Here you can see the last automatically transmitted ACARS messages of AF447:
Personally, I think this incident was caused by a combination of factors.
ALL speculations tread on thin air before the CVR and FDR are recovered, but based on current data I would QUESS:
-it is a dark, stormy night with no horizon or any landmarks visible
-160km/h updraft brings moist air to a much higher flight level than usual
-this causes sudden icing of the pitot tubes
-this causes the flight computer to think that the plane is in danger of stalling, and it lowers the nose automatically
-the crew switches auto-pilot and flight envelope protection partially off, or a (positive) lightning strike disables them
-the crew has no good idea about the true speed and orientation of the plane
-inside the cumulonimbus' horrible gusts the crew over-stresses the composite flight controls while fighting turbulence
-the place exceeds it's maximum speed and/or structural load (G-) limits
-two-three minutes later the agony of the 228 souls on board finally ends as the slowly disintegrating plane hits the sea near the speed of sound, instantly ripping them to stamp-sized bits
Here's more detailed speculation about possible causes and a crude analysis, taken from Usenet:
1. Terrorism or other malicious use of explosives
A bomb explodes in the cargo hold, crippling the aircraft's control systems or starting a structural break-up that eventually leads to loss of control.
Supportive evidence: According to Wikipedia, a bomb threat had been made on an earlier flight. Lack of communications from the flight crew indicates either a sudden event or something which lead to significant problems that the crew had to focus on. This would be consistent with the effects of a bomb. The automatic messages about computer system failures sent by the aircraft could be interpreted either as indications that the aircraft's movements have exceeded the limits that the systems can handle, or as indications about direct damage to the systems. A flash of light has been seen by other aircraft in the area.
Evidence against: While terrorist organizations exist both in France and Brazil, there has been no recent activity. No organization has claimed responsibility for the act. There is no specific evidence about a bomb. Nothing is known about any individuals or organizations who would have non-terrorism related reasons for malicious acts. It seems too big of a coincidence that a bomb would go off at the same time as the aircraft flies through very rough weather. Finally, what we know about the sequence of ACARS messages indicates that loss of cabin pressure was the last message in the sequence. This appears to rule out an explosion, unless it was contained in the hull and only damaged internal structures and components. This seems unlikely. The flash of light was apparently seen from too far to be caused by AF 447 related problems.
Open questions: Where are the cargo holds that are used to carry the passengers' luggage? Are they physically close to the computer and navigation systems that ACARS messages reported as failing? And obviously, physical evidence would be useful.
Verdict: Can most likely be ruled out
2. Explosion or other rapid, harmful reaction from the cargo
The sequence of events is as in the terrorism theory.
Supportive evidence: The sequence of events fits this theory, as it does the terrorism theory. The cargo might have shifted at the time of turbulence, initiating the reaction.
Evidence against: See the evidence regarding the malicious use of explosives. In addition, there is no information that the cargo could have contained something harmful.
Open questions: More information is needed about what was in the cargo, and who cargo was taken from.
Verdict: Can most likely be ruled out
Fire starts in cargo hold, in systems, or somewhere else in the aircraft. Eventually the fire either disables a sufficient number of aircraft systems leading to a loss of control, or incapacitates the crew.
Supportive evidence: Systems turning themselves off one by one could be caused by fire. Cabin pressure warning might have been generated through the crew's actions to let the air out of the cabin in an effort to put out the fire. The location of the debris indicates that the crew might have turned back in an effort to get to the nearest airport (but it has now become uncertain if any debris has actually been found). The long duration of the ACARS messaging (4 minutes) seems to indicate a slowly progressing event like fire, as opposed to sudden event like a bomb or structural failure. The fire might have been initiated when the aircraft hit heave turbulence 10 minutes before the problems began.
Evidence against: It seems unlikely that the fire could have spread fast enough that the crew would not have notified air traffic control. Why would fire happen exactly at the same time as the aircraft flies through extreme weather? Wouldn't a cargo hold fire lead to an ACARS message?
Open questions: More information is needed about the ACARS messages and whether they could have been caused by fire.
Verdict: Possible, but there are several open questions
Turbulence breaks up the plane, or causes it to enter an uncontrollable dive.
Supportive evidence: Very high winds and turbulence were detected in the area. Tim Vasquez's analysis indicates that there were strong updrafts in the exact area that the flight was expected to fly through. The crew manually sent a report of turbulence 10 minutes before the accident.
Evidence against: other flights in the same time frame and route did not encounter significant turbulence.
Verdict: Possible, but there are several unexplained questions. In particular, why was only AF 447 affected? Or did several issues contribute to the outcome, along with turbulence?
References: Tim Vasquez's analysis: http://www.weathergraphics.com/tim/af447/ AP reports that the pilots sent a report about turbulence and timing of the various events: http://hosted.ap.org/dynamic/stories...CJAC&SECTION=HOME&TEMPLATE=DEFAULT
Lightning hits the aircraft and causes either structural or systems damage.
Evidence against: Tim Vasquez's analysis speaks against this.
Verdict: Unlikely, by itself
Hail hits the aircraft and causes either structural damage, shuts the engines down, or breaks cockpit windows and incapacitates the crew.
Supportive evidence: There has been rumors about ACARS icing messages.
Evidence against: Tim Vasquez's analysis speaks against the occurrence of significant hail. Unless the crew was incapacitated, mere engine shutdown would have lead to the crew contacting the air traffic control. What we know about the sequence of ACARS messages indicates that the cabin pressure loss happened last, which speaks against the accident beginning with the breakage of the cockpit windows.
Open questions: The sequence of ACARS messages needs to be known in more detail.
Icing causes the aircraft to lose its flying capabilities.
Supportive evidence: See the evidence for the hail theory.
Evidence against: Tim Vasquez's analysis speaks against the existence of supercooled water at the flight level and conditions that the aircraft was flying through.
4.5. Sensor icing
Speed or other sensors are iced over or malfunction in some other way, leading to the computers or pilots taking incorrect action, causing the aircraft to stall and/or break up.
Supportive evidence: There are rumors of icing-related ACARS messages.
Evidence against: Normally the sensors (such as the pitot tubes) are heated. Icing should not affect them, unless the crew failed to turn the heating on, particularly severe icing conditions existed in the strom cell that they flew through, or hail hit the sensors. Also, Tim Vasquez's analysis speaks against the existence of suitable icing or hail conditions.
Open questions: More data is needed about the actual ACARS message sent by the aircraft.
References: BBC article: http://news.bbc.co.uk/2/hi/americas/8083474.stm
5. Systems malfunction
A repeat of the Qantas incident: ADIRU failures cause the aircraft control systems to believe that a steep dive and/or nose-up is needed, leading to a loss of control and/or aircraft break-up.
Supportive evidence: See GlobeEx's posts.
Evidence against: It seems unlikely that the crew would have been unable to control the situation at FL35, when crews in the previous incidents were able to regain control with the loss of few hundred feet of altitude. There are some reports that AF uses ADIRUs from a different manufacturer. The long duration (4 minutes) of the automatic messaging speaks against a very sudden event.
Open questions: At this point, we do not yet know if AF used the same manufacturer's ADIRU (Litton) as was involved in the Qantas incident. Rumors indicate that the equipment comes from a different manufacturer, namely Honeywell. No previous incident is known on a Honeywell ADIRU.
Verdict: Unlikely, though more information is needed. Perhaps combined with a pilot error and the effects of the bad weather this theory is more likely, e.g., maybe the recovery was not executed in time.
Collision with another flying object such as aircraft or meteor.
Supportive evidence: This could cause similar effects as a bomb, so in some sense it fits the sequence of events. A flash of light was seen, which might also be from a meteor
Evidence against: There is no known collision with another aircraft. Meteor collision is extremely improbable. The flash of light was apparently seen from too far to be caused by AF 447 related problems.
Verdict: Very, very unlikely
References: White flash of light: http://ca.news.yahoo.com/s/reuters/0...orld/international_us_france_plane
7. Fuel tank explosion
A TWA800 -like event.
Supportive evidence: Fuel tank explosion, like other explosions, would fit the sequence of events observed through ACARS and lack of a mayday message from the crew.
Evidence against: By this time, the center fuel tank would have been very cold. TWA800 accident required a warm fuel tank for the vapors to develop in suitable quantity.
Verdict: Can be ruled out.
8. Explosive decompression
A cargo door or other part departs the aircraft, leading to an explosive decompression.
Supportive evidence: ACARS message about cabin pressure.
Evidence against: The sequence of messages is wrong, as cabin pressure drop was not the first one.
Verdict: Can be ruled out
9. Crew error
9.1. Flying into severe weather
The crew flies directly into the most severe part of the storm
Supportive evidence: Tim Vasquez's analysis of the weather system indicates that it would have been very hard for AF447 to avoid the weather system alltogether, but there is no evidence to suggest that they flew directly into the worst part.
Open questions: Exact flight path is needed.
Verdict: They did fly into the weather, but it is unknown if this was avoidable or if they flew to the worst part. Note that in combination with another problem, such as radar failure the likelihood of this theory would be much higher.
9.2. Flying too high
Trying to fly over the weather system and exceeding limits of the aircraft, leading to a "coffin corner" situation and an unrecoverable stall.
Supportive evidence: None, except perhaps the diffult and hard-to-avoid weather system. Tim Vasquez's analysis points to very high updrafts, which might have contributed to an involuntary gain in elevation.
Open questions: Additional information about the aircraft's altitude is needed.
Verdict: Unlikely, it is very hard to believe that an experienced crew would on purpose violate the aircraft's limits
9.3. Excessive corrective action
The crew over-corrects the effects of turbulence or high wind in a manner similar to AA587.
Supportive evidence: This might have lead to separation of control surfaces, which in turn might have lead to autopilot and other control systems giving up. The sequence of ACARS messages might match this situation.
Evidence against: A330 is a fly-by-wire aircraft which may not allow excessive control input. On the other hand, the first ACARS message received indicated disengaging the autopilot and the control systems entering the alternate law mode. This mode has less protection against exceeding the safe flight envelope.
Open questions: More information is needed about the exact ACARS messages and in what conditions they will be sent. More information is needed about the effects of excessive control inputs in the A330 fly-by-wire system.
Verdict: Very unlikely, unless something else happened, disabling the autopilot and then the crew accidentally exceeded the structural limits of the aircraft.
10. Prior damage from a ground collision
The aircraft in question collided in 2006 with an Airbus A321 aircraft. An undetected problem in the wing of the A330 might have stayed dormant until the airframe was stressed in severe turbulence.
Evidence against: The maintenance procedures for inspecting and correcting collision damage are quite extensive, and it is hard to image something was missed.
11. Combination of factors
11.1. Systems failure and crew error
A systems failure initiates an event and the crew fails to respond properly or in time. For instance, a radar failure might lead to the crew flying blind into the worst part of the weather. Or disengaging the autopilot or Airbus flight envelope protection leads the crew to overstressing the airframe.
Supportive evidence: There is no strong evidence of any single event that is know to bring down the airplane. There is evidence of some of the control systems (autopilot, ADIRU) shutting down. While these events should be recoverable, it is easy to imagine a crew error that exacerbates the situation, particularly in bad weather.
Evidence against: There is no information about any errors by the crew. Many of the systems, such as the radar electronics, have backups, and are not expected to completely fail.
Open questions: CVR and FDR are needed to find out more.
11.2. External event and crew error
Icing, turbulence, lightning or lightning causes an upset, and the crew reacts in an inappropriate manner or fails to recover in time. Or, alternatively the crew makes an error in setting the proper speed for the type of weather (turbulence, updrafts) which later leads to a stall or a spin.
Supportive evidence: There is no strong evidence of any single event that is know to bring down the airplane. There is evidence of conditions that are potentially harmful (like turbulence), but that are known to be experienced by a large number of aircraft, so its not clear why these events alone would have caused the accident. But it is easy to imagine a combination of two problems, or one problem and a crew error to lead to a disaster.
Evidence against: There is no information about any errors by the crew.
Open questions: Again, CVR and FDR are needed.
References: Incorrect speed has been mentioned in the press as a possible cause of the accident: http://ca.news.yahoo.com/s/reuters/0...orld/international_us_france_plane
11.3. External event leading to fire
Turbulence rocks the plane, leading to a short circuit and fire. Lightning hits the plane and causes a fire. The fire eventually spreads to the whole aircraft.
Supportive evidence: The crew reported turbulence, and at least turbulence and maybe even lightning is known to exist in the area. For the other evidence, see the fire theory.
Evidence against: See the evidence regarding fire; it is odd that no ACARS or crew message would have been received from a slowly progressing event such as fire in the cargo hold.
Open questions: Again, CVR and FDR are needed.
Verdict: Possible, but there are open questions