have been an alarm going off in the cockpit to warn them, and the investigators would have heard it if it had gone off, but there was no alarm. If a plane hasn’t been configured for takeoff an alarm always sounds, so why didn’t it sound here?
They had the tapes analyzed by sound experts and found that the alarm did in fact sound, but only on one side of the cockpit and at very low volume, so the pilots most probably never heard it. So, why wasn’t the alarm working properly?
The investigators studied and questioned pilots of similar planes while they prepared for takeoff and played the sound made by an alarm only sounding on one side and faintly to them and asked if they knew how that could be made to happen. One pilot answered that most pilots manually deactivated the circuit breaker for that alarm by pulling it out whilst taxiing because it had a tendency to go off if the throttles were pushed a little too hard and that distracted and even annoyed them.
The investigators also noticed that the circuit breakers on most MD-80’s had often been deactivated because they had grease and other marks around them which had obviously accumulated from the fingers of the many pilots who had deactivated them over the years. So there was more than a good chance that the pilot of the plane that crashed had done the same thing.
But why had they forgotten the checklist? After all, if they had done the checklist there would have been no reason for the alarm to go off and everything would have been fine.
They contacted NASA, whose Shuttle checklists are enormously long and complex and asked them to analyze the checklists on planes and how they were designed. Their experts found that if pilots got distracted during takeoff preparations they often forgot where they had left off. But regulations ensured that pilots had to begin again at the top of a page after interruption so that should have ensured that the checklist was done.
But they found that in practice pilots didn’t always go over the whole checklist again because the number of items on each checklist – over 20 – took too long to repeat. So NASA recommended that checklists be fragmented into lists of about 5 items so that pilots would only have to repeat the five items on the checklist they were stopped from doing when they got back to them.
Also, the alarms were redesigned to make it impossible for them to sound during taxing, so pilots didn’t have to switch them off any more.
In other words, if the checklists and alarm systems had been better designed they would not have been misused by pilots, the pilots in the crashed plane would not have made the error they did and the accident would not have occurred. The circuit breaker wouldn’t have been pulled, the checklist would have been completed, and even if the breaker had been pulled the checklist would have been done and if the checklist wasn’t done the alarm would have made them abort the takeoff.
So what people call “pilot error” is almost never the sole fault of the pilot. Moreover, bad maintenance is almost never the sole cause either, nor is a malfunction in the plane, nor is almost anything else. That’s what is so fascinating about these investigations.
They find all the elements that cause an accident and make changes happen for the future. They may be painstakingly long and complex to carry out, but they are one of the biggest reasons why air safety has improved to the point it has over the years.
----------
(Dunno why I wrote this blog. Guess I just like everything to do with planes…..Oh, and here's a link to a documentary on that accident)

blog:9963:6::0
Latest News
Top News