About the complex events, which lead to the crash of AF 447 into the Atlantic Ocean in just over four minutes, you can read here:
1. Flight Air France 447 - Beginning, Disappearance, Reconstruction
1.1. Crew, Passengers, the Airliner, and the Intended Flight
The Crews in the Cockpit and in the Cabin
The most experienced pilot on board, Marc Dubois, 58, 10988 flight hours, acted as the captain of this flight. David Robert, 37, a seasoned pilot with 6547 flying hours, served as a relief first officer in order to allow for rest time. He had crossed the equator several times already. Hence, he was familiar with the cirumstances in the tropics.
The least experienced pilot on this flight was Pierre-Cédric Bonin, 32, with 2936 flying hours. He was used as a first officer in the right seat in the cockpit as the pilot flying (PF). In the cabin, nine flight attendants commenced their work.
The Passengers
There were 216 passengers from 33 nations on board. Most of them came from France, Brazil, and Germany. Along with 82 women and 126 men, eight children travelled on this plane.
An Airbus A330-203
The Airbus A330-200 is a modern and very safe aircraft. It has a "glass cockpit" in which all indications are displayed digitally. The pilots direct their plane using a system known as fly-by-wire, through which the input signals are no longer transmitted mechanically but electronically and are then carried out on the control surfaces by computers. Such an Airbus usually carries between 210 and 250 passengers.
The airliner used on flight AF 447 was registered F-GZCP. It was the newest A330 of the airline, delivered in 2005 and overhauled on 16 April 2009. At the time of the accident, it had accumulated about 18870 flight hours.
The Intended Flight from Brazil to France
In the evening of 31 May 2009, the aeroplane took off from Rio de Janeiro-Galeão International Airport at 19:29 local time. It was scheduled to arrive at the International Airport Paris-Charles de Gaulle on 1 June 2009 at 11:03 local time.
1.2. Over the Atlantic, AF 447 Disappeared From the Radar Screens
Flight Along the Brazilian Coast and Missing Flight Plan
After take-off, the pilots followed the Brazilian coast keeping radio contact with the respective sections of Brazilian Air Traffic Control (ATC) on their way. The last of these sections was ACC ATLANTICO (ACC stands for area control centre).
For their flight over the Atlantic Ocean, the crew had already received the frequency of the Senegalese ACC DAKAR Oceanic. Several attempts by ACC DAKAR Oceanic to make automated contact with the aeroplane (without radio contact with the pilots) were not successful, because the flight plan was missing in the ATC´s "Eurocat" system.
The air traffic controller in Dakar contacted ATLANTICO and created the flight plan for AF 447 later based on the planned route. The plan remained virtual, however, because there was no contact between AF 447 and ATC in Dakar, neither on the radio, nor via an automated system. It should be noted that in the section over the Atlantic Ocean, for which ACC Dakar is responsible, radio contact is known to be patchy.
ACC DAKAR Still Without Contact With AF 447
The Senegalese air traffic controller also talked to ACC SAL on the Cape Verde Islands at 02:47 UTC. He reported that he had not had contact with AF 447 and transmitted the flight plan he had created.
The air traffic controller on the Cape Verde Islands told his colleague in Senegal at 04:08 that he had the later flight AF 459 on his radar screen already in contrast to flight AF 447. ATC in Dakar had apparently not been alarmed by the fact it had not had contact with AF 447 over the entire period of time.
The Disappearance of AF 447 Was Finally Noticed
A series of consultations between ATLANTICO, SAL, DAKAR, but also CANARIAS ensued and can be read in detail in the Final Report by the BEA starting from page 52 (BEA: Bureau d’Enquêtes et d’Analyses pour la sécurité de l’aviation civile, Bureau of Enquiry and Analysis for Civil Aviation Safety).
Several attempts from various parties, amongst them the crew of flight AF 459 and Air France itsself, to make contact with AF 447 were futile. Taken together, it remains to be noted that ATC in Senegal was alarmed very late. It took several hours until the first aircraft, triggered by emergency messages coming from the control centres in Madrid and Brest, took off to a search and rescue operation.
1.3. The Discovery of the First Pieces of Wreckage, the Ongoing Search, and the Reconstruction of the Accident
The search was started at the last known position of the aircraft. The first pieces of wreckage were discovered in the vicinity of the Saint Peter and Saint Paul Archipelago on 2 June 2009. The first vicitms were found and recovered on 6 June 2009.
Further work continued for many months, however. Considerable resources concerning personnel, financing, and equipment were employed involving several nations, especially for the underwater exploration.
Finally, employees with the Woods Hole Oceanographic Institution made a breakthrough. The institution had also played a major role in the location of the wreckage of the Titanic. On 3 April 2011, they managed to spot a large part of the fuselage, as well as further victims on the sea bed using autonomous underwater vehicles.
After the cockpit voice recorder (CVR) and the flight data recorder (FDR) had been recovered on May 1st and 2nd 2011, respectively, the reconstruction of the course of events began. We will explore it in the following subchapters.
The original recordings of the CVR have never been published, but a transscript has been read and has later been cited in several publications. Amongst them is the truly informative film "Fatal Flight 447: Chaos in the Cockpit" of the British Channel 4 (September 2012), which has also been uploaded onto YT.
2. Weather Phenomena, System Outages, Startle, and Confusion
2.1. Thunder Storms in the Intertropical Convergence Zone and Other Phenomena
The Intertropical Convergence Zone, ITCZ
The ITCZ is an area of low pressure, which circles the Earth like a belt along its equator. Seasonal shifts are due to the position of the sun. The ITCZ moves to the North during the summer months on the Northern Hemisphere and to the South in the winter months.
In the ITCZ, in which the trade winds converge, the formation of large clouds, and severe thunder storms, torrential rain and hail prevail. The rain can freeze in the usual cruise altitudes. Pilots have to be prepared for these particularly dynamic weather conditions and to be alert during their flight in order to cross the ITCZ safely.
How the Team Shared Their Responsibilities in That Stormy Night
In the night of the flight, several crews from other airlines decided to fly around the thunder storms. Right before AF 447 headed towards the first of them, Dubois, the captain on this flight, left the left seat in the cockpit for his break. Robert, who returned from his break, took this seat. Bonin, who was sitting in the right seat, remained pilot flying. The more experienced Robert monitored his colleague´s actions and the flight path, amongst others.
At first, Bonin stayed on course and steered 12 degrees to the left only upon Robert´s suggestion in order to avoid one of the thunder storms ahead of them. Due to turbulence, the seatbelt signs had been switched on.
St Elmo´s Fire and the Smell of Ozone
St Elmo´s fire is a phenomenon of light caused by electrical charges. In addition, it occurs when frozen rain drops hit the windscreen of the cockpit. Also on flight AF 447 St Elmo´s fire appeared in the form of a blue light due to icy rain.
A little later, Bonin noticed a strong smell akin to chlorine, which he did not recognise at first. Robert explained to him that this was the smell of ozone, a normal occurence in the tropics.
2.2. Weather Related Effects on Decisive Measuring Systems
A system, consisting of two parts, is mounted to the fuselage to measure the airspeed. One of these two parts are the so-called pitot tubes which are located at the aircraft´s nose. They face forward and are open at the front. If they become blocked, the airspeed indication drops off.
On flight AF 447, the pitot tubes were covered with ice and did not deliver valid data any more. Subsequently, the autopilot and the auto-throttle disengaged at 02:10. Suddenly, the pilots had to take manual control of their flight.
2.3. Surprise, Startle, and Confusion
In the present case, the surprise following the freezing of the pitot tubes was triggered by the following factors:
- mainly due to the disengagement of the autopilot and the auto-throttle
- the indicated airspeed dropped
- an alledged drop in altitude of 330 feet was indicated
- the airliner banked considerably to the right, because the autopilot did not counterbalance the turbulences any more
As a first reaction to the surprise, Bonin took over the manual control of the aeroplane stating the formal "I have the controls" to which Robert responded "d´accord" (ok). In order to climb and correct the banking, Bonin pulled the sidestick back and to the left abruptly (the sidesticks are located next to the pilots´ seats and used on the airbus instead of a classic control column). The fierceness of his movement could be interpreted as a sign of a startle.
What happened during the next four minutes, was a mixture of inadequate flying manoeuvres, indications of flight parameters, which only confused the pilots even more, and warnings from different on board systems, of which they misinterpreted one in particular. This mixture lead to complete confusion and stress. The pilots were not capable of analysing their situation and reacting accordingly any more.
3. Subsequent Manoeuvres of the Pilots and Their Effects
3.1. Pulling Back the Sidestick, the Stall of the Aircraft, and an Unnoticed Descend
The Mechanism of a Stall
When a pilot pulls the sidestick back, the nose of the aeroplane is raised and the angle between the wings and the horizontal becomes larger. This so-called angle of attack (AOA) may only reach a certain degree, because otherwise a stall of the aeroplane happens.
The stall occurs, because the air flows in a much smaller quantity over the wing when the AOA is increased too much resulting in a reduced lift. If the lift is insufficient, the plane will not fly any more but start to lose altitude.
First a Steep Climb, then a Rapid Descend
On flight AF 447, the initial pull on the sidestick first resulted in a steep climb to an altitude of almost 38 000 feet because of the rise of the aeroplane´s nose. As the air is thinner at this altitude, the mechanism of the stall was even intensified, and the airliner started to descend rapidly.
Robert advised his colleague repeatedly to stabilise the plane and to reduce the altitude. Bonin, however, pulled the sidestick back most of the time. In addition, the plane continued to roll left and right. Bonin tried to correct these movements with corresponding inputs on the sidestick.
Warnings Sounded
Vocalising the term "stall", the automatic warning system alerted the crew several times to the state of their aircraft. It is not clear, why Bonin and Robert did not interpret the warning correctly. It seems that they did not deem the stall real. The following two facts may have contributed to this.
At 02:10 and 35 seconds, the pitot tubes were deiced again, and an airspeed of 220 knots was indicated to the pilots, in line with the climb. The two of them, however, seemed not to be able to assess whether this value was trustworthy.
The two pilots had selected maximum power. At an altitude of almost 38 000 feet, the engines, however, could not deliver the thrust technically any more, and the descend began.
3.2. Not Even All Three Pilots Together Could Solve the Problems
Strengthening the Team
In this intransparent situation, Robert used an important resource of the crew and called the captain back into the cockpit using the crew call button at 02:10 and 46 seconds. The latter returned less than one minute later and sat down on the jump seat behind the two first officers.
At this point in time, Bonin and Robert were not capable of reporting in a structured manner what was going on. Instead, they admitted in colloquial language that they had tried everything and understood nothing.
Further Flight Manoeuvres
At 02:11, Robert tried to take control over the flight, but without announcing this formally. Hence, following Bonin´s next action on his sidestick, these inputs were transmitted to the control surfaces again.
The rapid descend brought about a tremendous aerodynamic noise akin to the sound of strong wind gusts. Perhaps also due to that, Bonin assumed that they were flying far too fast. He wanted to employ the speedbrakes, but Robert defied this immediately. It would have been a wrong manoeuvre during the descend anyway.
The Angle of Attack and the Stall Warning
At 02:12, Bonin did the right thing to stabilise the airliner. In a moment without the frequent acoustic warning "stall, stall", he selected maximum power again and pushed the sidestick forward.
However, as soon as the airliner was in a slightly reduced AOA, paradoxically, the warning "stall, stall" sounded again. It seemed to be triggered exclusively by an AOA that was not too extreme. Understandably feeling insecure, the pilot did not have the courage to pursue this manoeuvre.
The Descend Continued
Following that, all three crew members concentrated on keeping the wings level. At the same time, they did not agree as to whether they were climbing or descending. They did not trust their instruments, and they still had no visual orientation at night.
At times, both pilots used their sidesticks which triggered the warning "dual input". Whilst in reality the descend continued fast, the nose of the plane moved up and down. Most of the time, however, the AOA exceeded the critical value.
Finally, it Was Noticed that the Sidestick Had Mostly Been Pulled Back
At 02:13, the crew realised that they were losing height fast. Furthermore, Robert noticed after a new "dual input" warning that he did not have the control over the plane with his sidestick.
Upon Robert´s request at 02:13 and 39 seconds to pull up ("climb, climb, climb, climb"), Bonin answered that he had pulled back the sidestick for a while already. Robert finally asserted himself with his demand to take over the control and obtained it also formally.
The ocean was now approaching fast, also the ground proximity warning system detected its surface, and the alarm "pull up" sounded. At this point in time, it was too late to recover the airliner from the stall by lowering its nose in order to reduce altitude and by maximising the thrust. Robert enhanced the power maximally again to pull up. However, at 02:14 and 26 seconds, the recordings on the CVR stopped.
4. AF 447 in the Light of CRM
In retrospect it is simple, but still very informative, to look in-depth at the facts in relation to CRM aspects. Apparently, the crew was not capable of falling back on its CRM training in the acute situation.
4.1. The Starte Effect Is an Important CRM Theme
An unexpected, potentially dangerous event triggers the age-old, biologically programmed startle response, which first leads to freezing and, following that, prepares for flight or fight. If an emotional burden, a stressful workday, or fatigue preceds it, the startle response is enhanced because of that.
It is a subfield of CRM, to prepare specifically for the possibility of such events in order to develop the resilience to be adequately able to think and act after the initial startle.
4.2. Preparation, Situational Awareness, Decsion-Making, Teamwork
Preparation
We do not know to what extend the pilots had discussed the crossing of the ITCZ during their pre-flight planning. It is not recognisable in the recordings that they in the acute situation fell back on a prepared plan, which could have contained measures in case of likely events, such as icing rain or turbulence.
Situational Awareness
The situational awareness of the cockpit crew was considerably impaired after the autopilot had disengaged. Spacially, they did not know where they were flying at what speed. Moreover, they did not realise the stall resulting from the high angle of attack. The lack of visual points of reference aggravated their spacial situational awareness even more.
In addition, they lacked the situational awareness to notice their actions (the pulled back sidestick), and to correctly process pieces of information and the warnings.
Analysis of the Facts and Decision-Making
A calm and structured analysis of the facts as part of the decision-making process did not take place. The pilots could not find out what flight parameters and warnings were real and how they should react to them.
Furthermore, they did not use checklists (first of all, the "Unreliable Speed Indication Checklist"), or the quick reference handbook, which is used in emergencies. They did not carry out the procedure designed to deal with unreliable speed indication, either.
There was no clear decision-making according to a model, which provides structure, such as FORDEC (facts, options, risks and benefits, decision, execution, check and control). Instead, they reacted in an uncoordinated manner and instantaneously to the presumably most urgent problem.
The Teamwork During the Decisive Minutes
The team did not succeed in establishing a shared mental model of the situation as it unfolded. This was mostly because it developed too fast with too many things happening at the same time. Hence, the pilots were overstrained and could not understand the state of affairs.
The team did not perform a team time out, for example, a 10-for-10 (ten seconds for the next ten minutes). A team time out is conducted by the whole team in order to analyse the facts and use them to establish the aforementioned shared mental model, as well as to decide on the next steps.
The allocation of roles between the captain and the first officers was also discussed in the final report on the accident, highlighting the question whether it was suitable for the crossing of the ITCZ (page 172 and following).
Workload Management
During the four minutes, the workload was high. The many visual and acoustic stimuli added to that. Due to the initial startle, stress and confusion, the crew could not manage the workload. This, in turn, minimised their chance to solve the chaos.
4.3. The Communication Lost Its Formality
Simply due to their effort to comprehend the chaos despite their stress, the pilots did not violate the sterile cockpit rule insofar as they only discussed issues pertaining directly to the situation. More and more expletives appeared in their communication, though.
Closed communication loops cannot be heard in the recording which had been read aloud. Robert omitted the important announcement that he had taken over the controls during his first attempt. After the captain had returned to the cockpit, the two first officers could not report in a structured manner what was happening.
Taken together, the speech became more and more informal, it showed the pilots´ stress and loss of control clearly. Perhaps, this severly impaired communication without standardised phraseology made it even more difficult to identify the problem.
5. Summary of Important Points
Here, we will shed light on how the cause of the crash can be defined and what role the training may have played, which pilots, who fly highly modern aircraft, attend. We will also elucidate, what interrelation today´s automation and the startle effect may have.
5.1. First the Cognitive Loss of Control, then the Physical Loss of Control Over the Airliner
Taken together, the crash of AF 447 was caused by a loss of control, to which not only the facts mentioned below have contributed, but also a few of those, which are mentioned in the article "Loss of Control" . On the whole, a loss of control is the most common cause of a deadly air disaster.
Summary of vital factors, which have contributed to the loss of control on flight AF 447:
- weather phenomena which at least did startle Bonin
- the captain left for his break right before the crew entered the cloud layer
- thunder storms, icy rain, freezing of the pitot tubes, the autopilot and the auto-throttle disengaged
- indicated parameters did not make sense
- it was necessary to suddenly take manual control over the aeroplane in turbulent weather conditions
- surprise and startle effect
- initially, the least experienced pilot on board remained pilot flying
- inadequate pulling back of the sidestick and the stall of the aircraft
- the necessary procedure in case of an unreliable indication of airspeed was not performed
- stress and confusion due to the high workload, alarms, and flight parameters, which did not make sense to the pilots
- further strain caused by the pilots´ inability to understand the chaos
- subsequently, they were not able to take the appropriate actions
- in a downwards spiral, the state of affairs deteriorated until Robert presumably realised that the problem was the stall, but it was too late to save the flight
- the flight took place in the middle of the night, performance is reduced in the early morning hours
With respect to CRM, the following issues possibly played a role in the loss of control:
- the startle effect may not have been recognised as such
- it is not clear whether the crossing of the ITCZ was particularly mentioned during pre-flight planning
- no shared mental model of the situation
- loss of situational awareness
- checklists were not applied
- an objective analysis of the facts and a structured decsion-making process were not possible any more
- it is unclear whether the allocation of roles within the team was optimal
- no team time out (for instance, a 10-for-10)
- deteriorating abilities to communicate in the course of the events
- the tasks at hand could not be shared and carried out in a meaningful way
5.2. Automation and the Training of Manual and Cognitive Flying Skills
Also in connection with AF 447, it was discussed that automation degrades pilots´ manual flying abilites. In particular, it was addressed that even the training in the simulator is not suitable to learn how to master an unexpexted situation at high altitude with the autopilot suddenly disengaged. It should explicitly be noted that all three pilots were fully trained and certified. However, they were not prepared to react appropriately to the situation, which they encountered.
5.3. Surprise and Startle in Times of Automation
Automation ensures that many systems in modern aircraft run reliably and with few failures only. Due to that, pilots are confronted with unexpected circumstances less frequently. In conjunction with the in this respect insufficient simulator training, the infrequent exposure enhances the surprise and the subsequent startle effect.
6. Consequences of AF 447 for Equipment, Training, and More
Due to their sheer numbers, we can only look at a selection of the recommendations and implementations concerning the topics below. All of the remaining, which, amongst others, pertain to search and rescue operations, as well as air traffic control, can be found at the end of BEA´s Final Report on the accident. Not only the BEA, but also EASA and the FAA, Airbus, and Air France reacted with analyses, recommendations and innovations.
6.1. The Pitot Tubes
The pitot tubes "Thales AA", installed on the Airbus A330-203 F-GZCP, have been discussed. Already before the crash of AF 447, failures of the airspeed indication had occurred, luckily without any fatal accidents. Air France had already started before 1 June 2009 to replace "Thales AA" on the aircraft of its fleet by "Thales BA". This was accelerated after the air accident. Following a directive by EASA, "Thales BA" were later replaced by Goodrich pitot tubes, which perform more reliably in case of ice crystal formation.
6.2. Amendments of Pilots´ Training
The BEA recommended in its Final Report on the accident that EASA introduce mandatory training for pilots, during which manual flying and stabilising an aircraft after a stall, especially at cruise altitude, is practised (page 204).
In fact, a new session was introduced into simulator training, during which particularly the handling of unreliable indication of airspeed, as well as managing a stall of the airliner, also at cruise altitude, is rehearsed.
In remembrance of the 228 people who lost their lives on flight AF 447 from Rio de Janeiro to Paris on 1 June 2009.
The bodies of 74 victims were never found. It aggravates the relatives´ mourning process tremendously that they could not have a grave for them.
7. In the Next Article: Is Automation the Solution?
What happens if an automated system is installed in order to solve a technical problem without informing the ones involved? How must these people, who were concerned, have felt as they were confronted with considerable malfunctions, but had no chance to intervene sufficiently? We will reveal in August what, unfortunately, has happend not only once.
Author: Eva-Maria Schottdorf
Date: July 9th, 2023
On my blog page , I have linked more blog articles for you.
After one year, the articles will be removed from the blog page, updated, augmented and turned into books. You can purchase these directly on the page "A Special CRM Book Series" . The first volume is already waiting there for you.