Emergent Wisdom

Thoughtful Observations from Business and LIfe

The Mystery of Flight AF447 - Deliberately avoiding the inconveniently obvious

I'm beginning to suspect that in yet another disaster where the true facts would be 'inconvenient', an effort is being made to find both technical and human scapegoats.


On 1st June 2009 Air France flight AF447 disappeared over the Atlantic on a flight from Brazil to France. There was no mayday call and initially it seemed to have just disappeared without any trace, until it was revealed that an on-board automatic system had been sending computer-generated technical messages describing a sequence of problems occurring on the flight and revealing its last known location. Eventually of course, a considerable amount of debris was recovered indicating the flight had at the very least crashed into the sea. But the flight data and cockpit recorders were still missing, presumed to be at the bottom of the Atlantic, along with most of the wreckage.


From the little information available and especially without the crucial recorders it was impossible to come to a definitive conclusion. But from the evidence that was available early on including the automated messages sent from the plane, known issues with certain pitot tubes used for measuring air speed, and awareness of the kind of weather encountered by the plane, some educated guesses could be made. It was provisionally deduced that the originating cause of the accident could well have been due to the pitot tubes icing up, leading to lack of information available to the on-board computers and the air crew, and thus contributing to incorrect decisions being made by the crew leading to a stall and crash into the sea.


On 30th May 2010, almost exactly a year after the accident and also almost exactly a year before writing this article, a programme speculating on the causes of the accident was broadcast on BBC2 called "Lost - Mystery of Flight 447". This contained a demonstration of a phenomena that seemed to have previously flown beneath my own personal radar: The significance of supercooled water freezing on contact with 'impurities' or rough handling. While on reflection I can recall years earlier Judith Hann on Tomorrow's World demonstrating supercooled water freezing when she took the top of the bottle, I was under the impression that supercooled water only occurred in the laboratory where it could be made especially pure. I didn't realise it can also happen in the earth's atmosphere at heights habituated by planes on long-distance flights.


Essentially, the problem was that flight AF447 was that it was likely to have gone through possibly not one but a series of storms. Whereas the first storm likely to be encountered may not have looked very big to the radar, weather patterns in the area indicate that it could have been hiding a much larger storm behind it, in the same way as the moon can eclipse the much-larger sun when seen from a person standing on the ground, simply because the moon is nearer. At present weather radar is not able to see through one storm to see what is behind it, and it is quite possible that they entered one stom thinking it to be relatively small and not particularly severe, only to find themselves confronting a much larger storm behind it.


But there is a crucial issue with the type of weather encountered and it's that they can contain large amounts of supercooled water. Supercooled water is water at a lower temperature than normally freezes at ground level. But then water at ground level nearly always has significant levels of impurities in it, providing seeds for ice crystals to form. Once water has started to crystalise into ice, the ice crystals themselves become the seed for adjacent supercooled water molecules to crystallise also. Thus, a significant amount of supercooled water can suddenly - ie, within a few seconds - become ice.


The significance of this to flight AF447 was that it was known that pitot tubes - a forward facing open tube to designed to determine the speed by measuring the air pressure in the tube - could become blocked by ice at high altitudes. Although as of For this reason, according to my recollection of the documentary, they were heated, to melt the ice, as a quick internet search confirms. But the Thales ones fitted to F-GZCP were found to be not sufficiently effective at melting ice fast enough and a few incidents similar to what was presumed to have happened to flight AF447 had been previously reported. For this reason a non-obligatory recommendation had been made to change them for more powerfully heated ones made by Goodrich, a change which hadn't yet been made to F-GZCP.


The programme didn't answer all the questions though. Even if the pitot tube icing up led to the problems experienced by the crew, why were normal fall-back procedures in those situations not effective at keeping the plane in the air? If a stall occurred, how come they were unable to recover all the way from the normal flying height of 35,000 feet?


Now of course the flight data and voice recorders have been recovered from a great depth, and very luckily in good condition. The preliminary examination of the contents confirms that the speed indications became completely unreliable, and that the pilots increased thrust and gained height, going through a couple of stall warnings up to 38,000 feet. Bearing in mind that it was dark, over the atlantic, in a storm, one realises that when there is no reference for basics such as air speed, it becomes very difficult to know how they could have flown at all. It is not difficult to get someone to feel they are being turned upside down and all over if they are blindfolded and merely being rotated in a chair.


But it was watching the programme on BBC 2 that provoked an obvious question from me, and one that was not even asked during the programme. Before I mention this, let's go back to 1992 when USAir flight 405 failed to take off effectively from La Guardia in New York and quickly crashed into Flushing Bay after only gaining a few metres of height, killing 27 people. On that occasion is was unambiguously decided that for a combination of reasons ice had built up on the wings before take-off leading to loss of lift and unstable handling. This was despite previous de-icing and some efforts the aircrew had takent to determine if any ice had re-formed on the wings. In that instance it had already been recognised prior to the accident that on a Fokker F28 only a small amount of ice was needed to seriously disrupt the aerodynamics of the wing and to create an unrecoverable roll, and not taking that into account may have led to the crew not sufficiently taking seriously the need to ensure the wings were ice-free before take-off. It certainly became clear because almost immediately on becoming airborne there appeared to be mutual agreement among the crew that the plane was not going to fly. The effort put into ensuring plane wings are de-iced before take-off at airports clearly indicates the critical effect of ice on a planes capacity to be a plane and not simply a plane shapped piece of metal.


Going back to AF447, what astonished me most in the programme was the speed at which supercooled water could become ice. Pouring supercooled water onto a pitot tube, the water became an almost instant ice-sculpture completely, covering it. In no time at all, the amount of ice built up on the pitot tube completely dwarfed the pitot tube itself. A demonstration of how rapidly that can happen is here:

{youtube popup="true" divid="popupsuper-cooled_water"}fSPzMva9_CE{/youtube}

{avrpopup type="lightbox" id="popupsuper-cooled_water"}super-cooled_water{/avrpopup}


Now, it's fair to say that this was water being poured out of a bottle, which wouldn't be encountered while flying, but instead a plane travels at several hundred miles an hour and in a cloud loaded with supercooled water, will be encountering substantial amounts of water in a short period of time. The question I had was, if that amount of ice can build up on a pitot tube, and even on a heated one, enough to block it and stop it working, how much could build up on the wing?


The effect of air heating on a wing is well known, and in fact Concorde had a special cooling system to cool the window glass in the cabin to stop passengers burning their hands on it because the outside skin temperature would reach 100 degrees c in flight at Mach 2. The heating effect would be considerably less at mach 0.8. As it appears that the plane may have been travelling through a prolonged stretch of intense weather, leading to the heated pitot tubes apparently icing up, is it not conceivable also that the wings may have been building up ice also? After all, despite the heating effects of air speed, a prolonged encounter with supercooled water could have cooled the leading edges of flying surfaces sufficiently to have enabled supercooled water to have formed into ice.


What concerned me about the programme was that this possibility that seemed obvious to me was not even discussed. In fact a glance through discussion on the internet on this subject almost always comes back to pitot tubes, and of the small proportion of the pages I looked through of the 43,000 that responded to the words 'AF447' and 'wing', I was not successful in finding any that discussed icing on the wing in the same way that the pitot tubes ice up.


I also have a problem with the notion of blaming the crew for poor decision-making in the circumstances. It's now known that the captain, who had been resting, had been fetched back to the cockpit, did not re-take command from the pilot in charge at that time. This essentially undermines the likelihood of two other human factors, that were seperately responsible for the world's worst ever accidental air crash, in Tenerife in 1977, and in the Kegworth crash in 1989.


The Tenerife disaster involving flights KLM 4805 and Pan-Am 1736 was essentially caused by an almost inexplicable decision of the KLM pilot to take off despite no definite clearance having been given, seemingly in denial of both risks and reality, and the co-pilot who was clearly aware of the error and the danger it involved but felt unable to challenge the captain's decision due to his deference to the captain's authority. As a result of this disaster there was theoretically at least a cultural change change of approach in the industry and theoretically at least, decisions made by a person with one set of responsibilities have become more easily challengeable by others. 


The Kegworth air crash was caused primarily by the beliefs of the crew overriding the information being presented to them by the display. The left-hand engine developed a serious problem, and the dashboard indicator told them this. However, the pilots believed that on the particular model of aircraft they were flying the air conditioning was driven only by the right-hand engine, and as they could smell oil in the air, beleived that it must have been the right-hand engine that was faulty instead, and disasterously shut that down. Incidentally, as a separate factor again, having told the passengers that the right-hand engine had been shut down because it was faulty, passengers who had seen flames and sparks coming out of the left-hand engine made no efforts to alert the crew, believing that 'they must know what they're doing'. **Update 20/04/2016: Information has just come to my attention, from someone who ought to know, that the official report is incorrect on this point and one or more people did inform the cabin crew of the discrepancy between what they had seen and the Captain's comments


But both of these primarily causal human factors do not explain why AF447 crashed. In this instance, it wasn't the captain that was in charge, it was one of the co-pilots, and even though the captain was brought back into the cockpit, he did not re-take command. This undermines the notion that it was a bad decision by one pilot that others present felt unable to challenge, and from what has been revealed so far this suggests instead that the more senior captain felt that the pilot was doing the right thing. Neither does it suggest that there was simply a mis-reading or misinterpretation of some technical information because the acknowledgement of failure of certain critical information led to the crew apparently adopting fall-back procedures. In addition, the captain being in the cockpit but not re-taking control suggests so far he was likely to have been taking a more 'stand back and re-evaluate' approach, which would involve making his own interpretations of information being given by the plane. There have been several incidents of plane engines failing in mid-flight, such as British Airways flight 9, where all four engines of the 747-200 failed after flying through volcanic ash and took several minutes to get restarted, and Air Transat flight 236 that ran out of fuel at 33,000 feet. Both of these managed to eventually land relatively safely, flight 9 by eventually managing to get the engines restarted after losing a lot of height, and flight 236 by gliding to the Azores. This raises the obvious question: How is it that three pilots were between them unable to get a modern plane, not apparently physically damaged and with all engines working, to stay in the air from a height of 38,000 feet?


The overall blame for the accident still seems to be focussed on two things: The pitot tubes, that have now been replaced, and errors the pilots seemingly made in the circumstances. Relatively little attention seems to have been paid to the ineffectiveness of weather radar resulting in what looks like the pilots being unaware that they were heading for a series of storms instead of just the one, and no mention of wing icing from supercooled water. What makes me uncomfortable about this is that from my analysis of many different types of disaster over the years, there is a pattern of either finding individuals who have died at fault, or using some easily-resolvable technical factor as a scapegoat that can easily be fixed. There is a widespread tendency to avoid finding 'difficult to solve' problems responsible.


A prime example of people not wanting to know unpleasant facts or to face unplatable possibilities was the Columbia space shuttle disaster. During the launch a suitcase-sized piece of rigid insulating foam fell from the booster tank and hit the leading edge of a wing. The implications were that if the tough, temperature-resistant leading edge was breached, the intense temperatures encountered when re-entering the eath's atmosphere on returning to earch would almost certainly have meant the wing failing due to excessive heat getting in, which would have meant inevitable destruction of the craft and loss of the whole crew. But when engineers wanted an investigation into potential damage to the leading edge before it returned to earth, NASA officials actively obstructed their requests, refusing their request for either a spacewalk or a Department of Defense satellite to photograph it for them. The main reason given was that there was nothing that could have been done to fix it even if it had been broken. So in other words, the officials at NASA didn't want the possibility of having to tell the crew that they were all doomed, or going through some tough and lengthy recovery procedure, so decided they'd rather not know. The evidence suggests that they falsely used technical data as a screen to justify not finding out, while in reality suspecting that the worse case scenario feared by the engineers was the reality.


There also seems to be significant doubt about the conclusion that the fatal Concorde accident (fatal to everyone on board and ultimately to Concorde's commercial viability) was not necessarily caused by the titanium metal strip as such but by other faults. There were compelling and credible witnesses who said the plane was already on fire before it reached the point on the runway where the metal strip was encountered. Some people were deeply suspicious that the conclusion of the case finding continental and one of its engineers guilty was to avoid finding the French government-owned Air France at fault. Much better to blame foreigners. A similar thing happened in the 1981 Spanish 'toxic oil' poisoning case, where importers of cheap vegetable oil were found to be guilty of passing off rapeseed oil inteded for industrial use as cooking oil. This was despite compelling and heavily-suppressed evidence undermining that evidence as being the cause of the problem and that the symptoms and deaths were caused by over-use of organophosphate pesticides on tomatoes from a region where the Spanish government was trying to support the establishment of an agricultural industry.


The effect of blaming individuals that have died or relatively minor technical problems that are easily fixed is that it reduces the cost of making things apparently safer and reassures the public that that particular error won't happen again. This means that there is less impact on business than if the main causes are found to be problems for which there remains no effective solution. From where I'm sitting, while being relatively untutored about the finer technicalities of aviation, if a pitot tube requires powerful heating to stop it icing up from encountering supercooled water in certain conditions, while exposed to the same airflow as the wing, then I can't see why a wing that is heated only by friction from air travelling at the same speed as that striking the pitot tube is not also going to be seriously susceptible. This may be a contributory explanation as to why the pilots, potentially trying to climb above the weather, seemed to have been unable to recover from a stall all the way down from 38,000 feet.


It may be then that the real ultimate causes might have been:

  1. Ineffective weather radar that does not warn pilots of the depth of heavy weather ahead, and

  2. Aerodynamic instability caused by ice forming from a substantial quantity of supercooled water, sufficient to also freeze up a heated pitot tube.


That's not to suggest a frozen pitot tube is a red herring, because the more accurate information the pilots have about the situation and what the plane's doing, the better, obviously. Plenty of commentators have indicated though that the pitot tube and its direct consequences does not really explain why the pilots were not able to keep the plane flying. I'm just suggesting that even if the pilots had known exactly what speed they were going and that they hadn't been overwhelmed by messages coming from the in-flight avionics, weather severe enough to freeze up a pitot tube may have also sufficiently upset the aerodynamics of the flying surfaces that it rendered them unable to recover from a stall, or even caused it to stall in the first place.


The problem is, there doesn't yet appear to be an effective solution for inadequate weather radar, nor wing icing occurring in flight that doesn't involve incurring massive expense to retrofit current planes at the very least. Even now with the information from the black boxes being analysed as we speak, the preliminary information put out by the investigating team so far seems to be priming the public for, essentially, blaming the pilots for failing to not being able to cope with the failing pitot tube while ignoring problems that are more unplatable to acknowledge. so let's put all the blame the small, cheap pitot tubes that have now all been replaced, and the pilots who struggled to keep the plane under control who are dead, to distract the public's attention from the fact that if the wings had become iced up to, there is currently no easy of avoiding those similar circumstances happening again. Convenient.


UPDATE 9th March 2016

As I had predicted, the blame for the crash of AF447 was essentially split between the tiny and relatively inexpensive-to-replace pitot tubes, that had already been replaced by the time the final report came out, and the pilots. So essentially the public could be persuaded that the problem was solved.

A former professional commercial airline pilot I consulted with discounted the likelihood of wing icing that I suggested might be a cause of the plane's behaviour, given that wings have de-icing systems too - on the leading edges. However, I have just watched an air crash investigation documentary about the American Eagle disaster near Chicago on October 31st 1994. In this accident, that was fatal to everyone on board, the cause was concluded to be icing on the wing surface behind the de-icing boot on the leading edge, as a result of repeatedly flying through conditions likely to contain supercooled water. In fact, ice build-up on the wing was implicated retrospectively in a number of previous incidents, both fatal and non-fatal for that type of aircraft, a propellor-driven Aerospatiale ATF, resulting in an airworthiness directive from the FAA in the US.

Notably, the sequence of events as the disaster unfolded indicated that aerodynamic changes resulting from ice building up on the wing behind the leading edge resulted in, in the first instance, the plane pointing nose up with the plane climbing until it reached a stall situation and then falling almost continuously until it hit the ground, with on one period in the middle of the pilots temporarily regaining some kind of control. Readers are invited to compare this to the apparent action of the Airbus 330, which climbed until, essentially, it then fell into the sea in an unrecoverable stall. And an obvious reason for a stall being unrecoverable is the wings still being covered in ice, which wouldn't melt immediately and would result in, at least for a while, significantly altered dynamics.

Given that in the AF447 crash, it was concluded that the tiny, heated pitot tube had iced up to the point of non-functioning, it would hardly be surprising if the vast and unheated area of wing surface behind the de-icing boots had become mis-shapen due to ice forming from the super-cooled clouds it is now agreed they were flying through. But this possibility is still ignored.