I write from the perspective of a long retired pilot who only flew on old fashioned round dial types (Britannias, VC10s, 707s and 747s). This does not mean that I decry the modern Airbus and Boeing systems – far from it, I think the Airbus philosophy has made a great contribution to air safety.
The views that follow are drawn from my own personal experience, from reading the BEA’s (Bureau d’Enquetes et d’Analyses) reports and from sifting through various aviation websites. This article is an attempt to discuss in layman’s language what may have happened on AF447 over the South Atlantic in 2009, to explain some of the complications involved, and to pose a number of questions. Those of you with long memories will remember the DH Comet accidents in the 1950s and how the lessons learned improved the structural integrity of all subsequent civil aircraft. When the full story of the AF447 is finally revealed, I hope this accident may act as a catalyst for changes to the way regulators, airlines and pilots conduct training on advanced highly automated aircraft.
The aviation industry has a very good safety record. We learn from every accident and incident. Therefore, if procedures and training are improved as an outcome of this tragic accident and safety is further improved as a result, those 228 people will not have died in vain.
At this point it is right to emphasise that speculation in the absence of facts is of little use. However, in the light of the reports below, certain conclusions can be drawn. But even more important they raise questions that it is hoped will soon be answered.
The BEA has issued three interim reports. The first, issued soon after the accident, discusses the pitot tubes and gives the information on how they function and how they are connected to the flight system.
The second, issued 30 Nov 2009, describes how pitot tubes are certified and the previous experiences of icing on pitot tubes manufactured by Thales and Goodrich.
The third (an update issued 27 May 2011) gives some information from the FDR and CVR regarding the last moments of the flight.
The crew were all qualified and experienced on type. The captain had a total of 10,988 flying hours, with 1,474 on type. The senior co-pilot had the necessary licence endorsements to act as a replacement for the captain during his rest period, and had a total of 6,547 flying hours, with 4,479 on type. The more junior co-pilot had a total of 2,936 flying hours, with 809 on type.
Weather Considerations
The aircraft had no significant defects and the weather on route was normal for that time of year, with a significant belt of thunderstorms across the route, typical of the ITCZ (Inter Tropical Convergence Zone) which stretches around the Earth:
The picture above is derived from satellite information and shows the main thunderstorm cells existing at the time.
The crew can be assumed to have used their weather radar to deviate as necessary around the active cells. There were no radio messages stating that they were deviating off track, but, the crew were clearly discussing a deviation, as shown by the comments on the Cockpit Voice Recorder (CVR).
Thunderstorms should be treated with great caution. The turbulence within the active cells can be sufficient to cause structural failure. Large hailstones (which can cause severe damage) almost always exist within these cells and can also be thrown out of the top and sides of the storm such that it is recommended to avoid active cells by at least 20nms. Also supercooled water can be encountered at normal operating altitudes in and around thunderstorms. This is a particularly hazardous phenomenon. Water normally freezes at 0 deg C but, in the absence of any nuclei (minute specks of dust) for ice crystals to form on, water can remain in a liquid state down to around – 40 deg C. The moment this water touches anything it instantly turns to ice. If an aircraft flies into supercooled water the leading edges of the airframe become covered with ice, and it can also affect the engines and the pitot tubes.
Pitot tubes and airspeed measurement
Pitot tubes are used to measure airspeed by comparing the ram air pressure coming into the front of the tube with the static pressure air pressure as measured by a static port on the side of the fuselage. The A330 has three independent pitot/static systems. Both the static ports and the pitot tubes can become blocked for a variety of reasons (ice, dust, insects, paint which has not been removed, etc). One of the most important pre-flight checks is to ensure that there are no obvious blockages and the protective covers have been removed. Also, the pilots do a comparative check between their airspeed indicators during take-off, and will abort the take-off if necessary.
For obvious reasons pitot tubes and static ports are electrically heated.
Here is a simple description of the system showing what happens when various parts become blocked.
The pitot tube captures ram air, which is sensed by an Air Data Module in the A330, and converted to airspeed. The drain tube allows water to pass without clogging the pitot. To do that, it passes air all the time.
The pitot opening is a specific size, say 5mm dia, and the drain is maybe 2mm. That makes the pitot about 20 sq. mm area, and the drain about 3 sq. mm area, for a ratio of about 7:1. This means the ram air pressure at the sensor is about 14% less than actual ram air pressure.
When it comes to clogging at high altitude, there are three possibilities:-
1. Pitot clogged, drain clogged at same time:- Indicated Air Speed (IAS) will not change with change of aircraft speed, but IAS will increase with increase of altitude.
2. Pitot open, drain clogged:- IAS will increase 14%. 14% excess IAS will remain regardless of actual airspeed or altitude.
3. Pitot clogged, drain open:- IAS will decrease toward zero as the drain bleeds off the ram air pressure to ambient. Increased altitude will not affect IAS.
Shortly before control was lost a number of automatic reports were sent from the aircraft via satellite using the Aircraft Communications Addressing and Reporting System (ACARS) indicating that there were abnormal speed indications and that various systems had been affected such as unavailability of the autopilot, of the auto-thrust and also of some of the ‘fly by wire’ protections. The purpose of automatic reporting of such failures is purely for maintenance purposes so that engineering staff can have the necessary spares available and so make repairs as quickly as possible. They are not designed for accident investigation purposes, nonetheless they can be useful.
Thus, in the first and second interim reports a number of deductions could be made from these messages and from the pieces of wreckage recovered from the surface of the sea. Namely; (1) unreliable speed indications and consequent system failures via ACARS; and (2) the fact that the aircraft had hit the water in a near horizontal attitude with negligible forward speed but with a high vertical velocity as deduced from the structural damage and the injuries to the bodies.
Therefore, suspicion immediately surrounded the performance of the pitot probes (there had been previous problems and Thales, the manufacturer, had advised replacement with an improved model). Also, it was fairly certain that the aircraft had hit the water in a stalled condition and was probably undamaged before impact.
Aircraft should not crash following failure of the airspeed indications. Since problems with pitot/static systems are not uncommon, information is provided on the checklists giving a set of attitudes and power settings to achieve specific speeds. From basic physics, for a given mass in level flight, a specific attitude at a specific power setting will give a specific speed since the thrust and the drag will be in balance. Doing this at night in turbulence is not easy but pilots are trained to do this.
It was essential, therefore, in order to determine why a mainly serviceable aircraft had crashed following unreliable speed indications, to recover the Flight Data Recorder (FDR) and the Cockpit Voice Recorder (CVR). This has now been done and a highly edited set of information has been published in the BEA’s third preliminary report.
My first set of questions
Did the crew have all the necessary weather information before departure? Answer; almost certainly yes. There was nothing unusual and other aircraft on the same route made successful deviations.
Did the crew use the weather radar to best effect? Answer; we don’t know. They discussed the need to deviate (see 3rd report). But it is fair to say that knowledge of the best use of weather radar amongst airline pilots is not always as good as it should be, and neither is the training.
Were the pitot tubes functioning correctly? Answer; almost certainly not. In fact, due to problems with these tubes back in 2001, Thales had issued a recommendation in 2007 stating that the pitot probes should be replaced with a later model, and Air France had started a replacement programme in April 2009.
Should Air France have started to replace the pitot tubes earlier? Answer; probably yes, but there was no legal requirement to do so at that time since an AD had not been issued. No doubt this will be a major issue of contention.
Stall training
All pilots are taught very early in their initial training about the hazards of stalling, how to recognise the symptoms, and how to recover control again. But it is not always dealt with adequately later on in their conversion training onto new types. There has long been a debate on the necessity and depth of this training. On the one hand the view is that with all the modern warning systems pilots should be able to recognise when they are getting near the stall and therefore training is only required for an incipient stall. On the other hand it is known that airliners get very near the stall and actually do stall with sufficient frequency (when I was flying about 1 in 100,000 flights) that training for the recovery from a full stall should be part of every training syllabus. I strongly believe the latter view is correct.
Stalling occurs when the wing reaches an Angle of Attack (AoA), the angle at which the air hits the leading edge of the wing, when the airflow over the top of the wing breaks down completely and becomes turbulent. At that point, the break point, lift is lost until the AoA is reduced to a point where normal airflow over the wing is re-established. During certification test flying a lot of very brave flying is done to ensure that the aircraft exhibits sufficient warning (increasing airframe buffet) before the break point, and that at the break point the aircraft pitches nose down so that it re-establishes a normal AoA. If this does not occur naturally, artificial devices are fitted (stick shakers) that warn the pilot of an approaching stall, and stick pushers which induce a nose down pitch at the appropriate moment.
Airbus, with their ‘fly-by-wire system’, went one stage further by using triplicated computer control systems to prevent a full stall by flying the aircraft at the best AoA (apha max) even with a full nose-up control input. This works very well in what is called Normal Law, but in Alternate Law some of these protections are degraded until, when in Direct Law, the computers have no protections, with the control surfaces following the pilots’ commands from the side stick – in other words much the same as an old fashioned aircraft. I don’t know enough about the Airbus systems to know exactly which protections were available in the AF447 accident. We do know that the autopilot and auto-thrust dropped out, and that the aircraft reverted to Alternate Law due to the unreliable airspeed indications. This means that the pilots had to fly using the power and attitude tables to maintain a safe airspeed, that basic stall warnings were still available, but that probably the full stall protection was unavailable.
My second set of questions
Exactly what stall protections were available to the pilots? Answer; stall warning was still available as this is sensed by the AoA vanes, independent from the pitot/static system. In Normal Law AoA protection is available and full nose-up input will cause the aircraft to pitch nose up to alpha max, just short of the stall. However, in Alternate Law, this protection is lost. In the BEA’s third report, at 2h 10m 16s the Pilot Not Flying (PNF) is recorded as saying’ “…so, we’ve lost the speeds” then … “alternate law…” Therefore it is almost certain that the aircraft was in Alternate Law. We will know more when the final report becomes available
Should the pilots have been able to detect the stall and recover? This is the nub of the whole problem. No one is sure yet exactly what information was being provided by the flight instruments, and we won’t know this until all the CVR and FDR information has been analysed. And even then there will be some essential data missing, e.g. the FDR does not record the airspeed indications given on the co-pilot’s side and, as he was the Pilot Flying (PF), we need to know this (see later). There is another recorder, the Quick Access Recorder (QAR) which is used for maintenance and quality control purposes and which records thousands of parameter but it is not crash proof. (However, it has probably been recovered amongst the contents of the electronics bay and the information may be readable – I certainly hope so.)
Is airline training appropriate and adequate? Answer; I am not sure that all airlines give sufficient in-depth stall training. There is some debate between the test pilot community, who want training to cover a variety of different stall conditions, versus the airline training community who want a simplified standard stall recovery procedure. I tend towards the test pilot view but also have some sympathy with the latter. Airbus have already amended their stall recovery advice. My own personal view is that are four basic situations (as a minimum) that need to be taught:- (1) The approach to the stall in a trimmed condition so that all the indications (attitude, buffet, warnings, etc.) can be observed and recovery made with minimum height loss. (2) An inadvertent stall caused by unnoticed loss of airspeed with autopilot connected so that the aircraft is in a large nose up trim condition; pilots need to experience the problems associated with the out of trim condition and the nose up pitching moment when thrust is applied with ‘underslung’ engines; ideally this recovery should be done in two ways (a) with minimum height loss when near the ground, and (b) at altitude with a positive pitch down to ensure a reduced AoA. (3) A windsheer recovery where the aircraft is flown just short of the stall with minimum height loss. (4) A stall at high altitude when already operating very close to the buffet boundary. Obviously details will vary depending on the aircraft type and the degree of automatic protections – but even on those with full fly-by-wire protections pilots must experience the full stall. BUT – see next question:-
Can simulators be used for stall training? Answer, to a reasonable extent, yes. But only within a fairly ‘normal’ set of flight conditions. Simulators can only be programmed with aerodynamic data obtained from test flights, beyond that the data is extrapolated using wind tunnel information and aerodynamic equations. To get extended flight data would require highly dangerous flight testing. Therefore flight instructors need to be very aware of the limitations of simulators when exploring flight characteristics and recovery from extreme situations.
Jet Upsets
Jet upsets and consequent loss of control can occur for a variety of reasons, such as spatial disorientation of the pilot; loss of essential flight instruments when in cloud; encountering extreme turbulence; flying at too high a Mach number; flying too slowly and stalling; or varying degrees of structural failure and/or loss of flight controls. The latter are particularly difficult to deal with, nevertheless there have been some spectacular ‘saves’ following complete loss of flight controls.
In the case of AF447 the aircraft definitely stalled; there was a loss of some speed indications (there may also have been others – we don’t know yet); and there may have been contributory elements of turbulence and spatial disorientation – witness the fact that the pilot flying pulled back on the side stick.
Stalling occurs when the Angle of Attack (AoA) exceeds the critical point, usually around 15 degs nose up. Once the AoA goes beyond this point the air flow separates, causing buffeting and loss of lift.
The only way to unstall the wing is to pitch the nose down so as to reduce the AoA to less than 15 degs. However, the critical AoA at higher Mach numbers can be considerably less than this.
But there are several problems that pilots do not always adequately understand. The first is the deep stall which can happen on T-tailed aircraft – a DH Trident and a BAC 1-11 were both lost during the manufacturer’s test flights, and a VC10 came close. The deep stall occurs when the turbulent air flow from the wings blanks the elevators such that they lose all effectiveness.
The Airbus A330 has a conventional tail configuration but nevertheless seems to have descended in something that looked like a deep stall. Whether or not it can become ‘locked’ into this condition I do not know, and, for obvious reasons this extreme case is not explored in either the manufacturers’ test flying programme or during certification testing. All natural flying and artificial aids (stick shakers, stick pushers, other warnings and the Airbus fly-by-wire protections) are designed to prevent the aircraft from ever getting into such a situation.
However, if the aircraft is in a deep stall, it can be difficult for the pilot to identify this because he has only attitude information on his flight instruments and not AoA information. The last figure in the diagram below shows that even with the nose pitched down the AoA can still exceed the critical figure of around 15 degs.
Nevertheless the deep stall should be fairly obvious from other symptoms, namely a very high rate of descent with very low forward speed, and very sluggish response (if any) to the controls. Although this was talked about endlessly when I was flying VC10s, and the aircraft was very well protected by the stick shaker and stick pusher it is not something that features very highly in training on those type which have conventional configurations. And on Airbus aircraft, with all the computer protections, I would imagine it does not feature at all.
However, AF447 hit the water in a stable nose up attitude and with a very high rate of descent, probably as shown here:
Some modern wings, designed to operate efficiently at high Mach numbers, can exhibit a nose up pitch tendency after the aircraft has penetrated well into the stall. Again, I don’t know enough about the aerodynamics of the Airbus wing to comment.
Next, although AoA is used for stall warning and identification, I don’t know of any current civil airliner which shows it on the flight instruments. AoA is sensed by vanes on the side of the fuselage and is completely separate from the pitot/static systems:
There will no doubt be a debate as to whether in future AoA should be displayed. Instead the pilot has to use his attitude display which looks like this:
In the picture above the aircraft is rolled 5 degs to the right, with a nose up pitch of 6 degs.
In the picture below the aicraft is rolled 30 degs to the right in a level attitude. AoA can be deduced from the difference between the nose level attitude and the flight path vector symbol. It shows that the AoA is around 5 degs. But this not a compelling display and may be difficult to interpret in the heat of the moment when all hell is breaking loose!
The next picture shows the whole Primary Flying Display (PFD), with the aircraft on the point of stall, at a speed of 202kts (left hand scale,) at the bottom end of alpha protection as shown by black/amber striped tape Alpha Protection band. It shows a nose up attitude of around 7 degs with wings level. The flight path vector symbol is on the horizon showing that despite the nose up attitude the aircraft is in level flight at an altitude of 35,000ft (right hand scale). This is corroborated by the Vertical Speed Indicator (VSI) on the extreme right showing zero rate of descent.
FDR and CVR information
The BEA’s third (update) report gives only very basic information obtained from the recorders, presumably because they are only willing at this stage to release totally verifiable information, but it is worth reading the full report. It states that, ‘after the autopilot and auto-thrust disengaged… (at 2h 10m 05s)…(presumably because of the speed anomalies)… ‘the pilot flying (PF) said “I have the controls” and made a left nose-up input. The stall warning sounded twice in a row. The recorded parameters show a sharp fall from 275kts to 60kts in the speed display on the left PFD, then a few moments later in the speed display on the Integrated Standby Instrument Systems (ISIS).’
‘The airplane’s pitch attitude increased progressively beyond 10 degs and the plane started to climb. The PF made a nose-down control input and alternately left and right roll inputs. The vertical speed, which had reached 7,000 ft/min, dropped to 700ft/min and the roll varied between 12 degs right and 10degs left. The speed displayed on the left side increased sharply to 215kts (mach 0.68). The airplane was then at an altitude of about 37,500ft and the recorded AoA was around 4 degs.’
Before the speed anomalies the aircraft was cruising at 35,000ft at Mach 0.80 (the turbulence penetration speed) with a pitch attitude of 2.5 degs. Unfortunately the FDR does not record the speed indications shown on the right side where the PF was sitting, therefore we don’t know what he saw in the way of speed information, and neither can we be certain what attitude information he was seeing (but there is no reason to believe the latter was abnormal). Since the QAR has probably been recovered from the sea bed, together with much of the equipment in the electronics bay, we can hope that the data recorded may be readable. If so, a vast amount of additional data will become available.
At the time of the anomalies the pilot flying (PF), the more junior of the 2 co-pilots, was in the right hand seat. The pilot not flying (PNF), the more senior and more qualified of the two, was acting as relief pilot in the left hand seat while the captain was taking his rest. There is nothing abnormal about this arrangement, and it was in accordance with AF procedures.
My third set of questions
When the autopilot and auto-thrust disconnected why did the PF make a nose-up control input? Answer; we don’t know, and we don’t know what speed indications he was responding to on his side.
Was the nose-up control input made accidentally because he was concentrating too much on the lateral control? Answer; some think this a possibility as lateral control when in Alternate Law is very sensitive. But the fact that the input was sustained for so long makes it unlikely.
How long after his nose-up input, and after the first stall warning, did he reverse this to a nose-down input? Answer; we don’t know as the report does not state the exact time interval but it was probably around 10 to 15 secs.
More FDR and CVR information
The report then goes on to say, ‘At 2h 10m 51s, the stall warning was triggered again. The thrust levers were positioned in the TO/GA detent and the PF maintained nose up inputs. The recorded AoA, of around 6 degs at the triggering of the stall warning, continued to increase. The trimmable horizontal stabiliser (THS) passed from 3 to 13 degs nose-up in about 1 minute and remained in the latter position until the end of the flight.’
‘Around 15 secs later, the speed displayed on the ISIS increased sharply towards 185kts; it was consistent with the other recorded speed. The PF continued to make nose up inputs. The airplane reached its maximum altitude of about 38,000ft, its pitch attitude and AoA being 16 degs.’
‘At around 2h 11m 40s, the captain re-entered the cockpit. During the following seconds, all recorded speeds became invalid and the stall warning stopped.’
‘The altitude was then about 35,000ft, tha AoA exceeded 40 degs and the vertical speed was about -10,000 ft/min. The airplane’s pitch attitude did not exceed 15 degs…’
My fourth set of questions
When the stall warning was triggered for the second time, about 35 secs after the PF made the nose-down input he applied max power (TO/GA) and made a nose-up input again, why did he do this? Answer; again we don’t know, it is against all normal stall recovery techniques. However, it is consistent with training for windsheer events and stall recovery when very near the ground. In both these cases, when full stall protection is available in Normal Law, application of TO/GA (max) power with a maximum nose-up input will allow the fly-by-wire system to fly the maximum AoA with minimum loss of altitude. BUT STALL PROTECTION IS NOT AVAILABLE IN ALTERNATE LAW and therefore this method was inappropriate in this case.
Therefore, has the industry as a whole been making this sufficiently clear in stall training, or did the PF make a mistake? Answer; we don’t know but I am sure everyone is examining very carefully every aspect of stall training. The distinction between stall recovery in Normal Law when near the ground and stall recovery at altitude (and especially in Alternate Law) needs to be made very clear. It is possible that the PF did not distinguish between the two cases.
So, is this pilot error, or inappropriate training, or something else? Serious questions have already been raised and there has been much debate on many aspects of airline training for recovery from upsets and stalls. This is an industry wide issue, and there are no easy answers.
The effect of trim and power application on elevator control
The Airbus system applies pitch trim automatically when a nose-up input is maintained, in this case, when full nose-up elevator was applied the Trimmable Horizontal Stabiliser (THS) also ran nose-up to assist the elevator input. In old fashioned aircraft the pilot would do this for himself. Stabiliser trim is a large very powerful control which should be used with caution. See the relative size versus the elevator in this picture:
Also, with engines fitted on pylons below the wing, application of full power gives a very strong nose-up pitch moment. The combined effect of 13 degs nose-up trim and full power may have meant that there was much reduced elevator authority to pitch the nose down even with full nose down elevator applied.
My fifth set of questions
Were the pilots aware of the trim of the aircraft? Answer; they may not have noticed the trim wheel turning when the THS was trimming automatically to a nose-up position. The trim wheel (A) is well marked and can be adjusted manually but hardly ever used in this way since it is almost always adjusted automatically. See picture below:
Note that the vernier scale (B) is rather small. Airlines do not like pilots adjusting the trim during recovery from upsets, whereas test pilots always put the aircraft in trim first to ensure that there is sufficient elevator authority. See this paper on the differing views between the test pilot and airline training communities on Airplane Upset Recovery Training.
Final discussion
The BEA’s third update report is very short on detail. This is understandable since they will not want to publish anything that has not been established beyond all reasonable doubt. Therefore one can only make limited deductions.
Question:- Perhaps the first question to be asked is should the pitot tubes been changed earlier in accordance with Thales’ recommendation? This is a matter of judgement. As I understand it before the AF447 accident it was only a recommendation, after the accident it became an AD which requires compliance. With hind sight it is clear that the pitot tubes should have been replaced sooner. But pitot failure should not cause an accident.
Next, I have to make it clear that all my experience is on old fashioned aircraft only, I have no fly-by-wire experience other than an informal morning spent flying the A320 simulator in Toulouse with an Airbus training pilot. My conclusion was that the protections designed by Airbus are entirely appropriate for modern public transport aircraft. There are a number of areas of difference compared to non fly-by-wire aircraft, all of which are easy to use, and one or two I did not like but these were minor issues.
Question:- Given adequate training the aircraft is straight forward to operate, and the accident record shows that the protections provided by fly-by-wire have improved safety. Airbus has made a feature of this in their sales and training promotion but regulators have not yet changed their training requirements to cover the higher levels of automation, and I wonder whether airline training departments have used these features to save money on some aspects of training? It appears from a number of recent accidents the pilots did not fully appreciate what the automatics were doing. BUT this was not a feature of the AF447 accident. Nevertheless there needs to be a thorough examination of the amount of training given and line pilots’ understanding of the automatics on modern aircraft. I understand the FAA is completing just such a study at the moment.
In this accident the aircraft was fully serviceable except for the unreliable airspeed indications. The autopilot and auto-thrust disconnected and the control computers reverted to Alternate Law in which there is no alpha max protection. In other words, stall warning and stall identification will still be given via the independent AoA vanes – but the aircraft can still be stalled.
Question:- Have airlines in the past given sufficient attention to the various types of stall and made the necessary distinctions between some aspects of the recovery procedures? Most training stalls are demonstrated in a correctly trimmed state with the pilots fully prepared for what is about to happen. But most inadvertent stalls occur unexpectedly with the aircraft often very much out of trim, as was the case in this accident. Have airline training departments trained pilots to cope with the reduced elevator authority in the out of trim condition and when exacerbated by the nose-up pitch moment caused by applying full power on underslung engines? Reading the note written by an Airbus test pilot I think probably not. In addition, given the so called non-stallable properties advertised by Airbus, have these same training departments been of the view that stall training can be reduced?
The only way to unstall an aircraft is immediately to reduce the AoA. Pilots tend to think in terms of minimum safe airspeeds, but although this serves the purpose very well, in reality it is AoA that matters.
Question:- Should manufacturers add AoA indication to the flight instruments? Those who have used AoA instruments have liked them. There needs to be a sufficiently detailed debate about the desirability, technical viability, reliability and certification of such indications. This will be a complex subject affecting many aspects of the industry’s approach to training. In addition should automatic stall recovery systems be developed? Technically it could be done, but it would be fraught with difficulty bearing in mind the different types of stall and the difference between recovery at low altitude (on the approach or just after take-off) versus a stall at high altitude.
Problems with pitot/static systems are rare but do occur with sufficient frequency that emergency checklists contain tables giving specific attitudes to fly, and specific power settings to apply in order to obtain the required airspeeds. In the calm of the training environment this is a reasonably easy exercise. But it is a different matter on a dark and dirty night, in cloud with no visible horizon, in turbulence, and when it occurs suddenly when one is at a low point in one’s circadian rhythms.
Question:- Airbus have designed the display below which is based on AoA information (not airspeed information) and which could be used following loss of reliable airspeed indications. It is a standard fit on the A380, should it be a required retrofit on the other Airbus types?
I have discussed some aspects of training regarding pilots’ knowledge of the automatics and recovery from the stall. But there is also the question of atrophying basic flying skills. Many (most?) airlines discourage their pilots from hand flying during normal line operations on the basis that fuel will be saved, passengers will have a more comfortable ride and, more importantly, the pilots will have more mental capacity left for coping with the very busy traffic situations in the congested airspace of Europe and the USA. The latter is a valid point. But it does lead to an over-reliance on the automatics to the detriment of instrument flying skills. This means that when systems fail suddenly and the pilot is faced with hand flying in difficult conditions he may be unprepared.
Question:- Should pilots be encouraged to hand fly their aircraft more often? I believe most certainly they should. Obviously one needs to use common sense and to choose one’s times carefully. But one should also take advantage of and to use the automatics as much as possible in very busy traffic situations. In the case of AF447, basic flying skills should have been sufficient when suddenly faced with unreliable airspeed indications, and, if no other parameters are changing (pitch attitude, bank angle and power), one should leave well alone while assessing the situation. To make any sudden change of attitude, power or trim is to invite another problem, the reason being that if the aircraft was in a stable state before the anomoly it is likely to continue in stable flight for a while thereafter. I know this is a counsel of perfection, but one should not leap into action and start making control inputs too soon. In the case of AF447, the pilot flying (PF) made a control input almost immediately after the autopilot and auto-thrust disconnected following the unreliable speed indications. The whole question of why the PF did this is a big unknown as we do not yet know what was being displayed on his flight instruments (the FDR only recorded the captain’s instruments).
Finally, we must bear in mind that the whole accident took place in bad weather conditions and within a very short time scale. The speed anomalies and the disconnect of the autopilot and auto-thrust occurred at 2h 10m 05s into the flight. The PF made a nose-up pitch input and the aircraft climbed showing that, despite the stall warning, it was not stalled at that time. It climbed to 37,500ft with the speed dropping to 215kts, during which the PF made a nose-down pitch input. At 2h 10m 51s (54secs after the disconnect) the stall warning triggered again and the PF applied max power and made another nose-up pitch input. From this point on the aircraft was stalled and remained stalled. 15 secs later (which would be at 2h 11m 06s) the speeds on the FDR showed 185kts (if they are correct the aircraft would still have been well and truly stalled) and the PF inexplicably maintained his nose-up inputs. At 2h 11m 40s the captain re-entered the flight deck, all recorded speeds became invalid and the stall warning stopped (because the speeds were below 60kts and the measured AoA is ignored by the system). At this point the aircraft was at 35,000ft with the AoA exceeding 40 degs nose up and the vertical speed 10,000ft/min down. It may still have been possible to recover from the stall at this point because at 2h 12m 17s the PF made a nose-down pitch input and the AoA decreased, the speeds became valid again but the stall warning triggered again as the speed parameters had returned to a valid condition, this would have been most confusing. However, the AoA, when it was valid, always remained above 35 degs. The aircraft hit the water with a vertical speed of 10,912ft/min down, a forward ground speed of 107kts and a nose up attitude of 16.2 degs, at 2h 14 m 28. This was 4mins 03 secs after the first speed anomalies, 3mins 17 secs after the first stall warning, and about 3 mins 02 secs after the aircraft was well and truly stalled, probably already in some kind of deep stall.
I am sure these are just some of the many questions that are exercising the various experts in the BEA, Airbus, Air France and throughout the whole aviation industry.