Tailstrike and runway overrun

An undetected incorrect take-off weight inadvertently entered into the electronic flight bag during the pre-departure preparation led to the use of erroneous take-off performance parameters.

Some certification standards, procedures design, distraction management training and crew qualifications issues were found

Emirates A345 (2)

Photo (C) Alexei. Planespotters.net

Tailstrike and runway overrun

Melbourne Airport, Victoria

20 March 2009 – 22:25 local time

A6-ERG, Airbus A340-541

Australian Transport Safety Bureau

Publication date: 16 December 2011

Excerpts from ATSB final report

FACTUAL INFORMATION

History of the flight

On the night of Friday 20 March 2009, 257 passengers, 14 cabin crew and 4 flight crew boarded an Airbus A340-541, registered A6-ERG, for a scheduled passenger flight from Melbourne, Victoria, to Dubai, United Arab Emirates-UAE (The duration of the flight meant that an augmenting flight crew (captain and first officer) had to be carried to provide the operating flight crew with rest breaks during the flight.). The flight, operating as Emirates flight EK407, was scheduled to depart Melbourne at 2225 Australian Eastern Daylight-saving Time and had a planned flight time of 14 hours and 8 minutes (Unless otherwise annotated, the 24-hour clock is used in this report to describe the local time of day, Australian Eastern Daylight-saving Time, as particular events occurred. Australian Eastern Daylight-saving Time is Universal Coordinated Time (UTC) + 11 hours). The pre-departure preparation included the use of an electronic flight bag -EFB laptop computer in calculating the performance parameters for the takeoff from runway 16. The EFB calculation required the input of a range of data: wind speed and direction; outside air temperature; altimeter setting; take-off weight; flap configuration; air conditioning status; anti-ice selection; runway surface condition; and aircraft centre of gravity.

A base take-off weight figure (361.9 tonnes) was taken from data in the aircraft’s flight management and guidance system – FMGS. An additional tonne was added to that figure to allow for any minor last-minute changes in weight, making a total figure of 362.9 tonnes. When entering that take-off weight into the EFB, however, the first officer inadvertently entered 262.9 tonnes instead of 362.9 tonnes and did not notice that error.

Based on the weight and other input information, the EFB calculated take-off performance parameters (including reference speeds and engine power settings) for entry into the aircraft’s flight systems. The incorrect weight and the associated performance parameters were then transcribed onto the master flight plan for later reference. At about this time, the captain and first officer discussed an aspect of the standard instrument departure that appeared to cause some confusion between the flight crew.

The EFB was handed to the captain to check the performance figures before he entered them into the aircraft systems. While the captain was checking the figures entered into the laptop, the first officer was confirming the departure clearance with air traffic control. There were also activities taking place that involved other persons in the cockpit and forward galley area.

The captain’s checks were required to include a verbal check between the captain and first officer to compare the take-off weight in the FMGS with that used in the take-off performance calculation. That verbal check did not take place in this instance.

The captain entered the EFB performance figures into the FMGS and crosschecked them with the first officer against the values that were previously transcribed onto the flight plan.

The captain handed the EFB back to the first officer, who stowed the EFB before they both completed the loadsheet confirmation procedure. During that procedure, the first officer correctly read the weight from the FMGS as 361.9 tonnes but, when reading from the flight plan, stated 326.9 tonnes before immediately ‘correcting’ himself to read 362.9 tonnes (the amended figure that included a 1-tonne allowance for last minute changes). Among the other checks in the loadsheet confirmation procedure, the first officer read out the green dot speed of 265 kts from the FMGS. The captain accepted that speed and the procedure was completed (The aircraft’s best lift to drag ratio speed in the clean configuration (flaps and landing gear retracted). The speed is affected by aircraft weight and altitude only).

The flight crew completed the pre-departure preparation and at 2218:28, the aircraft was pushed back from the terminal 7 minutes ahead of schedule and was taxied to the northern end of runway 16 for takeoff. At 2230:46, ATC cleared the aircraft to line up and then cleared it for takeoff in front of an aircraft that was on final approach. The thrust levers were set to the take-off position and the aircraft accelerated along the runway.

At 2231:53, when the aircraft had reached the calculated rotation speed, the captain called ‘rotate’. The first officer, who was the pilot flying, applied a back-stick (nose up) command to the sidestick, but the nose of the aircraft did not rise as expected. The captain again called ‘rotate’ and the first officer applied a greater back-stick command. The nose began to rise, but the aircraft did not lift off from the runway. The captain selected take-off/go-around (TO/GA) thrust on the thrust levers. The engines responded immediately, and the aircraft accelerated as it passed off the end of the runway, along the stopway and across the grassed clearway. The aircraft became airborne 3 seconds after the selection of TO/GA but, before gaining altitude, it struck a runway 34 lead-in sequence strobe light and several antennae, which disabled the airport’s instrument landing system for runway 16.

Shortly after, the crew were alerted to a tailstrike by an automated message in the cockpit and a radio call from air traffic control (ATC). The crew decided to return to Melbourne to assess the damage.

After stabilising the aircraft in a normal climb, the captain informed ATC of the intention to climb to 5,000 ft and the need to jettison fuel prior to returning for landing. ATC cleared the crew to climb to 7,000 ft and radar vectored them over water to facilitate the fuel jettison.

At 2237, about 5 minutes after lift-off, the crew commenced planning for the approach and landing. The first officer retrieved the EFB from its stowage to carry out the landing performance calculations and determine a suitable landing weight. The EFB was still in the take-off performance module and the crew noticed that the weight used for the take-off calculations was about 100 tonnes below the aircraft’s actual take-off weight.

At 2239, while climbing to 7,000 ft, the augmenting first officer informed the flight crew that the aircraft was not pressurising. The captain asked the augmenting first officer to locate the procedures for action in the event of a tailstrike in the aircraft’s operational documentation. After reviewing the documentation, the augmenting first officer informed the captain that he was unable to find the procedure for a tailstrike (The operator’s Flight Crew Operating Manual did contain a procedure in the case of a tailstrike. The procedure specified that in the event of a tailstrike warning, the flight crew were to limit flight to 10,000 ft to minimise the stress on the airframe and return to an airport for damage assessment as soon as possible).

At 2246, the captain contacted ATC and declared a PAN. All four flight crew then discussed an appropriate landing weight and decided to jettison fuel for a landing weight of 280 tonnes. Although above the aircraft’s maximum landing weight, the crew chose 280 tonnes as a precaution in case several approaches were required. To ensure that there were no further performance calculation errors, the flight crew made three independent calculations of the landing performance using two different references – the EFB and the quick reference handbook (QRH).

At 2311, ATC informed the crew of debris and runway surface damage found during an inspection of the runway and surrounding area. Later, ATC updated the crew on the damage, informing them that the operator’s ground engineers had inspected some of the items retrieved and that they should expect ‘significant damage to the tail’.

During the flight, the flight crew communicated with the cabin crew primarily via the intercom system, although the purser was provided with a detailed briefing in the cockpit. Communication was predominantly with the purser; however, the captain also contacted the senior flight steward in the rear of the cabin to ask about the cabin crew’s observations during the takeoff.

The captain gave the passengers two briefings over the passenger address system. The briefings included basic information on the situation and advice on the fuel jettison and return to Melbourne.

On completion of the fuel jettison, the flight crew prepared for the approach and commenced a descent from 7,000 ft to 5,000 ft. At 2327, as they were passing through about 6,500 ft and slowing the aircraft, the captain heard an unusual rumbling sound. The sound was unexpected and caused a degree of concern among the flight crew. Moments later, the senior flight steward at the left rear door contacted the flight crew to advise that he could see and smell smoke in the rear cabin. The first officer contacted ATC, informing them of smoke in the cabin and requested clearance for an immediate approach. ATC cleared the flight crew to descend to 3,000 ft and, subsequently, for the approach to runway 34. The first officer briefed the purser on the possibility of an evacuation after landing.

At 2332, the crew changed to the Melbourne Tower radio frequency. At the request of the flight crew, the Melbourne Tower controller organised for the aviation rescue and fire fighting (ARFF) vehicles to be on the tower frequency to allow direct communication with the flight crew. As there were several ARFF vehicles involved, there was a significant amount of radio communication between ATC and ARFF vehicles during the latter stages of the approach. The first officer reported that the additional radio communication resulted in some distraction.

At 2336:29, 1 hour and 4 minutes after lift-off, the aircraft touched down on runway 34 and rolled to the end of the runway where it was met by the ARFF services vehicles. After the aircraft came to a stop on the runway, the captain made an announcement for the cabin crew to prepare for a possible evacuation.

The aircraft was briefly inspected by the ARFF services personnel for signs of smoke and fire. None were evident and the flight crew were cleared by ATC to taxi the aircraft to the terminal. The captain advised the cabin crew to revert to normal operations and taxied the aircraft back to the terminal where the passengers disembarked.

There were no injuries to the passengers or crew.

Damage to the aircraft

Inspection of the aircraft revealed serious damage to the underside of the rear fuselage, where the lower skin panels were abraded by contact with the runway surface. In some areas, the skin was worn through its full thickness and grass and soil was caught in the airframe structure. A service panel was dislodged and was found beyond the end of runway 16, along with numerous pieces of metal from the abraded skin panels.

The right side rear fuselage contained several contact marks. One contact mark, forward of the abraded area and immediately below the rear cargo door, was orange in colour consistent with the orange paint on the localiser near-field monitor antenna. Another contact mark was located adjacent to the skin abrasion and consisted of several fine, divergent marks running rearwards and slightly upwards. Numerous fuselage frames and stringers in the rear fuselage area were damaged by the abrasion and contact forces during the tailstrike. The damaged frames were deformed and several were cracked. The composite rear pressure bulkhead had cracked, and the bulkhead diaphragm support ring was deformed.

The inboard rear tyre on the left main landing gear had a scuff mark on its sidewall. The mark contained transferred material that was the same orange colour as the localiser antenna system.

The flight data recorder (FDR) was dislodged from its mounting rack immediately behind the rear pressure bulkhead and was found lying on the lower fuselage skin below and slightly to the rear of the mounting rack. The FDR was undamaged and contained recorded data from the commencement of the take-off roll until the dislodgement at 2232:05. The results of an examination of the FDR rack are in Appendix A on the final report.

Other damage

foto 1

Photo: Ground contact marks. From ATSB final report.

An inspection of the runway, stopway, clearway, and overrun areas identified multiple contact marks, consistent with the tailstrike and overrun. The aircraft’s tail contacted the runway at three locations, starting at 265 m, 173 m, and 110 m from the end of the runway, respectively. The contact marks contained white paint and metallic material, consistent with the construction and paint scheme of the underside of the aircraft.

There was a small drop-off at the end of the stopway that resulted in the fuselage losing contact with the ground until the point in the clearway, 67 m beyond the end of the runway. The final ground contact mark ended 148 m beyond the end of the runway.

To the south of the last ground contact mark, the fuselage contacted the runway 34 lead-in strobe light that was closest to the runway 16 end. That contact was slight and resulted in scrape marks on the support post and slight deformation of the lens shield. A small strip of white paint was located adjacent to the strobe light.

The aircraft struck the runway 16 localiser near-field monitor antenna and the main localiser antenna array. The localiser near-field monitor antenna support post was fractured at the base and fell in the approximate direction of takeoff. The antenna was damaged and the top of the support post was indented and exhibited white paint transfer.

Damage to the localiser antenna array was limited to one of the 16 antennae. The forward (runway) end of the damaged antenna was deformed and the composite cover was severely disrupted. The top of the forward end of the antenna had a black, rubber-like marking and the deformation of the antenna was consistent with an impact.

Personnel information

Operating flight crew

Captain

Table 1

The captain was rated on the Airbus A330-243, A340-313K and A340-541. In the preceding 90 days, the captain had operated only on A340-313K and A340-541 and was not current on the A330-243. The captain’s relevant qualifications and aeronautical experience are outlined in Table 1.

The captain had operated on flights to or from Melbourne on 18 occasions during the preceding 12 months, including four occasions as part of an augmenting crew.

First officer

Tabla 2

The first officer was rated on the Airbus A330-243, A340-313K and A340-541. In the preceding 90 days, the first officer had operated on all three aircraft types. The first officer’s relevant qualifications and aeronautical experience are outlined in Table 2.

The first officer had operated on flights to or from Melbourne on 14 occasions during the preceding 12 months, including four occasions as part of an augmenting crew.

Crew resource management

The communication between the operating crew members during the taxi and takeoff was in accordance with procedures and reflective of an open and effective cockpit environment. The first officer was the pilot flying during the takeoff and for the majority of the flight, so the captain conducted most of the communications with ATC and the cabin. The captain also asked the augmenting crew to calculate the landing data for the return to Melbourne and therefore the amount of fuel to be jettisoned.

Recorded information showed that the captain also included the augmenting crew in the discussions about the landing configuration and after-landing checks that would be required on their return to Melbourne. During that time, the captain made numerous decisions about the return to Melbourne, some of which were challenged by the crew before being resolved, in accordance with accepted crew resource management practices.

Flight crew alertness and fatigue

Fatigue can be defined as a state of impairment that can include physical and/or mental elements associated with lower alertness and reduced performance. Fatigue can impair individual capability to a level where a person cannot continue to perform tasks safely and/or efficiently.

The investigation examined the likelihood that the operating flight crew were fatigued at the time of the accident and the effect that fatigue may have had on their performance. Using sleep/wake data provided to the investigation by the operating flight crew, the fatigue biomathematical modelling system Fatigue Avoidance Scheduling Tool –FAST (Eddy, D.R; Hursh, S.R. (2001). Fatigue Avoidance Scheduling Tool (FAST). Brooks AFB, TX: AFRL/HEOA; 2001; Report No: AFRL-HEBR-TR-2001–0140) was used to assess the flight crew’s task effectiveness. FAST allows the user to input the quality of sleep as well as the duration of any sleep and work.

The calculations used by FAST were designed to produce a score denoting an individual’s task effectiveness. Both operating flight crew members had FAST scores that were towards the top of the effectiveness range. The operator also assessed flight crew fatigue using a different biomathematical modelling tool. The operator’s results for the operating flight crew from that tool correlated closely with the FAST scores.

The operating captain reported having 6 hours sleep in the 24 hours prior to the occurrence and 16 hours sleep in the 48 hours prior to the occurrence. The operating first officer reported having 8 hours sleep in the previous 24 hours and 12 hours sleep in the previous 48 hours.

There were no salient indications recorded on the CVR to indicate that either flight crew were fatigued; such as yawning and prolonged silence, or the disengagement of the crew from conversations.

Fire

Examination of the rear fuselage found no indications of a fire that could account for the smoke reported in the rear cabin.

The apparent smoke that was reported by the cabin crew was consistent with dust entering the fuselage through the abrasion when the tail was in contact with the ground beyond the end of the runway. The change in the pitch attitude while the aircraft was being configured for the approach altered the airflow around the rear fuselage, may have resulted in some of the dust collected in the abraded area entering the rear passenger cabin. The heat generated when the fuselage skins were abraded on the runway may have produced some vapours from the synthetic materials in that area (for example, paints and insulation materials) that produced an odour similar to combustion products.

Tests and Research

The inadvertent use of erroneous take-off data for performance calculations and subsequent takeoffs has been the subject of two research studies, one by the Laboratoire d’Anthropologie Appliquée (LAA) and the other by the Australian Transport Safety Bureau (ATSB). Both studies highlighted the widespread, systemic nature of this issue, with the ATSB paper identifying 31 occurrences from a 20-year period. In addition, the studies offered considerable insight into the factors influencing the use of erroneous data for takeoff and were used to conduct a more targeted comparison between the accident flight and similar events.

Take-off performance calculations

Accident flight scenarios

table 5

Following the accident, the operator provided one of the EFBs that was on board the aircraft during the accident flight.

The investigation entered the ambient conditions of the day and take-off weights of 262.9 and 362.9 tonnes respectively into that EFB to calculate the take-off performance parameter A summary of the results of those calculations is presented in Table 5.

Expected take-off performance

Tabla 6 y 7

The aircraft manufacturer calculated the expected take-off performance for the aircraft when a take-off weight of 362.9 tonnes was used in the performance calculations on the EFB, but when the actual aircraft weight was 361.9 tonnes. The resulting take-off speeds and distances are presented in Table 6.

Pos is the position along the runway measured from the point at which the brakes were released. The actual take-off distances travelled on the accident flight were calculated from the FDR data for comparison and are listed in Table 7.

Comparison of the A340-541 with the A330-243 and A340-313K

Table 8

The crew operated on the A330-243, A340-313K and A340-541. Table 8 provides a comparison of some of the key aircraft specifications that affect its take-off performance. (Engine thrust (kN each):  At take-off/go-around (TO/GA) thrust)

Variability of take-off performance parameters

The flight crew reported observing a wide range of take-off performance parameters during normal operations as well as significant variations in passenger loads across routes and aircraft types. Both the captain and first officer reported that this resulted in the take-off performance figures losing significance and becoming ‘just numbers’.

In order to examine the range of this variability, copies of the flight plans and load sheets for the 2 months prior to the accident were obtained from the operator for each of the four flight crew. There were 87 individual flights by the flight crew in that time…

There was significant variation in the take-off performance parameters during the 2-month period examined, and the erroneous parameters used during the accident flight lay within the range of values observed during that period. Furthermore, the following points were noted:

  • There was no direct correlation between an aircraft’s weight and the FLEX temperature.
  • Although the take-off reference speeds generally increased with increasing weight, the variation was not linear and the correlation was very weak.
  • The take-off reference speeds experienced by the crew varied by more than 50 kts.
  • All four flight crew had experienced take-off parameters in the A340-541 that were very similar to the erroneous values used on the accident flight.

Flight crew perception of the take-off acceleration during the accident flight

All four flight crew reported that their perception of the aircraft’s take-off acceleration was typical of a heavy A340, particularly a heavy A340-313K. The operating flight crew reported that they did not realise there was a problem with the aircraft’s acceleration until they had nearly reached the end of the runway, and the red runway end lights became more prominent. Both operating flight crews reported that during operations from some runways at other airports, it was common to see the red runway end lights as the aircraft lifted off.

Pre-departure preparation

Following arrival at the airport, the first officer proceeded directly to the aircraft to prepare for the flight. The captain completed a number of other tasks before proceeding to the aircraft, and the augmenting flight crew remained with the cabin crew while they completed their pre-departure briefing. Following that briefing, the augmenting flight crew went to the cockpit to assist the operating flight crew with their preparations.

The augmenting captain then proceeded to the crew rest station to check the intercom while the augmenting first officer completed the external checks of the aircraft. On returning to the cockpit, the augmenting captain sat in the second observer’s seat and the augmenting first officer waited in the forward galley because the ground engineers were in the cockpit and one was occupying the first observer seat.

Preparation for the flight included the initialization and configuration of the aircraft’s systems and the entry and review of the flight plan in the navigation computers. At 2155, noticing that the first officer had not configured the overhead panel, the captain completed the required actions. Those actions included activating the cockpit voice recorder (CVR). A review of the CVR found that the take-off performance calculation and associated actions were captured by the recording.

The timing of, and actions during the pre-departure preparation are outlined in the following tabulated format.

CVR1

CVR2

 

CVR3

CVR4

CVR5

CVR6

The leading ‘3’ in the flexible take-off weight value (FLTOW) that was transcribed on the flight plan was not consistent with the other ‘3’ numerals that were also transcribed by the first officer on the flight plan. This number appeared to have been changed from a ‘2’ to a ‘3’ (Figure 31). The effect of that alteration was to change the recorded flexible take-off weight value from 262.9 to 362.9 tonnes.

Figure 31

Initially, the first officer reported that he believed he had changed the FLTOW value, but could not recall when it was changed. Later, the first officer listened to a replay of the CVR recording at the ATSB audio laboratory. After hearing the verbal slip that was made when reading the flexible take-off weight during the loadsheet confirmation procedure, discussed above, the first officer recalled that was the time when he altered the ‘2’ to a ‘3’. He also reported that, at the time he made the alteration, he believed that he had transcribed the value incorrectly from the EFB onto the flight plan.

At about the time the loadsheet confirmation procedure was completed, the augmenting first officer entered the cockpit. At that point, the preparation was ahead of schedule.

The augmenting captain reported that, although not required by the operator’s procedures when in the augmenting captain role, he had a personal habit of checking the take-off performance calculations on the EFB. He did that by either reviewing those entered by the first officer or by obtaining the second EFB and completing the calculations himself. He reported that he did not have the opportunity to do this on the accident flight due to the first officer using the primary EFB and the number of people in the cockpit during the pre-departure preparation blocking access to the second EFB.

At 2218, following the completion of passenger loading and closure of the doors, the aircraft was pushed back from the terminal ahead of schedule. At 2218:36, the engine start procedure was commenced and all four engines were started. After obtaining taxi clearance from ATC, the crew taxied the aircraft to the northern end of runway 16.

Emirates A345

Photo (C) Zacair. Airplanes-Pictures.net

ANALYSIS

Introduction

The 20 March 2009 tailstrike and runway overrun involving the Airbus A340-541 aircraft, registered as A6-ERG, was the result of the inadvertent use by the flight crew of an erroneous data figure of 262.9 tonnes that was input during the take-off performance calculations. That input error produced erroneous take-off speeds and engine thrust settings that were used for the takeoff.

This analysis begins with an examination of the occurrence events, before discussing the individual actions and local conditions that affected the performance of the flight crew. The risk controls to prevent such an occurrence are then discussed. Several other topics of interest, including the use of the electronic flight bag (EFB) and fatigue are also considered.

Occurrence events

The investigation identified two occurrence events that contributed to the development of the accident. Those events included the over rotation, leading to a tailstrike, and a long take-off roll, leading to a runway overrun. The following discussion examines the factors in the over rotation and long take-off roll.

1. Over rotation and subsequent tailstrike

The damage in the rear lower fuselage and marks on the runway indicated that the aircraft sustained a tailstrike during takeoff. The smooth, positive backstick command by the first officer to raise the nose resulted in a rotation rate of about 3° per second, which was within the normal range. Therefore, it was unlikely that the first officer’s rotation technique contributed to the tailstrike.

The use of take-off reference speeds that were too low for the aircraft’s actual weight or flap configuration of 1+F, meant that the wings did not produce sufficient lift to raise the aircraft off the ground before the geometric pitch limit was reached, and the tail contacted the runway. The only relevant cockpit indication provided to the flight crew was the tailstrike pitch limit indicator on the primary flight display (PFD). It is unlikely that the flight crew had time to recognise that the aircraft had not lifted off at the expected pitch angle of about 8° before the aircraft reached the geometric limit of 9.5°. It was, therefore, unlikely that the flight crew could have identified the over rotation using the PFD indicator, given that the rotation rate of 3° per second would have given them about half a second to perceive the information and react.

The rotation maneuver was initiated at a speed lower than necessary for the aircraft’s weight and this meant that the wing was unable to provide sufficient lift for the aircraft to lift off as expected at the normal pitch attitude. The investigation concluded that the over rotation and tailstrike were due to the incorrect rotation speed and flap configuration for the actual weight of the aircraft.

2. Long take-off roll and subsequent runway overrun

The additional distance travelled by aircraft to reach V1, VR and VLOF was due to the acceleration being lower than required. The operation of the engines was normal and the thrust being produced was appropriate for the calculated FLEX temperature. Given there was no indication of a retarding force, such as a locked brake or excess aerodynamic drag, the low acceleration was the result of a lower thrust setting than that required for the actual aircraft weight.

The crew’s lack of awareness of the low acceleration until towards the end of the take-off roll meant that, by the time the captain selected Take-off/Go-around (TO/GA) thrust, a runway overrun was inevitable. The increased thrust from that selection increased the aircraft’s acceleration and resulted in the aircraft becoming airborne and climbing away from the ground much earlier than it would have otherwise. The captain’s selection of TO/GA, therefore, reduced the likely significant adverse consequences of the runway overrun.

Individual actions and local conditions 

There were a number of actions taken by the flight crew during the pre-departure phase that contributed to the accident, two of which directly influenced the occurrence events. Those were the:

  • use of erroneous performance data for the takeoff
  • lack of recognition of the degraded take-off performance until very late in the take-off run

1. Use of erroneous performance data

A direct comparison of the erroneous take-off reference speeds that were used in the takeoff with those derived by the investigation for the aircraft’s actual weight was not possible due to the differences in aircraft configuration associated with the different take-off weights. However, as previously discussed, the reduced take-off reference speeds and incorrect take-off configuration adversely impacted on the lift available for the takeoff. Compounding that reduction in lift, the significantly higher FLEX temperature resulted in a much lower engine thrust setting than necessary for the takeoff.

Although the recorded information showed that the erroneous figures were entered into the Flight Management and Guidance System (FMGS) during the pre-departure phase, there was no indication that the flight crew were aware that the take-off performance figures were incorrect until after the tailstrike.

2. Incorrect take-off weight entered into the EFB

The introduction into service of the EFB resulted in the take-off performance calculation changing from an interactive process of referencing charts and tables to a simple data entry and retrieval exercise. All of the calculations were performed by the computer and the results presented to the crew. The relatively simple actions of data entry and retrieval probably resulted in the process becoming quite automatic, with little conscious oversight by the crew members. Because of the automatic nature of this process, the crew member entering the data into the EFB would be unlikely to detect any errors made unless the software provided an error message or if there was a significant and unusual result.

There are a range of explanations for the entry of the erroneous take-off weight of 262.9 tonnes into the EFB by the first officer, including confusion with the zero fuel weight figure of 226.6 tonnes or a mental slip while adding the last-minute changes to the take-off weight in a busy, distracting environment. It was, however, considered most likely that the first officer made a typing slip, where the ‘2’ key was accidentally pressed instead of the adjacent ‘3’ key, and that he did not detect the error.

3. Erroneous take-off weight undetected

Three factors were identified as contributing to the non-detection of the take-off weight data entry error. These were the:

  • non-adherence to standard operating procedures
  • first officer reading out the correct weight during the loadsheet confirmation procedure

• first officer amending the take-off weight figure that was recorded on the flight plan to the correct weight, without investigating the discrepancy.

Research into human error has shown that we are capable of making errors across a variety of tasks, and safety investigations aim to identify how such errors remain undetected by a system’s risk controls and/or defences. Errors generally do not occur in isolation and there is usually a series of events/actions that combine within a particular context to produce an error.

In this accident, a series of actions and omissions reduced the effectiveness of the procedural checks and resulted in the crew not detecting the difference between the actual take-off weight and that entered into the EFB.

The first officer’s reported focus of attention to the right of the take-off weight figure, combined with the routine nature of transcribing the value from the EFB onto the flight plan, meant that it was probable he saw the ‘2’ in the place of the ‘3’ but did not detect that it was erroneous. He also reported that while he thought the FLEX temperature appeared to be high, he became distracted and did not investigate this further.

The fact that the values read out by the crew during the pre-departure checks matched the values on the aircraft systems and loadsheet, reduced the chance the crew would detect the error with the EFB entry weight.

The operator’s pre-departure procedures included five checks that were intended to detect take-off weight data entry errors in the performance calculation. Those checks were included in the:

  • Take-off performance error check
  • Take-off data check
  • Loadsheet confirmation procedure.

Take-off performance error check

The take-off performance error check included a check of the data input into the EFB that was performed silently by the captain, and a verbal check by the captain and first officer of the EFB ‘result’ take-off weight against the FMGS INIT B page take-off weight.

As the captain’s EFB input data check did not require verbal crosschecking, the investigation could not determine conclusively from the recorded information whether or not the captain completed that check. Whereas the cockpit voice recorder (CVR) recorded the captain commencing the check after he received the EFB from the first officer, the required verbal comparison between the captain and first officer of the take-off weight in the FMGS INIT B page with the EFB ‘result’ weight did not occur.

That omission might have been explained by the large amount of activity in the cockpit at that time. Research into aural distraction has found that such distraction significantly degrades a person’s ability to apply their full attention to a task, and any distraction of the captain’s attention away from the performance error check increased the risk that it would be missed.

The discussion of an apparently confusing aspect of the planned standard instrument departure (SID) procedure would have added to the workload as the captain checked the EFB and the first officer conducted the pre-departure clearance (PDC) readback. The discussion of the SID may have drawn the captain’s attention to the first officer’s PDC readback, distracting him from checking the EFB input data. That was consistent with the captain’s statement that he ‘copied’ aspects of the PDC from ATC, indicating that his attention was on that communication.

The above distractions may have reduced the captain’s available attentional resources for the take-off performance error check. If the distractions did interrupt that check, the captain may have inadvertently resumed it after the take-off weight verification. He may also have not completed the check after becoming distracted, instead commencing the next action of entering the data into the FMGS, not realising that the take-off performance error check was incomplete.

In turn, the first officer’s attention on the PDC readback may have distracted him from participating in the take-off performance error check. That would explain the recorded gap in the first officer’s involvement until he began assisting the captain with the data entry confirmation.

At the completion of the data entry confirmation, the captain’s action to not transcribe the take-off weight and green dot speed onto his copy of the flight plan and his reading out of the green dot speed in the busy cockpit, negated one of the operator’s defences that might normally have detected the error. The first officer’s likely automated response of ‘checked’ to the captain’s verbalisation of the green dot speed from the EFB was consistent with him not comparing it to the value displayed on the FMGS. The effect of that response may have been to influence the captain to incorrectly accept the different green dot speed during the loadsheet confirmation procedure as this speed had previously been ‘verified’ by the first officer’s response.

The captain’s experience of the reliability of the EFB-derived take-off performance figures may have established an expectation that the results would most likely be correct. In combination with the in-cockpit distractions, that may have reduced the captain’s level of attention to the checking process, thereby reducing its effectiveness.

Although not required by the operator’s procedures, had the augmenting captain the opportunity to perform his own check of the take-off performance calculations, he may have detected the take-off weight entry error.

Take-off data check

The take-off data check was to be carried out following receipt of the final load sheet, and included a requirement for both crew members to silently verify that the take-off weight displayed on the INIT B page was greater than the FLEX-limiting take-off weight previously recorded by the first officer on the master copy of the flight plan.

The take-off data check could be interpreted in two different ways as a result of its wording and sequencing. The use of the term ‘CHECK/REVISE IF REQ’D’ could lead flight crew to think that the check was only required if there was a change to the zero fuel weight (ZFW) on the FMGS INIT B page, as identified by the captain’s review of the loadsheet. Given that the procedure required the take-off performance calculation to be carried out following the receipt of the revised ZFW, it is probable that the take-off weight does not often change from that used in the take-off performance calculation. This would act to reduce the significance of the check, and make it appear to be a superfluous repeat of the take-off performance error check carried out shortly before. On the accident flight the take-off performance calculation was made after the loadsheet had been received and printed.

The lack of a requirement to verbally verify the two take-off weights prevented the investigation from confirming whether this check was carried out by the crew. However, if it was carried out, it was ineffective and neither the captain nor the first officer detected the erroneous take-off weight.

Loadsheet confirmation procedure

The loadsheet confirmation procedure provided the final two procedural defences to detect the error in the take-off weight. The first was when the first officer read out the take-off weight from the FMGS INIT B page and then the ‘result’ take-off weight from the master flight plan.

The first officer read the weight from the INIT B page correctly as 361.9 tonnes, but when he read the value from the master flight plan he read 326.9 tonnes. He was then heard to immediately change this to 362.9 tonnes, even though the value recorded on the master flight plan was 262.9.

It is likely that, having just read the weight as 361.9 tonnes from the INIT B page, and knowing that this was correct, the first officer automatically started to say the ‘three’ (of 362.9) when reading the ‘result’ weight from the master flight plan because this was, logically, the next value. However, on seeing 262.9 he verbalised the value as 326.9, before, upon realising the transposition of the ‘2’ and ‘6’, ‘correcting’ it to 362.9. This was consistent with his understanding of the actual take-off weight. The first officer reported that he changed the number on the flight plan from a ‘2’ to a ‘3’ at this point during the procedure because he thought that he had made a simple transcription error when recording the values from the EFB on the flight plan. Since he believed he had made a simple transcription error, the first officer did not investigate the discrepancy, thereby removing the opportunity to detect the original data entry error in the EFB.

There was no specific requirement for the captain to refer to the ‘result’ weight on his copy of the flight plan. It is reasonable to expect that the captain would only have been comparing the values verbalised by the first officer, and those values satisfied the requirements of the check. That was, the INIT B take-off weight did not exceed the verbalised ‘result’ weight.

The final opportunity for the flight crew to detect the data entry error was the gross error check that compared the green dot speed values obtained from the EFB and FMGS. That check required the captain to compare the green dot speed read out by the first officer from the FMGS PERF TAKEOFF page with that calculated by the EFB and recorded by the captain on his copy of the flight plan. A difference of 3 kts or more indicated a weight input discrepancy and had to be resolved by the crew.

The captain’s hesitation and then non-standard response of ‘yes’ when the first officer read out the value of 265 kts from the FMGS INIT B page suggested that the captain was thinking about the value, rather than directly comparing it to a figure that was written down to confirm its acceptability. At that time, the captain had been crosschecking the first officer’s verbalised figures against the load sheet, and therefore may not have had his transcribed EFB green dot speed readily available for comparison. Instead, the captain may have relied on his recollection of the value calculated by the EFB.

Given that both green dot speeds had the same first and last number (that is ‘2-other value-5’), and the emphasis of the criteria was that the speeds had to be within 2 kts of each other, it is possible that the captain’s attention was drawn to the last digit, as he expected any difference to occur there. Because both numbers ended in a 5, it may have appeared to the captain that the 2 kts criterion was satisfied.

The flight crew’s reported trust in the performance calculation adversely affected their critical analysis of the results obtained. This trust in the standard operating procedures as a defence was likely reinforced by their previous experience of those procedures routinely detecting such errors.

In summary, the crew’s non-detection of the erroneous take-off weight entry in the EFB was multifaceted, and reduced the effectiveness of the procedural checks that could, individually, have detected the error. It is possible for errors to pass undetected through various checks, which is why most procedures incorporate multiple independent checks to verify critical information.

Such errors are not confined to any particular aircraft type, operator or type of operation. It is likely that, given that the current risk controls across operators are procedural in nature, these errors will continue to occur in normal operations throughout the world.

4. Degraded take-off performance not detected

All the calls made, and actions taken by the flight crew were typical of a normal takeoff until the point where the captain called a second time for the first officer to ‘rotate’. There were no indications from the communications or actions in the cockpit that any of the flight crew were aware of, or able to detect, that the aircraft’s performance was insufficient for a safe takeoff. It was not until the aircraft approached the end of the runway, without lifting-off as expected, that the captain realised there was a problem and applied TO/GA thrust.

Flight crew monitoring of take-off performance is based on a set of reference speeds during the take-off roll and does not include the monitoring of the aircraft’s acceleration. Therefore, if the take-off reference speeds are incorrect, or the acceleration insufficient, flight crew have no reliable indication of any problem. Accordingly, it is difficult for crew to identify that take-off performance is degraded. Two items of information are required for flight crew to determine degraded take-off performance:

  • a measure of the aircraft’s actual acceleration, in real time
  • a reference, or expected, level of aircraft acceleration.

There was no indication of the actual aircraft acceleration available to the flight crew on the night of the accident. The only sources of information on the aircraft’s take-off performance were the airspeed indication on the primary flight display, information from the engine instruments, and the pilots’ perception of the acceleration. As previously discussed, airspeed alone provides no indication of acceleration and the engine instruments provide an indication of engine thrust and other parameters. Flight crew have to derive engine- related problems from those parameters. A human’s ability to determine acceleration is neither an accurate nor reliable means to assess take-off performance. Furthermore, that accuracy and reliability is further degraded in darkness.

At the time of the accident, an indication of the expected acceleration was not provided to the crew, nor was it required to be. The take-off performance philosophy was based on the aircraft accelerating at a rate commensurate with the performance calculations.

Without a quantitative method for assessing the actual acceleration attained during the take-off roll, or having a ‘reference’ acceleration to compare with the actual acceleration, the flight crew could only judge the aircraft’s acceleration in comparison with their previous experience. All four flight crew reported that they ‘felt’ that the aircraft’s acceleration was consistent with a ‘heavy’ A340, specifically an A340-313K and were not alerted to the low acceleration.

All four flight crew members had encountered a large variation in take-off performance due to: the use of reduced thrust takeoffs; operating a variety of aircraft with significant differences in take-off weight (due to differing routes and passenger/cargo loads); and differences in runway lengths and ambient conditions. The result was that there was no experience-based acceleration ‘datum’ against which the crew could measure the takeoff. That was consistent with the recorded data, which showed that there was no direct correlation between acceleration and take-off weight. For example, the take-off weight for the previous flight from Auckland to Melbourne was 8% greater than the flight from Melbourne to Auckland, but the acceleration was about 80% lower.

None of the four flight crew members raised any concerns regarding the aircraft’s acceleration during the take-off roll, demonstrating the inherent difficulty in detecting degraded take-off acceleration.

5. Large variations in take-off weight

In the previous 2 months of operations, the flight crew were exposed to take-off weights that varied from about 150 to 370 tonnes. This large variation probably affected the conspicuity of the erroneous first ‘2’ in the take-off weight that was displayed in the EFB as it, in itself, was not abnormal. Both the captain and the first officer had operated the A340-541 with take-off weights in the 200 to 300 tonne range, and observing a take-off weight of 262.9 tonnes would not have been sufficiently conspicuous to alert the crew to the possibility of the data entry error.

The crew’s experiences of differing take-off weights would have been further complicated by their mixed fleet flying. Exposure to large take-off weight ranges makes it difficult for flight crew to form an expected ‘normal’ weight, and has been observed as a factor in other erroneous take-off performance incidents and accidents.

6. Variations in take-off performance parameters

The large variability in take-off performance experienced by the crew over the previous 2 months, and the lack of a simple, effective correlation between the weights and parameters, meant that crews were unable to develop mental models, or ‘ballpark’ figures, to assist them in detecting whether one or more of the parameters in a given set of take-off performance figures were anomalous. This was reflected in the flight crew’s comment that the take-off performance figures had lost significance and had become ‘just numbers’.

To further complicate the situation, all four flight crew had experienced parameters in the A340-541 that were very similar to the erroneous values experienced on the night. As such, the take-off performance figures were not sufficient in themselves to alert the crew to the erroneous take-off weight used to calculate the figures.

Another complicating factor for the crew’s ability to comprehend erroneous parameters was the use of the OPT CONF (optimum configuration) option in the EFB, which selected the high-lift device configuration that gave the lowest take-off speeds. Small changes in ambient conditions could result in a change in the take-off configuration, and associated take-off speeds. That increased the difficulty for flight crews to correlate the parameters, even from an airport from which they commonly operated, such as their home port of Dubai.

The factors affecting the crew’s ability to determine the ‘reasonableness’ of the take-off performance parameters is discussed further in section 6.4.4 of this report, titled Reasonableness self-check.

Emirates A345 (4)

Photo (C) Ronny Busch. Planespotters.net

Risk controls

1. Distraction management

Research on distraction and interruptions has identified their detrimental effect on the formation and detection of errors. The research has also highlighted that the majority of errors occurred during the pre-departure phase of a flight. Thus, it is important to manage distraction during this flight phase to minimise the potential for errors to be formed and not detected until they have effect.

The calculation and checking of the take-off performance was critical to the safety of the flight, yet there was no guidance provided by the operator on the management of distraction during that process. The operator had identified other flight phases as critical to the safety of flight, such as taxi, takeoff and climb, and had a sterile cockpit rule for those phases. There was no such management practice to reduce the potential for distraction during the take-off performance calculation and checking process. Together with the operator’s requirements for its flight crews to cooperate with all other personnel involved in a flight, including ground staff, this increased the risk that flight crew would be distracted by other personnel during those interactions.

The lack of clear direction on the role of, and required input from the augmenting crew during the pre-departure preparation further increased the distraction risk to the operating flight crew. That was consistent with the reports that the presence of augmenting crew in the cockpit during the pre-departure phase created a distraction for the operating crew.

By not including a component on the management of in-cockpit distractions in the operator’s training program, the operator effectively left it to flight crews to develop their own distraction management practices based on their operational experiences and the environment in which they were operating. Without ongoing, formal reinforcement, such as through simulator exercises, it could be expected that the importance placed by crews on distraction management might diminish, potentially increasing their acceptance of continued interruptions from ground crew during the pre-departure phase.

The provision by the operator of briefings to flight crews on distraction management in the months prior to the accident appear to have been ineffective in this accident. Ongoing, formalised training might have alerted the captain to the distraction risk of the non-linear task completion risk represented by his check of the EFB input while the first officer was carrying out an ATC readback.

The prevalence of distraction as a contributor or influence in error development is well documented in human factors research. The challenge for operators is to develop and implement training and standard operating procedures that enable flight crew to manage distractions during safety-critical tasks, especially during the pre-departure phase.

2. Standard operating procedure design/usability

The conduct of the take-off weight comparisons within the takeoff performance error check, take-off data check, and loadsheet confirmation procedure, within a relatively short period of time, may have been perceived by flight crew as redundant. Given that on the accident flight, the take-off performance calculation was based on the final, and therefore unchanging weight figures, the risk that the three, close proximity checks might appear superfluous was heightened. That might explain to some extent why only the final loadsheet confirmation procedure was completed.

Standard operating procedures are typically designed on the basis that information flow into the cockpit is sequential and the procedures are conducted in a linear fashion based on this sequential information flow. Research has shown that the information flow into the cockpit during line operations typically does not follow the sequence upon which the procedures are based. This increases the likelihood that, following a distraction, the flight crew will re-enter a procedure at an incorrect point. The sequence of delivery of information may also lead the crew to believe that a check is no longer required.

The reported normal practice for flight crew to add 1,000 kg to the take-off weight in an A340-541 before it was entered into the EFB, to allow for last minute changes to the load, appears to be a strategy used by flight crew to avoid having to recalculate the EFB figures in the likely event the final weight figures differed to those initially used. It is probable this strategy developed from the regularity of last minute changes, and 1,000 kg covered all possible changes that did not require the issue of a new loadsheet.

3. Documentation design

Given the captain’s deviation from the requirement to record the green dot speed on his copy of the flight plan, and the wide variation noted in the documentation obtained for the preceding 2-month period, it seems likely that the lack of a specific position on the flight plan for recording the green dot speed led crew members to develop their own method for recording it.

While these individual methods did not strictly comply with standard operating procedures, they did comply with the intent, which was to note the speed in order to conduct a subsequent comparison during the load sheet confirmation procedure. However, the variation by crews in recording the green dot speed, and therefore lack of a consistent and predictable information source, increased the risk that any EFB data entry errors would remain undetected.

4. Reasonableness self-check

A number of factors influenced the flight crew’s ability to determine the ‘reasonableness’ of the take-off performance figures calculated by the EFB. One of the main factors was the variation of those parameters as experienced by the flight crew during normal operations. The normalcy of that variation in parameters increased the difficulty for flight crew to recognise inappropriate outputs from the EFB. The reasons for this variation have been discussed previously.

This problem is not unique to this accident. Previous investigations into similar data entry error and tailstrike occurrences have highlighted the inability of flight crew to conduct a ‘rule of thumb’ or reasonableness check of speeds when moving between aircraft types. Furthermore, an unintended consequence of mixed fleet flying appears to be a reduction in a flight crew’s ability to build a model in long-term memory to facilitate recognition of ‘orders of magnitude’, or a ‘rule of thumb’, in respect of take-off performance data. Because the figures that are quite reasonable for one variant may not be reasonable for another variant, the flight crew would need to build a model for each variant that they operate.

There was no specific guidance in the regulatory or operator’s documentation to assist flight crew in forming appropriate mental models regarding the weight and corresponding take-off performance parameters for a particular flight.

Other safety factors

1. Electronic flight bag/operational procedures ergonomics

An ergonomic review of the EFB was carried out to determine the current and optimal flows of information into, and out of the EFB in the context of the operator’s procedures.

This included a review of the flow of information into the EFB, from the EFB to the flight plan, from the EFB to the FMGS, and from the FMGS to the final check against the flight plan. Figure 36 shows the link analysis for the flow of information from the EFB to the FMGS and then to the flight plan for the final check.

The analysis found the flow of information into the EFB and onto the flight plan was clear and simple. Because the EFB and flight plan mirrored each other with regard to the layout of information, the flow was easy to follow and sequential.

The analysis of the information flow from the EFB to the FMGS MCDU revealed a different situation. It was more complicated, less sequential, and required the focus of the user’s attention to move around the screen. The checking process, which required flight crew to verify information from the FMGS against the flight plan, was more difficult because the values were not printed on the flight plan in the same sequence in which they were read out from the FMGS. Although this did not occur on the accident flight, this complexity increased the risk of errors in data entry and checking.

In addition to the flow of information, a number of other issues were highlighted. The first related to the inconsistency in the weights entered into the EFB and recorded on the flight plan, which varied between tonnes and kilograms. The possibility of transposition errors would be reduced if the units were consistent.

The second issue was related to the initial entry of data into the EFB. The EFB required the user to enter the take-off weight and not the individual ZFW and fuel load figures. The previous incidents highlighted the number of times that the zero fuel weight was entered into the EFB instead of the take-off weight. If the user was required to enter the ZFW, the fuel weight and the take-off weight, the EFB could perform an independent check of the figures to reduce the likelihood of a data entry error.

The final issue related to last-minute changes. To minimise the possibility of conducting last-minute recalculations of take-off performance parameters, it was common practice for users to enter a take-off weight that included an additional weight to account for the maximum permissible last minute change. This created a potential problem because, by adding this margin the flight crew could, inadvertently, enter an incorrect take-off weight into the EFB, or be less likely to identify an error in a weight value entered into the EFB because the original value had been deliberately altered during entry.

Other information

1. Fatigue

Consistent with the results from the operator’s examination of the operating crew’s fatigue, the location of both operational crew members’ effectiveness towards the top of the Fatigue Avoidance Scheduling Tool effectiveness range indicated that they were not significantly impaired by fatigue at the time of the accident. That assessment was also supported by the crew providing data that indicated they both had probably obtained sufficient rest during their layover in Melbourne. The layover time was greater than 36 hours and the captain and first officer reported that they did not feel unusually fatigued when they commenced their duty period. Moreover, there was no sound on the CVR of any crewmember yawning, and no prolonged silence or disengagement of crew from conversations (other than when necessary for operational reasons) that might be linked with crew fatigue.

The investigation determined that it was unlikely the operating flight crews’ performance was impaired by fatigue at the time of the accident.

Summary

There were a number of similarities between the circumstances of this accident and other erroneous take-off performance data-related occurrences. In all cases examined, it was found that the manner in which the errors occurred, and went undetected were varied and was not particular to any aircraft type, operator, or procedure. However, there were two core factors that all the occurrences had in common:

  • individual actions rendered operator’s procedures and controls ineffective
  • degraded take-off performance remained undetected until very late in the take-off roll, if at all, as there was no specific requirement or system for monitoring an aircraft’s acceleration.

A number of safety recommendations have been made by several investigation agencies regarding automated take-off performance monitoring to assist flight crews during the take-off roll.

FINDINGS

Context

From the evidence available, the following findings are made with respect to the tailstrike and runway overrun at Melbourne Airport, Victoria on 20 March 2009 that involved Airbus A340-541, registered A6-ERG and should not be read as apportioning blame or liability to any particular organisation or individual.

Although there are a number of factors identified directly relating to this accident, the accident needs to be taken in the context of the long history of similar take-off performance events identified by this investigation. Even though the events leading to this accident may be particular to this case, the previous events highlight that there are a multitude of ways to arrive at the same situation, placing the aircraft and passengers in an unsafe situation before the aircraft has even been pushed back from the terminal. The preferred safety actions will be those that address the whole situation, not just those that address the specific factors identified in this accident.

Contributing safety factors

  • The first officer inadvertently entered the incorrect take-off weight into the electronic flight bag to calculate the take-off performance parameters for the flight.
  • The captain was distracted while checking the take-off performance figures in the electronic flight bag, which resulted in him not detecting the incorrect take-off weight.
  • During the pre-departure phase, the flight crew did not complete all of the tasks in the standard operating procedures, which contributed to them not detecting the error.
  • When conducting the loadsheet confirmation procedure, the first officer called out 362.9 tonnes as the FLEX take-off weight, rather than the 262.9 tonnes that was recorded on the master flight plan, which removed an opportunity for the captain to detect the error.
  • The first officer changed the first digit of the FLEX take-off weight on the master flight plan during the loadsheet confirmation procedure, believing it had been transcribed incorrectly, which removed an opportunity for the flight crew to detect the error.
  • The lack of a designated position in the pre-flight documentation to record the green dot speed precipitated a number of informal methods of recording that value, lessening the effectiveness of the green dot check within the loadsheet confirmation procedure. [Minor safety issue]
  • The flight crew’s mixed fleet flying routinely exposed them to large variations in take-off weights and take-off performance parameters, which adversely influenced their ability to form an expectation of the ‘reasonableness’ of the calculated take-off performance parameters.
  • The operator’s training and processes in place to enable flight crew to manage distractions during the pre-departure phase did not minimise the effect of distraction during safety critical tasks. [Significant safety issue]
  • The rotation manoeuvre was commenced at an airspeed that was too low to permit the aircraft to become airborne but sufficient to overpitch the aircraft, resulting in the tailstrike.
  • The application of the calculated (high) FLEX temperature during a reduced thrust take-off led to a reduced acceleration, an extended take-off roll, and the subsequent runway overrun.
  • The flight crew did not detect the reduced acceleration until approaching the end of the runway due to limitations in human perception of acceleration, which was further degraded by reduced visual cues during a night takeoff.
  • The existing take-off certification standards, which were based on the attainment of the take-off reference speeds, and flight crew training that was based on the monitoring of and responding to those speeds, did not provide crews with a means to detect degraded take-off acceleration. [Significant safety issue]

Other safety factors

  • The design of the flow of information from the electronic flight bag into the aircraft systems and flight documentation was complex, increasing the potential for error.
  • The available Cross Crew Qualification and Mixed Fleet Flying guidance did not address how flight crew might form an expectation, or conduct a ‘reasonableness’ check of the speed/weight relationship for their aircraft during takeoff. [Significant safety issue]
  • The failure of the digital flight data recorder (DFDR) rack during the tailstrike prevented the DFDR from recording subsequent flight parameters. [Minor safety issue]

Other key findings

  • It was unlikely the operating flight crew were unduly affected by fatigue.
  • The captain’s selection of Take-off/Go-Around (TO/GA) thrust during the rotation manoeuvre very likely limited the adverse consequences of the runway overrun.
  • The inability of the cabin crew member at door R2 to reach the interphone handset that was located at seat R2A degraded the flow of communication between cabin crew members.

SAFETY ACTIONS

As a result of the accident, the operator and aircraft manufacturer have taken a number of safety actions. In addition, the Australian Transport Safety Bureau (ATSB) has issued a safety recommendation to the United States Federal Aviation Administration and a safety advisory notice to the International Air Transport Association and the Flight Safety Foundation in an effort to minimise the likelihood of future similar events.

REFERENCES

Excerpted from Tailstrike and runway overrun – Melbourne Airport, Victoria –20 March 2009 – A6-ERG, Airbus A340-541, Australian Transport Safety Bureau final report

FURTHER READING

  1. When the error comes from an expert: The Limits of Expertise
  2. Multitasking in Complex Operations, a real danger
  3. The Organizational Influences behind the aviation accidents & incidents
  4. Pilot performance in emergencies: why can be so easy, even for experts, to fail

**********************

Last week, on Apri 25th, 2017, the BEA announced opening investigation on an Air France A343 abnormally long takeoff at Bogota (SKBO-BOG) which has been classified as a serious incident. That incident reminded some of us of this accident.

**********************

minime2By Laura Victoria Duque Arrubla, a medical doctor with postgraduate studies in Aviation Medicine, Human Factors and Aviation Safety. In the aviation field since 1988, Human Factors instructor since 1994. Follow me on facebook Living Safely with Human Error and twitter@dralaurita. Human Factors information almost every day 

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s