FLIGHT SAFETY BLOG: October 2010

Factual Information

History of the flight

On Thursday 22 March 2007 at 1554, the crew of A6-EBC, an Emirates Boeing 777-300ER, call sign Emirates 419, were carrying out their final preparations for a scheduled flight from Auckland International Airport to Sydney. The aircraft was parked at the international terminal and on board for the flight were 357 passengers, 16 cabin crew and 2 pilots.

The first officer on Emirates 419, whose first language was English, contacted the air traffic delivery controller and received a route clearance for the aircraft. The pilots said that they had listened to the current automatic terminal information service (ATIS) broadcast, Romeo, and had a digital printout of the previous ATIS, Quebec, from an aircraft on-board system. After checking that this information did not differ significantly from Romeo, they used the ATIS Quebec weather information for the aircraft operational performance tool software program of the electronic flight bag2 to calculate the take-off performance.

Included near the beginning of the verbal ATIS broadcast were the words, “active runway mode normal operations…” and advice that the active runway was 05 Right (05R). Near the middle of the ATIS were the words, “reduced runway length eastern end refer NOTAM B/1203”. The first officer did not acknowledge that the pilots had received ATIS Romeo, nor did the controller ask whether the pilots had done so, as was normal practice. The first officer did not request the western extension3 to runway 05R for the take-off.

The first officer contacted the aerodrome apron control and requested push-back and start-up clearance. The apron controller issued the appropriate instructions and, because it was not a requirement of apron control, did not confirm whether the pilots had received other relevant information such as ATIS or knew that the runway length had been reduced.

About 2 hours earlier, the pilots had flown the aircraft from Sydney to Auckland as Emirates 418. The captain advised that their normal procedure was to use digital ATIS printouts if they were available, and where possible to listen to the first part of the voice ATIS broadcast for any updates before requesting a printout. They had the then current ATIS Papa, which advised that runway 05R was in use, and near the middle of the ATIS were the words, “reduced runway length eastern end refer NOTAM B/1203”.

The pilots had NOTAM B/1203 (see NOTAM at the end), and at Sydney during their pre-departure stage they had planned for a reduced length landing at Auckland. In the meantime the full length of the runway had been made available temporarily to a departing long-haul flight to Singapore5. For traffic sequencing, the aerodrome controller held the Singapore-bound aircraft at the runway holding point and cleared the Emirates pilots to land their aircraft first. Because the full length of the runway was temporarily available, the aerodrome controller advised the pilots that the full length of the runway was available for their landing. The pilots landed the Boeing 777 with an automatic brake setting appropriate for a long landing rollout and taxied normally to the gate.

At 1609 the first officer contacted the aerodrome ground controller for taxi instructions for Emirates 419’s return flight to Sydney. The controller gave the pilots instructions to taxi for runway 05R and to hold at taxiway A10 (see figure above). The first officer read back the instructions correctly and the controller responded, “That’s correct, and confirm you will depart from alpha ten [A10] reduced length?”. The first officer replied, “That’s correct.”

At 1613 the first officer advised the aerodrome controller that Emirates 419 was approaching taxiway A10 and was ready for take-off. The controller instructed the pilots to hold the aircraft at A10. A short time later the controller cleared Emirates 419 for line-up on runway 05R, and at 1615 cleared the aircraft for take-off.

The first officer was the pilot flying and the pilots set the thrust that they had determined was necessary for an assumed temperature-reduced thrust departure using the full length of the runway from intersection A10. The aircraft began its take-off and accelerated normally.

The pilots said that when the aircraft was nearly halfway down the runway they saw vehicles ahead in the distance on the eastern end of the runway, so the captain immediately applied take-off go-around (TOGA) thrust.

According to data from the aircraft quick access recorder, and separate runway position information, TOGA was applied when the aircraft had travelled approximately 1327 metres (m); i.e. about 41% of the length of the full runway or 61% of the length of the reduced runway. The recorded airspeed at the time was 149 knots. Within 4 seconds the aircraft accelerated to the pilots’ predetermined take-off decision speed (V1) of 161 knots. The first officer later said that immediately after reaching V1 the captain called “rotate” when the rotation speed (VR) of 163 knots was achieved. The aircraft became airborne approximately 190 m before the end of the reduced runway and climbed away steeply.

The vehicles on the closed eastern portion of the runway were an airport safety officer’s utility vehicle and a 3 m high rubber-removal truck, about 460 m beyond the end of the reduced length of the runway. The vehicles were authorised to be on the closed portion of the runway for programmed work.

The airport safety officer had put his vehicle between the truck and the aircraft and watched Emirates 419 accelerate towards him. When the aircraft was closer than usual to the end of the available runway and still not airborne compared with other similar-sized aircraft, he drove his vehicle to the side of the runway as a precaution. The work crew on the runway also moved to the side but the truck remained on the runway.

The aerodrome controller said that he became concerned when the aircraft was still on the runway past the usual take-off point, and when it got close to the truck after take-off. A report from the safety officer suggested the aircraft might have passed some 20 m above the truck.

Calculations using quick access recorder data and vehicle position information showed that the aircraft wheels were about 103 feet (31 m) above the runway as the aircraft flew over the vehicles, or about 93 feet (28 m) clear of the truck. The aircraft climb rate was 2040 feet (622 m) per minute and increased to 5608 feet (1710 m) per minute 14 seconds later, before it reduced.

After contacting Auckland Control the pilots reviewed the ATIS and NOTAM information and saw the text in the ATIS about the reduced runway length, along with information about aircraft go-around ability after entering the aerodrome circuit.

The first officer asked the controller about the work vehicles and their distance from the end of the runway. The controller advised the NOTAM take-off distance available from the A10 intersection. The pilots then notified the operator of the incident. The aircraft continued to Sydney where the pilots landed it safely about 3 hours later. The operator stood the pilots down from duty and returned them to its headquarters in Dubai.

No one was injured in the incident and no damage occurred.

Pilot information

The captain was aged 40 years and used English as his first language. He held an Airline Transport Pilot Licence valid until 10 August 2007. His associated medical certificate was valid until 30 June 2007. He had flown approximately 14 000 hours.

The captain had been a Boeing 777 pilot with the operator for about 3.5 years. He had been a captain on the Boeing 777 for the previous 18 months. He had previous captain and first officer experience flying Boeing 737, 757 and 767 aircraft. His last pilot proficiency check had been completed on 4 March 2007.

A review of the results of the previous 2 proficiency checks found there were no significant concerns, and included a comment on the most recent check that it had been completed to a very good standard.

At the time of the incident on 22 March 2007, the captain had been on duty for approximately 6 hours and had flown about 3 hours. On 21 March he had been off duty on a layover in Sydney. On 20 March he had been on duty for approximately 11.5 hours and had flown 9 hours. On 19 March he had been off duty on a layover in Bangkok.

The captain said he had flown to and out of Auckland more than 30 times in the previous 3.5 years, and that he had flown all modes of approach and departure using the different coloured pages in the airport approach and landing publications.

The captain advised that the turnaround at Auckland was unrushed, and he and the first officer had gone through the briefing package for the return flight and prepared the cockpit in accordance with standard procedures.

The captain said he and the first officer had received the ATIS broadcast Romeo for departure, which he said was presented in the same way as when they had arrived from Sydney. He said they had the previous printed copy of the digital ATIS but did not “pick” the line that referred to reduced runway length. After reviewing the ATIS they used the operational performance tool software program of the electronic flight bag to calculate the aircraft take-off performance and thrust required for the full length of runway 05R from the A10 intersection. He said they did not see any vehicles on the end of the runway until after the aircraft had begun its take-off roll.

The captain said that after seeing the vehicles on the runway and applying TOGA, the first officer rotated the aircraft at the proper speed and that clearing the vehicles was not a problem.

The first officer was aged 48 years. He held an Airline Transport Pilot License valid until 23 June 2010. His associated medical certificate was valid until 1 September 2007. He had flown 12 664 hours.

The first officer had been with the operator since May 2006, flying the Boeing 777. His other flying experience included captaincy and first officer experience on Boeing 737, McDonnell Douglas MD 82 and MD 83 aircraft. His last pilot proficiency check had been completed on 19 January 2007. A review of the results of the previous 2 proficiency checks found they were unremarkable.

At the time of the incident on 22 March 2007, the first officer had been on duty for approximately 6 hours and had flown about 3 hours. On 21 March he had been off duty on a layover in Sydney. On 20 March he had been on duty for approximately 11.5 hours and had flown 9 hours. On 19 March he had been off duty on a layover in Bangkok.

The first officer had previously flown to and out of Auckland on 12 August 2006, 9 February 2007 and 7 March 2007.

The first officer said that during the turnaround at Auckland he had had some food before returning to the flight deck and preparing the aircraft for the flight to Sydney. He said that he and the captain had checked the current verbal ATIS broadcast.

The first officer said that it was not until after the aircraft had begun to accelerate down the runway during departure that his and the captain’s attention had been drawn to the far end of the runway. At that point they realised there was something on the runway but still some distance away, so the captain applied TOGA thrust. He said they made a normal take-off and “handily” cleared the vehicles.

The pilots advised that because the full runway length had been available when they landed, they shared a belief that it would also be available for their departure. They said that the ATIS words “normal operations” near the start of the verbal broadcast Romeo helped to reinforce that belief and they said they subsequently overlooked some information near the middle of the ATIS. The captain said that the same words in the earlier verbal ATIS to which the pilots had listened before they landed at Auckland caused them to think that NOTAM B/1203 was not active, and that their belief was confirmed when the aerodrome controller told them the full length of the runway was available for landing. They commented that the non-applicable information about the ability of a certain category of aircraft to enter the circuit in the event of a go-around had cluttered the information in the ATIS.

The aircraft operator advised that crews often requested the digital ATIS printout some time before departure in order to begin calculating the take-off performance data. As long as the ATIS information immediately prior to departure did not differ significantly from the earlier ATIS, no change to the data was required. The pilots had a digital printout of the previous ATIS Quebec, and the captain said he had used its information for the calculation of the take-off performance figures once they had the aircraft’s actual zero fuel weight, which was obtained about 30 minutes before departure. After listening to the first part of ATIS Romeo and hearing the words “normal operations” in the runway mode information, and confirming that the information for calculating take-off performance had not altered, they had not requested a digital printout. The pilots commented that because the digital ATIS printout from the aircraft ran lines together, it made text interpretation difficult.

The pilots said that when the ground controller asked, “… and confirm you will depart from alpha ten reduced length?” they thought she was referring to the western extension to the runway that they had elected not to use.

Meteorological information

At the time of the incident: the runway was dry; visibility was 50 kilometres; there were a few clouds at 1800 feet (550 m) and some scattered cloud at 2500 feet (762 m); the temperature was 22° Celsius and the dew point 15° Celsius; the surface wind was variable at 5 knots (9 kilometres per hour); and the QNH was 1022 hecto-pascals.

Analysis

The event

The flight was a routine scheduled service between Auckland and Sydney. The pilots were qualified, experienced international pilots, who were familiar with Auckland International Airport. There was no evidence that the pilots were under any undue pressure or suffering from fatigue or stress-related issues. There were no transceiver issues, and the radio communications were clear and coherent. The weather was not considered to have been a factor because the flight was during the afternoon in good visibility.

Despite the above and the fact the correct information about the reduced runway length was accessible by normal means and the pilots had that information, they did not fully scrutinise the ATIS information and made an error in that they believed the full runway length was available for their take-off. Consequently, they began their take-off with a larger reduced-thrust setting and different optimum flap setting on the aircraft than were required for the available runway length, which was about 1060 m less than the full length of the runway.

Because of the length of the runway, the pilots did not see the vehicles on the end of the runway until the take-off was well advanced. Once they saw the vehicles, they immediately recognised that something was amiss, and the captain’s quick action in applying full thrust ensured the aircraft became airborne about 190 m before the end of the available runway, and cleared the work vehicles by about 28 m.

At no stage did the aircraft fly below the obstacle clearance fan at the end of the runway and endanger the work crew. However, the airport safety officer’s concern and his actions are understandable because there was uncertainty over whether the aircraft would be airborne before the end of the available runway. The controller’s worry that the aircraft had not rotated at the expected point on the available runway also demonstrated that the take-off appeared to be irregular.

Large reduced-thrust take-offs could appear unusual to the observer because the aircraft may use a significant portion of the runway. The evidence in this case showed that had the captain not increased thrust, the aircraft would have taken approximately 2 further seconds to reach its planned rotation speed, and been about 30 m from the end of the available runway when it got airborne. If the full-length runway had been available, the take-off would have been normal. Had the pilots planned for a reduced-length take-off, the aircraft would have become airborne some distance earlier than it did.

The most significant threat to the safety of the aircraft related to the pilots rejecting the take-off before achieving the planned V1 of 161 knots, which was more hazardous than continuing the take-off. They had not determined the correct V1 of 143 knots because of their error in believing a longer runway distance was available. After they saw the work vehicles, the pilots had no ability to recalculate V1 and reconfigure the aircraft. Even if they had been able to determine the proper V1 speed, it would have been invalid because of the aircraft configuration and because more runway had been used in accelerating to that speed with the reduced thrust. Had they rejected the take-off between the point on the runway where the correct V1 was achieved until the planned V1 of 161 knots, the aircraft would have overrun the end of the available runway and entered the closed portion of the runway where work vehicles and personnel were present.

Given the situation in which the pilots found themselves after recognising their error, with less available runway than planned and the speed of the aircraft approaching the planned VR, the captain had little choice but to take the action he did by applying full thrust, continuing the takeoff and having the first officer rotate the aircraft at the planned VR of 163 knots. The relative light weight of the aircraft and its surplus engine thrust ensured that the aircraft climbed away quickly from the runway.

Theoretically, there was also a risk of the aircraft encroaching into the obstacle clearance safety fan at the end of the available runway had an engine failure occurred between the predetermined V1 and when the aircraft was safely above the runway. However, because the aircraft was rotated only seconds after full thrust was selected, and did safely clear the work zone, that risk was slight.

Understanding the pilots’ error

What needs to be understood is why the 2 experienced pilots made the information-processing error they did, which then compromised the safety of the aircraft, its occupants and the ground personnel.

The pilots had the correct flight information, including the relevant charts, ATIS and NOTAM information. They had studied NOTAM B/1203 for their flight earlier that day from Sydney to Auckland, were familiar with its contents and knew that the available runway length at Auckland could be reduced. Accordingly they expected a reduced-length landing on runway 05R. An updated ATIS when Emirates 418 was en route to Auckland advised that the NOTAM was active, but the words “normal operations” at the start of the verbal ATIS caused the pilots to think there were no runway restrictions.

Before Emirates 418 landed, the airport operator had returned the runway to its full length temporarily in response to a request by the crew of a departing long-haul aircraft. For traffic sequencing ATS held the departing aircraft so that Emirates 418 could land first. Because there was a temporary change to the information contained in the current ATIS broadcast that could have affected Emirates 418, the controller advised the pilots of the change, i.e. that the full runway length was available for their landing.

The pilots subsequently completed a normal full-runway-length landing using sufficient brake for a long roll-out. Because of the words in the verbal ATIS and the change to full runway length, the pilots formed a mindset that the runway restrictions had been removed for the day, and that the full runway length would be available for their return flight to Sydney about 2 hours later.

The pilots’ turnaround at Auckland was routine, but during that time the available runway length was again reduced. The ATIS broadcast was updated during that time, but it continued to advise that the available length of runway 05R was reduced at its eastern end and to refer to NOTAM B/1203. Because of the pilots’ mindset, they might have subconsciously believed the ATIS was updated to reflect a removal of the runway length restriction, following the controller’s advice that the full runway length was available for landing.

The words “active runway mode normal operations…” at the start of the current verbal ATIS broadcast reinforced the pilots’ mistaken belief that the full runway length was available. This misled the pilots into thinking that operations were normal and inadvertently to overlook the information in the middle of the ATIS that advised reduced runway length was in effect. Consequently, they believed they had no reason to apply the NOTAM information about reduced runway length.

Airways believed that in this situation it was restricted by the word choice in the AIP arrival and departure information about Auckland International Airport to use “normal operations” in the ATIS broadcast, because the instrument approaches and landing threshold were unaltered, therefore operations were “normal”. This was a matter of interpretation, because the AIP did imply that reduced runway lengths were sometimes “special operations”. The difficulty was that if “special operations” was referred to, pilots could use incorrect approach charts for landing on runway 05R because its threshold was not displaced.

Although the situation may have seemed normal to ATS, clearly the situation facing the pilots for their take-off was non-normal. What the pilots needed to break their mindset was some early confirmation that things were “not normal”, rather than the unintentional reinforcement the ATIS provided that they were “normal”.

The ATIS also contained about 11 seconds of permanent information about an aircraft category that was not relevant to the pilots’ flight. This information, which appeared immediately before the critical information about the reduced length at the eastern end of the runway, helped to clutter an already busy ATIS and to obscure the critical runway length information. While the printed copy of the ATIS broadcast did not contain the runway mode “normal operations” reference, its word order and presentation, with no punctuation or line separation, made it potentially misleading and difficult to read, and thus obscured the critical runway length

information. Because the verbal ATIS broadcast was about twice as long as the 30 seconds recommended by ICAO, it created the potential for pilots to filter information, which appeared to be the case with the Emirates 419 pilots.

When the pilots did not diligently follow standard checking procedures and listen to and read all of the ATIS information as they prepared the aircraft and planned for the return flight, they circumvented a defence against error. Had they properly scrutinised the ATIS, the error could have been avoided. Instead they shared a mistaken view about the runway length and prepared the aircraft for departure on the basis of that flawed view.

The action the aircraft operator took with the pilots following the incident should have heightened their awareness regarding their pre-flight responsibilities for their identification of potential threats, and helped them to avoid similar errors.

Although the pilots had a responsibility to follow all procedures and ensure they received, understood and applied the relevant ATIS information, from a human factors perspective their error was understandable. The ATIS broadcast was the usual means of alerting the pilots to the critical piece of information that runway restrictions were in effect, but it was not robust enough in its word choice and construction to break their mindset to ensure they understood that critical information. Instead, the verbal ATIS broadcast inadvertently reinforced their mistaken belief that things were normal.

Because the pilots had not identified any runway threats, they were denied the opportunity to apply the principles of threat and error management and manage correctly the threat posed by a reduced-length-runway departure.

To help reduce the potential for similar errors, ATS should ensure that ATIS information, regardless of the means of transmission, has a clear word and sentence structure, is unambiguous, contains only information that is of a critical and non-permanent nature and complies as closely as possible with the ICAO-recommended length. When operations are non-normal, the ATIS broadcasts should not contain information that suggests, or encourages pilots to believe, they are normal.

Because Airways published the AIP information that it believed restricted its word choice for an ATIS broadcast at Auckland International Airport, it could enhance the AIP information so that the words “normal operations” are not selected for use in ATIS broadcasts when any non-normal operations are in effect.

While the pilots’ omission to acknowledge that they had indeed received ATIS Romeo, and the delivery controller’s oversight in not questioning them to ensure they had received it, did not contribute to the incident, these omissions meant that the normal closed communications loop concerning receipt of the ATIS was not achieved as it should have been before the pilots proceeded. Because the ATIS was an internationally used method of improving controller effectiveness and relieving frequency congestion, it contained essential information and needed verification that it had been received.

Although the ground controller used her initiative in an attempt to ensure the pilots knew the runway length was reduced, the pilots misunderstood the controller’s question and thought the reference was to the runway extension behind them. As a result, the additional defence the controller provided was frustrated. To help prevent similar occurrences, ATS should introduce standardised procedures that require controllers, as a back-up to the ATIS information, to always warn pilots when runway restrictions are in effect, using phraseology that ensures there can be no misunderstanding. This action will aid controllers and help pilots to avoid similar

errors.

The runway works in place at the time were correctly documented, notified and programmed in accordance with normal processes. Although the airport operator fitted in as much runway work as practicable during the less busy air traffic periods, some work such as the work at the end of the runway could be carried out only during the day and the operator needed the flexibility to achieve this. Runway rubber removal was an ongoing process, and although the rubber removal at the time of the incident was in addition to the work at the end of the runway, the operator’s decision to use the opportunity to do this work was understandable.

The airport operator could help to reduce the likelihood of similar occurrences by reducing the opportunities for error by ensuring runway work is kept to only essential items during normal heavy air traffic periods. Whether a curfew period for heavier periods of maintenance could be introduced is something the airport operator should continue to explore as a means of enhancing safety.

The airport operator’s safety action (see 4.2) in having the apron tower personnel advise pilots whenever there are runway works or reduced runway length in effect will provide an additional defence for international flights against similar occurrences at Auckland.

Findings

Findings are listed in order of development and not in order of priority.

The pilots were appropriately qualified, rested, authorised and fit for the flight.
The aircraft was serviceable and its records indicated that it had been maintained in accordance with its schedules.
An information processing error by the pilots led them to believe the full runway length was available for departure when it had been reduced, so they began their take-off with less thrust than required for the available length of runway.
The pilots’ error stemmed from a mindset they had developed from their landing a few hours earlier that the runway length restrictions had been removed, and their insufficiently thorough pre-flight checking process.
The captain’s prompt application of full thrust after the pilots saw the obstruction on the end of the runway ensured a safe outcome to the take-off.
The pilots’ error compromised their ability to reject the take-off or maintain obstacle clearance safely in the event of a loss of engine thrust, and exposed the flight and ground personnel to unnecessary risk.
The current ATIS broadcast did not properly fulfil its intended purpose of conveying essential but routine information to the pilots because:

• it was about twice as long as that recommended by ICAO

• it contained permanent information

• its words “normal operations” contradicted the reference later in the ATIS that advised“reduced runway length”.

Although the ATIS broadcast contained the correct information about the runway length restrictions, its word choice helped to reinforce the pilots’ mindset that the runway length was normal when it was not, and its construction was not robust enough to break that mindset.
The Auckland AIP arrival and departure information inadvertently restricted ATS’s word choice for ATIS broadcasts so that it implied that operations were normal when a non-normal runway restriction was in effect.
The format of the printed version of the ATIS broadcast made it confusing to read, and some interpretation was required to determine the length and content of each sentence. This format contributed to the pilots’ information-processing error.
ATIS broadcasts were the only normal means used to alert pilots to critical runway information, such as when length restrictions were in effect. Had another defensive layer been established that required controllers to use standardised, unambiguous phraseology to warn pilots whenever runway restrictions were in effect, the pilots’ mindset would likely have been broken.
While it did not contribute to this incident, the pilots and the controller inadvertently removed a defence against error when they did not confirm the pilots had received the current ATIS broadcast. Confirmation of receipt of the current ATIS was necessary to ensure that pilots had correct aerodrome information to plan safely for a departure or landing.
Although the runway works and their scheduling did not directly contribute to this incident, any unnecessary works during normal traffic periods at major airports increase the opportunities for errors and the potential for an occurrence.