Wednesday, June 03, 2009

Air France 447

The disaster involving Air France flight 447 is a very odd one. A widebody airliner, running a regular scheduled oceanic flight, suddenly dissapears with no distress calls. A debris field has been found, so it seems certain that the aircraft is destroyed - whether this happened prior to or after contact with the water is to be determined.

I've been sifting the web - there are a million theories out there, but here's a super-condensed version of what seems to be known, and the most reasonable conjectures out there.

Firstly - what seems to be known (with UTC times).

A couple of notes - aircraft fly 'routes' in the sky that are marked with 'waypoints' at specific locations. Over the ocean, there is no radar coverage (it only extends a certain distance from land due to the horizon). So normal procedure is to radio in to Air Traffic Control as you pass a waypoint, advise them of where you are, and advise them of your estimated time at the next waypoint. These reports from aircraft allow ATC to track aircraft they cannot see, and to keep adequate separation between aircraft over the ocean.

The waypoint TASIL mentioned below is the halfway waypoint between South America and Africa. At this point, the flight would be 'handed over' from Atlantico ATC (CINDACTAIII - covering the South American side) to Dakar ATC (covering the African side). HF radio communications over the ocean can be a bit tricky, and other pilots have indicated problems in establishing communications when transitioning from Atlantico to Dakar, so just because no radio transmissions were received, does not mean that none were made.

22:03 (19:03 local) The Aircraft (An Airbus A330) left Rio de Janeiro, heading for Paris Charles de Gaulle

01:33 Flight reports to Atlantic Area Control as it overflew the INTOL waypoint, indicating it expects to arrive at the TASIL waypoing at 02:20

01:48 AF447 leaves the radar of CINDACTAIII at Fernando de Noronha, as expected. The aircraft is flying at 35,000 feet, and 453 knots as it leaves radar coverage.

02:20 The aircraft fails to report at waypoint TASIL. Atlantico ATC raise the alarm, and inform Dakar ATC also.

05:33 No further radio or radar contacts from the aircraft. A search is initiated by the Brazilian authorities.

The above, seems to be fact. What follows is less solid - but it seems to me the "most useful" info and the "most likely" scenarios that I can glean from the torrent of information, theory, supposition, and wild-eyed conspiracy flying around.

According to the Aviation Herald, "sources within Air France" suggest that the aircraft emitted a stream of ACARS messages, starting at 02:10 and ending at 02:14. (ACARS is a system for transmission of simple, short messages between the aircraft and ground, which can also be automatic). The unconfirmed sequence of ACARS messages is alleged to be:

02:10 Autopilot had disengaged, fly by wire system had changed to Alternate Law
02:11-02:13 A flurry of messages issued indicating faults with both ADIRU and ISIS systems
02:13 PRIM 1 and SEC 1 faults indicated
02:14 An advisory is issued regarding "cabin vertical speed".

The final ACARS report on "cabin vertical speed" pretty much speaks for itself. The most interesting one is the first one - the deselection of Autopilot, and the selection of "Alternatve Law" for the flight control system. Normally on the A330 (and other Airbuses), the flight is controlled by the pilot, via flight control computers (two of which - PRIM 1 and SEC 1 - started to get cranky at 02:13). Normally there are quite a few protections built in to these computers - they stop you getting the aircraft in too wierd positions, basically.

In "Alternate Law" mode, a lot of the protections are removed - you can stall the aircraft, you can put it in to extreme banks left or right, and you can fly at extreme angles of attack.

What is interesting is, under what circumstances would Alternate Law be engaged? It would seem that it's caused by "multiple failures of redundant systems". One other site suggested that Alternate Law can also be engaged if the aircraft gets in an attitude that is abnormal (i.e. pointed way up or way down, or banked left or right to an extreme degree).

So IF Alternate Law was in fact one of the first problematic ACARS transmissions, it could be supposed that the aircraft was in a highly unusual position *before* the other failures (ADIRU, ISIS, PRIM 1, SEC 1) started.

What could have gotten the plane in a highly unusual position? There's a fantastic weather analysis by Tim Vasquez here (it's a must-read if you are curious about this disaster) which for me comes the closest so far to putting a finger on a likely start to the problems. The plane had to fly through the ITCZ, and the thunderstorms there are not your average animal.

If you combine Vasquez's moving animation of the expected aircraft track, together with the times of the error messages, it does start to look more likely that extreme weather may have been where this problem started.

People are also pointing to the possibility of a dual engine flamout. On another forum, a pilot alleges that the aircraft had CF6 engines, and that they have a history of rollbacks (uncommanded reductions in power), including ones caused by super-cooled water droplets in the vicinity of thunderstorms.

Could it have been a double-engine failure?

Others are pointing out the possibility of a rudder failure, which Airbus may have had a couple of in the past, particularly when extreme rudder deflections are called for from the flight deck. Mabye. In response to extreme weather?

Finally - as of writing it seems to be a debris field of up to 35 miles in length. If that's the case, it strongly suggests the aircraft broke up in the air, not upon impact with the water. That's an interesting investigative point, and also it might provide some small solace to the families of those involved. If the aircraft broke up in the air, it would (together with the very quick flurry of ACARS error messages) point to something going wrong very very quickly.

One thing is for sure - I don't like 2-engined aircraft flying ETOPS long haul flights, no matter how reliable modern engines are. And I definitely don't like fly-by-wire either. Not only is the computer modifying what the pilot is trying to get the aircraft to do, but if for example the pilots were still concious and attempting to control the plane in the 02:10 - 02:14 timeframe, then not only were they trying to cope with extreme weather, and a huge aircraft getting out of control in the dark, but their systems were bombarding them with error messages - Alternate Law engaged, ADIRU faults, ISIS faults, PRIM 1 fault, SEC 1 fault - all this at the same time as trying to figure out what way up they were (it was dark) and how they could rescue the plane.

The area where the debris has been found is several thousand feet deep, and the locating transmitters on the Flight Data Recorder and Cockpit Voice Recorder only function for 30 days. Let's hope they can find the two boxes before that time is up, and a bit more light can be shed on what happened in the last few minutes of the flight, and doing everything possible to prevent a repeat.


Anonymous said...

I'm surprised that no-one has mentioned the possibility of this awful accident being related in some way to the spate of incidents that have involved uncommanded upwards and downwards pitching of the A330 aircraft. Last year's near-disastrous Qantas accident, which resulted in many injuries, is a prime example.

Also,if it turns out that turbulence was a factor in the Air France accident one is drawn to wondering whether the rudder could have failed, as was the case of the American Airline's A300 crash in Queens, NY,back in 2001?

Anonymous said...

What is curious to me is the total lack of radio contact. Even if (for example) both engines flamed out or something like that- why no calls or contact. It seems to me that the problem may have been within the cockpit itself. We keep hearing storms and lightning. Could a lightning strike have blown off the front of the plane? That would account for the breakup pattern over a wide area, the lack of radio contact, and the autmatic messages being sent back.

Anonymous said...

I have flown this track many times and can attest that HF reliability is very spotty in the ITCZ. Couple that with the cascading fault messages through the ACARS system and there is little liklihood of a successful MAYDAY call ...

Anonymous said...

Anonymous said...

Updates -

Anonymous said...

I understood that all ACARS messages are tagged with a time, longitude and latitude position as part of the communication package transmitted. Why is it then so hard to locate the approximate location of the downed airbus?

Anonymous said...

I know pilots who have lost "fly by wire" and the aircraft reverted to "Alt Law". This was due to a lightning strike to the pitot mast taking out the CADC completely on one side and badly damaging the others. The pilots were well seasoned with thousands of hours in the air and well over 2000 in the A320 in which this happened. They said is incident was the worse one in their career. It was day light with some storms in the ATC area of greater New York.(LGA,EWR&JFK)traffic. Put this senario on the Air France crew in the dark with a "super storm" over the ocean.